Optimal Two-Tier Forecasting Power Generation Model in Smart Grids

oil dependence and greenhouse gases emissions, yielding environmental improvements [8,9]. Many researchers have been developing solutions for PHEV recharging. Clement-Nyns et al. proposed in [6] a coordinated charging to minimize power losses and maximize the main grid power factor. To perform optimization, the authors use quadratic and dynamic programming, since an exact forecast of household loads is not possible. Coordinated recharging and linear programming are used to minimize PHEV recharging costs in [10]. It also investigates the consequences of inserting renewable generation into the grid. The interesting results obtained render the losses monitoring as a good strategy for reactive power, voltage level and branch loading control. The authors in [11] use Monte Carlo simulations in a coordinated recharging process to prove that maximizing power factor and also minimizing load variance are equivalent to power losses minimization. In [12] a real-time smart load management control strategy is proposed and developed for the coordination of recharging process based on real-time minimization of total cost of generation plus the associated grid energy losses. In [13], a power losses minimization is performed by applying a time coordinated optimal power flow. They show that, through the control of PHEV storage units and on-load tap changers transformers, electric network operator can influence savings in energy losses. Besides those approaches applied to losses minimization problem on [6,7,10–13], evolutionary techniques are also used, as particle swarm optimization (PSO), which finds the optimal solution using a population of particles. However, this technique has some convergence problems, reaching non-optimal solutions. The mutation feature, present in Artificial Immune Systems (AIS), can avoid this problem [14].

•Bill checking to verify that you are only charged for the energy you use •Aggregation of energy consumption, including WAGES, and costallocation per area, per usage, per shift or per time within the same facility •Energy cost and usage analysis per zone, per usage or per time period tooptimize energy usageThe PM2000 series meter is a next-generation energy and power meter that offers all the measurement capabilities required to monitor an electricalinstallation in a single 96 x 96 mm unit. The PM2000 series offers simplicity and reliability for basic energy cost and network management applications in both a LED or LCD display option.This multifunction energy and power meter is the easy choice for quality, value, and reliability. Essential features, such as configurable data logging, industry- leading measurement accuracy, and power quality analysis. And best of all, it’s easy, safe and fast to choose, purchase, and install the right PM2000 model for your application.The PM2000 series meters help to help maximize energy cost savings, optimize operational efficiency, and improve business performance.Cost Management Applications•Metering of electrical parameters to better understand the behavior of your electrical distribution system•Power quality analysis of THD and individual harmonics up to 15th and 31st order for Volts and Amps, per-phase•Measurement of True PF and Displacement PFNetwork Management Applications•PM2100 series:LED display type: Intuitive navigation with self- guided, three buttons,bright red color LEDs of 14.2 mm height. Two columns of LEDs indicate the parameter name chosen for display.•PM2200 series:LCD display type: Monochrome graphical LCD of 128 x 128 resolution lets users read all three phase values simultaneously. The bright display enables easy reading even in extreme lighting conditions and viewing angles with intuitive menus, multi-language text, icons and graphics.PM2000 meters are available in LED and LCD display variants.Versatile, multi-function power and energy metersSelection Guide•Performance without compromise- exceptional reliability in its class •Assured quality that you can trust to meet your needs throughout its life-cycle•Outstanding value for an optimized feature set•Simple to choose and easy to install and use for many applicationsSee your Schneider Electric representative for complete ordering information.PM2000 Series Features© 2020 Schneider Electric. All Rights Reserved. All trademarks are owned by Schneider Electric Industries SAS or its affiliated companies. Document Number 3000HO2003Schneider Electric6700 Tower Circle Suite 700Franklin, TN 37067/us/powerandenergy+4 Through communication only.+5Any one IO module can be used at a time with PM2130 or PM2230 meter. The control power range for PM2130 & PM2230 meters with or without IO module shall be 72 to 304 V AC L-N or 90 to 304 V DC.+6 Battery backup duration 3 years when meter is in continuous or accumulated Power OFF condition.。

An engineering approach to the optimal design of distributed energy systems in ChinaZhe Zhou,Pei Liu,Zheng Li*,Weidou NiState Key Laboratory of Power Systems,Department of Thermal Engineering,Tsinghua University,Beijing100084,Chinaa r t i c l e i n f oArticle history:Received11December2011 Accepted31January2012 Available online8February2012Keywords:Energy systems engineering Distributed energy system Distributed energy resource Optimization a b s t r a c tChina faces enormous challenges as it strives to meet its increasing energy demands.Distributed energy systems(DESs)provide a good opportunity to tackle these challenges by integrating various forms of distributed technologies,of which internal combustion engines,solar photovoltaic,and absorption chillers being the typical ones.However,fluctuation in energy demands and supplies,the large number of available distributed technologies and various possible combinations of these technologies make the design of DES a formidable task.The potential benefits of DES might be hampered by their impropriate design and operation.This paper provides a generic energy systems engineering framework toward the optimal design of DES in China,with the purpose of obtaining optimal combination of technologies and capacity of equipment for a given area with given energy demands.A hotel in Beijing is selected as an illustrative example to demonstrate the key steps and features of the proposed approach.Results show that the optimal configuration of such a DES is a rather complex system equipped with various technologies.It is more efficient and economic than conventional centralized energy systems as well as distributed combined cooling,heating and power systems.Ó2012Elsevier Ltd.All rights reserved.1.IntroductionA distributed energy system(DES)refers to an energy system where energy is produced close to end use,typically relying on a number of modular and small-scale technologies.DES has a number of advantages such as high overall efficiency,small losses on energy transmission,and small negative environmental impacts [1,2].In recent years,China faces enormous challenges in meeting its immense and rapidly increasing energy demands under more and more stringent environmental constraints,due to the fast growth of its economy[3].Therefore,interest has been intensifying in China in the development of DES,which is an efficient and environmental friendly alternative or complement to the conventional centralized energy system(CES).In October2011,the National Energy Administration(NEA)issued a regulation entitled“the Guidance on the Development of Natural Gas Distributed Energy Systems”.It suggests installing up to one thousand distributed natural gas energy stations,i.e.,combined heating cooling and power(CCHP), during the period of the12thfive-year plan,and the total capacity of DES is expected to reach50,000MW by2020.In this context, a large number of demonstration CCHP projects have been started in different areas of China,as listed in Table1[4].These CCHP projects share a similar configuration which generally includes a prime mover(gas turbine or gas engine),a waste heat boiler and an absorption chiller.The regulation and demonstration projects indicate that Chinese government has a strong interest in promoting DESs in China. However,it can also be observed that the government is more interested in classic CCHP systems at the current stage,although possibility of integration of renewable energy with CCHP systems is also mentioned in the Guidance.From a systems engineering viewpoint,focusing on a single or certain types of technologies,for instance,CCHP systems,may not be the best option,as there exist many other alternative energy resources and corresponding conversion technologies for various types of configurations of DES.These systems,with appropriate design and integration work,may exhibit even better performances than a classic CCHP system.For instance,integration with renew-able energy resources may lead to more environmental benefits such as reduced conventional air pollutants and lower greenhouse gas(GHG)emissions.However,optimal design of DES is not a trivial task as integra-tion of more types of energy resources and conversion technologies*Corresponding author.Tel.:þ861062795739;fax:þ861062795736.E-mail address:lz-dte@(Z.Li).Contents lists available at SciVerse ScienceDirect Applied Thermal Engineeringjournal ho mepage:www.elsevier.co m/locate/apthermeng1359-4311/$e see front matterÓ2012Elsevier Ltd.All rights reserved.doi:10.1016/j.applthermaleng.2012.01.067Applied Thermal Engineering53(2013)387e396may increase the complexity of the system,and the aforemen-tioned benefits may be partly or even totally compensated.In this paper,wefirstly present an energy systems engineering framework toward the optimal design of DES,with which energy efficiency and complexity of a system can be quantitatively evaluated.For a specific district,given its electricity,space heating,cooling and hot water loads,energy prices,and technical andfinancial infor-mation about optional technologies,the proposed framework helps to minimize the overall annual cost.We then illustrate the appli-cability of the proposed modeling and optimization framework in a case study of a hotel in Beijing.In this case study,the energy, economic and environmental performances of the optimal config-uration obtained by the proposed method are evaluated and compared with those of a classic CCHP system and a conventional CES.2.Superstructure based model for the optimal design of DESA number of studies have been reported on the optimal design of DES,but most of them focus on distributed CCHP systems[5e18]. Ren et al.and Liu et al.developed models for optimal design of CCHP systems integrated with solar energy and biomass[19e23]. However,these models do not consider certain technologies,for instance,energy storage technologies.Moreover,National Renew-able Energy Laboratory of the United States developed a design tool, namely HOMER for the optimal selection of distributed power generation technologies,which encompassed renewable energy technologies such as solar photovoltaic and wind turbines as well as energy storage technologies such as batteries[24].However, HOMER aims for the design of distributed power generation systems,therefore heating and cooling needs are not considered in the model.In summary,most existing mathematical models for optimal design of DES cannot capture all alternative configurations due to case-by-case modeling strategy.A generic mathematical model,which can ideally include all alternative distributed gener-ation technologies,is therefore desired to address design issues of DES.In this paper,we propose a superstructure based model for the optimal design of DES,featuring simultaneous determination of the optimal configuration of a DES and its optimal operating conditions via mathematical programming[25].It wasfirstly proposed to address process synthesis issues in heat exchanger networks [26e28],and widely used in a broad range offields thereafter,such as process synthesis[29]and supply chain network design[30,31]. It has been proved to be an effective method to address issues associated with optimal planning of energy systems,such as bio-energy processing systems[32],polygeneration energy systems [33e35]and hydrogen infrastructure[36].As a DES involves a variety of alternative technologies,superstructure based modeling is well suited for its optimal design.A superstructure representation of the optimal design problem of DES is shown in Fig.1.It consists of an energy generation section, an energy conversion section and an energy storage section.In the energy generation section,a number of technologies are consid-ered,and via these technologies heat and electricity can be produced from various primary energy resources.The secondary energy carriers,i.e.,heat and electricity,are then converted to different forms of tertiary energy carriers,i.e.,heat,cooling and electricity,through different energy conversion technologies.For instance,air-source chiller is a typical energy conversion tech-nology that can produce cooling with electricity.For a more effi-cient use of energy,an energy storage section comes after to shift energy generated during low usage periods for consumption during peak hours.The framework includes six primary energy resources,namely natural gas,diesel,wind,solar energy,biomass,geothermal,four energy demands,namely electricity,space heating,cooling and hot water,and twenty types of equipments,namely internal combus-tion engine,gas turbine,natural gas boiler,fuel cell,diesel engine, wind turbine,solar PV,solar thermal collector,biomass boiler, biogas engine,biogas boiler,electric heater,absorption chiller/heat pump,heat exchanger,water-source chiller/heat pump,air-source chiller/heat pump,ground-source chiller/heat pump,battery,ice storage equipment and thermal storage tank.These equipments in the model constitute the most common and mature distributed energy technologies available in China,while the model can be easily extended to cover more technologies when they become commercially available.3.Mathematical formulationA mathematical model based on mixed-integer linear programming(MILP)optimization is illustrated in this section.We use the symbol e to represent all energyflows and its superscripts indicate its position in the superstructure while the subscripts represent the types of energyflows.With respect to superscripts, thefirst superscript,i.e.,g,c and s,stands for the section that this energyflow belongs to,while the second one,i.e.,in and out, represents whether it is an input or an output.The naming rules for the energyflows in the optimization model are shown in Fig.2.For example,a variable named with e c;inh;hctmeans the amount of energy carried by a heatflow coming into the energy conversion section and converted by a certain heat conversion technology.The nomenclature of all parameters,variables,superscripts and subscripts are presented in Appendix at the end of this paper.3.1.Energy generation sectionEnergy generation section comesfirst in Fig.1.In this section, various types of primary energy are converted to electricity and heat by different energy generation technologies.The primary energy consumptions are restrained by local resources availability at any point in time as follows:Table1Demonstration projects of CCHP systems in China and their configurations. No Location Main equipments1Pudong International Airport,ShanghaiGas turbine,Waste heat boiler 2Huangpu Central Hospital,Shanghai Gas turbine,Waste heat boiler 3Minhang Central Hospital,Shanghai Gas turbine,Waste heat boiler4Shanghai Institute of Technology, Shanghai Micro-turbine,Absorption Chiller, Waste heat boiler5Shuya Liangzi Ministry of Health,ShanghaiDiesel engine,Waste heat boiler6Zizhu Science-based Industrial Park,ShanghaiMicro-turbine,Absorption chiller 7Shanghai Jinqiao Sports Center,ShanghaiGas engine,Waste heat boiler8Shanghai Global Financial Hub,ShanghaiGas engine,Waste heat boiler9Beijing Gas Group MonitoringCenter,BeijingGas engine,Absorption Chiller 10Ciqumen Station Building,Beijing Micro-turbine,Absorption Chiller 11Zhongguancun Software Square,BeijingGas turbine,Absorption Chiller 12International Trade Center,Phase3,BeijingGas turbine,Waste heat boiler13International Business Center,BeijingGas turbine,Waste heat boiler14Olympic Energy Exhibit Center, Beijing Micro-turbine,Absorption chillerZ.Zhou et al./Applied Thermal Engineering53(2013)387e396388e g;in f;m;hr e maxf;m;hr(1)where,e g;inf refers to the primary energy consumption;e maxfis themaximum amount of energy available for energy use;f stands for different types of primary energy resources,including natural gas, diesel,wind energy,solar energy and biomass;m stands for different months in a year and hr stands for different hours in a day. The optimization in this study is conducted on an hourly basis for a typical day of each month,taking into account daily and seasonal fluctuations in energy supplies and demands.We assume that each day in a month follows the same demand pattern of the repre-sentative day of this month.The demand pattern of a representa-tive day is averaged through that month.The time intervals are therefore reduced from8760to288(12month and24h a day).The daily grouping to months can significantly reduce the size of the model,thereby improving its computational efficiency.Each type of primary energy can be consumed by any tech-nology that uses it as the fuel.The total consumption of each type of primary energy resource at any point in time is given bye g;inf;m;hr¼Xegte g;inegt;f;m;hr(2)where,the subscript egt stands for different types of energy generation technologies.The production of secondary energy carriers,i.e.,electricity and heat,by various energy generation technologies depends on the consumptions of primary energy resources,and the efficiency of the on-site generation technologies.It can be characterized as follows:e g;outsec;m;hr¼Xegte g;inegt;m;hr$h egt;sec(3) Fig.1.The superstructure representation of a distributed energysystem.Fig.2.The naming rules for the energyflows in the optimization model.Z.Zhou et al./Applied Thermal Engineering53(2013)387e396389where,sec stands for secondary energy carriers,which could be either electricity or heat;sec˛{p,h},p and h refer to electricity and heat;e g;outsecrefers to the production of secondary energy;h egt,sec refers to the efficiency of converting a primary energy resource into a secondary energy carrier.For the purpose of model simplification,we presume that effi-ciencies of all technologies do not change when these facilities are operated at off-design points.In other words,the energy efficien-cies keep constant in this study.For technologies that produce electricity and heat simultaneously,there is afixed relationship between the amount of recoverable heat and electricity based on the manufacturer’s technical specifications.Electricity can also be directly purchased from the power grid. Therefore,the total amount of electricity as input to the energy conversion section is the sum of electricity drawn from the grid and the electricity generated in the energy generation section.This can be denoted ase c;in p;m;hr ¼e g;outp;m;hrþe gpp;m;hr(4)The other type of energy carrier,heat,can only be generated by different energy generation technologies,as follows:e c;in h;m;hr ¼e g;outh;m;hr(5)where,e c;insecis the amount of secondary energy for conversion in the energy conversion section;e gp p denotes the amount of electricity drawn from the power grid.3.2.Energy conversion sectionIn the energy conversion section,secondary energy,namely electricity and heat,are converted by energy conversion technolo-gies to tertiary energy carriers,namely electricity,heat and cooling. Accordingly,the consumption of each type of secondary energy is equal to the sum of energy consumed by energy conversion tech-nologies which convert this type of secondary energy,as follows:e c;in sec;m;hr ¼Xecte c;inect;sec;m;hr(6)where,ect stands for different types of energy conversion technologies.Energy conversion technologies such as air-source heat pump/ chiller usually have two working modes:heating mode and cooling mode.The coefficient of performance(COP)of an energy conver-sion technology differs when working in the heating mode and cooling mode.The productions of these tertiary energy carriers can be given bye c;out ter;m;hr ¼Xecte c;inect;sec;m;hr$j heating$h heatingect;sec;terþj cooling$h coolingect;sec;terð7Þwhere,ect stands for different types of energy conversion tech-nologies;ter stands for tertiary energy carriers,i.e.,electricity,heatand cooling;ter˛{p,h,c},c refers to cooling;e c;outter refers to theproduction of tertiary energy in the energy conversion section;h heating ect;sec;ter and h coolingect;sec;terare COPs of energy conversion technologiesoperating in heating mode and cooling mode;j heating and j cooling are binary variables.As an energy conversion technologies cannot produce heat and cooling simultaneously,j heating and j cooling are restrained by j heatþj cooling¼1(8) 3.3.Energy storage sectionFor a more efficient use of energy,the energy conversion section is followed by an energy storage section.The energy inputs of the energy storage section are given bye s;inter;m;hr¼e c;outter;m;hr(9)where,e s;interdenotes the input of an energy storage equipment.Energy storage technologies in this section allow shifting energy generated during low usage periods for use during peak hours.Each end use energy product is equal to the product generated in the energy conversion section plus the net energy flow from the energy storage unit,namely energy stored in the storage unit less released energy,at any point in time,which is denoted ase s;outter;m;hr¼e s;inter;m;hrþeÀter;m;hrÀeþter;m;hr(10)where,e s;outterrefers to the output of an energy storage device;eÀter refers to energy released from the energy storage device;eþter refers to energy stored into the energy storage device.The amount of energy stored at the beginning of each time interval is equal to the energy stored at the beginning of the previous time interval plus the net energyflow,given byE ter;m;hrþ1¼E ter;m;hrþeþter;m;hrÀeÀter;m;hr=h es;ter!0(11) Where,E ter represents the inventory of a storage device;h es,ter refers to the efficiency of an energy storage technology.Within a certain period of time,T,the total net energyflow from the energy storage units sums to zero as given byX ThrE ter;m;hr¼0(12)We set T as a day,which means the energy storage technolo-gies can only deal with short-term,dailyfluctuations.This assumption is reasonable,because energy storage technologies are costly and therefore their use tends to be limited to small-scale applications.Finally,each type of end use energy demand product must be satisfied by its corresponding tertiary energy carrier coming from the energy storage section,which is given bye s;outh;m;hr!D sh;m;hrþD hw;m;hre s;outp;m;hr!D p;m;hre s;outc;m;hr!D c;m;hr(13)where,D refers to end user’s demands;sh and hw represent space heating and hot water.3.4.Facility capacityThe capacities of all energy generation,storage and conversion technologies must be large enough to meet the maximum amounts of energy generation,conversion and storage over the whole period of time.The capacities of various energy generation and conversion technologies are given byCap tech!max e out tech(14)Z.Zhou et al./Applied Thermal Engineering53(2013)387e396 390where,tech refers to each technology appears in the energy generation section and energy conversion section,i.e., tech˛{egt,ect}.The outputs of different technologies are constrained within their lower and upper bounds of capacity via introduction of binary variables,which can ensure that these technologies are running in reasonable ranges,as follows:k tech;m;hr$Cap Mintech e out tech;m;hr k tech;m;hr$Cap Maxtech(15)where,k is binary variable;Cap Min and Cap Max are minimal and maximal capacities of different technologies.The capacity of energy storage technologies depends on the maximum inventory of a storage device,which can be denoted as: Cap es!max E ter;m;hr(16)3.5.Economic behaviorThe total annual cost is selected as the objective function.It consists of the capital cost,the operation&maintenance(O&M) cost and the fuel cost.The optimization model obtains the most suitable configuration of DES by minimizing the total annual cost, as follows:Min C total¼C capitalþC O&MþC fuel(17) All the associated costs are given byC capital¼Xtech Inv tech$Cap tech$I1Àð1þIÞÀY tech(18)C O&M¼Xtech OM tech$XmXhre gtech;m;hr!þXesOM es$Cap es(19)C fuel¼Xf XmXhre g;inf;m;hr$P f;hrþe gp m;hr$P gp;hr(20)where,Inv tech refers to the unit investment cost;Y tech is the life-time;I denotes the discount rate,specified at10%in this study; OM tech and OM es represent the unit operation costs;P f,hr and P gp,hr refer to the hourly fuel cost and electricity price.3.6.Model linearizationThe optimization model(Eqs.(1)e(20))presented above is nonlinear due to the multiplication of a binary variable and a continuous variable in Eq.(7).As other parts of the problem are linear,it is desirable to convert the nonlinear terms into linear terms so that a linear model could be obtained.The following linearization techniques are employed:Let d be a binary variable,while x and y be continuous variables, thus the following equation is nonlinear:y¼d$x(21) However,this equation is identical to the following set of equations:xÀð1ÀdÞ$M y xþð1ÀdÞ$M(22) y d$M(23) where,M is a large number.By this means,the optimization model can be converted toa MILP problem.4.Case study4.1.Model inputsA case study has been conducted to illustrate the modeling and optimization framework for the design of DES.The targeted energy consumer is a hypothetic large hotel in Beijing with an area of 30,000m2.The required inputs for the optimization model are the demand information and resource availability of this region,as well as the price information and technological and economic information.4.1.1.Demand informationThe electricity,heating,cooling and hot water loads are obtained through comprehensive investigations about energy demands of hotels in Beijing[37,38].The average electricity,space heating, cooling and hot water loads are listed in Table2and the hourly demand data of this hotel in some representative days are shown in Fig.3.4.1.2.Resource availabilityIn the optimization model,the possible resources include natural gas,wind power,solar energy,biomass and geothermal energy.For a real project,the developer of a DES has to provide information about the availability of these resources as the inputs of the optimization model.For the hypothetic case in this study,we artificially determined the resource availability of different resources as follows:Natural gas and geothermal energy are unlimited resources in our analysis.Solar energy and wind energy are obtained through meteoro-logical data[39].Assuming that maximum areas for receiving solar energy and installing wind turbines are both6000m2. Assuming that biomass is rare in urban areas.Wind energy is highlyfluctuating and difficult to predict.In this study,we adopt a probability distribution approach,complemented by the average wind power density data,to predict the volatility of wind energy.Currently,there are many types of probability distributions available for the simulation of wind energy,such as Weibull distribution,Rayleigh distribution,lognormal distribution and etc.[40].In this study,we adopt a three parameter generalized extreme value(GEV)distribution to predicate the hourly wind energy density data,as follows:fðx j k;m;sÞ¼1s1þkðxÀmÞsÀ1Àð1=kÞeÀ1þkðxÀmÞsÀ1=k(24)Where,k equals0.081;m equals0.654;s equals0.498[41].The average wind energy density in Beijing is about75W/m2[39].Table2The average energy loads of the hypothetic hotel in Beijing.Load,W/m2Electricity Heat Cooling Hot water Winter20.0938.33013.16 Transitional seasons22.7012.158.1 Summer27.73040.0110.35Z.Zhou et al./Applied Thermal Engineering53(2013)387e3963914.1.3.Price informationIn China,current prices of natural gas,diesel and grid power are determined by the National Development and Reform Commission (NDRC)[42],as shown in Fig.4.4.1.4.Technological and economic dataThe technological and economic data used in this study,including energy ef ficiency,unit investment cost,operation and maintenance cost,and lifetime of each technology,are listed in Tables 3e 5[4,12,43e 45].4.2.Setting of scenariosIn this study,we provide three different solutions to meet the energy demands in this area.Scenario 1is a conventional central-ized energy system;Scenario 2is a distributed CCHP system;Scenario 3is a DES proposed by the optimization model.The detailed information about these scenarios are described as follows:Scenario 1:Conventional centralized energy system.In this system,the electricity load is satis fied by the power grid;the cooling load is met by electric chillers;the space heating and hot water loads are met by gas boilers.In the conventional system,the electricity is transmitted and distributed via power grid.Scenario 2:Distributed CCHP system.The demonstration projects in Table 1indicate that the existing projects in China are all distributed CCHP systems.This paper chooses one of the typical con figurations of CCHP systems as an representative example of these projects.In this system,the core facility is an internal combustion engine,which can produce electricity andheat simultaneously.The de ficiency of electricity and heat is supplemented by the power grid and gas boilers.The cooling demands are satis fied by absorption chillers,utilizing waste heat recovered in power generation unit.This scenario can represent a classic CCHP system that the current policy is promoting.Scenario 3:DES proposed by the optimization model.In this scenario,the con figuration of the energy supply system and its operating strategy are both determined by the optimization model.5.Results and discussionsIn this study,we conduct the optimization process for the optimal design of DES on the platform developed by GAMS Development Corporations [46].The proposed MIP problem (Scenario 3)involves 13,377continuous variables,6624binary variables and 28,906constraints.It takes 0.39s of CPU time to solve this problem on an Intel(R)Core(TM)2Duo P8600CPU of 2400MHz with a memory size of 2G.The results of the scenario analysis are presented next.5.1.System con figuration of a complex DESThe optimal system con figuration and the capacities of adopted technologies are shown in Fig.5.The optimization model chooses only several technologies,including internal combustion engines,solar heat collectors,wind turbines,various chillers/heat pumps,thermal storage tanks and ice storage equipments,which indicates that the purpose of including various technologies intheFig.3.The hourly energy loads of the hypothetic hotel in Beijing.Z.Zhou et al./Applied Thermal Engineering 53(2013)387e 396392optimization model is to improve the generality of the model rather than proposing a more complex system.As the annual cost is the objective,the model tends to select inexpensive technologies.For instance,among the natural gas-fired energy generation technologies,only internal combustion engine is employed because it is much cheaper than gas turbine and fuel cell.Similarly,the model selects solar collector instead of solar PV to achieve an economic utilization of solar energy.In the energy conversion section,water-source,air-source and absorption chillers/heat pumps are adopted to enhance the production of heat and cooling.Energy storage technologies,such as thermal storage tank and ice storage equipment,are also included in the optimal con figuration to achieve a more proper use of energy.5.2.EconomicsThe annual costs of these three scenarios are shown in Fig.6.DES reveals a much better economics than both the CCHP systemand the centralized energy system.The economic bene fits of DES may stem from:The introduction of renewable energy,which can reduce the fuel costLower investment cost because the model can select inex-pensive technologies to achieve lower capital expenses than that of CCHP systemsSynergy from the integration of various distributed technologies5.3.Total energy consumption and GHG emissionsAs the potential energy system for this case has multiple inputs and outputs,it is hard to de fine an indicator to evaluate its ef fi-ciency.In this study,we use the same energy demands for all these scenarios and convert all kinds of primary energy consumptions to an equivalent amount of natural gas consumption as follows:TE ng equivalent ¼TE ng þTE gp =0:55(25)where,TE refers to annual total energy consumption;ng refers to natural gas;gp refers to grid power.The above equation converts the consumption of grid power to the equivalent natural gas consumption with a ratio of 0.55,as modern gas turbine plants with a triple-pressure heat recovery steam generator can reach ef ficiencies above 55%,which can represent the most ef ficient way of power generation from natural gas [47].GHG emissions of these scenarios are also accounted by multi-plying the energy consumption with a reliable CO 2emission factor.The CO 2emission factor of natural gas is speci fied at 0.19kg CO 2/kWh and the emission factor of the northeast power grid in 2009is 0.89kg CO 2/kWh [45].The annual energy consumptions,the equivalent natural gas consumptions and GHG emissions of three scenarios are shown in Fig.7.It can be found that grid power is an important energy source for DESs (Scenarios 2&3).Without the support of the power grid,the capacities of distributed energy technologies must be larger to meet the peak electricity load,which will incur a largerinvestmentFig.4.Current energy prices in Beijing.Table 3The technological and economic data of energy generation technologies.TechnologyInvestment,yuan/kW O&M cost,yuan/kWh Ef ficiency LifetimeElectricity Heat Internal combustion engine43570.0720.350.530Gas turbine79000.0680.250.5530Natural gas boiler 8510.00200.920Fuel cell37,2150.20.360.4510Diesel engine 43570.0720.350.520Wind turbine 60000.080.35020Solar PV 20,7000.010.12030Solar heat22500.0100.415Biomass boiler 13870.0800.8515Biogas engine 14,5840.080.280.415Biogas boiler28500.0030.7220Table 4The technological and economic data of energy conversion technologies.TechnologyInvestment,yuan/kW O&M cost,yuan/kWhCOPLifetime Heating mode Cooling mode Electric heater12001020Air-source chiller/heat pump12000.00973320Water-source chiller/heat pump20000.00974520Ground-source chiller/heat pump30000.0097 4.4520Absorption chiller/heat pump12280.0081 1.220Heat exchanger2050.00220.9820Table 5The technological and economic data of energy storage technologies.TechnologyInvestment,yuan/kW O&M cost,yuan/kWh capacity Ef ficiencyLifetimeBattery17828.30.7513.5Hot water storage 900.180.920Ice storage 1900.20.6520Z.Zhou et al./Applied Thermal Engineering 53(2013)387e 396393。

VIAVI SolutionsData SheetVIAVIT-BERD/MTS Quad OTDR ModuleFor T-BERD®/MTS-2000, -4000 V2, -5800, CellAdvisor 5G and OneAdvisor-800 PlatformsThe VIAVI Quad OTDR module is the ideal test tool for installers/contractors, wireless service providers, or any user dealing with both single-mode and multimode applications every day. It is perfect for use in installing, turning up, and maintaining premises and enterprise, access, metro, and wireless fronthaul/backhaul networks.Key Featuresy Up to 37 dB dynamic range insingle-mode and 26 dB in multimode y Quad-wavelength version with850, 1300, 1310, and 1550 nm and a dual-wavelength version with 850 and 1300 nmy Integrated continuous wave (CW) light source and power meter y TIA/IEC pass/fail thresholds y Propagation delay measurement in multimode (TIA-568-C)y Optimized for 10 MB to 40 GE testing y Certifies Tier 2 premises networks** y IEC 61280-4-1-compliant using an external modal controllery Ready for SLM, FTTA-SLM, and FTTH-SLM intelligent optical application softwareThe VIAVI Quad OTDR module features fast acquisition time, sharp resolution, up to a 26 dB multimode dynamic range, and up to a 37 dB single-mode dynamic range for installing and maintaining fiber links. Its integrated light source and power meter, accessible through both OTDR ports (multimode and single-mode), let users quickly identify fiber without switching ports and conduct a full range of fiber certification tests.The Quad module’s optical performance combined with the T-BERD/MTS, CellAdvisor 5G and OneAdvisor-800 platform’s complete suite of features ensures that testing is done right—the first time.Standard test features include: y Automatic macrobend detectiony Summary results table with pass/fail analysis y Bidirectional OTDR analysisy FastReport on-board report generation*Compatible with models -5811P/L and -5822P.**For Tier 1 certification, see the VIAVI Certifier40GT-BERD/MTS-2000 one-slot handheld modular platform fortesting fiber networksT-BERD/MTS-4000 V2 Two-slot handheld modular platform fortesting fiber networksT-BERD/MTS-5800Handheld test instrument for testing 10 G Ethernet and fiber networksCellAdvisor 5GCell site test solutionOneAdvisor-800All-in-One Cell-site Installation and Maintenance Test SolutionSpecifications1. Using a mode conditioner2. Laser at 25°C3. The one-way difference between the extrapolated backscattering level at the start of thefiber and the RMS noise level after 3-minutes averaging4. Measured at ±1.5 dB down from the peak of an unsaturated reflective event5. Measured at ±0.5 dB from the linear regression using an F/UPC-type reflectanceOrdering Information Array Multimode and Quad OTDR Modules and OptionsMultimode 850, 1300 nm OTDR module E4123MMQuad 850/1300/1310/1550 nm OTDR module E4146QUADContinuous and Modulated Source option E41OTDRLSPower Meter option E41OTDRPMAccessoriesEF modal controller for 50 µm MM fiber−SC/PC EFJEF50CONSCPCEF modal controller for 50 µm MM fiber−FC/PC EFJEF50CONFCPCUniversal Optical ConnectorsStraight connectors (single-mode port)EUNIPCFC,EUNIPCSC,EUNIPCST,EUNIPCDIN,EUNIPCLC8° angled connectors (single-mode port)EUNIAPCFC,EUNIAPCSC,EUNIAPCDIN,EUNIAPCLCStraight connectors (multimode port)EUNIPCFCMM,EUNIPCSCMM,EUNIPCSTMM,EUNIPCDINMM,EUNIPCLCMMFor more information on T-BERD/MTS-2000, -4000 V2,-5800, CellAdvisor 5G and OneAdvisor-800 test platforms,please refer to their respective data sheets and brochures.Contact your VIAVI representative for additionalinformation regarding your specific needs.2 VIAVI T-BERD/MTS Quad OTDR Module© 2021 VIAVI Solutions Inc.Product specifications and descriptions in this document are subject to change without notice.Patented as described at /patentsquad-ds-fop-tm-ae 30168207 906 0721Contact Us+1 844 GO VIAVI (+1 844 468 4284)To reach the VIAVI office nearest you, visit /contactVIAVI Solutions。

张或，等：重塑工业废弃地集体记忆景观设计研究同时，配合种植先锋植物来改良土壤，由低级到 高级，自然的发生演替。

场地主要原生植物有芦苇、 狗尾草等耐盐碱植物，通过样方观测的方法确定新增 种苗比例，减少人工恢复植被种群的种间竞争。

在覆 盖土上层种植天津市长势良好的园林绿化植物，如白 腊，臭椿，榆树等乔木，金银木，女贞，紫叶小檗等花灌 木。

种植过程中出现长势不好或腐烂的植物可回填 土壤中改善生态。

在植被恢复的基础上，还应该促进生态系统中动物和微生物的正常生物循环，保证生物 群落的稳定发展。

随着植物的生长，逐渐产生了适合 鸟类、昆虫、爬行动物的生活空间，以及微生物的衍生 场所，食物网随着植物物种的丰富而逐步完善。

生态 系统中运输者，转化者，传播者的营养循环，使当地生 态系统良性循环^农村对于农田中蟋蟀、刺猬、田鼠 被封存的乡土记忆再次被唤醒，如图8所示。

图8 a . b .c .d 景观恢复过程（图片来源:作者自绘)4 结论自然体制的运行过程本身就是生态景观。

自然 体制的运行当然也包括人的参与，这样才能形成一个 更加完整的生态机制。

只有充分构建集体记忆、合理 考虑人与自然之间的资源配置关系，才能达到真正意 义上的人居环境饱和状态；只有人与自然和谐相处， 才能真正地找到此片土地正在消失的乡土记忆。

当 地居民构筑农家生态孕、廊，不仅是对传统休闲文化 的新诠释，同时也是对传统空间可持续发展的表现， 从而更好地复兴这片工业废弃地本身的历史人文场 所精神。

[2] 杰罗姆•特鲁克，曲公英.对场所的记忆和记忆的场所:集体记忆的 哈布瓦赫式社会——K 族，志研究[J ]. M 际社会科，余忐，2012,4(12) ：33 -46.[3] 阿尔多•罗挪国外城市规划与设计理论译丛^城I 丨础筑学[M ].北 京:中国建筑工业出版社，2006:34 - 68.[4] 常必.历史环境的再生之道——历史意识与设计探索[M ].北京：中国迚筑丨:业出版社，2008 ,67.[5] 陈小杳，乍洪远.1业废弃物堆积场地小态恢复与景观营造[J ].现 代國林，2013,10(11) :19 -26.作者简介：张或（1992 )，山西太原人，硕丨:研究生，研究方向；景观规 划设计、城丨1】•附洪管理(494799280@ qq. com) _参考文献：[1]菇纬.灌于文化景观视财下的£业遗产整介[J ].艺术4设讣，2011:124 -125.指导教师：嘗磊（1962 )，大汴人，吨七，教授，研究方向：景观规划设 计、以观生态技术。

42CHINA TODAYWHILE stepping up the development of new materials, innovative drugs, and other cutting-edge sectors, China will also focus on hydrogen in its quest forclean, safe, and efficient energy for industry and transportation, according to the 2024 Report on the Work of the Government. Hydrogen is a clean, safe, and efficient energy source that can power industries and transportation, which makes it an attractive fuel for the future. Consequently, devel-oping hydrogen power was one of the focal points of the discussions at the recent “Two Sessions.”Building a Hydrogen Ecosystem“Hydrogen leads the new round of global en-ergy transformation,” NPC deputy and President of Beijing Yihuatong Technology Co. (SinoHytec) Zhang Guoqiang told China Today in an exclusive interview. It is a key direction for strategic emerg-ing industries and crucial for promoting green and low-carbon energy transition, driving industrial upgrading, and shaping new quality productive forces, whose nature is green development, Zhang added.China’s first guideline on accelerating the green development of the manufacturing industry, released on March 1, 2024, focuses on the energy revolution and industrial transformation under the dual carbon goals – achieving carbon peaking by 2030 and carbon neutrality by 2060. The objective is to build a complete industrial chain of technolo-gies and equipment for hydrogen production, stor-age, transportation and utilization, and to improve the economic efficiency.By staff reporter ZHOU LINFuel of the FutureLawmakers call for stepping up hydrogenpower production to facilitate China’s clean and low-carbon energy transformation and create a new engine of growth.Special Report43April 2024hydrogen energy, and about 20 percent to transportation.Currently, there are over 1,000hydrogen refueling stations worldwide.SinoHytec is a national hi-tech enterprise focused on research and industrializationof hydrogen and fuel cells. Zhang said its two majorbusinesses are hydrogen and fuel cells for land transporta-tion. SinoHytec products havepowered over 4,900 vehicles whose cumulative operating mileage is over 200 million kilometers. The company has been holding the largest market share in China for seven consecutive years. Hebei Province has become a significant player in the industry. Last year it issued an action planfor accelerating the development of new energy as well as seven special plans to promote green andlow-carbon transformation of energy, with the goal of forming an integrated system for wind, solar,water, nuclear, and hydrogen energy production and storage. By 2023, a technology research and devel-opment, inspection and testing system covering the entire hydrogen energy industry chain had been formed. Core technologies such as purification ofindustrial by-product hydrogen are being promoted in fields such as transportation, steel and chemical industries, and communication base stations.Zhang said SinoHytec has created a hydrogen ecosystem in Zhangjiakou in Hebei, building the first plant in China that produces hydrogen from renewable sources and Zhangjiakou’s first hydro-gen refueling station. During the Olympic Win-ter Games Beijing 2022, over 700 fuel cell vehiclesequipped with SinoHytec’s products provided transportation for athletes as well as visitors. Zhang said Zhangjiakou already has 35 hydrogen enterprises, which have formed a preliminary hydrogen energy industrial chain.Energizing the Path to Net ZeroThe green hydrogen chemical industry is an-other important area where hydrogen energy can be used.Major countries across the globe have all rolled out national policies for hydrogen energy, with increased investment to boost advanced technol-ogy research and industrialization. According to Hydrogen Insights 2023, released by the Belgium-headquartered Hydrogen Council and McKinsey & Company, the global hydrogen economy is growing rapidly with accelerated development momentum. By 2030, direct investment in hydrogen energy is expected to reach US $320 billion. About half of the projects funded by these investments will be related to large-scale industrial employment of The global hydrogeneconomy is growing rapidly with accelerated development momentum.By 2030, direct invest-ment in hydrogen energy is expected to reachUS $320 billion.ZhangGuoqiang, an NPC deputy and president of Beijing Yi-huatong Tech-nology Co. (SinoHytec).A fuel cell vehicleequipped with Sino-Hytec’s productsprovides transpor-tation for athletesduring the OlympicWinter Games Bei-jing 2022.44CHINA TODAYXu Zixia, an NPC deputy and chief engineer of China National Chemical Engineering No. 13 Con-struction Co., said the chemical industry is both high-energy-consuming and emission-intensive. It is one of the eight key industries covered by the national carbon market. “The chemical industry needs urgent technological and industrial upgrad-ing to accelerate transformation and reach the goal of carbon reduction,” Xu said.In 2022, China’s chemical industry produced over 1.5 billion tonnes of carbon emissions, ac-counting for nearly 15 percent of the country’s total amount, she said. However, the chemical industry is a pillar of the national economy, with a large economic output, a long industrial chain and multiple products. It is key to the security and stability of the industrial and supply chains and people’s well-being. Therefore green and low-carbon development of the industry is of critical importance.“Hydrogen energy is an important part of production in the chemical industry. Using green hydrogen generated via water electrolysis for production can achieve carbon reduction at the source, during the process, and at the end of the production,” Xu pointed out. It is estimated that 10,000 tonnes of green hydrogen can reduce carbon dioxide emissions by more than 100,000 tonnes.“Green hydrogen” means conducting electrolysis by using renewable sources such as wind, hydro, solar, and nuclear power to produce hydrogen, with no carbon emissions produced during the process. Hebei, for example, has a large amount of renewable energy resources and industrial by-product hydrogen resources, which can be tapped to develop hydrogen energy. Zhangjiakou, Chengde, and regions along the Taihang Moun-tains in Hebei have abundant wind and solar pow-er, which in turn means abundant green hydrogen resources.In April 2022, a guideline was issued for high-quality development of the petrochemical and chemical industry, urging enterprises in the two sectors to develop and utilize green hydrogen in accordance with local conditions, and promote the coupling of refining and coal chemical industries with green electricity and green hydrogen.Key enterprises and projects have started using hydrogen energy. In June 2023, the first10,000-tonne hydrogen production project in China – the Narisong Photovoltaic Hydrogen Produc-tion Industry Demonstration Project in Ordos City, Inner Mongolia – began operation. Two months later, China’s largest photovoltaic green hydrogen project – the Xinjiang Kuche Green Hydrogen Demonstration Project – was launched. Xu called them a breakthrough in the industrial applicationof green hydrogen in China.Guo Jianzeng, an NPC deputy and director of the Science and Technology Committee of the 718th Research Institute of China Shipbuilding Industry Corporation.Special Report“The chemical industry needs urgent technolo-gical and indus-trial upgrading to accelerate transformation and reach the goal of carbon reduction.”UPGRADING45April 2024Green Hydrogen as a National StrategyGuo Jianzeng, an NPC deputy and director of the Science and Technology Committee of the 718th Research Institute of China Shipbuilding In-dustry Corporation, said green hydrogen produced from the electrolysis of water is a major solution to tackle global decarbonization. “It is vital to master the key technologies in this field,” Guo said.Water electrolysis uses renewable energy such as offshore wind energy and tidal energy for the mass production of green hydrogen. Advanced technologies are required for high-density stor-age and transportation of high-pressure hydrogen, liquid hydrogen, and metallic hydrogen. Besides enabling large-scale production, storage, and transportation, they also help lower costs. The hydrogen produced can be used in many scenarios – as fuel for ships, in the chemical industry, trans-portation, and construction.Guo pointed out the multiple advantages of having a marine hydrogen industry. It is an effec-tive way of meeting the carbon peaking and car-bon neutrality goals; it boosts the consumption ofoffshore renewable energy; it meets international green shipping requirements; and finally, it is vital to have independent and controllable technology and equipment throughout the hydrogen energy industrial chain.“Oceans are a source of abundant renewable en-ergy, which can be tapped through offshore wind, floating solar photovoltaic, and other emerging ocean energy technologies. Hydrogen obtained from these renewable energy sources is sustain-able and clean, which makes it a key component in the global energy transition,” Guo said. Hydro-gen can also be used for large-scale energy storage and peak shaving, which effectively addresses the challenges of long-term and large-scale storage of electricity, increases the flexibility of the power system, and promotes high-proportion consump-tion of renewable energy.Moreover, hydrogen is an effective intermedi-ate agent for producing green and clean fuels for ships that meet the decarbonization standards. Various forms of hydrogen energy consumption can be created, such as adding carbon, oxygen, or nitrogen into hydrogen to make ammonia and methanol, which can be an alternative fuel for the shipping industry to achieve large-scale decarbonization.“Hydrogen energy covers numerous products in preparation, storage and transportation and appli-cation, and is an important component of China’s strategic emerging industry chain. It is also a key field in China’s high-end equipment manufacturing industry,” Guo said. The electrolysis water hydro-gen production technology of the 718th Research Institute is a premier technology in China. By de-veloping hydrogen energy and promoting the local-ization of its equipment, China can both enhance its international competitiveness and ensure the security and stability of the entire hydrogen energy industry chain.After discussing all the pros and cons, the NPC deputies were unanimous that the development of hydrogen power should be stepped up so that it becomes a new engine of economic growth. It would facilitate China’s clean and low-carbon en-ergy transition and contribute to the development of new quality productive forces. CThe hydrogen stor-age tank installation area of Sinopec Xin-jiang Kuqa Green Hydrogen Pilot Project in Aksu, Xinjiang, China, on October 28, 2023.。

S C I论文写作中一些常用的句型总结(一)很多文献已经讨论过了一、在Introduction里面经常会使用到的一个句子：很多文献已经讨论过了。

它的可能的说法有很多很多，这里列举几种我很久以前搜集的：A.??Solar energy conversion by photoelectrochemical cells?has been intensively investigated.?(Nature 1991, 353, 737 - 740?)B.?This was demonstrated in a number of studies that?showed that composite plasmonic-metal/semiconductor photocatalysts achieved significantly higher rates in various photocatalytic reactions compared with their pure semiconductor counterparts.C.?Several excellent reviews describing?these applications are available, and we do not discuss these topicsD.?Much work so far has focused on?wide band gap semiconductors for water splitting for the sake of chemical stability.（DOI:10.1038/NMAT3151）E.?Recent developments of?Lewis acids and water-soluble organometalliccatalysts?have attracted much attention.（Chem. Rev. 2002, 102, 3641?3666）F.?An interesting approach?in the use of zeolite as a water-tolerant solid acid?was described by?Ogawa et al（Chem.Rev. 2002, 102, 3641?3666）G.?Considerable research efforts have been devoted to?the direct transition metal-catalyzed conversion of aryl halides toaryl nitriles. (J. Org. Chem. 2000, 65, 7984-7989) H.?There are many excellent reviews in the literature dealing with the basic concepts of?the photocatalytic processand the reader is referred in particular to those by Hoffmann and coworkers,Mills and coworkers, and Kamat.(Metal oxide catalysis,19,P755)I. Nishimiya and Tsutsumi?have reported on（proposed）the influence of the Si/Al ratio of various zeolites on the acid strength, which were estimated by calorimetry using ammonia. (Chem.Rev. 2002, 102, 3641?3666)二、在results and discussion中经常会用到的：如图所示A. GIXRD patterns in?Figure 1A show?the bulk structural information on as-deposited films.?B.?As shown in Figure 7B,?the steady-state current density decreases after cycling between 0.35 and 0.7 V, which is probably due to the dissolution of FeOx.?C.?As can be seen from?parts a and b of Figure 7, the reaction cycles start with the thermodynamically most favorable VOx structures（J. Phys. Chem. C 2014, 118, 24950?24958）这与XX能够相互印证：A.?This is supported by?the appearance in the Ni-doped compounds of an ultraviolet–visible absorption band at 420–520nm (see Fig. 3 inset), corresponding to an energy range of about 2.9 to 2.3 eV.B. ?This?is consistent with the observation from?SEM–EDS. (Z.Zou et al. / Chemical Physics Letters 332 (2000) 271–277)C.?This indicates a good agreement between?the observed and calculated intensities in monoclinic with space groupP2/c when the O atoms are included in the model.D. The results?are in good consistent with?the observed photocatalytic activity...E. Identical conclusions were obtained in studies?where the SPR intensity and wavelength were modulated by manipulating the composition, shape,or size of plasmonic nanostructures.?F.??It was also found that areas of persistent divergent surfaceflow?coincide?with?regions where convection appears to be consistently suppressed even when SSTs are above 27.5°C.(二)1. 值得注意的是...A.?It must also be mentioned that?the recycling of aqueous organic solvent is less desirable than that of pure organic liquid.B.?Another interesting finding is that?zeolites with 10-membered ring pores showed high selectivities (>99%) to cyclohexanol, whereas those with 12-membered ring pores, such as mordenite, produced large amounts of dicyclohexyl ether. (Chem. Rev. 2002, 102,3641?3666)C.?It should be pointed out that?the nanometer-scale distribution of electrocatalyst centers on the electrode surface is also a predominant factor for high ORR electrocatalytic activity.D.?Notably,?the Ru II and Rh I complexes possessing the same BINAP chirality form antipodal amino acids as the predominant products.?(Angew. Chem. Int. Ed., 2002, 41: 2008–2022)E. Given the multitude of various transformations published,?it is noteworthy that?only very few distinct?activation?methods have been identified.?(Chem. Soc. Rev., 2009,?38, 2178-2189)F.?It is important to highlight that?these two directing effects will lead to different enantiomers of the products even if both the “H-bond-catalyst” and the?catalyst?acting by steric shielding have the same absolute stereochemistry. (Chem. Soc. Rev.,?2009,?38, 2178-2189)G.?It is worthwhile mentioning that?these PPNDs can be very stable for several months without the observations of any floating or precipitated dots, which is attributed to the electrostatic repulsions between the positively charge PPNDs resulting in electrosteric stabilization.(Adv. Mater., 2012, 24: 2037–2041)2.?...仍然是个挑战A.?There is thereby an urgent need but it is still a significant challenge to?rationally design and delicately tail or the electroactive MTMOs for advanced LIBs, ECs, MOBs, and FCs.?（Angew. Chem. Int. Ed.2 014, 53, 1488 – 1504）B.?However, systems that are?sufficiently stable and efficient for practical use?have not yet been realized.C.??It?remains?challenging?to?develop highly active HER catalysts based on materials that are more abundant at lower costs. (J. Am. Chem.Soc.,?2011,?133, ?7296–7299)D.?One of the?great?challenges?in the twenty-first century?is?unquestionably energy storage. (Nature Materials?2005, 4, 366 - 377?)众所周知A.?It is well established (accepted) / It is known to all / It is commonlyknown?that?many characteristics of functional materials, such as composition, crystalline phase, structural and morphological features, and the sur-/interface properties between the electrode and electrolyte, would greatly influence the performance of these unique MTMOs in electrochemical energy storage/conversion applications.（Angew. Chem. Int. Ed.2014,53, 1488 – 1504）B.?It is generally accepted (believed) that?for a-Fe2O3-based sensors the change in resistance is mainly caused by the adsorption and desorption of gases on the surface of the sensor structure. (Adv. Mater. 2005, 17, 582)C.?As we all know,?soybean abounds with carbon,?nitrogen?and oxygen elements owing to the existence of sugar,?proteins?and?lipids. (Chem. Commun., 2012,?48, 9367-9369)D.?There is no denying that?their presence may mediate spin moments to align parallel without acting alone to show d0-FM. (Nanoscale, 2013,?5, 3918-3930)(三)1. 正如下文将提到的...A.?As will be described below（也可以是As we shall see below）,?as the Si/Al ratio increases, the surface of the zeolite becomes more hydrophobic and possesses stronger affinity for ethyl acetate and the number of acid sites decreases.（Chem. Rev. 2002, 102, 3641?3666）B. This behavior is to be expected and?will?be?further?discussed?below. (J. Am. Chem. Soc.,?1955,?77, 3701–3707)C.?There are also some small deviations with respect to the flow direction,?whichwe?will?discuss?below.（Science, 2001, 291, 630-633）D.?Below,?we?will?see?what this implies.E.?Complete details of this case?will?be provided at a?later?time.E.?很多论文中，也经常直接用see below来表示，比如：The observation of nanocluster spheres at the ends of the nanowires is suggestive of a VLS growth process (see?below). (Science, 1998, ?279, 208-211)2. 这与XX能够相互印证...A.?This is supported by?the appearance in the Ni-doped compounds of an ultraviolet–visible absorption band at 420–520 nm (see Fig. 3 inset), corresponding to an energy range of about 2.9 to 2.3 eVB.This is consistent with the observation from?SEM–EDS. (Chem. Phys. Lett. 2000, 332, 271–277)C.?Identical conclusions were obtained?in studies where the SPR intensity and wavelength were modulated by manipulating the composition, shape, or size of plasmonic nanostructures.?（Nat. Mater. 2011, DOI: 10.1038/NMAT3151）D. In addition, the shape of the titration curve versus the PPi/1 ratio,?coinciding withthat?obtained by fluorescent titration studies, suggested that both 2:1 and 1:1 host-to-guest complexes are formed. (J. Am. Chem. Soc. 1999, 121, 9463-9464)E.?This unusual luminescence behavior is?in accord with?a recent theoretical prediction; MoS2, an indirect bandgap material in its bulk form, becomes a direct bandgapsemiconductor when thinned to a monolayer.?(Nano Lett.,?2010,?10, 1271–1275)3.?我们的研究可能在哪些方面得到应用A.?Our ?ndings suggest that?the use of solar energy for photocatalytic watersplitting?might provide a viable source for?‘clean’ hydrogen fuel, once the catalyticef?ciency of the semiconductor system has been improved by increasing its surface area and suitable modi?cations of the surface sites.B. Along with this green and cost-effective protocol of synthesis,?we expect that?these novel carbon nanodots?have potential applications in?bioimaging andelectrocatalysis.(Chem. Commun., 2012,?48, 9367-9369)C.?This system could potentially be applied as?the gain medium of solid-state organic-based lasers or as a component of high value photovoltaic (PV) materials, where destructive high energy UV radiation would be converted to useful low energy NIR radiation. (Chem. Soc. Rev., 2013,?42, 29-43)D.?Since the use of?graphene?may enhance the photocatalytic properties of TiO2?under UV and visible-light irradiation,?graphene–TiO2?composites?may potentially be usedto?enhance the bactericidal activity.?(Chem. Soc. Rev., 2012,?41, 782-796)E.??It is the first report that CQDs are both amino-functionalized and highly fluorescent,?which suggests their promising applications in?chemical sensing.(Carbon, 2012,?50,?2810–2815)(四)1. 什么东西还尚未发现/系统研究A. However,systems that are sufficiently stable and efficient for practical use?have not yet been realized.B. Nevertheless,for conventional nanostructured MTMOs as mentioned above,?some problematic disadvantages cannot be overlooked.（Angew. Chem. Int. Ed.2014,53, 1488 – 1504）C.?There are relatively few studies devoted to?determination of cmc values for block copolymer micelles. (Macromolecules 1991, 24, 1033-1040)D. This might be the reason why, despite of the great influence of the preparation on the catalytic activity of gold catalysts,?no systematic study concerning?the synthesis conditions?has been published yet.?(Applied Catalysis A: General2002, 226, ?1–13)E.?These possibilities remain to be?explored.F.??Further effort is required to?understand and better control the parameters dominating the particle surface passivation and resulting properties for carbon dots of brighter photoluminescence. (J. Am. Chem. Soc.,?2006,?128?, 7756–7757)2.?由于/因为...A.?Liquid ammonia?is particularly attractive as?an alternative to water?due to?its stability in the presence of strong reducing agents such as alkali metals that are used to access lower oxidation states.B.?The unique nature of?the cyanide ligand?results from?its ability to act both as a σdonor and a π acceptor combined with its negativecharge and ambidentate nature.C.?Qdots are also excellent probes for two-photon confocalmicroscopy?because?they are characterized by a very large absorption cross section?(Science ?2005,?307, 538-544).D.?As a result of?the reductive strategy we used and of the strong bonding between the surface and the aryl groups, low residual currents (similar to those observed at a bare electrode) were obtained over a large window of potentials, the same as for the unmodified parent GC electrode. (J. Am. Chem. Soc. 1992, 114, 5883-5884)E.?The small Tafel slope of the defect-rich MoS2 ultrathin nanosheets is advantageous for practical?applications,?since?it will lead to a faster increment of HER rate with increasing overpotential.(Adv. Mater., 2013, 25: 5807–5813)F. Fluorescent carbon-based materials have drawn increasing attention in recent years?owing to?exceptional advantages such as high optical absorptivity, chemical stability, biocompatibility, and low toxicity.(Angew. Chem. Int. Ed., 2013, 52: 3953–3957)G.??On the basis of?measurements of the heat of immersion of water on zeolites, Tsutsumi etal. claimed that the surface consists of siloxane bondings and is hydrophobicin the region of low Al content. (Chem. Rev. 2002, 102, 3641?3666)H.?Nanoparticle spatial distributions might have a large significance for catalyst stability,?given that?metal particle growth is a relevant deactivation mechanism for commercial catalysts.?3. ...很重要A.?The inhibition of additional nucleation during growth, in other words, the complete separation?of nucleation and growth,?is?critical(essential, important)?for?the successful synthesis of monodisperse nanocrystals. (Nature Materials?3, 891 - 895 (2004))B.??In the current study,?Cys,?homocysteine?(Hcy) and?glutathione?(GSH) were chosen as model?thiol?compounds since they?play important (significant, vital, critical) roles?in many biological processes and monitoring of these?thiol?compounds?is of great importance for?diagnosis of diseases.(Chem. Commun., 2012,?48, 1147-1149)C.?This is because according to nucleation theory,?what really matters?in addition to the change in temperature ΔT?(or supersaturation) is the cooling rate.(Chem. Soc. Rev., 2014,?43, 2013-2026)(五)1. 相反/不同于A.?On the contrary,?mononuclear complexes, called single-ion magnets (SIM), have shown hysteresis loops of butterfly/phonon bottleneck type, with negligiblecoercivity, and therefore with much shorter relaxation times of magnetization. (Angew. Chem. Int. Ed., 2014, 53: 4413–4417)B.?In contrast,?the Dy compound has significantly larger value of the transversal magnetic moment already in the ground state (ca. 10?1?μB), therefore allowing a fast QTM. （Angew. Chem. Int. Ed., 2014, 53: 4413–4417）C.?In contrast to?the structural similarity of these complexes, their magnetic behavior exhibits strong divergence.?(Angew. Chem. Int. Ed., 2014, 53: 4413–4417)D.?Contrary to?other conducting polymer semiconductors, carbon nitride ischemically and thermally stable and does not rely on complicated device manufacturing. (Nature materials, 2009, 8(1): 76-80.)E.?Unlike?the spherical particles they are derived from that Rayleigh light-scatter in the blue, these nanoprisms exhibit scattering in the red, which could be useful in developing multicolor diagnostic labels on the basis not only of nanoparticle composition and size but also of shape. (Science 2001,? 294, 1901-1903)2. 发现，阐明，报道，证实可供选择的词包括：verify, confirm, elucidate, identify, define, characterize, clarify, establish, ascertain, explain, observe, illuminate, illustrate，demonstrate, show, indicate, exhibit, presented, reveal, display, manifest，suggest, propose, estimate, prove, imply, disclose，report, describe，facilitate the identification of?举例：A. These stacks appear as nanorods in the two-dimensional TEM images, but tilting experiments?confirm that they are nanoprisms.?(Science 2001,? 294, 1901-1903)B. Note that TEM?shows?that about 20% of the nanoprisms are truncated.?(Science 2001,? 294, 1901-1903)C. Therefore, these calculations not only allow us to?identify?the important features in the spectrum of the nanoprisms but also the subtle relation between particle shape and the frequency of the bands that make up their spectra.?(Science 2001,? 294, 1901-1903)D. We?observed?a decrease in intensity of the characteristic surface plasmon band in the ultraviolet-visible (UV-Vis) spectroscopy for the spherical particles at λmax?= 400 nm with a concomitant growth of three new bands of λmax?= 335 (weak), 470 (medium), and 670 nm (strong), respectively. (Science 2001,? 294, 1901-1903)E. In this article, we present data?demonstrating?that opiate and nonopiate analgesia systems can be selectively activated by different environmental manipulationsand?describe?the neural circuitry involved. (Science 1982, 216, 1185-1192)F. This?suggests?that the cobalt in CoP has a partial positive charge (δ+), while the phosphorus has a partial negative charge (δ?),?implying?a transfer of electron density from Co to P.?(Angew. Chem., 2014, 126: 6828–6832)3. 如何指出当前研究的不足A. Although these inorganic substructures can exhibit a high density of functional groups, such as bridging OH groups, and the substructures contribute significantly to the adsorption properties of the material,surprisingly little attention has been devoted to?the post-synthetic functionalization of the inorganic units within MOFs. (Chem. Eur. J., 2013, 19: 5533–5536.)B.?Little is known,?however, about the microstructure of this material. (Nature Materials 2013,12, 554–561)C.?So far, very little information is available, and only in?the absorber film, not in the whole operational devices. （Nano Lett.,?2014,?14?(2), pp 888–893）D.?In fact it should be noted that very little optimisation work has been carried out on?these devices. (Chem. Commun., 2013,?49, 7893-7895)E. By far the most architectures have been prepared using a solution processed perovskite material,?yet a few examples have been reported that?have used an evaporated perovskite layer. (Adv. Mater., 2014, 27: 1837–1841.)F. Water balance issues have been effectively addressed in PEMFC technology through a large body of work encompassing imaging, detailed water content and water balance measurements, materials optimization and modeling,?but very few of these activities have been undertaken for?anion exchange membrane fuel cells,? primarily due to limited materials availability and device lifetime. (J. Polym. Sci. Part B: Polym. Phys., 2013, 51: 1727–1735)G. However,?none of these studies?tested for Th17 memory, a recently identified T cell that specializes in controlling extracellular bacterial infections at mucosal surfaces. (PNAS, 2013,?111, 787–792)H. However,?uncertainty still remains as to?the mechanism by which Li salt addition results in an extension of the cathodic reduction limit. (Energy Environ. Sci., 2014,?7, 232-250)I.?There have been a number of high profile cases where failure to?identify the most stable crystal form of a drug has led to severe formulation problems in manufacture. (Chem. Soc. Rev., 2014,?43, 2080-2088)J. However,?these measurements systematically underestimate?the amount of ordered material. ( Nature Materials 2013, 12, 1038–1044)(六)1.?取决于a.?This is an important distinction, as the overall activity of a catalyst will?depend on?the material properties, synthesis method, and other possible species that can be formed during activation.?(Nat. Mater.?2017,16,225–229)b.?This quantitative partitioning?was determined by?growing crystals of the 1:1 host–guest complex between?ExBox4+?and corannulene. (Nat. Chem.?2014,?6177–178)c.?They suggested that the Au particle size may?be the decisive factor for?achieving highly active Au catalysts.(Acc. Chem. Res.,?2014,?47, 740–749)d.?Low-valent late transition-metal catalysis has?become indispensable to?chemical synthesis, but homogeneous high-valent transition-metal catalysis is underdeveloped, mainly owing to the reactivity of high-valent transition-metal complexes and the challenges associated with synthesizing them.?(Nature2015,?517,449–454)e.?The polar effect?is a remarkable property that enables?considerably endergonic C–H abstractions?that would not be possible otherwise.?(Nature?2015, 525, 87–90)f.?Advances in heterogeneous catalysis?must rely on?the rational design of new catalysts. (Nat. Nanotechnol.?2017, 12, 100–101)g.?Likely, the origin of the chemoselectivity may?be also closely related to?the H?bonding with the N or O?atom of the nitroso moiety, a similar H-bonding effect is known in enamine-based nitroso chemistry. (Angew. Chem. Int. Ed.?2014, 53: 4149–4153)2.?有很大潜力a.?The quest for new methodologies to assemble complex organic molecules?continues to be a great impetus to?research efforts to discover or to optimize new catalytic transformations. (Nat. Chem.?2015,?7, 477–482)b.?Nanosized faujasite (FAU) crystals?have great potential as?catalysts or adsorbents to more efficiently process present and forthcoming synthetic and renewablefeedstocks in oil refining, petrochemistry and fine chemistry. (Nat. Mater.?2015, 14, 447–451)c.?For this purpose, vibrational spectroscopy?has proved promising?and very useful.?(Acc Chem Res. 2015, 48, 407–413.)d.?While a detailed mechanism remains to be elucidated and?there is room for improvement?in the yields and selectivities, it should be remarked that chirality transfer upon trifluoromethylation of enantioenriched allylsilanes was shown. (Top Catal.?2014,?57: 967.?)e.?The future looks bright for?the use of PGMs as catalysts, both on laboratory and industrial scales, because the preparation of most kinds of single-atom metal catalyst is likely to be straightforward, and because characterization of such catalysts has become easier with the advent of techniques that readily discriminate single atoms from small clusters and nanoparticles. (Nature?2015, 525, 325–326)f.?The unique mesostructure of the 3D-dendritic MSNSs with mesopore channels of short length and large diameter?is supposed to be the key role in?immobilization of active and robust heterogeneous catalysts, and?it would have more hopeful prospects in?catalytic applications. (ACS Appl. Mater. Interfaces,?2015,?7, 17450–17459)g.?Visible-light photoredox catalysis?offers exciting opportunities to?achieve challenging carbon–carbon bond formations under mild and ecologically benign conditions. (Acc. Chem. Res.,?2016, 49, 1990–1996)3. 因此同义词：Therefore, thus, consequently, hence, accordingly, so, as a result这一条比较简单，这里主要讲一下这些词的副词词性和灵活运用。

Data SheetCisco Aironet 1130AG Series IEEE 802.11A/B/GAccess PointLow-profile enterprise-class access point with integrated antennas for easy deployment in offices and similar RF environments.Product OverviewCisco® Aironet® 1130AG Series IEEE 802.11a/b/g access points provide high-capacity, high-security, enterprise-class features in an unobtrusive, office-class design, delivering WLAN access with the lowest total cost of ownership. With high-performing dual IEEE 802.11a and 802.11g radios, the Cisco Aironet 1130AG Series provides a combined capacity of up to 108 Mbps to meet the needs of growing WLANs. Hardware-assisted Advanced Encryption Standard (AES) or temporal key integrity protocol (TKIP) encryption provides uncompromised support for interoperable IEEE 802.11i, Wi-Fi Protected Access 2 (WPA2) or WPA security. The Cisco Aironet 1130AG Series uses radio and network management features for simplified deployment, along with built-in omnidirectional antennas that provide robust and predictable WLAN coverage for offices and similar RF environments. The competitively priced Cisco Aironet 1130AG Series is ready to install and easy to manage, reducing the cost of deployment and ongoing maintenance.The Cisco Aironet 1130AG Series is available in two versions: unified or autonomous. Unified access points operate with the Lightweight Access Point Protocol (LWAPP) and work in conjunction with Cisco wireless LAN controllers and the Cisco Wireless Control System (WCS). When configured with LWAPP, the Cisco Aironet 1130AG Series can automatically detect the best-available Cisco wireless LAN controller and download appropriate policies and configuration information with no manual intervention. Autonomous access points are based on Cisco IOS® Software and may optionally operate with the CiscoWorks Wireless LAN Solution Engine (WLSE). Autonomous access points, along with the CiscoWorks WLSE, deliver a core set of features and may be field-upgraded to take advantage of the full benefits of the Cisco Unified Wireless Network as requirements evolve.The Cisco Aironet 1130AG Series delivers optimal value for offices and similar environments. Built-in antennas provide omnidirectional coverage specifically designed for today’s open workspaces. A multipurpose mounting bracket easily secures Cisco Aironet 1130AG Series access points to ceilings and walls. With an unobtrusive design, Cisco Aironet 1130AG Series access points are aesthetically pleasing and blend into their environments. Formaximum concealment, the access point may be placed above ceilings or suspended ceilings. The UL 2043 rating of the Cisco Aironet 1130AG Series allows the access point to be placed above ceilings in plenum areas regulated by municipal fire codes. Offered at a competitive price, and optimized for easy installation and operation, the Cisco Aironet 1130AG Series helps organizations attain a lower total cost of ownership.ApplicationsIn offices and similarly open environments, Cisco Aironet 1130AG Series access points may be installed on the ceiling to provide users with continuous coverage as they roam throughout a facility. In school buildings and similar facilities, the access points may be installed on the ceiling of each room and hallway to provide users with full coverage and high network availability. In areas where a ceiling installation may not be practical such as retail hotspots or similar small facilities, the access points can be mounted simply and securely on walls for complete coverage with minimal installation cost.Award-Winning SecurityThe Cisco Aironet 1130AG Series has achieved National Institute of Standards and Technology (NIST) FIPS 140-2 level 2 validation and is in process for Information Assurance validation under the National Information Assurance Partnership (NIAP) Common Criteria program. The Cisco Aironet 1130AG Series supports 802.11i, Wi-Fi Protected Access (WPA), WPA2, and numerous Extensible Authentication Protocol (EAP) types. WPA and WPA2 are the Wi-Fi Alliance certifications for interoperable, standards-based WLAN security. These certifications support IEEE 802.1X for user-based authentication, Temporal Key Integrity Protocol (TKIP) for WPA encryption, and Advanced Encryption Standard (AES) for WPA2 encryption. These certifications help to ensure interoperability between Wi-Fi-certified WLAN devices from different manufacturers.The Cisco Aironet 1130AG Series hardware-accelerated AES encryption supports enterprise-class, government-grade secure encryption over the WLAN without compromising performance. IEEE 802.1X authentication helps to ensure that only authorized users are allowed on the network. Backward compatibility and support for WPA client devices running TKIP, the RC4 encryption algorithm, is also supported by the Cisco Aironet 1130AG Series.Cisco Aironet 1130AG Series Access Points operating with LWAPP support Cisco Unified Intrusion Detection System/Intrusion Prevention System (IDS/IPS), a software feature that is part of the Cisco Self-Defending Network and is the industry’s first integrated wired and wireless security solution. Cisco Unified IDS/IPS takes a comprehensive approach to security—at the wireless edge, wired edge, WAN edge, and through the data center. When an associated client sends malicious traffic through the Cisco Unified Wireless Network, a Cisco wired IDS device detects the attack and sends shun requests to Cisco wireless LAN controllers, which will then disassociate the client device.Autonomous or unified Cisco Aironet 1130AG Series Access Points support management frame protection for the authentication of 802.11 management frames by the wireless network infrastructure. This allows the network to detect spoofed frames from access points or malicious users impersonating infrastructure access points. If an access point detects a malicious attack, an incident will be generated by the access point and reports will be gathered on the Cisco wireless LAN controller, Cisco WCS, or CiscoWorks WLSE.Features and BenefitsTable 1 lists features and benefits of Cisco Aironet 1130AG Series access points.Table 1. Features and Benefits of Cisco Aironet 1130AG Series Access PointsFeature BenefitDual 802.11a and 802.11g Radios ●Provides up to 108 Mbps of capacity in a single device for industry-leading capacity and backward compatibility with legacy 802.11b clients.Feature BenefitSupports 15 Nonoverlapping Channels ●Lower potential interference with neighboring access points simplifies deployment ●Fewer transmission errors deliver greater throughputIndustry-Leading Radio Design ●Provides robust signals to long distances●Mitigates the effects of multipath signal propagation for more consistent coverageVariable Transmit Power Settings ●Allows access point coverage to be tuned for differing requirements●Low—dBm setting supports closer spacing of access points in high-density deploymentsIntegrated Antennas ●Complete system is deployable out of the box without external antennas●Specifically designed to provide omnidirectional coverage for offices and similar radio frequency environments Hardware-Assisted AESEncryption●Provides high security without performance degradationCisco Unified IDS/IPS ●This integrated software feature is part of the Cisco Self-Defending Network and is the industry’s firstintegrated wired and wireless security solution. When a trusted client acts maliciously, the wired IDS detectsthe attack and sends shun requests to Cisco WLAN controllers, which will then disassociate the client device. Management Frame Protection ●This feature provides for the authentication of 802.11 management frames by the wireless networkinfrastructure. This allows the network to detect spoofed frames from access points or malicious usersimpersonating infrastructure access points. If an access point detects a malicious attack, an incident will begenerated by the access points and reports will be gathered on the Cisco wireless LAN controller, Cisco WCS,or CiscoWorks WLSE.IEEE 802.11i-Compliant; WPA2-Certified and WPA-Certified●Helps to ensure interoperable security with wireless LAN client devices from other manufacturersLow-Profile Design ●Unobtrusive design blends in to environment●“Quiet” LED does not draw attention to it when operating normally and no action is requiredMultipurpose and Lockable Mounting Bracket ●Installs easily to walls, ceilings, and suspended ceiling railways ●Accommodates standard padlock to prevent theftInline Power Support (IEEE 802.3af and Cisco Inline Power) ●Provides an interoperable alternative to AC power●Simplifies deployment by allowing power to be supplied over the Ethernet cable ●Compatible with 802.3af-compliant power sourcesCisco Green Bulk Packaging To reduce product packaging and preserve the environment, the Cisco Aironet 1130 Series may be ordered in abulk package that includes 10 access points and 10 mounting kits.Summary/ConclusionThe Cisco Aironet 1130AG Series provides the ideal enterprise access point for offices and similar environments. With two high-performance radios, these access points provide simultaneous support for the 802.11a and 802.11g standards, offering 108 Mbps of capacity for your growing WLAN. Incorporating AES encryption in hardware, the Cisco Aironet 1130AG Series complies with the 802.11i security standard and is WPA2-certified, helping to assure that your network employs the strongest security available while maintaining interoperability with products from other manufacturers. Additional design features, including diversity antennas with omnidirectional coverage and an unobtrusive form factor, along with an attractive price, provide low total cost of ownership.For office environments, the Cisco Aironet 1130AG Series is a cost-compelling solution for a high-capacity, high-security, enterprise-class WLAN.Product SpecificationsTable 2 lists the product specifications for Cisco Aironet 1130AG access points.Table 2. Product Specifications for Cisco Aironet 1130AG Access Points Item SpecificationPart Number for Individual Access Points●AIR-AP1131AG-x-K9 (Cisco IOS Software)●AIR-LAP1131AG-x-K9 (Cisco Unified Wireless Network Software)Note: The Cisco Aironet 1130AG Series may be ordered with Cisco IOS Software to operate as an autonomous AP with Cisco Unified Wireless Network Software using LWAPP. When the 1130AG is operating as a lightweight AP a WLAN controller is required.●Regulatory Domains: (x = Regulatory Domain)●A = FCC●C = China●E = ETSI●I = Israel●J = TELEC (Japan)●K = Korea●N = North America (Excluding FCC)●P = Japan2●S = Singapore●T = TaiwanCustomers are responsible for verifying approval for use in their individual countries. To verify approval and to identify the regulatory domain that corresponds to a particular country please visit:/go/aironet/complianceNot all regulatory domains have been approved. As they are approved, the part numbers will be available on the Global Price List.Part Number for Cisco Green Bulk Packaging●AIR-AP1131-x-K9-10 (Cisco IOS Software)●AIR-LAP1131-xK9-10 (Cisco Unified Wireless Network Software)Note: The Cisco Aironet 1130AG Series may be ordered with Cisco IOS Software to operate as an autonomous AP with Cisco Unified Wireless Network Software using LWAPP. When the 1130AG is operating as a lightweight AP a WLAN controller is required.●Regulatory Domains: (x = Regulatory Domain)●A = FCC●E = ETSI●Customers are responsible for verifying approval for use in their individual countries. To verify approval and toidentify the regulatory domain that corresponds to a particular country please visit:/go/aironet/complianceSoftware ●Cisco IOS Software Release 12.3(8)JA or later (autonomous).●Cisco IOS Software Release 12.3(11)JX or later (Lightweight Mode).●Cisco Unified Wireless Network Software Release 4.0 or later. Data Rates Supported ●802.11a: 6, 9, 12, 18, 24, 36, 48, and 54 Mbps●802.11g: 1, 2, 5.5, 6, 9, 11, 12, 18, 24, 36, 48, and 54 Mbps Network Standard IEEE 802.11a, 802.11b, and 802.11gUplink Autosensing 802.3 10/100BASE-T EthernetFrequency Band and Operating Channels Americas (FCC)●2.412 to 2.462 GHz; 11 channels●5.15 to 5.35, 5.725 to 5.825 GHz; 12 channelsChina●2.412 to 2.472 GHz; 13 channels●5.725 to 5.825 GHz; 4 channelsETSI●2.412 to 2.472 GHz; 13 channels●5.15 to 5.725 GHz; 19 channelsIsrael●2.432 to 2.472 GHz; 9 channels●5.15 to 5.35 GHz, 8 channelsJapan (TELEC)●2.412 to 2.472 GHz; 13 channels Orthogonal Frequency Division Multiplexing (OFDM) ●2.412 to 2.484 GHz; 14 channels Complementary Code Keying (CCK)●5.15 to 5.25 GHz; 4 channelsJapan–P (TELEC 2 (Japan2) Cnfg)●2.412 to 2.472 GHz; 13 channels Orthogonal Frequency Division Multiplexing (OFDM) ●2.412 to 2.484 GHz; 14 channels Complementary Code Keying (CCK)●5.15 to 5.35 GHz, 8 channelsJapan-Q●2.412 to 2.472 GHz; 13 channels Orthogonal Frequency Division Multiplexing (OFDM)●2.412 to 2.484 GHz; 14 channels Complementary Code Keying (CCK)●5.15 to 5.35 GHz, 8 channels●5.470 to 5.725 GHz, 11 channelsKorea●2.412 to 2.472 GHz; 13 channels●5.15 to 5.35, 5.46 to 5.72, 5.725 to 5.825, 19 channelsNorth America●2.412 to 2.462 GHz; 11 channels●5.15 to 5.35, 5.725 to 5.825 GHz; 12 channelsSingapore●2.412 to 2.472 GHz, 13 channels●5.15 to 5.35 GHz, 8 channels and 5.725 to 5.825 GHz, 12 channelsTaiwan●2.412 to 2.462 GHz, 11 channels●5.25-5.35 GHz, 5.725 to 5.825, 7 channelsNonoverlapping Channels 802.11a: Up to 19 802.11b/g: 3Receive Sensitivity (Typical) 802.11a:6 Mbps: -87 dBm9 Mbps: -86 dBm12 Mbps: -85 dBm18 Mbps: -84 dBm24 Mbps: -80 dBm36 Mbps: -78 dBm48 Mbps: -73 dBm54 Mbps: -71 dBm 802.11g:1 Mbps: -93 dBm2 Mbps: -91 dBm5.5 Mbps: -88 dBm6 Mbps: -86 dBm 9 Mbps: -85 dBm11 Mbps: -85 dBm12 Mbps: -84 dBm 18 Mbps: -83 dBm 24 Mbps: -79 dBm 36 Mbps: -77 dBm 48 Mbps: -72 dBm 54 Mbps: -70 dBmAvailable Transmit Power Settings (Maximum Power Setting Will Vary by Channel and According to Individual Country Regulations) 802.11a:OFDM:17 dBm (50 mW)15 dBm (30 mW)14 dBm (25 mW)11 dBm (12 mW)8 dBm (6 mW)5 dBm (3 mW)2 mW (2 dBm)-1 dBm (1 mW)802.11b:CCK:20 dBm (100 mW)17 dBm (50 mW)14 dBm (25 mW)11 dBm (12 mW)8 dBm (6 mW)5 dBm (3 mW)2 dBm (2 mW)-1 dBm (1 mW)802.11g:OFDM:17 dBm (50 mW)14 dBm (25 mW)11 dBm (12 mW)8 dBm (6 mW)5 dBm (3 mW)2 dBm (2 mW)-1 dBm (1 mW)Indoor (Distance AcrossOpen Office Environment):Outdoor:Range802.11a:80 ft (24 m) @ 54 Mbps150 ft (45 m) @ 48Mbps200 ft (60 m) @ 36Mbps225 ft (69 m) @ 24Mbps250 ft (76 m) @ 18Mbps275 ft (84 m) @ 12Mbps300 ft (91 m) @ 9 Mbps325 ft (100 m) @ 802.11g:100 ft (30 m) @54Mbps175 ft (53 m) @48Mbps250 ft (76 m) @36Mbps275 ft (84 m) @24Mbps325 ft (100 m) @18Mbps350 ft (107 m) @12Mbps360 ft (110 m) @11Mbps375 ft (114 m) @802.11a:100 ft (30 m) @54Mbps300 ft (91 m) @48Mbps425 ft (130 m) @36Mbps500 ft (152 m) @24Mbps550 ft (168 m) @18Mbps600 ft (183 m) @12Mbps625 ft (190 m) @9Mbps650 ft (198 m) @802.11g:120 ft (37 m) @ 54Mbps350 ft (107 m) @ 48Mbps550 ft (168 m) @ 36Mbps650 ft (198 m) @ 24Mbps750 ft (229 m) @ 18Mbps800 ft (244 m) @ 12Mbps820 ft (250 m) @ 11Mbps875 ft (267 m) @ 9Mbps900 ft (274 m) @ 6Mbps910 ft (277 m) @ 5.5Mbps940 ft (287 m) @ 2Mbps950 ft (290 m) @ 1Mbps6Mbps6Mbps 9Mbps400 ft (122 m) @6Mbps420 ft (128 m) @5.5Mbps440 ft (134 m) @2Mbps450 ft (137 m) @1MbpsRanges and actual throughput vary based upon numerous environmental factors so individual performance may differ. Compliance StandardsSafety●UL 60950-1●CAN/CSA-C22.2 No. 60950-1●UL 2043●IEC 60950-1●EN 60950-1●NIST FIPS 140-2 level 2 validationRadio Approvals●FCC Part 15.247, 15.407●RSS-210 (Canada)●EN 300.328, EN 301.893 (Europe)●ARIB-STD 33 (Japan)●ARIB-STD 66 (Japan)●ARIB-STD T71 (Japan)●AS/NZS 4268.2003 (Australia and New Zealand)EMI and Susceptibility (Class B)●FCC Part 15.107 and 15.109●ICES-003 (Canada)●VCCI (Japan)●EN 301.489-1 and -17 (Europe)Security●802.11i, WPA2, WPA●802.1X●AES, TKIP●FIPS 140-2 Pre-Validation List●Common Criteria (when running Cisco IOS software)Other●IEEE 802.11g and IEEE 802.11a●FCC Bulletin OET-65C●RSS-102Antennas ●2.4 GHz◦Gain 3.0 dBi◦Horizontal Beamwidth 360°●5 GHz◦Gain 4.5 dBi◦Horizontal Beamwidth 360°Security AuthenticationSecurity Standards●WPA●WPA2 (802.11i)●Cisco TKIP●Cisco message integrity check (MIC)●IEEE 802.11 WEP keys of 40 bits and 128 bits802.1X EAP types:●EAP-Flexible Authentication via Secure Tunneling (EAP-FAST)●Protected EAP-Generic Token Card (PEAP-GTC)●PEAP-Microsoft Challenge Authentication Protocol Version 2 (PEAP-MSCHAP)●EAP-Transport Layer Security (EAP-TLS)●EAP-Tunneled TLS (EAP-TTLS)●EAP-Subscriber Identity Module (EAP-SIM)●Cisco LEAPEncryption●AES-CCMP encryption (WPA2)●TKIP (WPA)●Cisco TKIP●WPA TKIP●IEEE 802.11 WEP keys of 40 bits and 128 bitsStatus LEDs External:●Status LED indicates operating state, association status, error/warning condition, boot sequence, and maintenancestatusInternal:●Ethernet LED indicates activity over the Ethernet, status●Radio LED indicates activity over the radios, statusDimensions (H x W x D) 7.5 in. x 7.5 in. x 1.3 in. (19.1 x 19.1 x 3.3 cm)Weight 1.5 lb (0.67 kg)Environmental Operating●Altitude: 0 to 2500m●32 to 104°F (0 to 40°C)●10 to 90% humidity (noncondensing)Non Operating●-40 to 158F (-40 to 70C)●Up to 95% humidity (noncondensing)System Memory ●32 MB RAM●16 MB FLASHInput Power Requirements ●100-240 VAC; 50-60Hz (power supply)●36-57 VDC (device)Power Draw 12.2W maximumWarranty One yearWi-Fi CertificationSystem RequirementsTable 3 lists the system requirements for Cisco Aironet 1130AG access points.Table 3. System Requirements for Cisco Aironet 1130AG Access PointsAccess Utilizing DescriptionBrowser Using the Web browser management GUI, requires a computer running Internet Explorer Version 6.0 or newer, orNetscape Navigator Version 7.0 or newer.Power over Ethernet (PoE)Power sourcing equipment (PSE) compliant with Cisco Inline Power or IEEE 802.3af, and providing at least 12.2W at48 VDC.Service and SupportCisco Systems® offers a wide range of services programs to accelerate customer success. These innovative services programs are delivered through a unique combination of people, processes, tools, and partners, resulting in high levels of customer satisfaction. Cisco services help you protect your network investment, optimize network operations, and prepare your network for new applications to extend network intelligence and the power of your business. For more information about Cisco services, visit Cisco Technical Support Services or Cisco Advanced Services.For More InformationFor more information about the Cisco Aironet 1130AG Series, visit /go/wireless or contact your local account representative.。

GeoHelix S Product Specifi cation V2 Iss 07-06 Sarantel Propriety Information1Specifi cations subject to changeSarantel UK***************** v2 Iss 10-09 © Sarantel Ltd. 2009Product DescriptionBuilt on patented PowerHelix ® fi lteringantenna technology, the SL1204 surface-mount GPS antenna is the smallest active quadrifi lar helix antenna available, providing high performance in diffi cult GPSapplications. The SL1204 integrates a high-performance, optimised gain, low-noise amplifi er with Sarantel’s GeoHelix ®-P2 antenna for receivers requiring an active input. The SL1204 has been optimally designed to workbest with the latest generation of GPS chipsets. The SL1204 antenna is ideal in applications where:• the device is handheld, body-worn, or otherwise surrounded by high-dielectric materials that would de-tune conventional antennas;• the antenna is tightly integrated with other antennas, e.g., Bluetooth®/GPS receivers or GPS/GSM mobile phones;• there are tight constraints on the size of the device or the amount of space allocated to ground planes;• the 2nd Generation GPS receiver requires optimised LNA gain for optimal GPS performance;• the orientation of the device is random; or • the antenna will be embedded in the device.The SL1204 antenna is balanced, which isolates it from the device and enables the antenna to reject common mode noise resident on the device ground plane. The construction and materials of the antenna constrain its near-fi eld to a very small volume, therefore materials near the antenna have negligible de-tuning effects and the antenna maintains its pattern and effi ciency in the presence of dielectric loading. As a dielectrically -loaded antenna, the SL1204 antenna effectively attenuates signals from common GSM and ISM frequencies, minimizing the need for additional fi ltering.The SL1204 antenna may be deployed in an external, “stub-style” confi guration, but it is also a simple antenna to embed due to its isolation properties.Specifi cationsMinimumTypical Maximum Unit Part Number SL1204(see page 4 for detailed part numbers)Each Type Quadrifi lar HelixFrequency 1573.421575.421577.42MHz Polarization Right-hand circular polarizedVoltage 1.8 3.0 3.6V Current 3.1 3.4 3.9mA Gain +17+18dBic Beamwidth 135Degrees Bandwidth (3dB)20MHz Axial Ratio <2.0@ZenithVSWR <2.0:1 2.3:1Impedance 50Noise Figure0.80.9dB Input 3rd Order Intercept Point -7dBm Operating Temperature -40+20+85°C Element Dimensions10 (diameter) x 17 (length)mm Overall Dimensions (w/radome)13.3 (dia) x 12.4 (width) x 34 (length)mm Weight (excl radome or sleeve) 7.0gramsMountingSMTSL1204 (GeoHelix®–M)2nd Generation Active GPS Antenna Product Specifi cationThe SL1204 , with optimised LNA gain and low-noisefigure, provides a typical gain and Noise Figure of 18dBicand 0.8dB, respectively.This antenna is designed to work effi ciently with the latestgeneration of GPS chipsets out in the market today.The strength of the PowerHelix® antenna technology is itsimmunity to de-tuning in the presence of dielectric loading,like human tissues. The SL1204 antenna retains effi ciencyand polarization near the human body. Conventionalantennas lose 5-10dB of gain in similar circumstances.Though it will not electrically couple with a ground plane,the SL1204 antenna can be expected to increase effi ciencyby up to 100% when mounted over a ground plane due tonear-fi eld signal refl ections. Confi guration and orientationof the ground plane with respect to the antenna will varyresults, but effi ciency will not decrease.Radiation Patterns (dBic)Frequency (MHz)S 12 (dB)860GSM 900-40970-371575.42GPS (L1)+181700GSM 1800-41800-71900-92450Bluetooth/Wifi -21Filtering ResponseAntenna Integration OptionsThe SL1204 antennas may be mounted externally or embedded within a device. When mounting externally, the groove in theradome should be used as a mechanical support. Internally embedded antenna requires support ribs in the plastic housing. For further information on integrating the SL1204, please refer to the integration guidelines.Ordering Guide for the SL1204 antennaPart Number Description MOQ Pack SizeM Antenna 400 400SL1204 GeoHelix®M Antenna with Radome 400 400SL1204R GeoHelix®M Antenna with Short Sleeve 400 400SL1204SB GeoHelix®Notesplaced for below MOQ or not in multiples of pack quantities will be subject to a $20 handling 1. Ordersfeefurther guidance on selecting the correct part number please contact Sarantel distributors or 2. ForMechanical Drawings - Overall DimensionsMechanical Drawings - Footprint and Board Profi leRoSH Compliance StatementApplication SupportSarantel are committed to our customers’ success, and sooffer a variety of support options for customers designing RFproducts.Check the Sarantel web site at /products forthe latest production specifi cations, technical notes, andapplication guides for solutions to the most common antennaintegration issues.Contact our applications support group by email at****************************************fi cations,including mechanical drawings, surface mount pad layout,embedding recommendations, and other applicationquestions not answered in the technical literature.For further support options, please contact your local salesrepresentative at /wheretobuy.About SarantelSarantel designs and manufactures dielectricallyloaded antennas based on patented PowerHelix®filtering antenna technology. Sarantel’s antennasare ideal for applications in which the radio deviceis small, handheld, or body-worn, or in deviceswith multiple transceivers and high levels ofcommon mode noise. Sarantel antennas can bemounted externally or easily embedded within adevice.Sarantel antennas are protected by US and othergranted or pending international patents.GeoHelix®, PowerHelix®, and the Sarantel logoare registered trademarks of Sarantel Ltd.Contact SarantelSarantel Ltd. (HQ)Unit 2, Wendel PointRyle Drive, Park Farm SouthWellingborough, NN8 6BAUnited KingdomPh: +44 1933 670560Fax: +44 1933 401155Email:*****************Web: Global Distributors & Representatives/wheretobuySarantel strongly believes in the value of intellectual property and the right of entrepreneurs to protect what they have created. Sarantel dem-onstrates its commitment to this principle by continuously developing its technology and then fi ling patents in a number of regions around the world. Additionally, Sarantel is constantly fi ling new patent applications and has a substantial portfolio of pending applications.A list of Sarantel’s granted patents;Australia; 707488, 716542, 720873, 769570, 2004223229 Austria; 0791978 Brazil; PI9508769-9 Canada; 2198318, 2198375, 2245882, 2250790, 2272389, 2357041, 2373941, 2521493 China; ZL00136656.4, ZL00803562.8, ZL00808144.1, ZL95195772.4, ZL97181567.4, ZL97193099.6,ZL97194742.2, ZL99816387.2 Denmark; 0777922, 1088367 Finland; 0791978, 0876688, 0935826, 0941557, 1088367, 1098392, 1147571, 1153458, 1196963, 1609213 France; 0777922, 0791978, 0876688, 0935826, 0941557, 1081787, 1088367, 1098392, 1147571, 1153458, 1196963, 1609213 Germany; 60003157.8-0, 60029538.9-08, 60034042.2-08, 602004010085.4-08, 69535431.0-08, 69722590.9-0, 69723093.7-0, 69726177.8-0, 69730369.1-08, 69923558.8-08, 69930407.5-08 India; 193515, 193751, 193929, 195085, 206740 Italy; 0777922, 0791978, 0876688, 0935826, 1081787, 1088367, 1153458, 1196963 Japan; 3489684, 3489775, 3923530, 3946955, 4052800, 4057612, 4077197, 4099309, 4147260, 4159749, 4188412 Malaysia; MY-112473-A, MY-119077-A, MY-119465-A, MY-123075-A Mexico; 199890, 205239, 213947, 220048, 213633, 232437, 232438, 231633, 259577 Netherlands; 0791978, 1081787 New Zealand; 291852, 334614 Philippines; 1-1995-51169, 1-1997-55284, 1-1997-55978, 1-1997-58557, 1-1999-03167 Russia; 2173009, 2210146, 2339131 Singapore; 37745, 54891, 56480, 116791, 131698 South Korea; 348441, 366071, 446790, 458310, 523092, 625638, 650620, 650621, 650622, 663873, 667216, 667221, 709688, 767329 Spain; 0777922, 0791978, 1088367, 1196963 Sweden; 0777922, 0791978, 0876688, 0935826, 0941557, 1081787, 1088367, 1098392, 1147571, 1153458, 1196963, 1609213 Switzerland; 0791978, 1081787, 1196963 Taiwan; 094978, 108488, 123671, 144801, 156702, 285980, M 312023 Thailand; 17812, 19360, 19570, 23745 United Kingdom; 1081787, 1147571, 2292257, 2292638, 2309592, 2310543, 2311675, 2321785, 2326532, 2326533, 2338605, 2351850, 2356086, 2367429, 2383901, 2399948, 2419037。

Optimization Model for Residential Tiered Electricity Pricing Based on Peak-Valley TOU PriceL. Yang, K.T. Chen, Z.F. Tan School of Economics and Management North China Electric Power University Changping District, Beijing, ChinaB.B. ZhouLiaoning Province Power Company Ningbo Road, Shenyang, Liaoning, ChinaAbstract—In recent years, with the shortage of energy supply and tremendous of environmental pressure in China, the price of primary energy such as coal continues to rise, as well as the electricity price. However, the adjustable amplitude and frequency for residential electricity are lower than other industries. In order to construct resource conserving and environment friendly society, to gradually reduce the price of cross-subsidization and guide residents to save electricity, it is necessary to implement the residential tiered electricity pricing. This paper optimizes the model of residential tiered electricity pricing through the introduction of peak-valley TOU price, which lays the foundation for promotion of power system reform.Keywords-Residential electricity price; tiered electricity pricing; peak-valley TOU price; modelI.I NTRODUCTIONElectric power industry is an important basic industry of the national economy, and is also an important part of social public utilities. After years of efforts, China’s power system reform has achieved initial results. The price reform is one of the core content of electricity market reform. Throughout the reform system, the tariff is an important economic lever. It not only can optimize the resources allocation, but also can adjust the relations among various interests [1, 2].Residential electricity consumption is an extremely important part for electricity sales market. The establishment of residential electricity price involves thousands of families, and is related to vital interests and living standards of the masses, as well as social efficiency, effectiveness and fairness [3]. Therefore, it is significant to establish a scientific and reasonable price system for the residents. This paper introduces peak-valley TOU price to design and optimize the model of residential tiered electricity pricing, and then obtains the optimization model for operation cost of the system.II.R ESEARCH S TATUS FOR R ESIDENTIAL T IEREDE LECTRICITY P RICINGIn China, the tiered pricing should be described to be the increasing price by ladder. It is refers to replace the current residential price of a single form with the approach of segmented pricing according to residential consumption, in this pricing method, the price of electricity increases gradually in a stepwise form along with the growth of electricity consumption [4,5]. The tiered electricity price divide power level of residential consumption in accordance with three stages of “meet the basic demand”, ”meet the normal and reasonable demand” and ”meet the high quality electricity demand”. Then, the corresponding price is set up. The tariff implements principle of stepping increment [6].Although the existing tiered electricity pricing system has been promoted and has achieved the electricity saving benefit, this system has not considered the distribution of load and lacked reaction for electricity value in peak or valley period. Actually, electric power often needs more standby unit for power generation to meet the residents' consumption in the peak period, so the generation cost is high. Additionally, the load demand is low in valley period, so many units have not been called and utilization efficiency of units is not high. If we can guide residents to use electricity in the valley time, the reducing peak and filling valley will be achieved, and the utilization efficiency of units will be improved [7, 8]. Therefore, this paper introduces the peak-valley TOU price in residential tiered electricity price, and then optimizes the sub-file number, sub-file power and block-tariff.III.D ESIGN M ODEL FOR R ESIDENTIAL T IEREDE LECTRICITY P RICINGTo guide residents using electricity rationally, tiered electricity pricing has been introduced. The upper tiered electricity pricing optimization model is constructed to differ from the traditional concept, targeting as the minimal electricity cost for residents. Optimization cycle is 24h in this paper in order to combine the tiered electricity pricing with the peak-valley price. Specific goal function is as follows.()()1,1,2,2,21,2,21,2,11minT It t i t i t i t i t i t i tt iC P Q P Q Q P Q Q−++===×+×−+×−∑∑(1)In this formula,1,tP refers to electricity price for thefirst-file resident at t time.1,tQ is electricity consumption for the first-file resident at t time.,i tP refers to electricity price for i-file resident at t time.,i tQ is electricity consumption for i-file resident at t time.Introducing tiered electricity price should consider the psychological strategy for residents. For the goal of reducing electricity costs and consumption, electricity costs of some users would rise after implementing tiered electricity price. In order to facilitate the expression, the comparison function()12,F x xis defined.()121211,,0,x xF x xx x>=≤(2)(1) Electricity cost constraintInternational Conference on Computer Information Systems and Industrial Applications (CISIA 2015)Electricity costs of some users should remain unchanged after implementing tiered electricity price. If total resident electricity is m, then:()11,Ii ii F C C Iλ=≥∑ (3)In this formula, i C and iC are electricity costs before and after introducing tiered electricity price for the resident i. 1λ is proportional coefficient which refers to the proportion of residents whose electricity cost is not higher than the cost before tiered electricity price.i i i C P Q =× (4)1Jiij ijj C P Q ==×∑ (5) In formulas above, i P and iQ refer to electricity cost and consumption before introducing tiered electricity price for the resident i. ij P and ij Q refer to electricity cost and consumption after introducing tiered electricity price for the resident i in j-file. At the same time, to ensure the implementation of tiered electricity price, electricity consumption for a certain percentage of residents should cross at least two intervals after implementing tiered electricity price, then: ()112,Ij i F Q Q I λ=≥∑ (6) In this formula, 2λis the proportion of residents whoseelectricity consumptions have crossed at least two intervals. To reduce residential reaction degree after introducingtiered electricity price, the basic electricity demand of residents should be limited, it is as follows.1i Q Q≥ (7)Q is the basic electricity demands of residents.After determining the number of sub-grade, the block-tariff *ij P and block-electricity *ij Q can be obtained by solving the model above.IV.D ESIGN M ODEL FOR P EAK -V ALLEY TOU P RICEFurther, TOU price is introduced to reflect the time valueof electricity. The main psychological factor of residents to respond TOU price is to minimize electricity costs. Therefore, residential peak-valley TOU price model is constructed as the goal of minimal electricity costs. The target function is as follows:11min T Iit itt i C P Q ===∑∑ (8)In this formula, it P and it Q are tariff and electricity consumption for resident i at t moment after introducing TOU price.In order to ensure the load shifting effect of peak-valleyTOU price, a certain proportion of residential electricity peak-valley difference should be decreased, the specific constraint is that:()()max min max min 13,mi i i i i F Q Q Q Q Iλ= −−≥∑ (9)In this formula, max iQand max iQare maximum load forresidents in peak period before and after TOU price. min iQ and min i Q are minimum load for residents in valley period beforeand after TOU price. 3λ is residential proportion whoproduce the load shifting effect./1p p i i P P < (10)/1v v i i PP ≥ (11) In the formulas, p i P and p iP are tariff for resident i in peak period before and after using TOU price. v i P and v iP are tariff for resident i in valley period before and after using TOU price.Considering different price characteristics of timesharingand multi-level in peak-valley step price, a concept of elastic coefficient with peak-valley and multi-level has put forward. For each resident, the tiered-TOU price in each levelcorresponds to an elasticity coefficient, so electricity consumption of the resident in each level is determined by peak-valley elasticity of multi-level, which can be described as follows.()()()()////p p p p i i ip i i v v v v i i iv i i Q Q e Q Q Q Q e Q Q ∆=∆∆=∆(12)In this formula, ijp Q and ijv Q are electricity for resident i at j level in peak and valley periods when implementing a single price (ijp Q =ijvQ ). After introducing peak-valley step price, change amounts of electricity for resident i at j level in peakand valley periods are that: ijp ijp ijp Q Q Q ∆=−，ijv ijv ijv Q Q Q ∆=− . e ij is the demand elasticity for resident i at j level.Solving the model above, we can get optimal tariffs and electricity consumptions in peak and valley periods. Combined with the tiered price in section 2, the residential peak-valley step price can be formed.V. O PTIMIZATION M ODEL FOR P EAK -V ALLEY S TEP P RICE The implementation of residential peak-valley step price will change electrical behavior and transfer using times of residents, which will directly affect the operation cost of the system. In order to ensure the stable operation of the system, the basic benefits of power system should be guaranteed. Therefore, an optimization model for peak-valley step price is constructed as the goal of maximizing system operation efficiency. The goal function is as follows.()111111max TTm m m m t t it st it it st it t t i i i i R R C c Q C c Q π====== =−=−−−∑∑∑∑∑∑ (13) In this formula, t R and tR are system profits before and after implementing peak-valley step price. st c refers to power purchase price for power company.In order to optimize the system operation, system power efficiency should be promoted when reducing residential electricity cost. Specific constraints are as follows.Firstly, changes of electricity sales profits for power companies are in a certain range. That is:(14)Secondly, electricity sales profits should be higher than system operating costs C op after implementing peak-valley step price:(15)Lastly, change magnitude of the total average price should be in a certain range:1121Ii i Iii C P P Q ξξ==−≤≤−∑∑ (16)In this formula, ξ1and ξ2 are constants.Although the objective function is to reduce power consumption, electricity should satisfy the following formula after the introduction of peak-valley step price in order to improve utilization rates of existing power generating equipment.1Igen i geni kQ Q Q =<<∑ (17)In this formula, Q gen refers to generation capacity. k （0<k <1）is a scaling factor to indicate a minimum utilization of power generation equipment.The characteristic of peak-valley step price determines the following relationship among tariffs:max1max10,1,2,,10,1,2,,1p p i i i v v i i i P P P i N P P P i N ++ <<≤=− <<≤=− (18)In this formula, the upper limit of tariff P is decided byelectricity consumption level, policy, economy and other factors, which results in some differences.VI.C ONCLUSIONSIn order to reflect the time value of electricity, this paper introduces peak-valley TOU price under the traditional residential tiered electricity price model, so peak-valley TOU price model is constructed. While the promotion of residential peak-valley TOU price directly influences the system operation cost, in order to ensure stable operation of thesystem, this paper constructs the optimization model forpeak-valley step price with the goal of maximizing system operation benefit.Optimization for residential tiered electricity price plays an important role in promoting the system to consume the random power resources. The final optimized model can promote random power resources into the grid, so the power output of thermal power unit will be reduced, as well as the cost of power generation system. Finally, the significant benefits would be created in economy and environment.R EFERENCES[1] Wu Jianhong. Pricing Model of Tiered Electricity Pricing for Residentsand Policy Simulation Based on Social Equilibrium [D]. North China Electric Power University, 2013, 4-7.[2] National Development and Reform Commission improves residentialtiered electricity price. National Energy Administration, /2013-12/26/c_132998400.htm[3] Wang Ruichun. Analysis of different pricing scheme under the guidanceof residential tiered electricity price [J]. Water Resources and Power, 2013, 31(1), 215-218.[4] Chen Minjun, Ci Xiangyang, Ping Zongfei. Residents Ladder ElectricityPrice Effect Evaluation Based on Analytic Hierarchy Process [J]. Science Technology and Industry, 2014, 14(9), 88-91.[5] Li Gen. Implementation Status and Improvement Strategy of China'sResidents Ladder Price[J]. Information and Communication, 2013, (2), 271-272.[6] ZENG Ming, LI Na, LIU Chao. Energy-Saving Assessment of TieredPricing Program for Residential Electricity Based on Utility Function [J]. East China Electric Power, 2011, 39 (8), 1215-1219.[7] HUANG Hai-tao. Design Model for Residential Block ElectricityPricing Structure and Level [J]. East China Electric Power, 2012, 40 (5), 0721-0727.[8] ZENG Ming. Assessment of Tiered Pricing Program for ResidentialElectricity Based on the Multiple Indexes [J]. East China Electric Power, 2012, 40 (3), 0359-0362.5R RRλ−≤ 1mi opi C C =≥∑。

基于双重注意力GRU与相似修正的光伏功率预测
何威;苏中元;史金林;吴炎琳;马昌流;王军
【期刊名称】《太阳能学报》
【年(卷),期】2024(45)3
【摘要】提出一种基于双重注意力机制GRU网络(dual-attention-GRU)及相似序列修正的光伏功率预测模型。

在Encoder-Decoder框架的基础上引入特征注意力以及时间注意力,能有效解决GRU网络对于输入特征及时间序列存在注意力分散的问题;采用相似功率序列的未来功率值对DA-GRU预测结果进行修正,能进一步改进预测结果。

算例采用DKASC数据进行验证,对比模型在不同预测步长下的表现,结果表明:相比于其他传统模型,DA-GRU在不同评价指标下具有最佳的预测表现,且相似序列修正方法能进一步提高其预测精度。

【总页数】8页(P480-487)
【作者】何威;苏中元;史金林;吴炎琳;马昌流;王军
【作者单位】东南大学能源与环境学院;浙江华东院控股有限公司;天长市中电建大桥新能源有限公司
【正文语种】中文
【中图分类】TM615;TP183
【相关文献】
1.基于综合相似日和功率相关性的光伏电站预测功率修正
2.基于特征挖掘的GRU-A光伏发电功率预测
3.基于PSO-GRU的光伏电站短期功率预测研究
4.基于K-
means分层聚类的TCN-GRU和LSTM动态组合光伏短期功率预测5.基于GRU-BLS的超短期光伏发电功率预测
因版权原因，仅展示原文概要，查看原文内容请购买。

TREE 3: POWER MANAGEMENT 2Supervisors & Voltage Detectors Unique Strengths (So What)Broad Portfolio(It's likely we have your part)Small Packages: SOT-23 and SC-70 (Saves space)Industrial Standard Crosses (Replace high priced and poor delivery suppliers)Battery Management Unique Strengths (So What)Wide variety of charging solutions for Li-Ion batteries(We have the solution for you)Small SOT-23, MSOP, DFN and QFN packages (Saves space)DC-DC Converter (So What)Low-voltage operation (Saves Power)PFM/PWM Auto switch mode (PFM at low loads reduces current, saves power)Small SOT-23 packaging (Saves space)Step-down, Step-up (Efficiently increase or decrease voltage) Charge Pumps (So What)Low-voltage operation (Battery operation)Small SOT-23 packaging (Saves space)Step-down, Step-up (Efficiently increase or decrease voltage)Doubling & Inverting (Meets V OUT needs) Low-Frequency capable (Reduces EMI)Low-Current Operation (Saves power)LDO Unique Strengths (So What)Hundreds of voltages, currents, packages (We have a match for the need)0.5% V OUT accuracy (Fills precision need)Up to 1.5A output current(Able to power high load applications)Op Amp Unique Strengths (So What)Low current versus GBWP (Saves power)TC and MCP6XXX devices RR-I/O (Expands usable voltage range)MCP604X 1.4V operation(Two alkaline cells 90% used =1.8V)MCP644X, 450 nA operation (Use the batteries even longer)Comparators Unique Strengths (So What)Low current versus propagation delay (Saves power)Integrated Features (Saves space)1.8V and 1.4V operation (That stuff about the batteries)Programmable Gain Amplifier Unique Strengths (So What)MUX inputWide bandwidth (2 to 12 MHz) (Reduces demand on MCU I/O)System control of gain(Changes easier through software configurable hardware)TREE 5: LINEARTemperature Sensor Unique Strengths (So What)Wide variety of solutions: logic, voltage and digital output products(Multiple sensor needs met)Small packages (Saves space)Low operating current(Saves power, smaller supply)Field or factory programmable (Low cost vs. flexibility)Programmable hysteresis (Stop system cycling)Multi-drop capability (Great for large systems)Beta compensation (Compatible with processor substrate diodes)Resistance error correction (Compensates for measurement error from long PCB traces)Fan Controllers Unique Strengths (So What)Closed loop fan control (Adjust to meet target speed even on aging fans)Integrated temperature sensing (Consolidate thermal management)Multiple temperature measurements drive one fan (Consolidate thermal management)Built-in ramp rate control and spin up alogorithm (Quick time to market, lower acoustic noise)Ability to detect/predict failure of less expensive 2-wire fans (Saves system cost)Unique solutions for extending fan life and reducing acoustic noise(Less power, nuisance and long fan life)TREE 6: MIXED-SIGNALADC Unique Strengths (So What)Low current at max sampling rate (Saves power, system cost)Small SOT-23 and MSOP packages (Saves space)Up to 24-bit resolution(Ideal for precision sensitive designs)Differential & single ended inputs (Able to cover various design needs)Up to 6 ADC per device(Save board space, system cost)DAC Unique Strengths (So What)Low Supply Current (Saves power)Low DNL & INL (Better accuracy)Extended Temperature Range(Suitable for wide temperature applications)Digital Potentiometers Unique Strengths (So What)64/256 tap (6-bit to 8-bit resolution)(Sufficient resolution for most applications)Non-volatile Memory(Remembers last wiper setting on power up)WiperLock™ Technology(Locks NV memory setting-better than OTP)Small SOT-23 and 2 × 3 DFN packages (Saves space)Low CostMOSFET Drivers (So What)4.5V up to 30V Supply voltages (Fills many application needs)Up to 12A Peak output current(Able to meet demanding design needs)Outstanding robustness and latch-upi mmunity (Ours work when the others burn up)Low-FOM MOSFETs(Support high-efficency applications)TREE 4: POWER MANAGEMENT 3LIN Unique Strengths(So What)Compliant with LIN Bus Specs 1.3, 2.0, 2.1 andSAE J2602 (Allows for reliable interoperability)High EMI Low EME (Meets OEM requirements)On-board V REG available(Saves space, allows for MCU V CC flexibility)CAN Unique Strengths(So What)Simple SPI CAN controller is an easy way toadd CAN Ports (Short design cycles)High speed transceiver meets ISO-11898 (Drop inreplacement for industry standard transceivers)Low-cost, easy-to-use development tools(Tools easy to buy/use, quick design)I/O Expanders Unique Strengths(So What)Configurable inputs (interrupt configuration flexibility)Interrupt on pin change, or change fromregister default (interrupt source flexibility)Can disable automatic address incrementingwhen accessing the device(allows continual access to the port)The 16-bit devices can operate in 8-bit or 16-bitmode (easy to interface to 8-bit or 16-bit MCUs)IrDA Unique Strengths (So What)IrDA protocol handler embedded on chip(Complex design issue solved)Low cost developer's kit available to assistInfrared design-in (Quick design cycle)Small, cost-effective way of replacing serial links(No more wires)Enables system to wirelessly communicatewith PDA (Wireless connectivity solution) TREE 8: INTERFACEAnalog & InterfaceQuestion TreesAnalog & Interface Development ToolsDemonstration Boards, Evaluation Kits and AccessoriesAnalog & Interface LiteratureADM00313EV: MCP73830L 2 × 2 TDFN Evaluation BoardADM00352: MCP16301 High Voltage Buck Converter 600 mA Demonstration BoardADM00360: MCP16301 High Voltage Buck Coverter 300 mA D2PAK Demonstration BoardADM00427: MCP16323 Evaluation Board (Supports MCP16321 and MCP16322)ARD00386: MCP1640 12V/50 mA Two Cells Input Boost Converter Reference DesignMCP1252DM-BKLT: MCP1252 Charge Pump Backlight Demonstration BoardMCP1256/7/8/9EV: MCP1256/7/8/9 Charge Pump Evaluation BoardMCP1630RD-LIC1: MCP1630 Li-Ion Multi-Bay Battery Charger Reference DesignMCP1630DM-NMC1: MCP1630 NiMH Battery Charger Demonstration BoardMCP1640EV-SBC: MCP1640 Sync Boost Converter Evaluation BoardMCP1640RD-4ABC: MCP1640 Single Quad-A Battery Boost Converter Reference DesignMCP1650DM-LED1: MCP165X 3W White LED Demonstration BoardMCP1726EV: MCP1726 LDO Evaluation BoardMCP73831EV: MCP73831 Evaluation KitMCP7383XEV: MCP73837/8 AC/USB Dual Input Battery Charger Evaluation BoardMCP7383XRD-PPM: MCP7383X Li-Ion System Power Path Management Reference DesignMCP7384XEV: MCP7384X Li-Ion Battery Chager Evaluation BoardMCP73871EV: MCP73871 Load Sharing Li-Ion Battery Charger Evaluation BoardTC1016/17EV: TC1016/17 LDO Evaluation BoardVSUPEV: SOT-23-3 Voltage Supervisor Evaluation BoardPowerManagementThermalManagementMCP9700DM-PCTL: MCP9700 Thermal Sensor PICtail Demonstration BoardMCP9800DM-PCTL: MCP9800 Thermal Sensor PICtail Demonstration BoardTC72DM-PICTL: TC72 Digital Temperature Sensor PICtail Demonstration BoardTC74DEMO: TC74 Serial Daughter Thermal Sensor Demonstration BoardTC1047ADM-PCTL: TC1047A Temperature-to-Voltage Converter PICtail™ Demonstration BoardSerial GPIODM-KPLCD: GPIO Expander Keypad and LCD Demonstration BoardMCP23X17: MCP23X17 16-bit GPIO Expander Evaluation BoardInterface MCP2515DM-BM: MCP2515 CAN Bus Monitor Demonstration BoardMCP2515DM-PTPLS: MCP2515 PICtail™ Plus Daughter BoardMCP2515DM-PCTL: MCP2515 CAN Controller PICtail Demonstration BoardMCP215XDM: MCP215X/40 Data Logger Demonstration BoardMCP2140DM-TMPSNS: MCP2140 IrDA® Wireless Temp Demonstration BoardLinear ADM00375: MCP6H04 Evaluation BoardARD00354: MCP6N11 Wheatstone Bridge Reference DesignMCP651EV-VOS: MCP651 Input Offset Evaluation BoardMCP661DM-LD: MCP661 Line Driver Demo BoardMCP6S22DM-PCTL: MCP6S22 PGA PICtail Demonstration BoardMCP6S2XEV: MCP6S2X PGA Evaluation BoardMCP6SX2DM-PCTLPD: MCP6SX2 PGA Photodiode PICtail Demonstration BoardMCP6SX2DM-PCTLTH: MCP6SX2-PGA Thermistor PICtail Demonstration BoardMCP6V01RD-TCPL: MCP6V01 Thermocouple Auto-Zero Ref DesignMCP6XXXDM-FLTR: Active Filter Demo BoardPIC16F690DM-PCTLHS: Humidity Sensor PICtail Demonstration BoardMixed-Signal MCP3221 DM-PCTL: MCP3221 12-bit A/D PICtail Demonstration BoardMCP3421DM-BFG: MCP3421 Battery Fuel Gauge Demonstration BoardMCP3551DM-PCTL: MCP3551 PICtail Demonstration BoardMCP355XDM-TAS: MCP355X Tiny Application Sensor Demonstration BoardMCP355XDV-MS1: MCP3551 Sensor Demonstration BoardMCP402XEV: MCP402X Digital Potentiometer Evaluation BoardMCP4725EV: MCP4725, 12-bit Non-Volatile DAC Evaluation Board (Preferred One)MCP4725DM-PTPLS: MCP4725, 12-bit Non-Volatile DAC PICtail Demonstration BoardADM00398: MCP3911 ADC Evaluation Board for 16-bit MicrocontrollersCorporate Microchip Product Line Card - DS00890Brochures Analog and Interface Product Selector Guide - DS21060Low Cost Development Tools Solutions Guide - DS51560Analog and Interface Guide (Volume 1) - DS00924Analog and Interface Guide (Volume 2) - DS21975Cards Analog Highlights Card - DS21972Microchip Op Amp Discovery Card - DS21947Analog & Interface Question Trees - DS21728Mirochip SAR and Delta-Sigma ACD Discovery Card - DS22101Software Tools MAPS - Microchip Advanced Product SelectorAnalog & Interface Treelink Products PresentationDesign Guides Analog-to-Digital Converter Design Guide - DS21841Digital Potentiometers Design Guide - DS22017Programmable Gain Amplifiers (PGAs), Operational Amplifiersand Comparators Design Guide - DS21861Interface Products Design Guide - DS21883Signal Chain Design Guide - DS21825Power Solutions Design Guide - DS21913Temperature Sensor Design Guide - DS21895Voltage Supervisors Design Guide - DS51548DS21728JPowerManagementLDO & SwitchingRegulatorsCharge PumpDC/DC ConvertersPower MOSFETDriversPWM ControllersSystem SupervisorsVoltage DetectorsVoltage ReferencesLi-Ion/Li-PolymerBattery ChargersUSB Port PowerControllersMixed-SignalA/D ConverterFamiliesDigitalPotentiometersD/A ConvertersV/F and F/VConvertersEnergyMeasurement ICsCurrent/DC PowerMeasurement ICsInterfaceCAN PeripheralsInfraredPeripheralsLIN TransceiversSerial PeripheralsEthernet ControllersUSB PeripheralLinearOp AmpsInstrumentationAmpsProgrammableGain AmplifiersComparatorsSafety & SecurityPhotoelectricSmoke DetectorsIonization SmokeDetectorsIonization SmokeDetector Front EndsPiezoelectricHorn DriversThermalManagementTemperatureSensorsFan Control& HarwareManagementMotor DriveStepper and DC3Ф BrushlessDC Motor DriverTREE 7: MOTOR DRIVE Stepper Unique Strenghts(So What)Industrial standard footprint(Footprint compatible to industrial leaders)Perfect PIC® MCU companion chip(Solid field support)Micro-stepping ready(Enhanced performance)Integration protections(Simplify software development)3-Phase BLDC Unique Strengths(So What)Full-wave sinusoidal(Quiet operation, low mechanical vibration)Sensorless operation (Minimum externalcomponents, no software required)Thin form factor(Fits space concerned applications)Information subject to change. The Microchip name and logo, the Microchip logo, dsPIC, PIC are registered trademarks and MiWi, PICtail and ZENA are trademarks ofMicrochip Technology Incorporated in the U.S.A. and other countries. All other trademarks mentioned herein are property of their respective companies.© 2012, Microchip Technology Incorporated. All Rights Reserved.。

第１３卷㊀第１１期Ｖｏｌ．１３Ｎｏ．１１㊀㊀智㊀能㊀计㊀算㊀机㊀与㊀应㊀用ＩｎｔｅｌｌｉｇｅｎｔＣｏｍｐｕｔｅｒａｎｄＡｐｐｌｉｃａｔｉｏｎｓ㊀㊀２０２３年１１月㊀Ｎｏｖ．２０２３㊀㊀㊀㊀㊀㊀文章编号：２０９５－２１６３（２０２３）１１－０１６６－０６中图分类号：ＴＰ３９１文献标志码：Ａ融合Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器的ＴＣＮ－ＳＡ－ＢｉＧＲＵ风电功率预测秦小晖，樊重俊，付峻宇（上海理工大学管理学院，上海２０００９３）摘㊀要：陆上风力发电作为主流清洁能源发电方式之一，预测其发电功率也是目前研究热点问题㊂本文提出融合Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器与基于自注意力机制的ＴＣＮ－ＢｉＧＲＵ风电功率预测模型㊂利用Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器对风电功率及相关特征数据进行降噪，随后将数据输入进由ＴＣＮ时域卷积神经网络㊁自注意力机制模块㊁双向门控循环单元网络所搭建的ＴＣＮ－ＳＡ－ＢｉＧＲＵ模型中，这些模块能够更深㊁更快挖掘数据特征㊂最终预测结果显示，融合了Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器的模型能够有效对数据降噪，并且相较于传统单一神经网络等模型，本模型的预测性能更高㊂关键词：风电功率预测；Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器；时域卷积神经网络（ＴＣＮ）；自注意力机制ＷｉｎｄｐｏｗｅｒｆｏｒｅｃａｓｔｉｎｇｉｎｃｏｒｐｏｒａｔｉｎｇＳａｖｉｔｚｋｙ－Ｇｏｌａｙ－ＴＣＮ－ＳＡ－ＢｉＧＲＵＱＩＮＸｉａｏｈｕｉ，ＦＡＮＣｈｏｎｇｊｕｎ，ＦＵＪｕｎｙｕ（ＢｕｓｉｎｅｓｓＳｃｈｏｏｌ，ＵｎｉｖｅｒｓｉｔｙｏｆＳｈａｎｇｈａｉｆｏｒＳｃｉｅｎｃｅａｎｄＴｅｃｈｎｏｌｏｇｙ，Ｓｈａｎｇｈａｉ２０００９３，Ｃｈｉｎａ）Ａｂｓｔｒａｃｔ：Ａｓｏｎｅｏｆｔｈｅｍａｉｎｓｔｒｅａｍｃｌｅａｎｅｎｅｒｇｙｇｅｎｅｒａｔｉｏｎｍｅｔｈｏｄｓ，ｐｒｅｄｉｃｔｉｎｇｔｈｅｐｏｗｅｒｇｅｎｅｒａｔｅｄｂｙｏｎｓｈｏｒｅｗｉｎｄｐｏｗｅｒｉｓａｌｓｏａｈｏｔｒｅｓｅａｒｃｈｐｒｏｂｌｅｍａｔｐｒｅｓｅｎｔ．ＴｈｉｓｐａｐｅｒｐｒｏｐｏｓｅｓａＴＣＮ－ＢｉＧＲＵｗｉｎｄｐｏｗｅｒｐｒｅｄｉｃｔｉｏｎｍｏｄｅｌｔｈａｔｃｏｍｂｉｎｅｓＳａｖｉｔｚｋｙ－Ｇｏｌａｙｆｉｌｔｅｒａｎｄｓｅｌｆ－ａｔｔｅｎｔｉｖｅｍｅｃｈａｎｉｓｍ．ＴｈｅＳａｖｉｔｚｋｙ－Ｇｏｌａｙｆｉｌｔｅｒｉｓｕｓｅｄｆｏｒｎｏｉｓｅｒｅｄｕｃｔｉｏｎｏｆｗｉｎｄｐｏｗｅｒａｎｄｒｅｌａｔｅｄｆｅａｔｕｒｅｓ，ａｎｄｔｈｅｄａｔａａｒｅｔｈｅｎｆｅｄｉｎｔｏａＴＣＮ－ＳＡ－ＢｉＧＲＵｍｏｄｅｌｂｕｉｌｔｂｙＴＣＮ，ｓｅｌｆ－ａｔｔｅｎｔｉｖｅｍｅｃｈａｎｉｓｍ，ａｎｄＢｉＧＲＵ，ｗｈｉｃｈａｒｅｃａｐａｂｌｅｏｆｍｉｎｉｎｇｔｈｅｄａｔａｆｅａｔｕｒｅｓｍｏｒｅｄｅｅｐｌｙａｎｄｑｕｉｃｋｌｙ．ＴｈｅｆｉｎａｌｐｒｅｄｉｃｔｉｏｎｒｅｓｕｌｔｓｓｈｏｗｔｈａｔｔｈｅｍｏｄｅｌｉｎｃｏｒｐｏｒａｔｉｎｇｔｈｅＳａｖｉｔｚｋｙ－Ｇｏｌａｙｆｉｌｔｅｒｉｓｅｆｆｅｃｔｉｖｅｉｎｎｏｉｓｅｒｅｄｕｃｔｉｏｎｏｆｔｈｅｄａｔａａｎｄｈａｓｈｉｇｈｅｒｐｒｅｄｉｃｔｉｏｎｐｅｒｆｏｒｍａｎｃｅｔｈａｎｔｒａｄｉｔｉｏｎａｌｍｏｄｅｌｓｓｕｃｈａｓｓｉｎｇｌｅｎｅｕｒａｌｎｅｔｗｏｒｋｓ．Ｋｅｙｗｏｒｄｓ：ｗｉｎｄｐｏｗｅｒｆｏｒｅｃａｓｔｉｎｇ：Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙｆｉｌｔｅｒ：ＴｅｍｐｏｒａｌＣｏｎｖｏｌｕｔｉｏｎａｌＮｅｔｗｏｒｋ：ｓｅｌｆ－ａｔｔｅｎｔｉｏｎ基金项目：２０２０教育部哲学社会科学重大课题攻关项目，２０２０－２０２３（２０ＪＺＤ０１０）㊂作者简介：秦小晖（１９９８－），男，硕士研究生，主要研究方向：人工智能理论及应用㊁数字经济㊂通讯作者：樊重俊（１９６３－），男，博士，教授，博士生导师，主要研究方向：数字经济㊁人工智能㊁电子商务㊂Ｅｍａｉｌ：ｆａｎ．ｃｈｏｎｇｊｕｎ＠１６３．ｃｏｍ收稿日期：２０２３－０５－３１０㊀引㊀言各国工业发展离不开资源消耗，也导致碳排放量数值上升，从而进一步引发温室效应㊁全球水平面上升等环境问题，所以随着技术的发展，风能㊁太阳能及潮汐能等清洁能源逐步取缔火力发电成为主要发电能源㊂由于风能具有可再生性㊁无污染性等特点，所以目前风力发电仍是国内主要发电方式之一，但研究表明风能短时间变化随机，其发电功率呈间歇性以及波动性，所以使用风力发电会产生发电功率不稳定的现象，而这一现象也会导致电网运行风险提高，造成事故［１］㊂因此准确预测风力发电功率能够提高风力发电的稳定性㊁减少不必要损失，从而提升经济效益㊂随着预测技术的不断发展，国内外学者在研究过程中所使用的预测方法也从传统的统计模型预测逐渐演变成包含智能算法㊁机器学习以及深度学习的组合预测模型［２］㊂早期的风电功率预测方法主要有：持续预测法㊁卡尔曼滤波法㊁灰色预测法以及随机时间序列法（例如ＡＲ㊁ＭＡ以及ＡＲＩＭＡ等模型），但是存在预测时间范围小㊁数据收集及处理难度较大以及预测结果不稳定等问题［３］，所以后续也产生了改进相关时间序列模型以及将滤波器与时序模型组合的预测方法㊂随着机器学习方法的面世，诸如支持向量机（ＳＶＭ）㊁决策树以及随机森林等方法陆续涌现［４］，其中以支持向量机（ＳＶＭ）的应用最为广泛，相较于传统的时序预测模型㊁马尔可夫等方法，ＳＶＭ在处理非线性等小样本时具有更好的适应能力，但是只采取ＳＶＭ单一模型进行预测的效果也并不好，所以在后续研究中，学者逐渐将ＳＶＭ与遗传算法相组合，利用算法对ＳＶＭ进行参数寻优以达到更好的预测效果［５］，例如粒子群算法㊁灰狼算法以及果蝇算法等等［６］，但这些算法多会存在收敛时间过长㊁出现局部最优解以及搜索步长设置繁复等问题，所以研究学者会对现有遗传算法进行优化后再对机器学习模型进行参数优化㊂目前，主流的风电功率预测方法仍然是以神经网络模型为主的组合模型，在ＬＳＮｅｔ神经网络㊁长短期记忆神经网络（ＬＳＴＭ）㊁门控循环神经网络（ＧＲＵ）等基础上针对数据特征添加算法模型与之组合㊂文献［７］经过研究发现ＧＲＵ网络相较于ＬＳＴＭ网络在风电功率预测上的作用更加高效且结构简单㊂而文献［８］建立融合注意力机制的Ｂｉ－ＧＲＵ模型对数据中所蕴含的信息进行深度挖掘，提高特征提取的效率与质量并进行预测㊂文献［９］则利用ＴＣＮ神经网络与ＬＳＴＭ神经网络进行组合对时序数据加以预测，ＴＣＮ网络可以较好地提取间隔较长和非连续时序数据的特征信息㊂本文在已有研究的基础上，搭建融合Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器的ＴＣＮ－ＢｉＧＲＵ模型㊂为了有效降低拥有噪声的风电功率数据，选取Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器对风电功率相关特征数据进行降噪，并将预处理后的数据输入进ＴＣＮ－ＳＡ－ＢｉＧＲＵ模型中，通过ＴＣＮ网络对相关数据进行特征提取，最后将有效信息输入进ＢｉＧＲＵ模型中进行预测㊂本文选取西班牙某地区的风电功率数据以及其他历史相关数据以验证该组合模型的预测有效性㊂１㊀模型原理介绍１．１㊀时间卷积神经网络时间卷积神经网络（ＴｅｍｐｏｒａｌＣｏｎｖｏｌｕｔｉｏｎａｌＮｅｔｗｏｒｋ，ＴＣＮ）最早于２０１６年由Ｌｅａ等学者提出，随后由Ｂａｉ等学者在文章中正式提出㊂ＴＣＮ在卷积神经网络（ＣＮＮ）的基础上做出改进，增加因果卷积以及空洞卷积㊂ＴＣＮ中的因果卷积为单向结构，能保证提取信息特征时的因果性，即输入序列ｘ１，ｘ２， ，ｘｔ预测ｙ１，ｙ２， ，ｙｔ时，如果需要预测ｔ时刻的值ｙｔ时只能利用已存在并观察到的ｘ１，ｘ２， ，ｘｔ－１，而不能使用ｘｔ＋１，ｘｔ＋２， ，所以ＴＣＮ中的因果卷积可以保证在时序预测过程中并不会受到未来信息因素的干扰，但是单纯的因果卷积受制于卷积内核大小，在抓取依赖信息特征时需要堆叠过多层数，所以ＴＣＮ中又采用膨胀卷积增加网络中的感受野，对上一层输入信息增大采样范围以使得ＴＣＮ能够提取间隔较长以及非连续性数据的时序特征㊁即在更少层数的情况下能够抓取更多信息特征，同时ＴＣＮ在各网络层之间采用残差进行连接，防止梯度消失爆炸问题的产生［１０］㊂膨胀因果卷积结构如图１所示，残差模块图如图２所示㊂O u t p u td=4H i d d e nd=2H i d d e nd=1I n p u ty0y1y2yt-2y t-1y tx0x1x2x t-2x t-1x t图１㊀膨胀因果卷积图Ｆｉｇ．１㊀ＳｔｒｕｃｔｕｒｅｓｏｆＣａｕｓａｌａｎｄＤｉｌａｔｅｄＣｏｎｖｏｌｕｔｉｏｎｓI n p u tO u p u t1?1卷积（o p t i o n a l）d r o p o u tR e L u激活函数批量标准化膨胀因果卷积d r o p o u tR e L u激活函数批量标准化膨胀因果卷积残差模块(k,d)图２㊀残差模块图Ｆｉｇ．２㊀Ｒｅｓｉｄｕａｌｍｏｄｕｌｅ１．２㊀自注意力机制模块由于在神经网络结构中所接收到的输入向量大小不一，同时这些向量之间所蕴含的信息可能存在一定关系，所以在模型训练中因为忽略了这些向量之间的关系而导致模型训练结果很差，而自注意力机制可以针对全连接神经网络中的相关输入变量较好地建立起数据相关性，从而提高了特征有效提取能力㊂其公式如下：Ｑ＝ＸＷＱ（１）Ｋ＝ＸＷＫ（２）７６１第１１期秦小晖，等：融合Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器的ＴＣＮ－ＳＡ－ＢｉＧＲＵ风电功率预测Ｖ＝ＸＷＶ（３）ＡＱ，Ｋ，Ｖ()＝ｓｏｆｔｍａｘ（ＱＫＴｄ）Ｖ（４）㊀㊀其中，Ａ为Ａｔｔｅｎｔｉｏｎ，即注意力权重；Ｑ为查询向量矩阵（ＱｕｅｒｙＶｅｃｔｏｒ）；Ｋ为键向量矩阵（ＫｅｙＶｅｃｔｏｒ）；Ｖ为值向量矩阵（ＶａｌｕｅＶｅｃｔｏｒ）㊂１．３㊀双向门控循环单元门控循环单元（ＧａｔｅｄＲｅｃｕｒｒｅｎｔＵｎｉｔ，ＧＲＵ）是循环神经网络（ＲＮＮ）的一种，由学者Ｃｈｏ等学者在２０１４年提出㊂ＧＲＵ与长短期记忆神经网络（ＬＳＴＭ）相类似，是在ＬＳＴＭ基础上提出的衍生网络结构［１１］，缓解了ＲＮＮ存在的梯度消失㊁且无法捕捉长期信息关联性等问题，并且ＧＲＵ的结构比ＬＳＴＭ简单，具有计算效率高㊁网络参数较少等优点［１２］㊂ＧＲＵ结构如图３所示㊂由图３可知，ＧＲＵ设置了２个门㊂其中，更新门ｚｔ用于控制ｔ时刻的新旧输入信息保留程度；重置门ｒｔ用于控制ｔ－１时刻每个位置输入信息保留程度㊂具体公式如下：ｚｔ＝σＷｚ㊃ｈｔ－１，ｘｔ[]()（５）ｒｔ＝σＷｒ㊃ｈｔ－１，ｘｔ[]()（６）ｈｔ－＝ｔａｎｈ（Ｗ㊃ｒｔ∗ｈｔ－１，ｘｔ[]）（７）ｈｔ＝１－ｚｔ()∗ｈｔ－１＋ｚｔ∗ｈｔ－（８）㊀㊀其中，Ｗｚ㊁Ｗｒ与Ｗ分别表示更新门㊁重置门以及隐藏层的权重矩阵㊂1-h th t -1σσ~t a n hx tz tr th t图３㊀ＧＲＵ结构图Ｆｉｇ．３㊀ＧａｔｅｄＲｅｃｕｒｒｅｎｔＵｎｉｔ㊀㊀ＧＲＵ在进行预测时的顺序都是由前到后，这种单一方向的时序数据预测会导致具有长关联性数据的信息发生遗漏，所以在ＧＲＵ的基础上加入双向学习得到双向门控循环单元（Ｂｉ－ＧＲＵ），其结构如图４所示㊂㊀㊀由Ｂｉ－ＧＲＵ结构图可观察到，从正向来看，正向传导单元可以捕捉数据中的历史信息，从反向来看，反向传导单元可以捕捉数据中的未来信息，这种双向的结构可以实现全局信息捕捉，从而提高时序特征提取效率［１３］㊂更新后的公式如下：ｈｔ＝ＧＲＵｘｔ，ｈｔ－１()（９）ｈᶄｔ＝ＧＲＵｘｔ，ｈᶄｔ－１()（１０）Ｈｔ＝Ｗｈｔｈｔ＋Ｗｈᶄｔｈᶄｔ＋ｂｔ（１１）㊀㊀其中，Ｗ为ｔ时刻隐藏层状态；ｈ为ｔ时刻隐藏层权重；ｂｔ为ｔ时刻隐藏层状态偏置㊂W 4W 6W 4W 4W 6W 6W 5W 5W 5W 5W 3W 3W 3W 3W 1W 1W 1W 2W 2W 2正向反向图４㊀Ｂｉ－ＧＲＵ结构图Ｆｉｇ．４㊀ＢｉｄｉｒｅｃｔｉｏｎａｌＧａｔｅｄＲｅｃｕｒｒｅｎｔＵｎｉｔ２㊀融合Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器的ＴＣＮ－ＳＡ－ＢｉＧＲＵ风电功率预测模型２．１㊀Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器由于来自传感器的原始数据包含过多噪声会影响模型性能，所以很少被作为预测模型的输入数据㊂因此，在进行实验分析时，需要首先对数据进行降噪处理㊂在本研究中，采用Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器来去除原始数据的噪声㊂Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器由Ｓａｖｉｔｚｋｙ和Ｇｏｌａｙ在１９６４年提出，是一种基于时域局部多项式最小二乘拟合的滤波方法，常用于数据流平滑降噪㊂作为一种有限脉冲响应（ＧＩＲ）数字滤波器，通过卷积操作对原始时域信号进行平滑处理，而滤波后的数据在去除噪声的同时，保持了相同的信号结构㊂作为光谱预处理中的常用方法，主要是对一定长度内的数据点进行ｋ阶多项式拟合后得到拟合结果㊂Ｓ－Ｇ滤波是一种移动窗口的加权平均算法，但是其加权系数不是简单的常数窗口，而是通过在滑动窗口内对给定高阶多项式的最小二乘拟合得出，该算法的最大特点就是在滤除噪声的同时可以确保信号的形状㊁宽度不变㊂公式如下：ｓ∗ｊ＝ðｍｉ＝－ｍＣｉＳｊ＋１Ｎ（１２）㊀㊀其中，Ｓ表示原始信号；ｓ∗表示降噪后的信号；Ｃｉ为第ｉ次的降噪系数；Ｎ为（２ｍ＋１）组数据的滑动窗口宽距；ｊ为数据集中第ｊ个样本㊂当将滤波器应用于时间序列数据进行平滑处理时，必须确定滤波器中的２个参数㊂第一个参数Ｎ，是Ｓ－Ｇ滤波器８６１智㊀能㊀计㊀算㊀机㊀与㊀应㊀用㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀第１３卷㊀中的滑动窗口宽距，一般来说，当Ｎ越大时会产生更为平滑的结果，但同时会降低峰值的平整度；第二个参数ｋ，表示Ｓ－Ｇ滤波器中的拟合阶数，通常ｋ值会设置在２到４的范围中㊂ｋ值越小，会产生更平滑的结果，但可能会引入偏差；而ｋ值越高，虽然会减少滤波器的偏差，但可能会发生过拟合现象，产生更大的噪声，所以调整Ｎ与ｋ的大小是实现有效降噪以及信号平衡的关键所在㊂２．２㊀Ｓ－Ｇ－ＴＣＮ－ＳＡ－ＢｉＧＲＵ模型结构及预测步骤本模型主要分为７个模块：输入层㊁Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器降噪层㊁数据拼接层㊁时域卷积层（ＴＣＮ）㊁自注意力机制模块㊁双向门控循环单元以及输出层㊂模型结构如图５所示㊂对模型中的各预测步骤将给出阐释分述如下㊂图５㊀融合Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器的ＴＣＮ－ＳＡ－ＢｉＧＲＵ模型Ｆｉｇ．５㊀ＴＣＮ－ＳＡ－ＢｉＧＲＵｍｏｄｅｌｉｎｃｏｒｐｏｒａｔｉｎｇＳ－Ｇｆｉｌｔｅｒ㊀㊀（１）数据输入㊂对样本按照时间点位进行采样并进行数据预处理，将处理后的数据作为输入㊂这里的数据包括风电功率数据以及相关特征数据（例如温度㊁湿度㊁大气压强㊁风速㊁风力等级㊁天气情况以及实际价格）㊂（２）降噪㊂假若全部数据进入Ｓ－Ｇ滤波器进行降噪则会造成数据失真，所以选取部分数据进行降噪，根据样本设置Ｓ－Ｇ滤波器中的滑动窗口宽距Ｎ以及拟合阶数ｋ㊂（３）数据拼接及划分㊂将未降噪的数据以及降噪后的数据在此层进行拼接，并按照比例划分训练集以及测试集㊂（４）模型训练㊂将训练集输入进ＴＣＮ－ＳＡ－ＢｉＧＲＵ网络中进行训练，首先利用ＴＣＮ时域卷积网络层对时序数据进行快速特征提取，再利用自注意力机制模块对特征进行权重调整，接着进入ＢｉＧＲＵ网络中再次进行信息提取以更深层次地挖掘时序数据中蕴含的数据相关性，最后得到训练完成的ＴＣＮ－ＳＡ－ＢｉＧＲＵ㊂（５）预测及分析㊂将测试集输入进已训练完成的ＴＣＮ－ＳＡ－ＢｉＧＲＵ网络中进行风电功率预测，并设置对比分析实验以检测该组合模型的优越性㊂３㊀实验３．１㊀实验环境实验所使用计算机配置见表１㊂表１㊀实验环境Ｔａｂ．１㊀Ｅｘｐｅｒｉｍｅｎｔａｌｅｎｖｉｒｏｎｍｅｎｔ实验环境配置计算机系统Ｗｉｎｄｏｗｓ１１ＣＰＵＡＭＤＲｙｚｅｎ５５６００ＨｗｉｔｈＲａｄｅｏｎＧｒａｐｈｉｃｓ＠３．３０ＧＨｚ运行内存１６ＧＢ开发语言Ｐｙｔｈｏｎ３．８深度学习框架Ｐｙｔｏｒｃｈ１．１０．０３．２㊀数据选取本实验数据来自Ｋａｇｇｌｅ数据网站上西班牙某地区２０１５年１月１日００：００至２０１８年１２月３１日２３：５９的风电功率数据㊂每日共设２４个采样点，采样时间间隔为１ｈ，即每日数据集为２４维风电功率数据㊂并且，陆上风力发电功率与一些其他的因素息息相关，所以本实验数据集包含了数据所属地的温度最大值及最小值㊁湿度㊁大气压强㊁风速㊁风力等级㊁天气情况以及电力实际价格㊂３．３㊀数据预处理３．３．１㊀数据缺失值处理对数据进行缺失值处理，由于处理过程中未见大范围数据缺失㊁且数据采样间隔时间较短，故采取均值插补法对缺失数据进行填充㊂３．３．２㊀数据标准化处理由于在处理数据问题的时候，实验数据来自多个维度，例如本实验中温度㊁风力等因素会对风电功率产生影响，这一情况会产生以下问题：数据量纲不同导致数量级差别大；过大的数值会产生数值问题引发特征贡献度失衡㊂所以为了避免以上问题，实验中选择对风电功率㊁温度以及压强等特征进行数据标准化处理，标准化公式如下：ｘᶄ＝ｘ－ｘｍｉｎｘｍａｘ－ｘｍｉｎ（１３）９６１第１１期秦小晖，等：融合Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器的ＴＣＮ－ＳＡ－ＢｉＧＲＵ风电功率预测㊀㊀其中，ｘᶄ为重新调节后的数据向量，范围值为［０，１］，ｘｍａｘ与ｘｍｉｎ分别表示数据集中的最大值与最小值㊂对于离散型数据 天气质量状况 则进行ＬａｂｅｌＥｎｃｏｄｅｒ编码处理，例如： 晴天 为１， 雨雪天 为２，等等㊂３．４㊀实验评价指标本文模型评价指标选取决定系数（ｃｏｅｆｆｉｃｉｅｎｔｏｆｄｅｔｅｒｍｉｎａｔｉｏｎ，Ｒ２）㊁平均绝对误差（ｍｅａｎａｂｓｏｌｕｔｅｅｒｒｏｒ，ＭＡＥ）以及均方误差（ＭｅａｎＳｑｕａｒｅＥｒｒｏｒ，ＭＳＥ）㊂（１）决定系数（Ｒ２）㊂一般用来表示模型预测拟合程度㊂计算公式如下：Ｒ２＝１－ðｍｉ＝１（ｙｉ－ｙ＾ｉ）２ðｍｉ＝１（ｙｉ－ｙｉ）２（１４）㊀㊀（２）平均绝对误差（ＭＡＥ）㊂用来准确反映实际预测误差的大小㊂计算公式如下：ＭＡＥ＝１ｍðｍｉ＝１ｙｉ－ｙｉ（１５）㊀㊀（３）均方误差（ＭＳＥ）㊂用来衡量观测值同真值之间的偏差㊂计算公式如下：ＭＳＥ＝１ｍðｍｉ＝１（ｙｉ－ｙ＾ｉ）２（１６）㊀㊀其中，ｙｉ表示原始数据；ｙ＾ｉ表示拟合数据；ｙｉ表示原始数据均值㊂决定系数Ｒ２取值范围为［０，１］，Ｒ２取值越接近１表示拟合效果越好，平均绝对误差ＭＡＥ以及均方根误差ＭＳＥ的取值则越小越好㊂３．５㊀实验结果及分析利用Ｓ－Ｇ滤波器对部分风电功率及相关数据进行降噪处理，由于滤波器中的参数无法对数据进行自适应处理，故在多次对比实验后设置Ｓ－Ｇ滤波器中滑动窗口宽度Ｎ为５９，拟合阶数ｋ为３㊂此后在拼接层将降噪完的数据与原输入未降噪后的数据进行拼接输入进ＴＣＮ－ＢｉＧＲＵ网络中，设置模型训练迭代次数为１００次，ＴＣＮ网络中的膨胀系数设置为以２的倍数进行叠加，ＢｉＧＲＵ网络中神经单元数为６４，其中ＢｉＧＲＵ网络层数为１层㊂为了更直观地展示该组合模型的预测精度，故设置对比实验以及消融实验㊂３．５．１㊀对比实验为了验证模型预测性能，该处选择传统循环神经网络（ＲＮＮ）㊁长短期记忆神经网络（ＬＳＴＭ）以及双向循环门控单元神经网络（ＢｉＧＲＵ）进行对比，各模型预测值与真实值对比结果见表２㊂表２㊀对比实验结果Ｔａｂ．２㊀Ｃｏｍｐａｒｉｓｏｎｏｆｅｘｐｅｒｉｍｅｎｔａｌｒｅｓｕｌｔｓ模型ＭＡＥＭＳＥＲ２ＲＮＮ０．１３３００．０１３４０．８９３３ＬＳＴＭ０．１６３１０．０１５６０．８８００ＢｉＧＲＵ０．１１２６０．０１１１０．９０５７Ｓ－Ｇ－ＴＣＮ－ＳＡ－ＢｉＧＲＵ０．１０２００．０１０２０．９１７５㊀㊀由表２中结果可以看出，融合Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器的ＴＣＮ－ＢｉＧＲＵ模型的预测值以及预测性能要比上述所提到的对比模型都具有优势㊂与传统循环神经网络（ＲＮＮ）相比，本组合模型的ＭＡＥ降低了０．０３１，ＭＳＥ降低了０．００３２；与长短期记忆神经网络（ＬＳＴＭ）相比，本组合模型的ＭＡＥ降低了０．０６１１，ＭＳＥ降低了０．００５４；同双向循环门控单元神经网络（ＢｉＧＲＵ）相比，本组合模型的ＭＡＥ降低了０．０１０６，ＭＳＥ降低了０．０００９；从拟合程度来看，本模型的拟合程度最高㊂综上可以表明，本组合模型在风电功率预测方面的结果是更为精确的㊂３．５．２㊀消融实验为了验证本文组合模型中各个模块的合理性，研究设置了消融实验：剔除ＴＣＮ时域卷积神经网络㊁剔除自注意力机制模块以及剔除Ｓ－Ｇ滤波器，实验结果如图６所示，实验指标对比见表３㊂S-G-T C N-S A-B i G R US-G-T C N-B i G R US-G-B i G R UT C N-S A-B i G R UT r u e V a l u e s160001400012000100008000600040002000050100150200250300350400t i m e s图６㊀消融实验预测值对比Ｆｉｇ．６㊀Ｃｏｍｐａｒｉｓｏｎｏｆｐｒｅｄｉｃｔｅｄｖａｌｕｅｓｆｏｒａｂｌａｔｉｏｎｅｘｐｅｒｉｍｅｎｔｓ表３㊀消融实验对比结果Ｔａｂ．３㊀Ｃｏｍｐａｒａｔｉｖｅｒｅｓｕｌｔｓｏｆａｂｌａｔｉｏｎｅｘｐｅｒｉｍｅｎｔｓ序号模型ＭＡＥＭＳＥＲ２（１）Ｓ－Ｇ－ＴＣＮ－ＳＡ－ＢｉＧＲＵ０．１０２００．０１０２０．９１７５（２）ＴＣＮ－ＳＡ－ＢｉＧＲＵ０．１１２２０．０１１４０．９１７１（３）Ｓ－Ｇ－ＢｉＧＲＵ０．１１０６０．０１０３０．９０８７（４）Ｓ－Ｇ－ＴＣＮ－ＢｉＧＲＵ０．１４９６０．０１３７０．８５９８㊀㊀从图６以及表３中的结果可以看出，模型（２）的ＭＡＥ相较于模型（１）高了０．０１０２，ＭＳＥ高了０．００３５；０７１智㊀能㊀计㊀算㊀机㊀与㊀应㊀用㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀第１３卷㊀模型（３）的ＭＡＥ相较于模型（１）高了０．００８６，ＭＳＥ高了０．０００１；模型（４）的ＭＡＥ相较于模型（１）高了０．０４７６，ＭＳＥ高了０．００３５；㊂综上，在分别剔除Ｓ－Ｇ滤波器㊁自我注意力机制模块以及ＴＣＮ时域卷积网络结构后，每一种模型的ＭＡＥ以及ＭＳＥ都有所上升，这也证明了本文所提出的组合模型的每个结构在模型预测性能方面都有一定的提升作用，进而表明组合模型的预测性能要优于单个模型㊂４　结束语（１）本文所使用的Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器在一定程度上对数据进行了有效降噪并提升了预测模型整体的预测精度，但该滤波器中的参数需要根据输入数据进行调节，所以在此研究基础上可进一步利用算法优化Ｓ－Ｇ滤波器的滑动窗口参数Ｎ以及拟合阶数ｋ，用来提高模型对于不同数据的适应性㊂（２）在风电功率预测方面， 融合Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器的ＴＣＮ－ＳＡ－ＢｉＧＲＵ模型的风电功率预测性能更高 这一实验目的在经过对比㊁消融实验后得到验证，模型结构中的ＴＣＮ㊁自我注意力机制模块以及双向学习策略结构不仅提升了预测性能，并且其提取数据特征速度更快，能够更好地挖掘时序数据中所蕴含的信息要素㊂（３）基于本模型的研究更多方面是展现模型的优化，而针对风电功率预测等实际应用场景应当进一步寻找合适的预测模型进行研究㊂参考文献［１］杨茂，代博祉，刘蕾．风电功率概率预测研究综述［Ｊ］．东北电力大学学报，２０２０，４０（２）：１－６．［２］樊重俊，等．人工智能基础与应用［Ｍ］．北京：清华大学出版社，２０２０．［３］丁明，张立军，吴义纯．基于时间序列分析的风电场风速预测模型［Ｊ］．电力自动化设备，２００５，２５（８）：３２－３４．［４］ＹＡＮＧＬ，ＨＥＭ，ＺＨＡＮＧＪ，ｅｔａｌ．Ｓｕｐｐｏｒｔ－ｖｅｃｔｏｒ－ｍａｃｈｉｎｅ－ｅｎｈａｎｃｅｄＭａｒｋｏｖｍｏｄｅｌｆｏｒｓｈｏｒｔ－ｔｅｒｍｗｉｎｄｐｏｗｅｒｆｏｒｅｃａｓｔ［Ｊ］．ＩＥＥＥＴｒａｎｓａｃｔｉｏｎｓｏｎＳｕｓｔａｉｎａｂｌｅＥｎｅｒｇｙ，２０１５，６（３）：７９１－７９９．［５］乔路丽．基于ＨＧＷＯ－ＳＶＭ的风电功率预测方法研究［Ｊ］．东北电力技术，２０２３，４４（３）：１２－１７．［６］赵新，刘冬生．基于改进果蝇算法优化ＳＶＭ的模拟电路故障诊断及对比分析［Ｊ］．电子测量与仪器学报，２０１９，３３（３）：７８－８４．［７］ＫＩＳＶＡＲＩＡ，ＬＩＮＺｉ，ＬＩＵＸｉａｏｌｅｉ．Ｗｉｎｄｐｏｗｅｒｆｏｒｅｃａｓｔｉｎｇ－ａｄａｔａ－ｄｒｉｖｅｎｍｅｔｈｏｄａｌｏｎｇｗｉｔｈｇａｔｅｄｒｅｃｕｒｒｅｎｔｎｅｕｒａｌｎｅｔｗｏｒｋ［Ｊ］．ＲｅｎｅｗａｂｌｅＥｎｅｒｇｙ，２０２１，１６３：１８９５－１９０９．［８］许阅，刘光杰．基于注意力机制的Ｂｉ－ＧＲＵ内容流行度预测算法［Ｊ］．电子测量技术，２０２２，４５（３）：５４－６０．［９］符杨，任子旭，魏书荣，等．基于改进ＬＳＴＭ－ＴＣＮ模型的海上风电超短期功率预测［Ｊ］．中国电机工程学报，２０２２，４２（１２）：４２９２－４３０３．［１０］孙隽丰，李成海，曹波．基于ＴＣＮ－ＢｉＬＳＴＭ的网络安全态势预测［Ｊ／ＯＬ］．系统工程与电子技术：１－１１［２０２２－０９－２３］．ｈｔｔｐｓ：／／ｋｎｓ．ｃｎｋｉ．ｎｅｔ／ｋｃｍｓ／ｄｅｔａｉｌ／１１．２４２２．ＴＮ．２０２２０９２２．０９１２．００２．ｈｔｍ．［１１］ＸＵＣｏｎｇｙｕａｎ，ＳＨＥＮＪｉｚｈｏｎｇ，ＤＵＸｉｎ，ｅｔａｌ．Ａｎｉｎｔｒｕｓｉｏｎｄｅｔｅｃｔｉｏｎｓｙｓｔｅｍｕｓｉｎｇａｄｅｅｐｎｅｕｒａｌｎｅｔｗｏｒｋｗｉｔｈｇａｔｅｄｒｅｃｕｒｒｅｎｔｕｎｉｔｓ［Ｊ］．ＩＥＥＥＡｃｃｅｓｓ，２０１８，６：４８６９７－４８７０７．［１２］蒲贞洪，朱元富．基于ＴＳＮＥ－ＢｉＧＲＵ模型短期电力负荷预测［Ｊ］．电工技术，２０２３（３）：５２－５７．［１３］ＣＨＥＮＫｕｉ，ＬＡＧＨＲＯＵＣＨＥＳ，ＤＪＥＲＤＩＲＡ．Ｄｅｇｒａｄａｔｉｏｎｐｒｅｄｉｃｔｉｏｎｏｆｐｒｏｔｏｎｅｘｃｈａｎｇｅｍｅｍｂｒａｎｅｆｕｅｌｃｅｌｌｂａｓｅｄｏｎｇｒｅｙｎｅｕｒａｌｎｅｔｗｏｒｋｍｏｄｅｌａｎｄｐａｒｔｉｃｌｅｓｗａｒｍｏｐｔｉｍｉｚａｔｉｏｎ［Ｊ］．ＥｎｅｒｇｙＣｏｎｖｅｒｓｉｏｎ＆Ｍａｎａｇｅｍｅｎｔ，２０１９，１９５（ＳＥＰ．）：８１０－８１８．１７１第１１期秦小晖，等：融合Ｓａｖｉｔｚｋｙ－Ｇｏｌａｙ滤波器的ＴＣＮ－ＳＡ－ＢｉＧＲＵ风电功率预测。

The concept of environmental protection and the dream of a green future have become increasingly significant in todays world.As the human population grows and industrialization expands,the impact on our planet has become a pressing concern.Here are some key points that could be included in an essay about the journey towards an environmentally friendly and sustainable future.Introduction:The importance of environmental protection in the21st century.The dream of a green future and its significance for the survival of our planet.The Current State of the Environment:The effects of pollution on air,water,and soil.The loss of biodiversity and the extinction of species.The impact of climate change and global warming.The Role of Governments and International Agreements:The role of the Paris Agreement in combating climate change.National policies and regulations aimed at reducing emissions and promoting renewable energy.Individual and Community Actions:The power of individual choices in reducing waste,conserving energy,and supporting sustainable products.Community initiatives such as tree planting,recycling programs,and local cleanup events.Technological Innovations:The development of clean energy technologies like solar,wind,and hydroelectric power. Advances in electric vehicles and public transportation to reduce reliance on fossil fuels.Education and Awareness:The importance of environmental education in schools and communities.Public awareness campaigns and the role of media in promoting green values.Challenges and Solutions:The economic challenges of transitioning to a green economy.Innovative solutions and the need for international cooperation.The Future of Green Living:Visions of smart cities and sustainable urban planning.The integration of green spaces and ecofriendly infrastructure.Conclusion:The collective responsibility of every individual,community,and nation to protect the environment.The hope for a greener,cleaner,and more sustainable future for generations to come.Call to Action:Encouraging readers to take steps towards a more sustainable lifestyle.The potential for each person to contribute to the global effort for environmental protection.By focusing on these points,an essay can effectively convey the urgency and importance of the environmental protection and the dream of a green future.It can inspire readers to consider their own role in this journey and to take action towards a more sustainable world.。

In the everevolving landscape of global competition, the role of technological innovation has become more pivotal than ever before. As a high school student who is deeply interested in the intersection of technology and global cooperation, I have observed the significant impact that collaborative efforts in innovation can have on both economic growth and societal advancement.Growing up in an era where technology is an integral part of our daily lives, I have witnessed firsthand how the sharing of knowledge and resources can lead to groundbreaking developments. One such example is the collaboration between different countries in the field of space exploration. The International Space Station ISS is a testament to what can be achieved when nations come together to pool their expertise and resources. This collaborative project has not only expanded our understanding of the universe but also led to numerous technological advancements that have found applications on Earth.Moreover, the fight against global challenges such as climate change has necessitated a collective approach to innovation. The development of renewable energy sources, for instance, has been significantly propelled by international research partnerships. Solar panels, wind turbines, and electric vehicles are just a few examples of technologies that have benefited from a collaborative approach, leading to more sustainable and environmentally friendly solutions.In the realm of healthcare, the recent global pandemic has underscored the importance of international cooperation in medical research anddevelopment. The rapid development of vaccines and treatments for COVID19 was made possible by the sharing of scientific data and the coordination of efforts across borders. This has not only saved countless lives but also highlighted the power of a united global community in the face of adversity.The benefits of technological innovation through collaboration extend beyond these highprofile examples. In the business world, partnerships between companies from different countries can lead to the creation of new products and services that cater to diverse markets. For instance, the collaboration between a tech giant from Silicon Valley and a manufacturing powerhouse in Asia can result in the production of cuttingedge electronic devices that are both affordable and accessible to a global audience.However, it is important to acknowledge the challenges that come with international cooperation in innovation. Issues such as intellectual property rights, cultural differences, and political tensions can sometimes hinder the progress of collaborative efforts. Nevertheless, the potential rewards of such partnerships far outweigh the risks, as they foster a culture of learning and adaptation that can drive progress in an increasingly interconnected world.As I look towards the future, I am excited about the possibilities that lie ahead in the realm of technological innovation. I believe that by embracing a spirit of collaboration and openness, we can overcome the barriers that stand in the way of progress. The exchange of ideas, the sharing ofresources, and the joint pursuit of knowledge can lead to a future where technology serves as a catalyst for positive change across the globe.In conclusion, the importance of strengthening technological innovation cooperation cannot be overstated. It is through the collective efforts of individuals, organizations, and nations that we can unlock the full potential of technology to improve our lives and address the pressing issues of our time. As a high school student, I am inspired by the achievements of those who have come before me and am eager to contribute my own ideas and energy to this ongoing journey of discovery and collaboration.。

Markov-optimal Sensing Policy for User State Estimation inMobile Devices∗Yi Wang†,Bhaskar Krishnamachari†,Qing Zhao‡,Murali Annavaram††Ming Hsieh Department of Electrical Engineering,University of Southern California,Los Angeles,USA{wangyi,bkrishna,annavara}@‡Department of Electrical and Computer Engineering,University of California,Davis,USAqzhao@ABSTRACTMobile device based human-centric sensing and user state recognition provide rich contextual information for various mobile applications and services.However,continuously capturing this contextual information consumes significant amount of energy and drains mobile device battery quickly. In this paper,we propose a computationally efficient algo-rithm to obtain the optimal sensor sampling policy under the assumption that the user state transition is Markovian.This Markov-optimal policy minimizes user state estimation er-ror while satisfying a given energy consumption budget.We first compare the Markov-optimal policy with uniform peri-odic sensing for Markovian user state transitions and show that the improvements obtained depend upon the underly-ing state transition probabilities.We then apply the algo-rithm to two different sets of real experimental traces per-taining to user motion change and inter-user contacts and show that the Markov-optimal policy leads to an approx-imately20%improvement over the naive uniform sensing policy.Categories and Subject DescriptorsC.3[Special Purpose and Application Based Systems]: Signal processing systemsGeneral TermsAlgorithms,Design,Performance,VerificationKeywordsEnergy efficiency,Mobile sensing,Optimal sampling policy, Markovian User state∗We’d like to acknowledge partial support for this work from Nokia Inc and National Science Foundation,through grants numbered NSF CNS-0831545and NSF ECS-0622200.Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on thefirst page.To copy otherwise,to republish,to post on servers or to redistribute to lists,requires prior specific permission and/or a fee.IPSN’10,April12–16,2010,Stockholm,Sweden.Copyright2010ACM978-1-60558-955-8/10/04...$10.00.1.INTRODUCTIONMobile devices permeate our daily lives and their utility is further enhanced with mobile applications and services.A growing number of these mobile services rely on accu-rate and automatic detection of user er state is obtained by sensing the user’s surroundings and deriving contextual information from the sensed data.Current gen-eration mobile devices integrate a wide range of sensing and networking features such as GPS,microphone,camera,light sensor,accelerometer,WiFi,Bluetooth,GPRS,and so on. These sensors collectively enable us to obtain user’s con-textual information,such as user’s activity,location,back-ground sound,and number of peering devices.Throughout our study,we use the term“user state”to represent human context information,which evolves over time.While incorporating user’s contextual information brings mobile application personalization to new levels of sophis-tication,a major impediment to collecting and determin-ing user context is the limited battery capacity on mobile phones.Since the integrated sensors can consume significant amount of energy,continuously collecting context informa-tion through sensor activation will make even the most com-pelling mobile application less than desirable to users.For example,our experiments show that a fully charged battery on Nokia N95device would be completely drained within six hours if its GPS is operated continuously[33].Therefore,en-ergy efficient mobile sensing is critical in order to maintain the lifetime of mobile device battery and to continue to take advantage of mobile applications and services.This paper addresses the critical issue of energy efficient mobile sensing by making the following two contributions: First,we formulate the problem of sampling Markovian user state sequence under energy constraint as a constrained op-timization problem.We develop a framework based on Con-strained Markov Decision Process(CMDP)in order to ob-tain the Markov-optimal sensor sampling policy.Second,we analyze and evaluate the Markov-optimal policy,and com-pare its performance against uniform periodic sampling on real user state traces.In our prior work[32],we studied how to energy efficiently sample a discrete time Markovian user state sequence by proposing a stationary deterministic sensor sampling policy. The policy assigns different butfixed duty cycles to the sen-sor at different user states.We investigated its performance in terms of the tradeoffbetween two conflicting performance metrics:the expected energy consumption and the expected user state estimation error.However,this prior work re-quires exponential run-time and computes a sub-optimalpolicy.Motivated by these shortcomings of the previous work,in this paper we address the following question:Given a certain device energy consumption budget and user state transition probabilities,what is the optimal sensor sampling policy that leads to the minimum expected state estimation error,while satisfying the energy budget constraint?We for-mulate this constrained optimization problem as an infinite-horizon CMDP.The expected average cost of the CMDP is considered as the cost criteria to be minimized(as compared to the discounted cost),with the sensor energy consumption serving as the constraint.Solving such CMDP yields a glob-ally optimal sensor sampling policy u∗which guarantees the minimum expected user state estimation error while satisfy-ing the given energy budget for Markovian user state tran-sitions(we thus name this policy“Markov-optimal”).The Markov-optimal policy is stationary and randomized(shown in[2,11,12])and can be obtained by solving the correspond-ing Linear Programming(LP)of the CMDP formulation, which requires only polynomial run-time for any number of user state inputs.To evaluate the performance of the Markov-optimal pol-icy,we compare it to uniform periodic sampling for Marko-vian user state transitions.The expected user state esti-mation error in both cases is computed for different transi-tion probabilities and energy budget values.The numeric comparison results show that the Markov-optimal policy al-ways leads to a non-negative gain over uniform sampling. We then apply both the Markov-optimal policy and peri-odic sampling to real user context data traces pertaining to user motion change(“Stable”vs.“Moving”)and the sta-tus of user contact(“In contact”vs.“Not in contact”).We show that although the user state transitions in these real data traces are not strictly Markovian,the Markov-optimal policy still leads to approximately20%improvement over uniform periodic sampling.The reason for energy efficiency improvement is that even when the state transitions do not strictly follow Markovian,Markov-optimal policy is able to reduce the expected estimation error by sampling more fre-quently when state transition is more uncertain,and on the other hand,increasing the length of the idle interval as state transition becomes more predictable.Although in this paper,we limit the analysis and theory development to discrete user state transitions,it is worth mentioning that continuous context such as location can be divided into discrete space and thusfits in our frame-work well.In addition,while the approach presented in this paper is centered around mobile sensing,the problem for-mulation can be generalized easily to many other monitor-ing/detection/probing problems where certain performance metric needs to be optimized with energy consumption limi-tations.For example,in thefield of environmental monitor-ing using wireless sensors,it is ideal to improve the detection accuracy while maintaining long device lifetime.The rest of this paper is organized as follows.In section2, we discuss relevant prior work.In section3,we introduce the background of our study,including the model and assump-tions,the mechanism that estimates user state for missing observations,the formal definition of expected energy con-sumption and expected user state estimation error,as well as the constrained optimization problem.In section4,we propose our CMDP formulation and the algorithm thatfinds the optimal sensing policy and associated state estimation error by solving the corresponding LP.We also show some of the properties of the optimal policy for special Markov transition probabilities.In section5,we compare the op-timal policy with uniform periodic sampling both through numerical analysis and by applying them to real user state traces.Finally,we conclude and present directions for future work in section6and7.2.RELATED WORKIn the context of using mobile device to sense and recog-nize user state,it is worth noting that Schmidt et al.[25]first propose to incorporate low level sensors to mobile PDAs to demonstrate situational awareness.Several works have been conducted thereafter by using the commodity cell phones as sensing,computing or application platforms[5,13,20,21, 27,34].For example,“CenceMe”[21]uses the integrated as well as external sensors to capture the user’s status such as activity,disposition,and surroundings and enables mem-bers of social networks to share their sensing presence.Sim-ilarly,“Sensay”[27]is a context-aware mobile phone that uses data from a number of sources to dynamically change cell phone ring tone,alert type,as well as determine users’“un-interruptible”states.The study of energy efficiency in mobile sensing has been conducted in works such as[18]and[17].Krause et al.[18] investigate the topic of trading offprediction accuracy and power consumption in mobile computing,and the authors showed that even very low sampling rate of the accelerom-eter can lead to competitive classification results while de-vice energy consumption can be significantly reduced.The hierarchical sensor management concept is explored by the “SeeMon”system[17],which achieves energy efficiency by only performing context recognition when changes occur dur-ing the context monitoring.More recently,energy manage-ment in personal health care system using mobile devices is studied by Aghera et al.[1].The authors aim to address the tradeoffbetween energy efficiency and data uploading latency by implementing a task assignment strategy capa-ble of selecting different levels of operation.Constandache et al.[9]study energy efficiency in mobile device based lo-calization,and the authors show that human can be profiled based on their mobility patterns and thus location can be predicted.The proposed“EnLoc”system achieves good lo-calization accuracy with a realistic energy budget.Besides in mobile sensing,energy efficient monitoring and event detection has been widely studied in a much broader context such as communication,data collection,and so on. For example,Stabellini and Zander[29]propose a new en-ergy efficient algorithm for wireless sensor networks to de-tect communication interference.Shin et al.[26]propose DEAMON,an energy efficient distributed sensor monitor-ing algorithm where sensors are controlled using a Boolean expression and energy is conserved by selectively turning on/offthe nodes and limiting communication operations. Similarly,Silberstein et al.[28]investigate suppression tech-nique(reducing the cost for reporting changes)and take ad-vantage of temporal and spatial data correlation to reduce data reporting therefore enlarge the lifetime of sensors.The work in this paper is originated from[33]and[32]. In[33],we present an Energy Efficient Mobile Sensing Sys-tem(EEMSS)that recognizes user states as well as detects state transitions.EEMSS significantly improves device bat-tery life,by powering a minimum set of sensors at any given time and applying appropriate sensor duty cycles.However,in[33],sensors still adopt pre-determined,fixed duty cycles whenever activated,which is not adjustable to different userbehaviors.We address this issue in[32],where we modelthe user state as a discrete time Markov chain(DTMC),and propose a stationary deterministic sensing policy to in-crease energy efficiency by assigning different sensor dutycycles at different user states.In this paper,we study how sensor duty cycles can be optimized in order to minimizethe expected user state estimation error,while maintainingan energy consumption budget.We propose an efficient al-gorithm that obtains the optimal sensing policy for Marko-vian user state,by formulating the constrained optimizationproblem as a infinite-horizon CMDP and solving its corre-sponding LP.CMDP[2]is a variant of Markov Decision Processes[4](MDP)which provides a framework for constructing opti-mal policies for a stochastic process.CMDP considers asituation where one type of cost is to be optimized whilekeeping others below some pre-defined bounds.For example, Lazar[19]used CMDP to analyze the problem of maximizingthroughput subject to delay constraints in telecommunica-tion applications.In traffic applications,Nain and Ross[23]studied the problem where different traffic types competefor the same resource.Some weighted sum of average de-lays of certain traffic types is to be minimized while some other traffic types need to satisfy certain delay constraints.More recent studies of CMDP include[10],where Dolgovand Durfee limit the optimal policy search to stationary de-terministic policies coupled with a novel problem reductionto mixed integer programming,and the method yields an algorithm thatfinds computationally feasible policies(i.e.,not randomized).It has been shown by Feinberg and Shwartz[11,12]thatstationary deterministic policies are not guaranteed to beoptimal due to the constraints added to the classical MDPmodel.In order to solve for the optimal stationary random-ized policy in polynomial time,Kallenberg[16],and Heymanand Sobel[14],propose the use of LP where adding con-straints with the same discount factor does not increase thecomplexity of the problem.In our study,we utilize this con-cept and rewrite the constrained optimization problem as anLP,and obtain the optimal randomized policy in polynomial time.3.BACKGROUND3.1Model and assumptionsWe assume that time is discretized and the user stateevolves as a N-state discrete time Markov chain(DTMC)with transition probabilities p ij(i,j∈{1,2,...,N})from state i to j.The discrete time horizon is denoted as T= {t,t=0,1,2,...}.At each time slot during the process,a sensor may be sampled,in which case user state is detectedwith100%accuracy and a unit of energy consumption isassociated.On the other hand,the sensor can also stayidle to save energy.This process is illustrated infigure1. We define O as the set of time slots when the sensor makes observation and O is thus a subset of T,as shown byfig-ure1.Ideally,if the sensor can be sampled in each time slot (O=T),the user state sequence can be obtained perfectly. However,in general,since sensor duty cycles are required to extend mobile device lifetime,there exist durations where user state needs to be estimated using only observeddata Figure1:Illustration of the Markovian user statesequence,sampled at certain time slots.and the knowledge of the user state transition dynamics.3.2User state estimationA user state estimation mechanism has been proposed in one of our previous works[32].Here we give a brief intro-duction of the method,which estimates the most likely userstate for each time slot between two subsequent observa-tions.In particular,given the sensor detects state i at timet m,and detects state j at time t n(t m,t n∈T and t m<t n), the probability of being in state k(k∈1,2,...,N)at time t(t m<t<t n),is given by:p[s t=k|s t m=i,s t n=j](1) =p[s t=k,s t n=j|s t m=i]p[s t n=j|s t m=i](2)=p[s t=k|s t m=i]·p[s t n=j|s t=k]p[s t n=j|s t m=i](3)=P(t−t m)ik·P(t n−t)kjP(t n−t m)ij(4)where s t denotes the true user state at time t.In order to select the“most likely”state(denoted by s t)at time t,the quantity given in(1)needs to be maximized:s t=argmaxkP(t−t m)ik·P(t n−t)kjP(t n−t m)ij(5)The expected estimation error for time slot t,denoted by e t,is defined as the probability of incorrect estimation, which is given by:e t=1−p[s t=s t|s t m=i,s t n=j](6)=1−maxkP(t−t m)ik·P(t n−t)kjP(t n−t m)ij(7)3.3Expected energy consumption E[C]and ex-pected user state estimation error E[R] Throughout our study,we address two important perfor-mance metrics:(a)the expected energy consumption and (b)the expected user state estimation error.These two in-trinsically conflicting metrics are fundamental performance measures of mobile sensing applications,as the former char-acterizes the device lifetime and the latter is directly related to the quality of the application in terms of providing high user state detection accuracy.The formal definitions of the two are given as follows:Definition 1.The expected energy consumption E[C]of the user state sampling process is defined asE[C]=1E[I](8)where E[I]is the expected sensor idle interval,i.e.,the av-erage number of time slots the sensor waits before the next sample.Definition 2.The expected user state estimation error E[R]is defined as the long term average of per-slot estima-tion error:E[R]=limn→∞nt=1e tn(9)3.4The optimization problemIn practical system operations,a device energy consump-tion budget is often specified in order to ensure a long enough device lifetime.We defineξ(where0≤ξ≤1)as the energy budget which is the maximum expected energy consumption allowed.Let u denote a sensor sampling policy,which spec-ifies how sensor duty cycles should be assigned when differ-ent user states are detected,i.e.,how long the sensor should stay idle after each observation.We investigate the follow-ing constrained optimization problem:Considering a long user state evolving process,given a device energy consump-tion budgetξand user state transition matrix P,what is the optimal sensor sampling policy u∗such that the expected user state estimation error E[R]is minimized,and the ex-pected energy consumption is maintained below the budget,i.e.,E[C]≤ξ?4.THE MARKOV-OPTIMAL POLICYWe formulate the optimization problem as a infinite-horizon CMDP with expected average cost.Solving such CMDP yields a Markov-optimal policy u∗which is stationary and randomized[2,11,12],i.e.,u∗does not vary over time and decision is randomized over several actions.Meanwhile,the algorithm thatfinds the Markov-optimal policy u∗is compu-tationally efficient,since u∗can be obtained by solving the corresponding LP,which is a polynomial time computation.4.1The CMDP formulationWe use an infinite-horizon CMDP to model the sensor duty cycle selection problem,whose major components are explained as follows:•Decision Epochs O.Decision is made immediatelyafter each sensing observation.The set of decisionepochs is defined as O={d,d=1,2,...}.Note thatthis is the same O defined in section3.1.•System State Space X.The system state at deci-sion epoch d is the detected user state s d.The set ofsystem states is denoted by X={1,2,...,N}.•Action Space A.The set of actions is denoted byA.An action a∈A is the number of time slots thesensor waits until making the next observation.Notethat the potential number of actions could be infinity,i.e.,A=N∗,the set of positive integers.However,inorder to facilitate the study,we limit the set of actionsto positive integers under some threshold a m in ourstudy.Asξdecreases,a m needs to be increased inorder to meet the energy budget constraint.•System Transition Probabilities P.Define P iajas the probability of moving from system state i to j,when action a is taken.Given the user state transitionprobability matrix P,it is easy to conclude that P iaj=P(a)ij.•Intermediate Cost c(y,a).The intermediate costc(y,a)is defined as the expected aggregate state esti-mation error when action a is taken on state y,wherey∈X and a∈A.c(y,a)can be calculated as:c(y,a)=jP(a)yj·e(a)yj(10)=jP(a)yj·a−1t=11−maxkP(t)yk·P(a−t)kjP(a)yjwhere e(a)yjis the aggregate state estimation error fora length-a observation interval starting at state y andending at state j.•Expected Average Cost E u[R].under policy u,the expected average cost of infinite-horizon CMDP isdefined as:E u[R]=limn→∞nd=1E u c(X d,A d)n(11)where n is the total number of decision epochs,andE u is the corresponding expectation operator of theintermediate cost,under the policy u.•Intermediate Sampling Interval d(y,a).The in-termediate sampling interval when action a is taken onstate y,satisfies d(y,a)=a.•Expected Sampling Interval E u[I].Similar to E u[R], the expected sampling interval E u[I]is defined as:E u[I]=limn→∞nd=1E u d(X d,A d)n(12)•Constraintξ.The maximum allowed expected en-ergy cost.Therefore E[C]≤ξor E[I]≥1/ξ.4.2Finding the optimal policy u∗Letρ(y,a)denote the“occupation measure”of state y and action a,i.e.,the probability that such state-action pair ever exists in the decision process.It is shown in[16,14,2] that the optimal policy of the CMDP with expected average cost criteria can be obtained by solving the corresponding LP.Inspired by these works,we rewrite the sensor duty cy-cle optimization problem defined above as the following LP (denoted as LP1):LP1:Find miny∈Xa∈Aρ(y,a)c(y,a)(13)subject to:y∈Xa∈Aρ(y,a)(δx(y)−P yax)=0,∀x∈X,(14)y∈Xa∈Aρ(y,a)=1,(15)ρ(y,a)≥0,∀y,a,and(16)y∈Xa∈Aρ(y,a)d(y,a)≥1ξ(17)where δx (y )is the delta function of y concentrated on x .The constraint given in (14)emphasizes the property for ergodic processes and describes that the outgoing and incoming rate for a state need to be the same.The constraints (15)and (16)define ρ(y,a )as a probability measure.The inequality constraint given in (17)guarantees that the expected energy usage is less than the energy constraint value ξ,by setting the expected idle interval greater than 1/ξ.Solving LP1yields the ideal occupation measure of each state/action pair that minimizes the quantity given in (13).The algorithm for selecting the optimal policy u ∗is given as follows:Algorithm 1Pseudo code for selecting the Markov-optimal sampling policy u ∗and calculating the corresponding ex-pected user state estimation error E u ∗[R ]1:Input:ξ,P2:Output:u ∗(ξ,P ),E u ∗[R ]3:Initialize:Select a large enough a m ∈N ∗such that the energy constraint can be sufficiently satisfied.Let A ={1,2,...,a m },X ={1,2,...,N }4:Solve LP1for the optimal occupation measure ρ∗(y,a ),∀y ∈X,∀a ∈A 5:Calculate ρ∗y = a ∈A ρ∗(y,a ),∀y ∈X6:Constructing u ∗:Assign ρ∗(y,a )ρ∗yas the probability of taking action a at state y ,∀y ∈X,∀a ∈A 7:Calculate the expected user state estimation error:E u ∗[R ]=Eu ∗[R ]1/ξ= y ∈X a ∈A ρ∗(y,a )c (y,a )1/ξIn Algorithm 1,the optimal occupation measure of each state-action pair can be obtained by solving LP1in step 4.Step 5of Algorithm 1calculates the overall probability of seeing a particular state y in the decision process under the optimal policy,and this quantity ρ∗y is used later in step 6as the normalization to the optimal occupation measure ρ∗(y,a )in order to compute the probability of taking ac-tion a whenever state y is detected,in the optimal sampling policy u ∗.This normalization step is needed to guarantee thata ρ∗(y,a )ρ∗y =1,∀y ∈X (18)In step 7,the optimal expected user state estimation errorE u ∗[R ]can be calculated once the optimal expected averagecost Eu∗[R ]of the CMDP is obtained.Here we illustrate the relationship between E [R ]and E [R ].Recall that E [R ]is defined as the long-term,average per-slot error,i.e.,the ratio of incorrect estimations over total number of time slots;therefore,under policy u ,it can be written as:E u [R ]=lim n →∞n d =1E uc (Xd ,A d ) n d =1E u d (X d ,A d )(19)=lim n →∞1n· n d =1E u c (X d ,A d )1n · n d =1E u d (X d ,A d )(20)where n denotes the total number of decision epochs,whereasE u c (X d ,A d )and E u d (X d ,A d )denote the aggregated state estimation error and the sampling interval (action)size for the d th decision epoch under policy u ,respectively.Con-sidering the Law of Large Numbers,equation (20)can befurther written asE u [R ]=Eu [R ]E u [I ](21)where E u [I ]is the expected sampling interval (action)size under policy u .Since the optimal policy u ∗needs to be fully utilizing the energy budget,its expected sampling interval size satisfies:E u ∗[I ]=1/ξ.(22)Algorithm 1is able to return the Markov-optimal policy u ∗in polynomial time for any number of states input.It can be seen that although step 4is the most time-consuming pro-cedure,it is able to obtain the optimal occupation measures in polynomial time.4.3A case study:Selectingu ∗for two-stateMarkov chainsWhile Algorithm 1returns the optimal policy u ∗for any number of user state inputs,in this section,as a case study,we apply Algorithm 1to two-state Markov chains and find the optimal policy u ∗by solving the corresponding LP in MATLAB.We test six different user state transition matri-ces,namely:0.90.10.10.90.90.10.50.5 0.90.10.90.1 0.50.50.50.5 0.50.50.90.1 0.10.90.90.1 Different energy budget constraint values ranging from 0to1are used as different inputs to Algorithm 1.The solution of the LP returns a randomization over all possible state-action pairs,which essentially constructs the stationary randomized optimal policy u ∗.Since it is difficult to visualize the randomization of multiple actions,we com-pute the mean value of actions taken at each decision epoch for different user states,i.e.,the expected average idle in-terval at each state according to the optimal policy u ∗,and show the result of the average actions in figure 2.It can be noticed that when p 12=0.1and p 21=0.9,the Markov chain degenerates to an independently and identi-cally distributed (i.i.d.)user state sequence,where the user state transition does not depend on any current or previous state information.In this case,the expected state estimation error E u ∗[R ]under the optimal policy u ∗can be quantified in terms of user state transition probability matrix P and the energy budget ξ.The following theorem holds.Theorem 1.For two-state Markov chain that degener-ates to an er state sequence,given its transition probabilities p ij (i,j ∈{1,2})and the energy budget ξ,the following holds for optimal sampling policy u ∗:E u ∗[R ]=(1−ξ)·min {p 12,p 21}(23)i.e.,E u ∗[R ]is linear in ξ.Proof.For i.i.d.transition probability matrix the fol-lowing holds:P =P (1)=P (2)=...=P (I ),∀I ∈N ∗(24)that is,the state transition probability matrix in I steps,denoted by P (I ),is equal to the original state transition probability matrix P ,for any positive integer I .Figure2:The average idle intervals in u∗for dif-ferent energy budgets.I1and I2are the average idle interval lengths when state1and2is observed, respectively.Let p,q be such that p=p12=1−p11=p22=1−p21= 1−q.As equation(24)holds,the term e t in equation(7)can be further written as:e t=1−maxp2p,q2q,pqp,pqq(25)=min{p,q}(26)=min{p12,p21}(27) i.e.,e t stays constant no matter what sample interval is used.This illustrates the fact that the estimation error for eachtime slot with missing observation is constant.Recall thatE[R]is defined as the expected per-slot estimation error,and that the optimal policy utilizes the full energy budget withexpected sensor sampling interval1/ξ,therefore,E u∗[R]can be expressed as:E u∗[R]=1/ξ−11/ξ·e t=(1−ξ)·min{p12,p21}(28)for optimal policy u∗.It can be easily concluded that for all Markov chains that degenerates to an er state sequence,there exist more than one optimal policy due to the fact that any randomiza-tion of actions that leads to the optimal sampling interval results in the same expected state estimation error.Another observation fromfigure2is that,whenever the two-state Markov chain is symmetric,i.e.,p12=p21,the average sampling intervals for both state1and2are equal. In fact,the following theorem describing the property of op-timal policy holds for all symmetric two-sate Markov chain: Theorem 2.For two-state symmetric Markov chain where p12=p21,there exists an optimal sampling policy that makes the same decision,i.e.,selects the same randomization ofactions at each decision epoch.Proof.Let S={s i,i=1,2,3,...}be the user state se-quence in the original decision process,where s i∈{1,2},∀i. Define T∗as the set of time slots where sensing is performed under the optimal policy u∗.We label the original process as S∗={s∗i,i=1,2,3,...},in which s∗i is defined as follows:s∗i=s i,if i∈T∗0,if i/∈T∗S∗is a process that characterizes the sampled process based on the policy u∗,by retaining the original state sequence when observations are made,and labeling“0”for the rest time slots.Note that the numbers of consecutive zeros in all estimation intervals need not necessarily be the same,due to the fact that u∗adopts randomized decision at each decision epoch u∗.We then define the process S∗∗={s∗∗i,i=1,2,3,...} wheres∗∗i=⎧⎨⎩s∗∗i=2,if s∗i=1s∗∗i=1,if s∗i=2s∗∗i=0,if s∗i=0S∗∗is a state sequence containing the“flipped”state from S∗for all observed time slots,and zeros otherwise.Since p12=p21,it immediately follows that P(i)12=P(i)21,∀i =1,2,3....Recall equation(10)where the aggregated error for an observation interval with a certain length is calcu-lated.It can be concluded that e(a)12=e(a)21and e(a)22=e(a)11, since the following equalities hold:maxkP(t)1k·P(a−t)k2P(a)12=maxkP(t)2k·P(a−t)k1P(a)21(29)maxkP(t)1k·P(a−t)k1P(a)12=maxkP(t)2k·P(a−t)k2P(a)22(30)where k∈{1,2}.It follows that the relabeled process S∗∗produces the same expected error as S∗,which implies that by swapping all decisions made at state1with all decisions made at state2in the original decision process,the expected state estimation error remains optimal.Therefore,the opti-mal policy decision at state1is able to be applied to state2, and vice versa,while maintaining the optimal error.This in-dicates that for symmetric two-state Markov chain,no mat-ter what state is observed at each decision epoch,there exists an optimal policy that makes the same decision.5.PERFORMANCE COMPARISON:MARKOV-OPTIMAL VS.UNIFORM Researchers have proposed many sophisticated sampling schemes such as[30]and[8],aiming at energy efficient event detection in mobile and wireless networks.However,in the context of mobile sensing for user state detection and esti-mation,there does not exist an established framework that aims to achieve the best tradeoffbetween energy consump-tion and state estimation error and select the best sampling intervals correspondingly.A baseline performance evalua-tion of the optimal policy u∗,therefore,is to compare it with a uniform policy¯u where the sensor simply performs periodic sampling.In this section,we compare the expected user state esti-mation error,and visualize the gain(which is the ratio of。