Structural method for Sensor Placement

Structural Health Monitoring Structural health monitoring (SHM) is a crucial aspect of ensuring the safety and integrity of various structures, such as buildings, bridges, and infrastructure. It involves the use of sensors, data analysis, and advanced technology to continuously monitor the condition of a structure and detect any potential issues or defects. The primary goal of SHM is to prevent structural failures, reduce maintenance costs, and extend the lifespan of the structure. However, there are several challenges and considerations that need to be addressed when implementing SHM systems. One of the key considerations in structural health monitoring is the selection and placement of sensors. The type and number of sensors required depend on the specific characteristics and usage of the structure. For example, a bridge may require different sensors than a high-rise building. Additionally, the placement of sensors is critical to ensure that they can effectively monitor the structural behavior and detect any changes or abnormalities. This process requires careful planning and analysis to determinethe optimal sensor configuration for each structure. Another important aspect of SHM is the data collection and analysis process. Sensors continuously collect data on various parameters, such as strain, temperature, and vibration, which is then analyzed to assess the structural condition. Advanced algorithms and data processing techniques are used to interpret the data and identify any potential issues or anomalies. However, the sheer volume of data collected can present challenges in terms of storage, processing, and interpretation. As a result, it is essential to have robust data management and analysis systems in place to effectively utilize the information collected by the sensors. Furthermore, the implementation of SHM systems requires a significant investment in terms of technology, equipment, and expertise. This can be a barrier for many organizations, especially smaller ones or those with limited resources. Additionally, there maybe challenges in integrating SHM systems with existing infrastructure and ensuring compatibility with other monitoring and maintenance systems. As a result, there is a need for collaboration between various stakeholders, including engineers, researchers, and industry professionals, to develop cost-effective and efficient SHM solutions that can be widely adopted. In addition to technical considerations,ethical and privacy concerns also need to be addressed in the implementation of SHM systems. The use of sensors to continuously monitor structures raisesquestions about the collection and use of data, as well as potential privacy implications. It is essential to establish clear guidelines and regulations regarding data collection, storage, and sharing to ensure that the rights and privacy of individuals are protected. Moreover, there is a need for transparency and communication with the public about the purpose and benefits of SHM, as wellas the measures taken to address any ethical and privacy concerns. Despite the challenges and considerations involved, the benefits of structural health monitoring are significant. By continuously monitoring the condition of structures, SHM systems can provide early detection of potential issues, allowing for timely maintenance and repairs. This proactive approach can help prevent catastrophic failures and ensure the safety of the public. Additionally, SHM can lead to cost savings by optimizing maintenance schedules and extending the lifespan of structures, ultimately resulting in a more sustainable and resilient built environment. In conclusion, structural health monitoring is a critical aspect of ensuring the safety and integrity of various structures. While there are several challenges and considerations to address, the benefits of SHM are significant in terms of safety, cost savings, and sustainability. By carefully planning and implementing SHM systems, and addressing technical, ethical, and privacy concerns, we can create a built environment that is more resilient and secure for future generations.。

A Thesis Submitted in Partial Fulfillment of the Requirementsfor the Degree for the Master of EngineeringDynamic Analysis and Vibration Sensing of Thin-Wall Annular WorkpieceCandidate : Liu WuguangMajor : Mechatronic EngineeringSupervisor : Prof. Lee Kok-MengAssoc. Prof. Guo JiajieHuazhong University of Science & TechnologyWuhan , Hubei 430074, P.R.ChinaMay, 2015独创性声明本人声明所呈交的学位论文是我个人在导师指导下进行的研究工作及取得的研究成果。

尽我所知，除文中已经标明引用的内容外，本论文不包含任何其他个人或集体已经发表或撰写过的研究成果。

对本文的研究做出贡献的个人和集体，均已在文中以明确方式标明。

本人完全意识到本声明的法律结果由本人承担。

学位论文作者签名：日期：年月日学位论文版权使用授权书本学位论文作者完全了解学校有关保留、使用学位论文的规定，即：学校有权保留并向国家有关部门或机构送交论文的复印件和电子版，允许论文被查阅和借阅。

本人授权华中科技大学可以将本学位论文的全部或部分内容编入有关数据库进行检索，可以采用影印、缩印或扫描等复制手段保存和汇编本学位论文。

保密□，在年解密后适用本授权书。

本论文属于不保密□。

（请在以上方框内打“√”）学位论文作者签名：指导教师签名：日期：年月日日期：年月日华中科技大学硕士学位论文摘要薄壁零部件（如航空发动机压气机盘）在加工时易受装夹力、切削力等因素影响，时空分布的工件变形和振动会导致加工质量和精度降低。

KIT 3628TModel M3628RStorage Bank DischargerAnd Resistor BankCustomer Reference ManualWeb: ●Tel: 615-244-2825 ●Email:*****************Bonitron, Inc.2Bonitron, Inc.Nashville, TNAn industry leader in providing solutions for AC drives.A BOUTB ONITRONBonitron designs and manufactures quality industrial electronics that improve the reliability of processes and variable frequency drives worldwide. With products in numerous industries, and an educated and experienced team of engineers, Bonitron has seen thousands of products engineered since 1962 and welcomes custom applications.With engineering, production, and testing all in the same facility, Bonitron is able to ensure its products are of the utmost quality and ready to be applied to your application.The Bonitron engineering team has the background and expertise necessary to design, develop, and manufacture the quality industrial electronic systems demanded in today’s market. A strong academic background supported by continuing education is complemented by many years of hands-on field experience. A clear advantage Bonitron has over many competitors is combined on-site engineering labs and manufacturing facilities, which allows the engineering team to have immediate access to testing and manufacturing. This not only saves time during prototype development, but also is essential to providing only the highest quality products.The sales and marketing teams work closely with engineering to provide up-to-date information and provide remarkable customer support to make sure you receive the best solution for your application. Thanks to this combination of quality productsandsuperior customer support, Bonitron has products installed in critical applications worldwide.Bonitron, Inc.3AC D RIVE O PTIONSIn 1975, Bonitron began working with AC inverter drive specialists at synthetic fiber plants to develop speed control systems that could be interfaced with their plant process computers. Ever since, Bonitron has developed AC drive options that solve application issues associated with modern AC variable frequency drives and aid in reducing drive faults. Below is a sampling of Bonitron’s current product offering.W ORLD C LASS P RODUCTSUndervoltage SolutionsOvervoltage SolutionsUninterruptible Power for Drives(DC Bus Ride-Thru) Voltage Regulators Chargers and DischargersEnergy StorageBraking Transistors Braking Resistors Transistor/Resistor ComboLine RegenerationDynamic Braking for Servo DrivesCommon Bus SolutionsPortable Maintenance SolutionsSingle Phase Power Supplies 3-Phase Power Supplies Common Bus DiodesCapacitor Formers Capacitor TestersPower Quality SolutionsGreen Solutions12 and 18 Pulse KitsLine RegenerationKIT 3628T & M3628R1.I NTRODUCTION (7)1.1.Who Should Use This Manual (7)1.2.Purpose and Scope (7)1.3.Manual Revision (7)Figure 1-1: KIT 3628T (7)1.4.Symbol Conventions Used in this Manual and on Equipment (8)2.P RODUCT D ESCRIPTION (9)2.1.Related Products (9)2.2.Part Number Breakdown (9)Figure 2-1: Example of Part Number Breakdown for KIT 3628T (9)Table 2-1: System Voltage Ratings for KIT 3628T (9)Figure 2-2: Example of Part Number Breakdown for M3628R (10)Table 2-2: Chassis Sizes for M3628R (10)2.3.General Specifications (10)Table 2-3: KIT 3628T General Specifications (10)Table 2-4: M3628R General Specifications (11)2.4.General Precautions and Safety Warnings (11)3.I NSTALLATION I NSTRUCTIONS (13)3.1.Environment (13)3.2.Unpacking (13)3.3.Mounting (13)3.4.Wiring and User Connections (13)3.4.1.Power Wiring (13)Table 3-1: M7001 Wiring Connections (14)Table 3-2: M7009 Wiring Connections (14)Table 3-3: P105/P115 Relay Wiring Connections (14)Table 3-4: HX200 Relay Wiring Connections (14)Figure 3-1: KIT 3628T Typical Wiring (15)Figure 3-2: Recommended External Temperature Sensor Placement for M3628R (16)4.O PERATION (17)4.1.Functional Description (17)4.2.I/O and Features (17)4.2.1.Indicators (17)4.3.Startup Procedure (17)5.M AINTENANCE, AND T ROUBLESHOOTING (19)5.1.Maintenance (19)5.2.Troubleshooting (19)Table 5-1: Troubleshooting Guide (19)5.3.Technical Help – Before You Call (19)6.E NGINEERING D ATA (21)6.1.Ratings (21)Table 6-1: KIT 3628T Ratings (21)6.2.Dimensions and Mechanical Drawings (21)Table 6-2: M3628R Chassis Dimensions (21)Figure 6-1: M3628R Chassis Dimensional Outline (21)Figure 6-2: M7009 24VDC Buffer Module Dimensional Outline (22)Figure 6-3: M7001 24VDC Power Supply Dimensional Outline (23)4Table of Contents Figure 6-4 P105/P115 Relay Dimensional Outline (23)Figure 6-5: HX200 Relay Dimensional Outline (24)5KIT 3628T & M3628RThis page intentionally left blank 6User’s Manual71. I NTRODUCTION1.1. W HO S HOULD U SE T HIS M ANUALThis manual is intended for use by anyone who is responsible for integrating, installing, maintaining, troubleshooting, or using this equipment with any capacitive energy storage system.Please keep this manual for future reference.1.2. P URPOSE AND S COPEThis manual i s a user’s guide for the KIT 3628T capacitor storage bank discharger. It will provide the user with the necessary information to successfully install, integrate, and use the KIT 3628T in a capacitive energy storage system.In the event of any conflict between this document and any publication and/or documentation related to the capacitive energy storage system, the latter shall have precedence.1.3. M ANUAL R EVISIONThe initial release of this manual is Rev 00a.Figure 1-1: KIT 3628TKIT 3628T & M3628R8 1.4. S YMBOL C ONVENTIONS U SED IN THIS M ANUAL AND ONE QUIPMENTEarth Ground or Protective EarthAC VoltageDC VoltageDANGER!DANGER: Electrical hazard - Identifies a statement that indicatesa shock or electrocution hazard that must be avoided.DANGER!DANGER: Identifies information about practices or circumstancesthat can lead to personal injury or death, property damage, oreconomic loss.CAUTION!CAUTION: Identifies information about practices or circumstancesthat can lead to property damage, or economic loss. Attentionshelp you identify a potential hazard, avoid a hazard, andrecognize the consequences.CAUTION!CAUTION: Heat or burn hazard - Identifies a statement regardingheat production or a burn hazard that should be avoided.User’s Manual2. P RODUCT D ESCRIPTIONCapacitors are taking the place of batteries as the preferred method of storing energy for many industrial applications that experience short term power outages. One major advantage capacitors have over batteries is that the energy can be completely drained for safer maintenance and shipping. The downside is that the capacitors can take a significant amount of time to self-discharge. It is for this purpose Bonitron developed the KIT 3628T capacitor discharger.KIT 3628T is a combination of a relay, 24V power supply, and 24V backup module. In conjunction with an appropriately sized M3628R discharge resistor. The system will discharge an attached capacitor bank to below 50V in 1 minute. Automatic discharge can be set up using an aux contact on the cabinet disconnect switch or breaker, or manual discharge can be triggered locally or remotely via PLC control.KIT 3628T is available in voltages up to 1000 VDC, and in peak currents up to 600 amps.Complementary M3628R resistors are available up to 4000 kJ, and can be paralleled for faster discharge times or for larger energy banks.2.1. R ELATED P RODUCTSM3460S ERIES R IDE-T HRU M ODULESVoltage regulators used for sag or outage protection of higher power systems.M3534S ERIES R IDE-T HRU M ODULESVoltage regulators used for sag or outage protection of lower power systems.M3528B ATTERY AND U LTRACAPACITOR C HARGERSChargers for high voltage storage strings.2.2. P ART N UMBER B REAKDOWNFigure 2-1: Example of Part Number Breakdown for KIT 3628TKIT3628T Y200B ASE M ODEL N UMBERV OLTAGE R ATINGC URRENT R ATINGB ASE M ODEL N UMBERThe base model number for all storage bank discharger kits is KIT 3628T.S YSTEM V OLTAGE R ATINGThe system voltage rating indicates the nominal voltage of the system the KIT 3628Tis intended to be a part of.Table 2-1: System Voltage Ratings for KIT 3628T9KIT 3628T & M3628R10C URRENT R ATINGThe current rating indicates the rated instantaneous peak current for the KIT 3628T, which is equal to the DC storage bus voltage divided by the M3628R resistance. For example, a 600 Amp peak KIT 3628T is 600.B ASE M ODEL N UMBERThe Base Model Number for all M3628 resistor banks is M3628R.R ESISTANCE V ALUEThe resistance value indicates the value of the M3628R resistor bank inohms. For example, a 3 ohm resistor bank is 03.0.K ILOJOULE R ATINGThe kilojoule rating indicates the energy rating of the M3628R resistor bank in kilojoules. For example, a resistor bank rated at 1500 kilojoules is 1500.C HASSIS S IZEThe chassis size indicates the chassis used for the M3628R. It is determined by the kilojoule rating of the M3628R.Table 2-2: Chassis Sizes for M3628R2.3. G ENERAL S PECIFICATIONSTable 2-3: KIT 3628T General SpecificationsTable 2-4: M3628R General Specifications2.4. G ENERAL P RECAUTIONS AND S AFETY W ARNINGSDANGER!CAUTION!ANY QUESTIONS AS TO APPLICATION, INSTALLATION OR SERVICE SAFETY SHOULD BE DIRECTED TO THE EQUIPMENT SUPPLIER.This page intentionally left blank3. I NSTALLATION I NSTRUCTIONSThe KIT 3628T is intended to be part of a larger variable frequency drive system and will require different hardware for interconnection based on the installation. An appropriate enclosure may need to be provided to protect personnel from contact and the systemfrom damage. The enclosure may also need to protect the equipment from theinstallation environment.Please read this manual completely before designing the drive system or enclosurelayout to ensure all required elements are included.3.1. E NVIRONMENTThe maximum ambient operating temperature of the KIT 3628T should not exceed50 C. Temperatures above this can cause overheating during operation.The M3628R discharge resistor is designed to dissipate energy in the form of heat andcan cause overheating of other components in the system if not cooled properly.Non-condensing, filtered air may be required to cool the system. This is especiallytrue if the M3628R is mounted inside the enclosure.3.2. U NPACKINGInspect the shipping crate of the KIT 3628T for damage.Notify the shipping carrier if damage is found.3.3. M OUNTINGMounting dimensions can be found in Section 6.The M3628R should be mounted with a clearance of 4 inches on each side with noequipment mounted above the resistor. Connection wires should not be above theresistor elements.3.4. W IRING AND U SER C ONNECTIONSReview this entire Section before attempting to wire the KIT 3628T.OWER IRINGDANGER!EBEFORE ATTEMPTING SERVICE OR INSTALLATION FAILUREDEATH!This section provides information pertaining to the field wiring connections of the KIT 3628T and M3628R. Actual connection points and terminal numbers of the storage capacitors will be found in the documentation provided with the drive system.Be sure to review all pertinent AC drive system documentation as well as the connection details listed below before proceeding.Table 3-1: M7001 Wiring ConnectionsTable 3-2: M7009 Wiring ConnectionsTable 3-4: HX200 Relay Wiring ConnectionsThe total wiring distance between the capacitor bank, KIT 3628T relay, and M3628R should not be longer than 50 feet. Wire in the discharge path can be sized for 2/3 of the peak discharge current (Vdc/R).Ensure correct polarity for the connection between M7001 and M7009, and the connection between the capacitor bank and M7001. Failure to do so can cause severe damage to the system.DANGER!AENSURE THEY ARE AT SAFE LEVELS BEFORE MAKING CONNECTIONSFigure 3-1: KIT 3628T Typical WiringFigure 3-2: Recommended External Temperature Sensor Placement for M3628RBOTTOMTOP*LOCAL SHUTDOWN TEMP SENSORTOPBACKFRONTACCESS PLATE FOR WIRINGVERTICAL MOUNTINGON BACK PLATEINSIDE RACK*REMOTE INDICATION TEMP SENSORLOCAL SHUTDOWN TEMP SENSORHORIZONTAL MOUNTING OUTSIDE, ON TOP OF CABINETACCESS PLATE FOR CONDUIT WIRINGUSE 0A220 (open at 220 c)WITH M3628T IGBT SWITCH*REMOTE INDICATION TEMP SENSOR1. USE TEFLON HIGH TEMP WIRE2. SECURE WIRES TO EXTERNAL MOUNT, PULLING THEM AWAY FORM CHASSIS DO NOT SECURE WIRE TO CHASSIS4. O PERATION4.1. F UNCTIONAL D ESCRIPTIONThe KIT 3628T discharger and M3628R resistor are used together to safely and quickly discharge a capacitor bank. The kit is composed of an M7001 power supply, a M7009 buffer module, and a normally-open relay. Since the M7001 runs directly off the capacitor bank, discharge can be accomplished even with no external power feed. During power-down conditions, the open relay and resistor are connected across the capacitor bank. When the capacitor bank begins charging and the voltage increases to approximately 190VDC, the M7001 power supply will begin operating, the power LED on its front panel turns on, and 24VDC is available at its output terminals. This 24VDC is fed directly to the M7009 buffer module. When the user wishes to initiate a discharge event, a user-defined contact closes the buffered 24VDC to the relay coil. This contact is typically either a PLC command or an auxiliary contact on the door handle.During a typical discharge cycle, the KIT 3628T would be enabled and the capacitor bank voltage would begin to decrease. When the capacitor bank voltage gets down to approximately 60VDC, the M7001 will deactivate. The M7009 will support the relay coil for approximately 60 additional seconds, allowing the system to complete discharging to a safe level.To prevent resistor overheating, a normally-closed temperature sensor may be connected in series with the relay coil.4.2. I/O AND F EATURES 4.2.1. I NDICATORS4.2.1.1. M7001 P OWER LEDThe Power light illuminates green when the internal power supply is operating.4.3. S TARTUP P ROCEDURECAUTION !∙∙DANGER !1. Ensure all system components are wired properly.2. Enable the charger, and allow capacitor bank voltage to increase.∙ Charge time is dependent on capacitor bank value and charge currentlevel.∙ M7001 “POWER” light should illuminate as DC bus increases past190VDC.3. After the capacitor bank has fully charged, remove power from charger and close24V to the discharge relay. This will turn the relay on and begin discharging the capacitor bank.∙Capacitor bank voltage should decrease.∙M3628R will begin to emit heat.4. M7001 will turn off when the capacitor bank voltage drops below approximately60VDC. M7009 will hold the relay closed for approximately 60 seconds past this point.5. M AINTENANCE, AND T ROUBLESHOOTING5.1. M AINTENANCEThe KIT 3628T and M3628R require no maintenance.5.2. T ROUBLESHOOTINGBelow are suggestions on how to check some common issues.If you continue to have problems after going over this list, please contact Bonitron.Table 5-1: Troubleshooting GuideCAUTION! RBEQUIPMENT BY PERSONNEL NOT APPROVED BY REMAINING5.3. T ECHNICAL H ELP –B EFORE Y OU C ALLIf possible, please have the following information when calling for technical help:∙Exact model number of affected units∙Serial number of unit∙Name and model number of attached drives∙Name of original equipment supplier∙Brief description of the application∙The AC line to line voltage on all 3 phases∙The DC bus voltage∙KVA rating of power source∙Source configuration Wye/Delta and groundingThis information will help us support you much more quickly. Please contact us at (615) 244-2825 or through This page intentionally left blankUser’s Manual216. E NGINEERING D ATA6.1. R ATINGS(1)calculated from the capacitor bank energy being discharged.6.2. D IMENSIONS AND M ECHANICAL D RAWINGSFigure 6-1: M3628R Chassis Dimensional OutlineKIT 3628T & M3628R22Figure 6-2: M7009 24VDC Buffer Module Dimensional OutlineUser’s Manual23Figure 6-3: M7001 24VDC Power Supply Dimensional OutlineFigure 6-4 P105/P115 Relay Dimensional OutlineKIT 3628T & M3628R24 Figure 6-5: HX200 Relay Dimensional OutlineUser’s Manual This page intentionally left blank25KIT 3628T & M3628RNOTES 26D_KIT_3628_CMAN_vall_00a 04/05/2016 521 Fairground Court ● Nashville, TN 37211 ● USATel: (615) 244-2825 ● Fax: (615) 244-2833 ● Web: ● Email:*****************。

重心法的英语一、“重心法”的英语：centroid method二、英语释义1. In mathematics and physics, the centroid method is a way to find the geometric center or average position of a set of points. It is often used in various fields such as engineering design, image processing, and logistics location problems.- 在数学和物理学中，重心法是一种找到一组点的几何中心或平均位置的方法。

它经常用于诸如工程设计、图像处理和物流选址问题等各个领域。

三、短语1. centroid calculation method 重心计算方法2. weighted centroid method 加权重心法3. geometric centroid method 几何重心法四、单词（与重心法相关的一些单词）1. centroid /ˈsentrɔɪd/ n. 重心；质心- The centroid of a triangle is the point where the medians intersect. 三角形的重心是中线相交的点。

2. method /ˈmeθəd/ n. 方法；办法- We need to find a more efficient method to solve this problem. 我们需要找到一种更有效的方法来解决这个问题。

3. location /ləʊˈkeɪʃn/ n. 位置；地点- The location of the new factory is determined by the centroid method. 新工厂的位置由重心法确定。

4 Concrete Placement Classes of Structural Concrete Concrete Plant InspectionPlantsTypesofTechnicianPlantPreparations for Concrete PlacementPlan ReviewPreparationsSiteRestrictionsWeatherConcrete DeliveryEquipmentDeliveryTicketsDeliveryField TestsRecording Test Results Concrete PlacementSegregationConcrete ConsolidationofJointsConstructionCasesSpecialFinishing Concrete SurfacesAreasFinishingBearingTreatments OtherSurfaceCuring Substructure ConcreteCold Water CuringMethod of Measurement and Basis for PaymentCHAPTER FOUR:CONCRETE PLACEMENTThe major topics to be discussed in this chapter are:1)Classes of concrete2)Concrete plants3)Preparations for concrete delivery4)Field tests5)Concrete placement6)Concrete finishing7)Curing methodsTechnicians are required to have a basic knowledge of concrete used inbridge construction and the plants that produce the concrete. Such anunderstanding helps to form a cooperative relationship between theTechnicians at the plant and the Technicians in the field.CLASSES OF STRUCTURAL CONCRETEStructural concrete is produced as Class A, Class B, and Class C. Thedifferences between the classes are in the cement and aggregate contentsand water/cementitious ratios.Bridge construction often requires the use of all three classes of concrete.For example, the plans may require Class B concrete for the footings,Class A for the piers and bents, and Class C for the decks and railings.The Notes section of the General Plan sheet lists the classes of concrete tobe used. The Bill of Materials section on the detail sheets also lists theseclasses. In most cases, the Contractor may substitute Class C concrete forClass A concrete and Class A or Class C concrete for Class B concrete.The materials used in the production of structural concrete includecombinations of the following:1) Fine and coarse aggregates2)Portland cement3)Fly ash (a coal by-product)4)Water5)Admixtures, including retarders, accelerators, waterreducers, and air entraining agentsCONCRETE PLANT INSPECTIONTYPES OF PLANTSINDOT categorizes concrete plants as captive plants or commercialplants. Captive plants are usually temporary plants (constructed on thejob-site) and are used primarily to produce concrete for a specificcontract. When the contract is finished, the plant is disassembled andmoved. Commercial plants (Figure 4-1), on the other hand, arepermanent installations.Figure 4-1. Commercial PlantConcrete plants are inspected and certified by the Office of MaterialsManagement. Commercial plants are inspected once a year. Captiveplants are inspected at the beginning of each construction season andwhenever they are moved to a new location.PLANT TECHNICIANThere are two reasons for having Technicians at concrete plants. The firstis to insure that INDOT receives the quality of materials the Contractorhas agreed to supply, and the second is to insure those materials aredelivered in the proper quantities.Plant Technicians are responsible for observing all weighing, batching,and mixing operations, except when mixing takes place away from theplant site. Technicians are required to ensure that all materials have beensampled, tested, and approved. The plant scales used for batching cementand aggregates are required to be checked for accuracy twice a day.Plant Technicians are required to maintain a cooperative relationship withthe Contractor and plant personnel. Being prepared for work and knowingthe requirements for the concrete are necessary to maintain thisrelationship.PREPARATIONS FOR CONCRETE PLACEMENTThe smooth delivery of concrete to the job-site is critical. Delays in thedelivery of the concrete or during the placement operation may causeproblems that are time consuming and costly to resolve. The fieldTechnician, the Contractor, and the concrete plant Technician are requiredto work together to ensure the correct concrete is delivered on time and inthe necessary quantities.PLAN REVIEWPreparing for the delivery of concrete begins with a review of the plans.The class or classes of concrete are required to be checked. The Bill ofMaterials section of the detail sheets is used to locate the estimatedquantities for each class of concrete. Substitutions of a higher class ofconcrete are generally allowed; however, the Contractor is never permittedto substitute an inferior class of concrete for the class of concrete requiredin the plans.The detail sheets also provide important information concerning theconcrete pour, such as the pour sequence and the locations and dimensionsof the construction joints and keyways.SITE PREPARATIONSThe Technician is required to ensure that the site has been adequatelyprepared for concrete placement. Such preparations include that:1)Excavations have been dewatered2)Forms have been checked for adequate bracing and properelevations and alignment3)Chamfer strips have been installed and are in good shape4)Trash and debris have been removed from all forms5)Reinforcement has been tied securely and checked forproper clearance and spacing6)The Contractor has adequate manpower and equipment tohandle the pour to include a sufficient number of vibratorsand backups.WEATHER RESTRICTIONSThe Technician is required to know the weather forecast for the concreteplacement operation. Weather conditions may influence everything fromthe timing and method of concrete delivery and placement to postponingthe operation altogether. Ideally, concrete is placed in temperaturesbetween 50 and 90° F, when there is no threat of rain, and when steps havebeen taken to protect the concrete from excessive wind.In general, when the temperature is 35° F or below, the temperature of theconcrete is required to be between 50 and 80° F at the time of placing.The Contractor may heat the water and/or aggregates used in the concretemix to achieve this range of temperatures; however, the heating is requiredto be done in accordance with the Specifications for cold-weatherconcrete. The Technician is required to use a dial thermometer to checkthe concrete temperature whenever the concrete is suspected to be near theSpecifications limits.CONCRETE DELIVERYNo concrete may be placed without a Technician on the job and anotherTechnician at the concrete plant. Prior to the beginning of concretedelivery, the Technician is required to contact the plant Technician todouble-check the following items:1) The class of concrete to be used2)The quantity of concrete needed for the pour3)The slump and air content requirements4)The proposed starting time of delivery5)The desired rate of delivery.DELIVERY EQUIPMENTConcrete is typically delivered to the job-site in mixer trucks, agitatortrucks, or in non-agitating equipment. All delivery trucks are required tocomply with the equipment Specifications designated in Section 702.Mixer trucks (Figure 4-2) are designed for mixing concrete at or on theway to the job-site. For this reason, mixer trucks always have a water tankon board and a measuring device that is capable of controlling the amountof water that is added to the mix. Agitator trucks deliver ready-mixedconcrete. Any water on the truck is for cleaning purposes only, not formixing.Figure 4-2. Mixer TrucksWhen mixer trucks are used, the following items are required to bechecked:1)Manufacturer’s rating plates are in place and legible2)Revolution counters are operating properly3)Mixing speed and the number of revolutions are incompliance with the Specifications. The number ofrevolutions of the drum at mixing speed is required to bebetween 70 and 100.4)Trucks are operated at or below their rated capacity5)Old concrete is removed from the drum6)Wash water is properly drained from the drumDELIVERY TICKETSAs the concrete is delivered to the job-site, the Technician collects adelivery ticket from each truck.When the concrete delivered to the job-site is produced at a commercial orcaptive plant, the Producer’s ticket is used to document delivery. TheProducer’s ticket for the first load of each class of concrete delivered eachday is required to contain the following information:1)The correct contract or project number2)The correct date3)The producer’s name4)The plant location5)The Contractor6)The class of concrete delivered7)The weights per cubic yard of all materials, includingadmixtures8)The number of cubic yards delivered9)The time of day that the water and cement were combined10)The plant Technician’s signatureAfter the first load is delivered, the Producer’s name, the plant location,the Contractor, and the weights of the various materials used in batchingmay be omitted from the Producer’s tickets for the rest of the day or untilthe mix design is changed. If the mix design is changed, the ticket for thefirst load of the revised mix is required to indicate all changes.FIELD TESTSConducting and/or observing concrete field tests is one of the mostimportant duties of a Technician. Typical field tests include slump, aircontent, yield, and water/cementitious ratio. The equipment used toconduct the tests is required to be clean, in good shape, and capable ofproviding accurate results. The air meters (Figure 4-3) used in air contenttests and the scales used in yield tests are required to be calibrated andapproved. All test equipment is provided by District Testing.Figure 4-3. Air Content MeterThe procedures for conducting slump, air content, flexural strength, andyield tests are detailed in the INDOT General Instructions to FieldEmployees. The required frequency for conducting all tests is listed in theFrequency Manual. That frequency of tests listed, however, is a minimumand may be increased as specified by the plans, the Special Provisions, orby the PE/PS. The following is a brief description of the purpose of eachtest.Slump tests are conducted to determine the consistency of fresh concreteand to check the uniformity of concrete from batch to batch. Typicalspecification limits for slump for structural concrete are between 1 and 4in. Unacceptable slump measurements usually indicate improper mixproportions, especially the water content. Contractors are not permitted toadd water simply to make the mix easier to pour. Any such change to themix design requires prior approval.Air content tests are conducted to determine how much air is contained inthe concrete. In most cases, air has been purposely added or entrained inconcrete to make the concrete more durable. Allowable air content mayrange from 5 to 8 %, depending on the maximum size of the aggregatesused in the mix. Results outside the specified limits indicate arequirement to adjust the amount of admixture in subsequent batches.Yield tests are conducted to determine the weight per cubic foot of freshconcrete which is used to determine the cement in barrels per cubic yard.Yield tests are not used to check the batching of any one mix component.Flexural strength is conducted to determine when forms and/or falseworkmay be removed from a structure or to determine when a structure may beput into service. This test requires placing fresh concrete in a beam moldand allowing the concrete to set and cure under the same conditions as theconcrete used in the structure. The concrete is then removed from themold and broken in a controlled environment by a beam breaker. The testresults may then be used to make certain assumptions regarding thestrength of the concrete used in the structure.Water/cementitious ratio is the ratio of the total amount of free water inthe aggregates, including all free water in the concrete, to the amount ofcement in the concrete.RECORDING TEST RESULTSThe results of the slump, air content, and yield tests are recorded andsubmitted on Form IT 652. Flexural strength tests are documented onForm IT 571A, and water/cementitious ratio is recorded on Form IT 628. CONCRETE PLACEMENTCompared to the preparations leading up to the pour and the testing, theactual placement operation is relatively simple; however, there are stillitems that may go wrong.SEGREGATIONSegregation occurs when the coarse and fine aggregates used in theconcrete separate and become unevenly distributed throughout the mix.The larger coarse aggregate sinks to the bottom while the fines rise to thetop. Segregation always leads to an inferior quality of concrete. For themost part, however, segregation may be prevented with the use of properplacement equipment and techniques.Concrete is required to be placed as close as possible to the location theconcrete occupies in the structure. The concrete should not be dumped ina central location and then spread to the location required in the structure(Figure 4-4).Figure 4-4. Concrete PlacementWhen possible, concrete is required to be deposited in layers no more than24 in. thick (Figure 4-5). Care is required be taken, however, to placeeach successive layer before the preceding layer has taken the initial set.This initial set is usually 45 minutes to an hour, depending on thetemperature. Too much time between the placement of layers usuallyresults in a cold joint which is a weak line of separation between thelayers.Figure 4-5. Concrete LayersDropping concrete from too great a height causes the finer particles in themix to splash away from the larger, heavier particles. In addition, theforce of the mix striking the reinforcing steel may shift bars out ofposition. The maximum drop height or allowable free fall is 5 ft. Hopperswith flexible chutes called tremies are required to be used to funnel themix down into tall, narrow forms. Workers may be stationed inside theforms to move the chutes around to ensure an even distribution of theconcrete. The hoppers may not rest on the reinforcing steel and arerequired to be supported by the formwork.CONSOLIDATION OF CONCRETEFresh concrete naturally contains air pockets or voids. If the concretewere left that way, the finished product would have a rough surface andhave questionable strength. To eliminate voids and to ensure a good bondto the reinforcing steel, the concrete is required to be consolidated to auniform density.The most common method of consolidating concrete is by vibrating theconcrete with a portable spud type vibrator (Figure 4-6). Most vibratorshave an effective radius of 18 in. all around. Once the vibrators areinserted, they consolidate an area approximately 3 ft in diameter.Figure 4-6. Spud VibratorAlthough the procedure is a simple operation, vibrating concrete is oftenconducted incorrectly. Some points that ensure a good job include:1)Vibrating is required to be done immediately as theconcrete is placed2)Vibrators are required to be inserted and withdrawnvertically and should not be dragged through the concrete3)Vibrators are required to be inserted and withdrawn within5 seconds. Over-vibrating forces the finer aggregates to thetop and drives the larger aggregates toward the bottom.4)When concrete is poured in layers, the head of the vibratoris required to penetrate through the top layer and partiallythrough the layer underneath (Figure 4-7).Figure 4-7. Depth of Vibration5)The workers are required to avoid contacting thereinforcing steel with the vibrator so that the bond betweenthe steel and the concrete is not broken6)The workers are required to avoid contacting the formwalls with the vibrator as that may loosen the forms andmay also cause honeycombing of the concrete surface7)The Contractor is required to have a backup vibrator onhand for larger pours in case of equipment problems CONSTRUCTION JOINTSThe purpose of a construction joint is to join a section of fresh concrete toa previously poured section that has already set. Construction joints arenecessary when a substructure unit is too large to pour in one continuousoperation or when rain, equipment problems, or other conditions interruptthe pour. Unless construction joints are specifically called for on theplans, the Contractor is required to have written permission to use thejoints. Some joints, however, may be described on the plans as optional, and may be used at the Contractor’s discretion. In addition, the Contractor may request the relocation or elimination of construction joints. Such a change is required to be approved by the PE/PS.To make a construction joint, either planned or unplanned, the Contractor is required to form a raised keyway or keyways in the section to be poured later. After the first section has hardened, the surface is swept clean with wire brooms and kept wet. If a Type A construction joint is specified (Figure 4-8), the surface is notched between the reinforcement. Immediately before the fresh concrete is placed, the Contractor draws the forms up tight against the concrete in place. To improve the bond between the sections, the Contractor may apply a bonding epoxy to the exposed surface before resuming the pour.Figure 4-8. Type A Construction JointTo resist shear and other forces, construction joints in footings and in abutments for arch bridges are required to be vertical. Horizontal construction joints are used in walls and columns. Joints that are exposed to view are required to be constructed straight, clean, and watertight. This is done by finishing the concrete with the underside of a straightand level strip of wood that is nailed to the form at the proper elevation.No construction joint is allowed to be made in areas where the reinforcingsteel has been spliced. Section 702.15 includes all of the requirements forthe use of construction joints. Sheet 724-BJTS-05 of the StandardDrawings for Bridges provides additional details.During the pour, care is required to be taken not to disturb the position ofany reinforcing steel that is used to tie the section being poured to asection that is poured later. If the bars are displaced, they are required tobe re-tied immediately in the proper position. Concrete that is splashed onthese bars is required to be cleaned off before the next section is poured toensure a good bond. If the steel is exposed to the weather for some timeafter the pour, the steel may require coating with a cement paste orequivalent to prevent the steel from rusting.SPECIAL CASESThe placement of concrete under water in footings and for foundationseals requires techniques outside the general rules given above. For acomplete explanation of the methods and requirements, see Section702.20(d-f).FINISHING CONCRETE SURFACESUnless otherwise authorized, the surface of the concrete is finishedimmediately after form removal. Only the minimum amount of coveringnecessary to allow finishing operations to be done is removed at one time.Subject to approval, metal ties may be left in the concrete for the purposeof supporting or bracing subsequent work. Such ties are required to be inaccordance with Section 702.13(b) and be of a type which uses a cone androd as both spreader and tie. Before final acceptance of the work, thecones are removed and the cavities filled, in accordance with Section702.13(b).All concrete surfaces are required to be given a finish immediatelyfollowing the removal of any forms.The concrete surfaces of pier and bent caps, the front face of mudwalls,and any other concrete surfaces specified are required to be sealed. Thematerial used for sealing is required to be in accordance with Section 709.The seal is applied to obtain a finished film thickness of at least 250 µm(10 mils). Mixing, surface preparation, and the method of application arerequired to be in accordance with the manufacturer's recommendations;however, the surfaces to be sealed are prepared in accordance with Section709 prior to applying the sealer.At the time of the removal of forms, the concrete surface is scraped toremove all fins and irregular projections. The surface is then powerground to smooth all joints and chamfers.After grinding is completed, a paste of grout is applied to the concretesurface with a sponge float to fill all air holes and small irregularities. Thepaste grout is required to be 6 parts of pre-mix mortar mix for masonryand 1 part white portland cement in accordance with ASTM C-150, Type1.After the paste grout takes the initial set, the surface of the concrete isscraped with a steel drywall knife to remove the paste from the surface. FINISHING BEARING AREASThe bridge seats and areas in between require special treatment at thefinishing stage. The tops of the bridge seats are required to be finished atexactly the right elevation and be completely level to ensure full contactwith the bottom of the bearing device. The areas in between the bridgeseats are required to be sloped or crowned slightly to ensure adequatesurface drainage. Both results are obtained through proper finishingtechniques.OTHER SURFACE TREATMENTSThe plans or Special Provisions may provide for the use of surfacetreatments other than the three classes of concrete finishes. For example,the plans may require the Contactor to leave a rough surface texture ofexposed aggregate. This may be done by blasting off the surface mortarwith a high-pressure water hose.The surfaces of pier and bent caps, the front face of mudwalls, and anyother areas specified are required to be sealed against moisture penetrationwith an approved concrete sealer. The surfaces to be sealed aresandblasted to remove form oil and other foreign matter and are requiredto be completely dry before the application. The sealant is applied in acriss-cross pattern and is required to be a thickness of 10 mils. No sealedsurface is rubbed. Section 709 describes this operation in detail.CURING SUBSTRUCTURE CONCRETEOnce the concrete is in place, the concrete is allowed to cure a certainamount of time to achieve the full strength. During the curing period, theconcrete is not to be placed under stress. The typical curing period of theconcrete is 96 hours after the initial set. The use of certain materials, suchas fly ash or Portland-pozzolan cement in the concrete, increases thecuring period to 120 hours.The Specifications describe two methods of curing concrete. The first iscalled the protective covering curing method. This method requirescovering the surfaces to be cured with canvas, straw, burlap, sand, or otherapproved material and keeping the concrete wet with water throughout thecuring period. The water prevents the concrete from drying out tooquickly. Surfaces that require a Class Two rubbing finish are required tohave the protective covering temporarily removed to allow the rubbing tocontinue. The covering is required to be restored as soon as possible.The other curing method requires the use of a membrane forming curingcompound. The curing compound may be applied after the concretesurface has received the specified finishing treatment. Up until then, theconcrete is required to be protected by the protective covering method or,in the case of vertical surfaces, simply by leaving the forms in place.Curing compound is applied at a minimum rate of one gallon for every150 ft2 of concrete surface. The application is done in two stages. Thefirst coat is applied immediately after stripping the forms or uponacceptance of the concrete finish. The surface is required to be wettedwith water and coated with the compound as soon as the water filmdisappears. The second application is required to begin after the first hasset and according to the manufacturer’s directions. During the curingoperation, all untreated areas are required to be kept wet.Finally, the plans may call for certain areas to be waterproofed. When theapplication of waterproofing material begins, curing of those areas is nolonger required.COLD WATER CURINGIn cold water weather (35° F and below), the Contractor is required tokeep the freshly poured concrete and the forms within a protectiveenclosure or covered with approved insulation material that is at least 2 in.thick. The air inside the enclosure or under the insulation is required to bekept above 50° F for at least 72 hours. If for any reason the temperaturedrops below 50° F within the enclosure, the heating period is required tobe extended. When dry heat is used to maintain the required temperature,the Contractor is required to devise a means of providing enough moisturein the air within the enclosure to prevent the concrete from drying out tooquickly. Heaters may be used to maintain the required temperature if theyprovide continuous operation and the Contractor has taken adequate fire-prevention and safety measures.When rubbing the concrete is required, the forms are removed and therubbing conducted during the protection period. Again, if this means thatthe concrete is exposed to temperatures below 50° F before the required72 hours has transpired, the period of protection and heating is required tobe extended.METHOD OF MEASUREMENT AND BASIS FOR PAYMENTConcrete is measured and paid for by the cubic yard placed in accordancewith the plans or as directed. Forms, falsework, and other miscellaneousitems required to complete the work are not paid for separately. The costsfor these items are included in the costs of the concrete.。

英语教学法术语achievement test 成绩测试acquisition 习得，语言习得active vocabulary 积极词汇，主动词汇affective filtering 情感筛选aim, objective 目的，目标analysis of errors 错误分析applied linguistics 应用语言学associative learning 联想性学习auditory perception 听觉audio-lingual method 听说法audio-visual method 视听法aural-oral approach 听说教学法，听说法aural-oral method 听说法basic knowledge 基本知识basic principle 基本原则basic theory 基本理论basic training 基本训练basic vocabulary 基本词汇behaviourism 行为主义bilingual 双语的bilingual education 双语教育blank filling 填空choral repetition 齐声照读，齐声仿读class management 课堂管理cloze 完形填空cognitive approach 认知法communicative drill 交际性操练communicative exercise 交际练习communicative language teaching 交际派语言教学法，交际教学法community language learning 集体语言学习法comparative method 比较法communicative approach 交际法comprehensible input 可理解性输入comprehensive method 综合法controlled composition 控制性作文course density 课堂密度course design 课程设计cramming method 灌输式cue word 提示词curriculum 课程，教学大纲curriculum development 课程编制，课程设计deductive learning 演绎性学习deductive method 演绎法demonstration 演示demonstration lesson 示范教学describe a picture in writing 看图说话describe a picture orally 描写语言学diagram 图解dicto-comp 听写作文direct method 直接教学法educational objective, aim 教育目的EFL 英语作为外语English as a Foreign Language 英语作为外语English as an InternationalLanguage 英语作为国际语言English environment 英语环境ESOL English for Speakers of Other Languages 供非英语民族使用的英语ESL Programme（English as a Second Language Programme）英语（第二语言）教程evaluation 评价examination 考试experimental method 实验法extensive reading 泛读extra-curriculum activity 课外活动extra-curriculum club, group 课外小组facial expression 面部表情feedback 反馈film projector 电影放映机filmstrip 电影胶片final stage 高级阶段first language 第一语言，母语frequency of word 词的频率gestalt style 格式塔式（学习），整体式（学习）gesture 手势getting students ready for class 组织教学global learning 整体式学习，囫囵吞枣式学习grammar lesson 语法课grammar translation method 语法翻译法group reading 集体朗读group training 集体练习guided composition 引导性作文heuristic method of teaching 启发式教学法heuristics 启发法；探索法humanistic approach 人本主义教学法idealism 唯心主义imitation 模仿immersion programme 沉浸式教学imparting knowledge 传授知识incomplete plosive 不完全爆破individualized instruction 个别教学individual training 个别练习inductive learning 归纳性学习inductive method 归纳法information processing 信息处理inner speech 内语言语in-service training 在职培训integrated approach 综合教学法，综合法intensive reading 精读intermediate stage 中级阶段interpretation 头口翻译International Phonetic Alphabet 国际音标juncture 连读，音渡key words 基本词，关键字kinesics 身势语，身势学knowledge structure 知识结构language acquisition device 语言习得机制language competence, or knowledge 语言知识language learning capability 语言学习能力language laboratory; lab 语言实验室language pedagogy 语言教育language performance 语言行为language test 语言测试learning by deduction 演绎性学习learning by induction 归纳性学习learning process 学习过程learning style 学习方式lesson plan 课时计划，教案lesson preparation 备课lesson type 课型linguistics 语言学linguistic competence 语言能力long-term memory 长期记忆meaningful drill 有意义的操练meaningful exercise 有意义的练习meaningful learning 理解性学习means of teaching 教学手段mechanical drill 机械操练mechanical exercise 机械练习mechanical memory 机械记忆mechanical translation 机器翻译memory 记忆，记忆力memory span 记忆幅度memorizing 用记记住method 方法methodology of teaching English 英语教学法microteaching 微型教学minimal pair 最小对立体（一种辨音练习）modeling 示范教学monitor hypothesis 语言监控说mother tongue 母语motivation 引起动机native language 本族语natural approach 自然教学法，自然法natural method 自然法needs analysis 需要分析notional approach 意念法notional-functional syllabus 意念-功能派教学大纲observation lesson 观摩教学oral approach 口语教学法，口语法oral exercise 口语练习order of acquisition 语言习得顺序organization of teaching materials 教材组织organs of speech 发音器官pair work 双人作业，双人练习passive vocabulary 消极词汇pedagogy 教育法peer teaching 同学互教perception 知觉placement test 分班测验principle of communication 交际性原则principle of teaching 教学原则productive exercise 活用练习productive vocabulary 活用词汇proficiency 熟练program designing 课程设计qualified teacher 合格教师question band 试题库questionnaire 调查问卷questions 提问rate of reading 阅读速度readability 易读性read by turns 轮读reading 阅读reading lesson 阅读课reading speed 阅读速度reading vocabulary 阅读词汇，阅读词汇量regression 回看，重读reinforcement 巩固reinforcement lesson 巩固课repetition drill 复述操练repetition-stage 仿照阶段response 反应retelling 复述review; revision 复习review（revise）and check up 复习检查review（revision）lesson 复习课rewriting 改写rhythm 节奏role-play 扮演角色rote learning 强记学习法，死记硬背scanning 查阅，扫瞄school practice 教学实习seminar 课堂讨论sentence completion 完成句子short-term memory 短期记忆sight vocabulary 一见即懂的词汇silent reading 默读silent way 沉默法，静授法simplification 简写simplified reader 简写读本simulation 模拟，模拟性课堂活动simultaneous interpretation 同声翻译situational method 情景法situational language teaching 情景派语言教学法，情景教学法situational syllabus 情景派教学大纲skimming 略读，济览slide 幻灯片slide projector 幻灯片soft ware 软件speech reading 唇读法speed reading 快速阅读，快读spelling 正字法spiral approach 螺旋式教学法，螺旋法spoken lauguang 口语stick drawing; match drawing 简笔画stimulus and response 刺激与反应stress accent 重音，重读structuralism 结构主义（语言学）structural method 结构法student-centered 学生中心student-centered learning 学生为主学习法student teacher 实习教师student teaching 教育实习submersion programme 沉浸式教程substitution 替换substitution table 替换表suggestopedia 暗示教学法syllabus 教学大纲syllabus design 教学大纲设计syllabus for middle school English 中学英语教学大纲synthetic approach 综合性教学法，综合法target language 目的语，译文语言teacher’s book教师用书teacher’s manual 教师手册teaching experience 教学经验teaching objective / aim 教学目的teaching procedure 教学过程teaching words in isolation 孤立教单词theory of teaching 教学理论time allotment 时间分配total physical response method 全身反应法transformation drill 转换操练translation method 翻译法transformational generativegrammar 转化生成语法unconscious 潜意识undergraduate 大学本科生undergraduate course 大学本科课程undergraduate school 大学本科学院upperclassman 高年级学生verbal association 词语联想video 电视，影象videotape 录象磁带visual perception 视觉visual aid 直观手段visit a class 听课visual memory 视觉记忆vocabulary control 词汇控制word association 词际联想word list 词表word study 词的研究word frequency 词汇重复率written language 书面语。

T e c h n i q u e s for S e n s o r-B a s e d DiagnosisMark S. Fox Simon L o w e n f e l d & Pamela K l e i n o s k yIntelligent Systems Laboratory, The Robotics Institute Westmghouse E lectric CorporationCarnegie Mellon University, Pittsburgh, Pennsylvania Pittsburgh, PennsylvaniaA B S T R A C TThis paper describes a system called PDS, a forward chaining, rule-based architecture designed for the online, realtime diagnosis of machine processes. Two issues arise in the application of expert systems to the analysis of sensor-based data: spurious readings and sensor degradation. PDS implements techniques called retrospective analysis and meta-diagnosis as solutions to these problems. These techniques and our experiences in knowledge acquisition in a large organization, and the implementation of PDS as a portable diagnostic tool are described.1 I n t r o d u c t i o nResearch in the field of Al diagnosis systems has been evolving rapidly since the first event based (Nelson, 1982) or surface (Hart, 1982) reasoning systems (Shortliffe, 1976; Pople, 1977; Fox & Mostow, 1977; Duda et al., 1978), to systems that have functional or deep knowledge of their domain (Davis et al., 1982; Genesereth, 1982; Underwood, 1982; McDermott & Brooks, 1982). Whatever the style of diagnosis, these systems assume that information is provided manually through a question asking/answering dialogue, or automatically by means of sensors, or other devices. In both cases, the information is handled in the same manner, which, we have found, should not always be the case. In applications where the sources of information may be errorful (e.g., sensors), we found that it is just as important for a diagnostic system to reason abo ut the sources of its information and their veracity, as it is to perform diagnosis based on the information.During the summer of 1981, we began the design and construction of a rule-based architecture, called PDS, for the on-line, realtime diagnosis of malfunctions in machine processes. Diagnoses would be based on information acquired from tens to hundreds of sensors attached to a process. During the applicationof PDS, a number of sensor related problems arose. First, the process sensors in our applications degrade over time, reducing their diagnostic value. Second, a properly operating sensor may provide spurious readings periodically due to factors exogenous to the process. Though the frequency of such malfunctions are small, their detection may result in substantial savings. For example, in the electrical power utilities, replacement costs of electricity lost due to sensor malfunction averages $500,00/year per plant (Meigeret al., 1981). Any diagnosis system which is to receive Its information directly from devices which possess these characteristics must be able to handle the information without providing incorrect diagnoses, or at least have its diagnosis degrade gracefully with the sensors. As a result, PDS was extended to deal gracefully with these problems. These extensions are the topics of this paper.In the following, the architecture of PDS is described. We then examine the facilities provided by PDS to deal with sensor problems. Following, we discuss our experiences in the acquisition and testing of knowledge in PDS, and describe its implementation.2 PDS: Basic R e p r e s e n t a t i o nPDS is a forward chaining rule-based system implemented in the Schema Representation. Language SRL (Fox, 1979; Wright & Fox, 1982). The representation and propagation of belief is similar to that found in MYCIN (Shortliffe, 1976). For each rule, there are schemata describing each constituent part of the rule's antecedent (or evidence), a schema describing the rule's consequent (or hypothesis), and a schema describing the relationship between the rule's evidence and hypothesis. The implementation of rules as schemata results in an inference net similar to that found in PROSPECTOR (Duda et al., 1978) Sensor readings, hypotheses, and malfunctions are represented as sub-types of the p d s-n o d e schema (figure 1).{{ p d s-n o d eM B: "level of belief in the node being true"M D: "level of disbelief in the node being true"C F: "level of certainty = mb - m d"SUPPORTING RULE S: "rules for which this node is hypothesis"SUPPORTED-RULES: "rules for which this node is evidence"SIGNAL: "contains signal schema name(s)"DESCRIPTION: "E nglish description of the node"HAS-IS-A: (or sensor hypothesis malfunction)}}F i g u r e 1: Generic Node in PDSThe following specializations (see HAS-IS-A slot) of p d s-n o d e s are defined:1. s e n s o r schemata represent the actual sensors. Twoadditional slots are defined to store the currentreading: R E A D I N G-V A L U E and RE ADING-TIME.M. Fox et a l. 1592. m a l f u n c t i o n schemata correspond to those states of the physical system which are indicative of the problem(s) to be diagnosed.3. h y p o t h e s i s schemata represent intermediate conclusions in the inference net.A b e l i e f -r u l e schema (figuie 2) represents those rules which are used by the inference program to propagate belief.{{ signalM I N R A N G E : <constant> MAX RANGE : <constant>MESSAGE: "text to be displayed" ACTION: <function-name>DESCRIPTION: "English text describing the signal" }}EVIDENCE is represented as a Boolean combination of nodes. The combination is explicitly represented by a n d , or, and not schemata. Each b e l i e f -r u l e has associated with it one or more contexts which are also represented as a Boolean combination. If the context is evaluated to be true, then the rule will be included in the diagnostic process.Belief propagation is similar to the one used by MYCIN. To summarize, each hypothesis and malfunction has associated with it a measure of belief MB, and disbelief MD. These measures are altered by the evaluation of supporting belief-rules. In a belief- rule, the sufficiency of evidence is determined by evaluating the contents of the SF-FUNCTION slot. The necessity of evidence is determined by evaluating the contents of the NF-FUNCTICN slot. Table 1 shows the direction of change of beliefs and disbeliefs as determined by evidence CF (the CF of disjunctive evidence (OR) is the maximum of the constituents, and the CF of conjunctive evidence is the weighted average of constituents (AND) or the mimimum of constituents (FAND)), and the rule's SF and NF (an up-arrow indicates an increase, a dash indicates no change).Actual updating is performed using MYCIN'S updating rule (e.g.,A p d s -n o d o may have one or more s i g n a l s attached to it (Figure 3) When the node with which a s i g n a l is associated has a certainty factor CF in the range specified in the signal definition, then the text in the ME SSAGE slot is displayed and the function whose name is in the action slot is evaluated (this may trigger an alarm, for example).Figure 3: s i g n a l SchemaA c o n t e x t (Figure 4) specifies an external condition which may be relevant to the execution of the diagnostics. For'example, when diagnosing a machine, different rules may apply at startup than found during normal operation. For each rule, a Boolean of contexts can be defined, such that the rule will fire only when the context specification is true.{{ c o n t e x tVALUE: <true| false |0| 1>DESCRIPTION: "English description of this context"}}Figure 4: c o n t e x t SchemaAs found in most knowledge representation systems, SRL provides the user with the ability to define type hierarchies of schemata using the is-a relation, and instances of types using the i n s t a n c e relation. Rules may refer to types of hypotheses, sensors, and malfunctions, rather than instances. When the knowledge network for a particular machine is created, instances of sensors, etc. inherit rules attached to nodes in their related type hierarchy. A library of rules may be created for different types of machines, and instantiated with little effort.160 M. F o x et a l.3 R e t r o s p e c t i v e A n a l y s i sSpurious readings do occur often enough in sensors to requiretheir detection and omission from the diagnostic process. Thesereadings may be due to factors exogenous to the process, or tosensor malfunction. Spurious readings are handled in mostdiagnostic systems before they reach the system: the readings aresmoothed or omitted, manually or by a preprocessor associatedwith the sensor. It quickly became apparent in our applications thatsuch an approach was not sufficient. First, external modification ofreadings prohibit the system from performing other types ofanalyses not anticipated in the design of the sensor and its pre-processor. Second, we found that retrospective analysis of theunaltered sensor data was important in order to refine the rulebase.Solutions to this problem were found to have much in commonwith other diagnostic techniques. In particular, a variety of timeseries analyses was found to be important. Rate of change (firstderivative), averages, filtering, and curve smoothing are examplesof the kinds of time domain analysis employed both at the front endof diagnostic systems, and during the diagnosis itself.To provide general retrospective analysis support, PDS providesthe ability to store and analyze successive readings of a sensor, orthe successive values of any other node. A r e a d i n g-s e t schema(figure 5) acts as a memory for PDS. Successive readings/valuesare stored in the RE ADING-LIST slot. Information about a reading,e.g., time of reading, may be attached directly to it using SRL'sfacility for attaching meta-schemata to a schema, slot, and/orvalue.Another type of rule, called r e a d i n g-t r a n s f o r m (figure 6), is thelink between some input node, usually, but not necessarily as o n s o r node, and a r e a d i n g-s e t node. When ar e a d i n g-t r a n s f o r m rule fires, its T R A N S F O R M function is applied tothe value found in the rule's EVIDENCE node. If the evidence node isa s e n s o r, the value is a sensor READING, otherwise it is the node'scertainty factor. The transform is useful for such things asconversions of engineering units, scaling, etc. The result of thetransformation is automatically appended to the RE ADING-LIST of ther e a d i n g-s e t node. When all the rules leading to a r e a d i n g-s e tnode have fired, the FUNCTION specified in the reading-set schemais applied to the RE ADING-LIST. This function does all the requiredsignal processing and the result is placed in the STATISTIC slot ofthe reading-set node The STATISTIC can, from this point on, beused as a "normal" sensor reading.4 M o t a-D i a g n o s i sThe sensor intensive applications of PDS share the problem ofsensor degradation; environments containing corrosive chemicalsand widely varying temperatures can reduce sensor performance.This problem has been solved partially at the machine level by theplacement of redundant, overlapping sensors. But at the diagnosislevel, the problem of recognizing and removing malfunctioningsensors from the diagnostic process has received little attention inthe Al literature.If the rules which used sensors as evidence only referred to asingle sensor, then it might be possible to summarize theacceptability of a sensor's reading by the CF associated with it.Setting the CF to zero would stop propagation of any belief ordisbelief by the supported rule. (The change in belief provided by arule is the product of either the rule's SF or NF and the evidence'sCF.) But in many applications sensors may be combined asevidence in a single rule. Using CF'S to reflect sensor degradationin such systems would have an unexpected result. Consider thefollowing:1.lf evidence is conjunctive and the fuzzy minimumoperator (FAND) is used to derive the evidential CF,then the existence of a zero CF would remove the entirerule though the other sensors may be operating andoverlap the redundant sensor.2. If evidence is conjunctive and a weighted average isused to derive the evidential C F, then the sensor(s) witha zero CF would reduce the rule's change in beliefsince its weight is not reduced (PDS uses a weightedaverage to derive the evidential C F).Neither approach solves the problem satisfactorily.We call the solution implemented in PDS meta-diagno sis. Thefirst step in meta-diagnosis is the detection of sensor degradation.This is accomplished through the use of rules which monitor asensor's behavior. The ultimate consequent of these rules is one orM. Fox et al. 161more sensor m a l f u n c t i o n schemata. The second step is the adaptation of rules to reflect the reduction in importance of amalfunctioning sensor in the diagnostic process. This is accomplished through the introduction of a p a r a m e t r i c -a l t e r a t i o n rule (figure 7). This rule provides the capability of altering the definition of any other node or rule in the system. A p a r a m e t r i c -a l t e r a t i o n rule may be attached to any node. When the E VIDE NCE SLOT in the node changers, the rule's TRANSFORM function is applied to the value, possibly altering the HYPOTHESIS schema.{{ c o m p o s i t e -s e n s o r is-A: pds node SENSOR: "list of sensors whose readings are to be combined" TRANSFORM: "function which generates a composite reading"READING-VALUE:}}Figure 8: c o m p o s i t e -s e n s o r schema[{ p a r a m e t r i c -a l t e r a t i o n is-A: ruleEVIDFNCE: "the schema monitored by this rule"EVIDENCE SLOT: "the slot whose value is monitored by this rule" TRANSFORM: "function whose result is placedin the hypothesis slot"HYPOTHESIS: "schema altered by this rule"HYPOTHESIS-SLOT: "slot in hypothesis altered by this rule" }}Figure 7: p a r a m e t r i c -a l t e r a t i o n SchemaAn example of the use of this type of rule is the reduction of a sensor's weight in the conjunctive evidence of a belief-rule. It is important to note that this is a metarule in the sense that it alters the rule base, while not performing actual diagnosis. Such rules must be used carefully as they may introduce cycles in the propagation of belief. 5 C o m p o s i t e S e n s o r sAn alternative method of analyzing and reacting to readings from redundant overlapping sensors, which falls short of altering the rule base as provided by meta diagnosis, is to combine multiple sensors into a single composite sensor. In many domains, techniques such as:• voting, and• auctioneering (i.e., reading)ignoring the lowest or highest are used to combine sensor readings. PDS supports the exploration of these types of techniques by means of a c o m p o s i t e -s e n s o r schema (figure 8).The TRANSFORM slot contains a lisp function which analyzes the individual readings of the sensors listed in the SENSOR slot, and fills the RE ADING-VALUE , M B , and MD slots with the composite reading. Standard functions can be provided to implement voting, auctioneering, and other multiple sensor techniques. 6 K n o w l e d g e A c q u i s i t i o n and T e s t i n gA number of systems capable of performing on line diagnosis in an industrial environment are under development at the Westinghouse E lectric Corporation. A working prototype of one ofthese systems is being installed and tested at this time and is the subject of the following discussion. Although many of the steps in the development of this prototype arc common ones to the creator of an expert system, they often brought unexpected results when carried out in the commercial environment of a single large company.The knowledge engineer responsible for the generation of rules for this system was actually a team. The person writing the rules was an engineer whose background was in the problem being diagnosed (a quasi-expert) and whose acquaintance with PDS consisted of a working knowledge of the general principles and those structures which applied specifically to the problem at hand. Working in close conjunction with this "quasi-expert" was an engineer whose background was in expert systems and who knew PDS in depth.As with the development of most systems of this type, the first step was the creation of a small system that we called a "root system". It began with ten sensors and used forty-four rules and twenty-nine intermediate hypotheses to indicate seven malfunctions. It took approximately one month to develop and test. This "root" served us in two ways. First, it acted as a vehicle for eliciting information and stimulating the thinking of the experts. Secondly, it was a tool to sell the experts on the feasibility of what they were being asked to undertake and, more importantly, to sell upper management on the probability of a return on what they were being asked to invest.The testing scheme developed for the "root system" was a general one, and with some expansion was used for the prototype itself. First, approximately 150 sets of test data were generated in four groups. Group one tested the interactions of the rules themselves. Group two tested the system's response to malfunctions previously diagnosed by the experts. Group three tested the experts' response to malfunctions diagnosed by the system. Group four comparison-tested the responses of the experts and the system to data neither had "s e e n " before.The second phase of the testing is currently under way, and involves the installation and operation of the sensors used in the diagnosis system in an industrial environment. The resulting data are fed directly to the diagnosis system and the presence of any malfunctions are noted, recorded and displayed.Preliminary results of this testing are promising. Although mathematical analysis of these results has not yet been performed, me expert analyses of the data have agreed quite closely with the computer analyses.162 M. Fox et al.No development project is without its problems and this one has been no exception. Our first major problem was in convincing the necessary experts to participate in the project. Older and more established engineers, most useful as exports, have been traditionally wary of computer systems and the perceived possibilities of being replaced by a machine. Also the necessary experts were distributed throughout the company, raising many organizational problems.Once a sufficient number of experts agreed to participate in the project, we discovered the second major difficulty; problems within the cognitive models of the experts themselves. When the experts were questioned about the if-then rules used in their thinking processes their first tendency was to go from a sensor reading to a final malfunction in one step. This is not to say that one-step diagnoses are not possible, just uncommon. After some work, the experts began to think of their diagnosis rules step-by-step, but then the difficulties of verbalizing the steps became evident; at this point the quasi-expert status of the knowledge engineer became very valuable. After the rules had been established to the satisfaction of the experts and the knowledge engineer, quantifying the steps in terms of sufficiency and necessity functions became the final hurdle. For this step of the process a script was developed using the form of the questions that, through trial and error, seemed to elicit the most consistent responses from the experts. This script was then used for all experts evaluating all rules. The fact that most engineers are very logic-oriented and generally resent what they consider to be a need to justify their decision-making processes should not be discounted at any step of the cognitive process.The third problem encountered was an actual gap in the knowledge of the experts. The expert diagnosis of the system under consideration is in itself a new problem and the knowledge of sensor malfunction and the forms it may take is exlremely limited. At this time the problem is being addressed by using the data from the on-line test as a basis for the o n g o i n g development of rules relating to sensor behavior.The sensors were also the source of another problem, this time in the testing of the system. Since the sensors being used for the on-line test malfunctioned more often than desired, it was difficult to obtain data to test anything but malfunction detection rules. It is expected that this problem can be overcome by the use of more efficient sensor systems currently being designed.The most difficult problem discovered during development was the disagreement of the experts on certain rules of diagnosis. It is difficult for a knowledge engineer to offer any solution to this problem. When the problem occurred an attempt was made to hold group meetings of the experts to discuss the disagreements and reach a consensus. If no agreement could be reached, the rule was modified, usually making it less effective. Fortunately, this occurred infrequently and did not unduly reduce the efficacy of the system.Some "rules of thumb" became obvious during the project and when followed gave optimum results for the time involved. It is probably best to present then in list form:I. Do not misrepresent the c a p a c i t i e s of the system. Itwill alienate the experts and raise managementexpectations to a level the system cannot deliver.2. Develop a script for use in questioning the experts. Ifthe same question is asked two ways, it will often elicitanswers as if two different questions had been asked.3. After a phase of development is completed (byjudgement of the knowledge engineer) turn the systemover to any interested experts. This often results ininformation that was missed during questioning.4. For best efficiency, the team approach to knowledgeengineering seems to be successful. The quasi expertknew when steps were left out and could offer theexperts assistance in verbalizing their thoughtprocesses, while the PDS expert insured that the mostefficient PDS structures were used throughout.7 I m p l e m e n t a t i o nAs mentioned earlier, PDS is written in SRL, which in turn is implemented in the Fran/. Lisp dialect of LISP running on a VAX-780, under the VMS operating system. The program consist of four parts: the knowledge base development functions, the input simulation functions, the inference mechanism and an explanation facility.The knowledge-base development part of PDS provides the "tools" that the knowledge engineer, or the sophisticated expert, can use to develop the rule base. Aside from the obvious functions that ease the addition, deletion and editing of rules, an extensive library of utility functions exists to list and describe the rules, to save and restore the rule base, to print hard copy listings, to initialize the system to a predefined state, and so on.Testing of the rules is facilitated by the presence of functions that allow the manual entry of sensor values and the cutting of specific contexts. Input data can be entered in lieu of actual sensor readings. An edit function is provided to allow the study of the effects of small modifications in sensor values on the propagation of belief process.The inference program performs forward propagation of belief from sensor nodes. If a parametric alteration rule fires, all nodes and rules directly or indirectly affected are re-evaluated.The explanation facility is quite primitive. It does not involve natural language generation, but rather it puts together sentences from "c a n n e d" fragments contained in the description slot of the various schemata. Plans exist to improve on this particular feature, both in the language generation aspect and in the range of questions it can answer (the only question at this time is "why ?"). The explanation facility can be connected to a text-to speech converter (the PROSE 2000 board by Telesensory.lnc), thereby being able to "speak" its explanations.A goal of this project is to provide machine technicians with a portable and inexpensive diagnostic tool. A microprocessor version of PDS is being implemented on a four board package (CPU, memory, graphics I/O and text-to-speech converter), that fits in a hand-carry suitcase. We call it the "E xpert in a Box" version of PDS. It has two modes of operation. In the independent mode, it can perform diagnosis only. In the alternative mode, It can connect,via a built-in modem, to the VAX. a nd thereby act as an intelligent remote terminal for the expert system. In this mode, it can also be connected to a loudspeaker, for voice explanations, and to a color graphic display. 8 C o n c l u s i o n sPDS is a forward chaining rule-based system for process diagnosis. It is being implemented in environments in which dataacquisition is totally automated, thereby limiting the amount of userinteraction. In pursuing its implementation in tnese environments,problems of spurious readings and general degradation had to be solved. We chose to solve ihese problems in three ways. First, raw data was introduced, stored, and retrospectively analyzed in the knowledge base in order to provide greater flexibility. Second,meta-diagnosis was performed to adapt the rule-base to changes inits physical environment (i.e.. sensor degradation); as sensors degrade, PDS focuses its analysis only on the sensors which provide reliable information. Third, redundant senior readings were analyzed and combined into a composite sensor. Theapproach reported here is just a step towards the general problemof the intelligent acquisition and analysis of sensor-basedinformation. Techniques such as sensor redirection and tuning remain to be investigated. 9 A c k n o w l e d g e m e n t sThis research was supported by the Westinghouse Research and Development Center. Chris Kemper, a domain expert and system user, provided valuable feedback on the design of PDS. R. Byfordand A.I Szabo provided continued management support. Manythanks to Brad Allen of the Intelligent Systems Laboratory for hisinvaluable help and support.10 R e f e r e n c e sDavis R., H. Shrobe, W. Hanscher, K Wieckert, M.Shirley, and S. Polit, (1982), "Diagnosis Based on Description of Structure and Function", Proceedings o f the American Ass o ciati o n f o r Artificial Intelligence, Aug. 1982, pp. 137-142.Duda R.O.. P.. Hart, P. Barrett, J.G. Gaschnig, K. Konolige,R Reboh, and J. Slocum, (1978), "Development of the Prospector Consultation System for Mineral E xploration: Final Report", Tech. Rep., SRI International, Menlo Park CA, Oct. 1978.Fox M.S., (1979), "On Inheritance in Knowledge Representation",Proceedings o f the Sixth Internati o nal J o int Conference on Artificial Intelligence, Tokyo Japan.Fox M.S. and D.J. Mostow, (1977), "Maximal ConsistentInterpretations of E rrorful Data In Hierarchically Modeled Domains", Fifth Internati onal J oint Conference on Artificial Intelligence, Cambridge MA, 1977.M. Fox et al. 163Genesereth M.R., (1982), "Diagnosis Using Hierarchical DesignModels", Pr o ceedings o f the Sec ond Conference o f the American Ass o ciati o n f o r Artificial Intelligence, Aug. 1982, pp. 278-283.Hart P., (1982), "Direction for Al in the ighties", SIGARTNewsletter, No. 79, Jan. 1982.McDermott D. and R.Brooks, (1982), "ARBY: Diagnosis with Shallow Causal Models", Pr o ceedings o f the Second C o nference o f the American Association f or Artificial Intelligence, Pittsburgh PA.Meijer C.H.. J.P. Pasquenza, J.C. Deckert, J.L. Fisher, D.B. Laning, and A. Ray, (1981), "Online Power Plant Signal Validation Technique Utilizing Parity-SpaceRepresentation and Analytic Redundancy", Electric Power Research Institute, TechnicalReport E PRI NP 2110, Nov. 1981.Nelson W.R., (1982), "RE ACTOR: An E xpert System for Diagnosis and Treatment of Nuclear Reactor Accidents", Proceedings o f the Sec o nd C o nference o f theAmerican Ass o ciati o n f o r Artificial Intelligence, Aug. 1982, pp. 296 301. Pople H., (1977), The Formation of Composite Hypotheses in Diagnostic Problem Solving: An xercise in Synthetic Reasoning. Pr o ceedings o f the Fifth International J o int C o nference o n ArtificialIntelligence, Cambridge, Aug. 77. Shortliffe E .H., (1976), C o mputer-Based Medical C o nsultati o ns:MYCIN, New York: American E lsevier.Underwood W.E ., (1982), "A CSA Model-based Nuclear PowerPlant Consultant", Pr o ceedings o f the Sec nd Conference o f the American Ass o ciati o n f o r Artificial Intelligence, Aug. 1982, pp. 302-305.Wright J.M., and Fox M.S., (1982), "SRL/1.5 User Manual",Robotics Institute, Carnegie-Mellon University, Pittsburgh PA.。