44.2 ABSTRACT Test Cost Reduction for SOCs Using Virtual TAMs and Lagrange Multipliers ￡

主要方法。

相关的成本核算人员需要在整个建筑行业的全部施工过程中对于每一步所需要的经费进行核算。

做好合理规划与资金投用的方向减少施工中不必要的成本损失。

施工过程中的成本核算，不仅能提高资金的使用效率，同时也能减少资金的浪费做到监督资金使用动向的作用，以此来保障企业的合理收入。

不仅要做好成本的核算，还应该从提高建筑物的质量入手，提升建筑物的质量，使建筑既符合标准，又符合大众的审美。

1.2保障建筑施工质量在建筑行业中，施工项目的成本核算能够更好地保障建筑物的质量。

在整个施工过程中，成本核算规划了在所有施工内容的前提下做出的资金核算。

成本核算大致地规划好了整个施工项目所需要用到的金额数目，在实际的施工操作中也会进行稍微调整。

比如，在实际的施工操作中遇到了预算金额不够的情况就很容易导致施工项目无法按照计划进行下去。

如果在实际的施工过程中没有用完规划好的金额数目，就很容易导致相关的单位出现过度投资的现象。

所以在整个工程项目进行成本核算的工作是非常有必要的，不仅能够保障整个项目顺利进行，而且还能最大限度地减少金额的损失，提高建筑物的质量，保障施工效果。

1.3保障建筑工程正常开展施工项目的成本核算在一定程度上也能够为建筑工程开展进度提供一定的保障。

有了成本的核算就能够提前掌握好施工的进度，确保每一步的施工内容都能够按照计划来进行。

如果开工前的成本核算阶段没有做到位，在实际的施工操作过程中出现资金不足引发安全问题。

如某些施工项目在动工的时候出现了成本预算不足的情况，就很有可能会导致建筑过程中所使用的建筑材料质摘要 建筑工程的造价是建筑工程行业中不可缺少的组成部分，这项工作为工程建筑施工过程中的成本计算和建筑物质量提供保障。

文章围绕建筑工程造价控制工作，阐述了成本核算的意义，论证了施工项目成本核算的关键问题，并对有效的成本核算措施进行分析、评价，以期为相关工作提供参考。

关键词 成本核算；建筑工程；造价控制中图分类号 TU723.3文献标识码 ADOI 10.19892/ki.csjz.2023.24.54Abstract The cost of construction engineering is an important and indispensable part of the construction engineering industry. This work provides guarantee for cost calculation and building quality in the process of engineering construction. This paper focuses on the cost control of construction projects, expounds the significance of cost accounting, demonstrates the key problems of cost accounting of construction projects, and analyzes and evaluates the effective cost accounting measures, which can be used as a reference.Key words cost accounting; architectural engineering; cost control如今，在我国经济水平不断提高的社会环境下，人们对物质生活的要求也越来越高，同时也给建筑行业一个重新的定义和构建。

英汉统计学常用词汇(SPSS)Aabsolute deviation 绝对离差absolute residuals 绝对残差acceptable hypothesis 可接受假设acceptable region 接受域actual frequency 实际频数adaptive estimator 自适应估计量addition theorem 加法定理additivity 可加性adjusted R square 调整判别系数admissible error 容许误差alphafactorin g α因子提取法alternative hypothesis 备择假设among groups 组间analysis of correlation 相关分析analysis of covariance 协方差分析analysis of regression 回归分析BBayesian estimation Beyes估计bell-shaped curve 钟形曲线best-trim estimator 最好切尾估计量beta distribution β分布between groups 组间的between measures 重复测量间的bivariate 双变量的bivariate correlate 二变量相关biweight interval 双权区间biweight M-estimator 双权M估计量block 区组/配伍组boxplot 箱线图Ccanonical correlation 典型相关case-control study 病例对照研究categorical variable 分类变量Cauchy distribution 柯西分布centering and scaling 中心化和定标central tendency 集中趋势chance statistics 随机统计量chance variable 随机变量chi-square distribution 卡方分布chi-square statistics 卡方统计量chi-square test 卡方检验classified variable 分类变量coefficient of skewness 偏度系数coefficient of variation 变异系数communality variance 共性方差compare means 均值比较分析complete association 完全正相关concomitant variable 伴随变量conditional likelihood 条件似然conditional probability 条件概率confidence limit 置信限consistency check 一致性检验consistent estimate 一致估计contingency tables 列联表continuous variable 连续变量control charts 控制图controlled experiments 对照实验conventional depth 常规深度correction coefficient 校正系数critical point 临界点critical ratio 临界比cumulative probability 累计概率curvature 曲率cyclist 周期性Ddata capacity 数据容量data deficiencies 数据缺乏1data handling 数据处理data reduction 数据简化分析data transformation 数据变换degree of precision 精密度degree of reliability 可靠性程度density function 密度函数density of data points 数据点的密度derivative matrix 导数矩阵description 描述descriptive 描述性的deviation from average 离均差Df. Fit 拟合差值df(degree of freedom)自由度dichotomous variable 二分变量discriminant analysis 判别分析discriminant coefficient 判别系数disproportional 不成比例的dissimilarity 不相似性distribution shape 分布形状disturbance 随机扰动项double logarithmic 双对数Eeffect 实验效应effects of interaction 交互效应efficiency 有效性eigenvector 特征向量enumeration data 计数资料equal size 相等的数量error of estimate 估计误差error type Ⅰ第一类错误error type Ⅱ第二类错误estimation 估计量Euclidean distance 欧氏距离expectation plane 期望平面expectation surface 期望曲面expected value 期望值experimental sampling 试验抽样explanatory variable 解释变量explore Summarize 探索-摘要EXSMOOTH 指数平滑方法extended fit 扩充拟合extra parameter 附加参数extreme observation 末端观测值extreme value 极值Ffactor score 因子得分factorial designs 因子设计factorial experiment 因子试验failure rate 失效率family of estimators 估计量族fatality rate 病死率finite population 有限总体finite-sample 有限样本first derivative 一阶导数first quartile 第一四分位数Fisher information Fisher信息量fitting a curve 曲线拟合fixed model 固定模型fixed variable 固定变量fluctuation 随机起伏fourth 四分点fractional error 相对误差frequency polygon 频数多边图frontier point 界限点F-test F检验function 函数function relationship 泛函关系Ggamma distribution 伽玛分布general census 全面普查geometric mean几何均值Gini’s mean difference 基尼均差goodness-of-fit 拟合优度gross-error sensitivity 大错敏感度group averages 分组平均grouped data 分组资料grouped median 组中值growth curve 生长曲线Hhalf-life 半衰期happenstance 偶然事件harmonic mean 调和均值hazard function 风险均数hazard rate 风险率Hessian array Hessian立体阵Heterogeneity 不同质heterogeneity 不齐性HOMAIS 多重响应分析homogeneity of variance 方差齐性homogeneity test 齐性检验Huber M-estimators Huber M 估计量hyperbola 双曲线hypothesis 假设hypothesis test 假设检验hypothetical universe 假设总体Iimage factoring 典型因子提取法impossible event 不可能事件independent samples 独立样本independent variable 自变量indirect standardization 间接标准化法infinitely great 无穷大information capacity 信息容量interclass correlation 组内相关inter-item correlation 样本内相关interpolation 内插法interquartile range 四分位距interclass correlation 组间相关inverse matrix 逆矩阵item means 样本均值L large sample problem 大样本问题Latin square 拉丁方Latin square design 拉丁方设计Least-square estimation 最小二乘估计L-estimator of location 位置L估计量level of significance 显著性水平leverage value 中心化杠杆值life expectance 预期期望寿命life table 寿命表life table method 生命表法light-tailed distribution 轻尾分布likelihood function 似然函数likelihood ratio 似然比likelihood ratio test 似然比检验linear relation 线性关系linear trend 线性预测值loading 载荷location invariance 位置不变性log rank test 时序检验logarithmic scale 对数尺度logic check 逻辑检查logistic 逻辑的logistic distribution logistic分布logit model logit模型logit transformation logit转换logarithms 对数lost function 损失函数lower limit 下限MMahal Distance 马氏距离main effect 主效应maintainability 可维护度matched data 配对资料matched distribution 匹配分布matrix 矩阵maximum 最大值mean 均值mean difference 均值差值mean square 均方mean sum of square 均方和measure 度量median 中位数median lethal dose 半数致死量median polish 中位数平滑m-estimator M估计midpoint 中值model specification 模型的确定modeling statistics 型统计models for outliers 离群值模型modifying the model 模型的修正Monte Carle method 蒙特卡洛法multiple comparison 多重比较multiple correlation 多元相关系数multiple response 多重响应multiple response sets 多重响应集合multiple solutions 多解multiplication theorem 乘法定理multi-response 多元响应multi-stage sampling 多阶段抽样multivariate 多元的multivariate analysis 多元分析mutual exclusive 互不相容mutual independence 互相独立Nnegative correlation 负相关nominal variable 名义变量nonlinear regression 非线性相关nonlinear regression 非线性回归nonparamtric statistics 非参数统计nonparametric test 非参数检验normal distribution 正态分布normal P-P 正态P-P图normal probability 正态概率normal Q-Q 正态Q-Q图normal value 正常值null hypothesis 零假设Oobjective function 目标函数observation unit 观察单位observed value 观察值one sided test 单侧检验one-sample 单样本one-tailed test 单侧检验one-way classification 单因素分类order statistics 顺序统计量ordered categories 有序分类ordinal 序数ordinal variable 有序变量origin 原点orthogonal 正交的orthogonal design 正交试验设计Ppaired observations 成对观测数据paired design 配对设计paired sample 配对样本parametric statistics 参数统计Pearson curves Pearson曲线P-estimator P估计量pie chart 饼(圆)图Pitman estimator Pitman估计量pivot 枢轴量pivot table 枢轴表polynomial regression 多项式回归population 总体positive correlation 正相关posterior distribution 后验分布preliminary analysis 预备性分析probability 概率probability density 概率密度probability of F F显著性概率probit analysis 概率分析product moment 乘积矩/协方差QQ-Q Plot Q-Q概率图quadratic regression 二次多项式回归quadratic term 二次项quality control charts 质量控制图quantitative analysis 定量分析quartile 四分位数RR square 判别系数random 随机random event 随机事件random number 随机数random sampling 随机取样random variable 随机变量randomization 随机化rank statistic 秩统计量rank sum test 秩和检验rank test 秩检验ranked data 等级资料ratio analysis 比率分析raw data 原始资料Rayleigh's test 雷氏检验reciprocal 倒数reject region 拒绝域rejection point 拒绝点relative dispersion 相对离散度relative number 相对数reliability 可靠性reliability analysis 可靠性分析reliability test 可靠性检验report summaries 报告摘要residual 残差residual sum of square 残差平方和response 响应root mean square 均方根rotation 旋转row effects 行效应run test 游程检验S S. E.mean 均值的标准差S.E.of Kurtosis 峰度的标准差S.E.of Skewness 偏度的标准差sample size 样本容量sample space 样本空间sampling design 抽样设计sampling distribution 抽样分布sampling error 抽样误差sampling inspection 抽样检验scatter diagram 散点图schematic plot 示意图/简图score statistic 得分统计score test 计分检验sensitivity curve 敏感度曲线sequential analysis 贯序分析sequential data set 顺序数据集serial tests 系列试验series mean 系列均值sign test 符号检验signed rank 符号秩significance digits 有效数字significance test 显著性检验significant figure 有效数字similarity 相似性simple regression 简单回归skewed distribution 偏态分布skewness 偏度small sample problem 小样本问题Smirnov test Smirnov检验specific factor 特殊因子specific factor variance 特殊因子方差standard deviation 标准差standard error 标准误standard residual plots 标准化残差图standardize 标准化standardized coefficients标准化系数standardized residual 标准化残差statistics 统计学(量)、统计图表std.predicted value 标准预测值Std.residual 标准残差stem and leaf display 茎叶图step factor 步长因子stochastic models 随机模型stochastic process 随机过程survival 生存分析symmetry 对称systematic error 系统误差systematic sampling 系统抽样Ttest criterion 检验判据test for linearity 线性检验test of goodness of fit 拟和优度检验test of homogeneity 齐性检验test oh independence 独立性检验test rules 检验法则testing function 检验函数testing of hypotheses 假设检验theoretical frequency 理论频数time series 时间序列tolerance 容忍度tolerance interval 容忍区间tolerance limits 容限tolerance lower limit 容忍下限tolerance upper limit 容忍上限total sum of square 总平方和total variation 总变异transfer function 转换函数Uunbiased estimation 无偏估计unbiasedness 无偏性unequal size 不等含量unweight 不加权upper limit 上限upward rank 升秩Vvalidity 有效性value 数值value of estimator 计值variability 变异性variable 变量variance components 方差成分variance ratio 方差比variation 变异various 不同的vector 向量WWeibull distribution 威布尔分布weighted mean square 加权平均方差weighted sum of square 加权平方和weighting coefficient 权重系数weighting method 加权法weighted average 加权平均值ZZ score Z分数Z test Z检验zero correlation 零相关Z-transformation Z变换6。

Optimization of Lighting Design Usign GeneticAlgorithmsGerson F.M.LimaJosimeire Tavares Faculty of Electrical Engineering Federal University of UberlandiaP.O Box593 Uberlandia,MG,Brazil38400-902 Email:gersonlima@,josycbelo@Igor S.PerettaKeiji YamanakaFaculty of Electrical EngineeringFederal University of UberlandiaP.O Box593Uberlandia,MG,Brazil38400-902Email:iperetta@,keiji@ufu.brAlexandre Cardosoand Edgard Lamounier Jr.Faculty of Electrical EngineeringFederal University of UberlandiaP.O Box593Uberlandia,MG,Brazil38400-902Email:alexandre@ufu.br,lamounier@ufu.brAbstract—This paper proposes a Genetic Algorithm to opti-mize lightning design parameters integrated with a Computer Aided Design(CAD)application.The algorithm offers a search optimization approach to achieve better design solutions when compared to traditional tools.The proposed CAD tool uses innovative concepts of Information plex3D representation and calculation is established through an intuitive and strategic conjunction.I.I NTRODUCTIONLighting design has an extreme importance due to the high cost of implantation and maintenance of the illumination installations.Best illumination dimensioning is indispensable to achieve an adequate lighting and reduced electrical power consumption.CAD tools are very important to the engineer design process to enable alternative simulations before thefinal solution,improving quality and maturity to the engineering projects.In the electrical installation area,these tools have been proving suitable in several design activities.However, recently studies show that the computational facilities are still dependent to the creativity and experience of the engineer[1], with hinders a better performance from the design to market. The illumination calculation method,commonly called Point-To-Point,is usually used in Lighting Design and is indicated to external places.Funded in this technique,it is possible to calculate the illumination level in each point of interest in a determinate work surface plan of study[2]. This method in general requires the use of computational applications to calculate a matrix of points in a determinate area plan,providing each point luminance values as result. The scalarfield visualization in3D is applied to Lighting Design through the graphic representation of the luminance field.Recently,some applications using Virtual Reality has proved to be a useful alternative for this kind of design application[3].Therefore,it is possible to develop algorithms to process and displayfields of illumination in a3D virtual environment for any required lighting system[4].An efficient lighting design requires good visibility and color reproduction,power economy,reduce the maintenance costs and must considerate to get lower market expends to the adopted solution.In luminaire equipment,illumination intensity emitted in a determinate point can be calculated by an algorithm routine using the IES(Illuminating Engineering Society)file format created for the electronic transfer of photometric data.It has been widely used by many lighting manufacturers and is one of the industry standards in photometric data distribution. Genetic algorithms[5]are a class of search techniques that use the mechanics of natural selection and genetics to conduct a global search of a solution space and constitute in an effi-cient optimization environment for the Design Optimization procedures[6],[7].This paper proposes a Genetic Algorithm to optimize light-ing design parameters integrated with a Computer Aided Design(CAD)application.Genetic Algorithm offers an opti-mization approach to perform a search to achieve better design solutions,and proposed CAD tool uses innovative concepts of Information Visualization within virtual environments.II.M ETHODSA.Design ParametersDuring the process of lighting design,an engineer needs to evaluate the illumination requirements.These requirements are important to specify the uniformity of illumination levels,glare levels,index of color reproduction,and color temperature.This methodology involves the following items:a Survey of activities of the local,physical dimensionsof the layout,materials used and the characteristicsof the grid in place(initial data of the project);b Determination of the goals of lighting and effects tobe achieved;c Choice of lamp types;d Choice offixtures(luminaire);e Analysis of the factors influencing the quality ofenlightenment(IRC and Color Temperature);f Calculation of general lighting;g Calculation of control;2010 9th IEEE/IAS International Conference on Industry Applications- INDUSCON 2010 -978-1-4244-8010-4/10/$26.00 ©2010 IEEEh Distribution of light;i Definition of points of light;j Calculation of directional lighting;k Evaluation of energy consumption;l Assessment of costs;m Calculation of profitability.In this paper,we propose to focus only on items[a],[c], [d],[f],and[h]as input parameters for the Genetic Algorithm (GA)application described in section II-D.The GAsfitness function deals with items[j],[k],and[l],automatically.The Genetic Algorithm generates the light source positions and the best pair of lamp type and the corresponding luminaire.B.Directed Lighting MethodThe directed lighting method(or point-to-point method) is the most often used method for illumination calculation in external areas[8]and also recommended by Illuminating Engineering Society of North America(IESNA).With this method,we obtain the luminance performed by one or more light sources,at any desired point or a mesh of dots.A light source is defined by the set of a luminaire and itsfixed lamps.A luminarie distributes,filters or converts the light emitted by one or more lamps.Part of the luminousflux emitted by the lamps is absorbed by the luminaire and does not contribute to the ambient lighting.Theflow balance is spent above and below a horizontal plane passing through the center of the luminaire.The beams of light irradiated directly on the work plan are the main contributors to the luminance.For each luminarie,the photometric data used for calculating luminance are obtained by their respective IESfiles(Illuminating En-gineering Society Standard),a international standard used in many existing softwares.The total luminance,given by adding the contributions of each light source,is represented by(1):E total=Ni=1E i(1)where:E i=luminance of the source i;N=number of light sources.The luminance E i,held by each light source individually, is given by(2)and shown by Figure1:E i=I P i·cosγd2(2)where:I P i=intensity of the light source i toward the point P;γ=angle between normal surface and direction of the considered point P;d=distance between the light source and the point P.The intensity emitted by a light source in a certain direction is given by the photometric distribution of the light source and varies for each type of luminaire and lamp used.The light distribution of a light source is obtained through laboratory measurements and is usually provided by the manufacturer in different forms:tables,diagrams,graphs or digitalfiles.Fig.1:Arrangement of a lamp:L=installation point,d= distance from the point of light,P=point of interest, n= normal to the plane,γ=angle between the vector near straight LP.The use of digitized curves and photometric standard(IES) allows automation for consulting the light intensity during calculations,which facilitates the process and makes it faster and more dynamic than when done manually.The point-to-point method takes as input the initial configu-ration of lighting(light sources with their position and height inside work plan environment)and provides output results that contain luminance values of each point of the mesh(2D or 3D).The average level of illumination is given by(3).It is calculated to allow a comparative value.E average=F util·F lux·N lamp·N lumW·D plum(3) where:E average=luminance average;F util=Utilization Fac-tor of the used luminaries(manufacturer);F lux=Luminous flow emitted by each lamp;N lamp=Number of Lamps per Luminary;N lum=Number offixtures used in the algorithm; W=Width of the Environment to be Enlightened;D plum= Distance between poles or luminariesC.Photometric CurveThe IESfile allows the creation of photometric curves,as shown in Figure2,that are polar diagram representations of light energy propagation provided by a lamp/luminary set. These diagrams usually provide values of light energy in discrete angles increment and,if is necessary,it is possible to calculate a value for an intermediate angle using an interpo-lation criteria.D.The Genetic Algorithm ImplementationA genetic algorithm(GA)is a stochastic search technique that,for a given problem,searches a solution space for a near-optimal solution.This search is done in a fashion inspired by Darwinian Evolution.Within a population of possible solutions,they could breed among themselves regarding their respectivefitness,forming new solutions that will belong to new generations.There will be evolution of possible solutions for many generations.When the algorithm reaches its end,the best solution found is returned.Fig.2:Photometric curve of the lamp HLF432SONT400W(Phillips)Fig.3:Chromosome representationGAs are often applied as a global optimization problemsolver.More information about GA can be found at Mitchell,1999[5],among other authors.To implement a GA,three mainsubjects are required:definition of the chromosome structure,definition of a Fitness Function to evaluate possible solutionsand to choose strategic genetic operators.1)Chromosomes:A chromosome is composed by variousgenes and represents a solution to a problem.Each generepresents a variable of the proposed problem.The valueassumed by each gene is called an allele.The best solution isused to the posterior CAD drawing and results.From the left to right,thefirst gene is an id that referees toa textfile which describes all the information about the typeof light associated with a lamp,the power intensity,the marketcost and the maintenance cost.The second gene is an id thatrepresents the type of pole(material composition),its heightand cost for installation and maintenance.Each of the next3gene sets represents the pole coordinates,the direction of the luminaire focus and its slope,as can beseen in Figure3.In this chromosome,n poles are represented. The luminaire direction(δ)varies by45o intervals,thusrunning the cardinal points:0o to East,45o to Northeast, 90o North,and so on.The luminaire slope(ζ)varies in15o intervals,with the angle of0o as the reference that indicates theluminaire is parallel to the work plan,i.e.,its focus coincideswith a perpendicular line to the plane.When tilted in45o,its preconfigured maximum tilt,the luminaire focus will becoincident with a45o line to the plane.These angles are shownin Figure4.It is important to note that in this implementation,the distance between the actual position of the luminaire andthe base of the corresponding pole(ΔdL)tends to zero,since Fig.4:Schematic of the guidelines: N=normal to the plane,γ=angle between the vector N and the line LP,ζ=slope of the light,δ=direction of the light,ΔdL tends to zero,h= height of the lamp(post),θ=vertical angle andα=horizontal angle,both for calculation of illuminancewe consider thefixing point of the luminaire with the same coordinate of the pole in the x-y plane.2)Fitness function:Thefitness function deals with each chromosome of the population evaluating how good is it as a possible solution.Thefitness function evaluates an illuminance matrix to the corresponding horizontal work plan,using the Point-To-Point method applying(2).Using average level of illumination,by applying(3),the function determines the shadow area given by the light source positions presented in each chromossome. Using the illuminance matrix and the average illumination level,the function also determines the over-illuminated area. The energy consumption and assessment of costs given by the chromosome are also calculated.These four evaluated features(shadow area,over-illuminated area,energy consumption,and assessment of costs)are the output results of thefitness function.A Paretos multi-objective strategy[9],[10]is then applied to optimize these parameters.3)Genetic Operators:The following operators were im-plemented in the system:ElitismNew generations will grant10%of the last generation best individuals to be reproduced.Crossover(Recombination)The tournament selection method was used with10%of the population.A canonical crossover was implemented in a way to respect the integrity of a light source entity.For pole coordinates,the Radcliff crossover operator was used. MutationIf the chosen gene for mutation is thefirst or the second one,then a different one is randomly chosen to replace it.If the chosen one belongs to a light source entity,a Gaussian mutation operator is applied,respecting minimum and maximum respective values.Fig.5:Thefigures represent the luminousflux due to impact all the lamp at a givenpointFig.6:Colors that are used to conceptualize the quantized points in the work planrmation VisualizationWith the objective to present the data for calculating illumi-nance in a simplified context and adequately informative,this work uses Information Visualization techniques[11].Figure5shows one area in which each point is represented by its luminance level quantized and displayed in different col-ors that indicate comparative intervals with a value previously calculated in(3).The thematizing colors limits are shown in Figure6(Por-tuguese version,’Acima’means Above and’Entre’means Between),where the value E i is compared to the value E average and colors are presented in accordance with the percentage of tracks comparison.Thus it is possible that the engineer has total clarity that every point is seen a defined light level and presented by the text.This information presented in colors,provide to the designer the information of light or dark compared to the average expected and recommended by the lighting standard. This type of presentation intends to apply the concept of Graphic Excellence,i.e.,communication of complex ideas clearly to maximize the efficiency of the engineer to the lighting design.Looking at Figure7,the yellow dots have a luminance above200%of recommended average(20Lux),representing a degree of excess which can be optimized to provideenergy Fig.7:Proposal Lighting poles calculated to3meters in height and Pressure Sodium Lamps70w viewed in3DFig.8:Parking lot80x10m street,the red area(1x80m each one)indicates where you can have polessavings.The red dots represent the level below50%of recom-mended average,which can mean points of sub dimensioning. Below,it is highlighted the following features found in the application for a proposed future work for feature comparison with other similar applications:•Allows the use of3D meshes;•Allows simultaneous viewing of3D CAD models and field luminance in a single environment;•Allows navigation and iteration;•Views by different techniques of scientific Visualization (or Information Visualization).Due to these aspects,the program proved effective in plan-ning the lighting to provide a method of viewing information in a simplified and intuitive design engineers allowing for the reduction process and also facilitating the decision-making to design better projects.Besides,by exploring Virtual Reality techniques,the system allows an engineer to immerse and navigate in the virtual environment with the feeling that he is actually in thefield with different views.This helps the user to configurate and to test different design alternatives.III.R ESULTSTo understand the feasibility of this project,were used a traffic street,in a parking lot[12],as it can be seen in Figure8. To this street,we consider a minimum illumination of20lux and only1%of acceptable shadow area(less than20lux incident).To set up the Genetic Algorithm,we use a population of 300individuals and50generations by each number of poles. We considered only6models of luminaries&lamps:•HLF432—SONT400W;•HLF432—HPLN250W;TABLE I:Processing time comparison Parking Lot street Handball blockVariables database 6light sources4highs of poles14light sources4highs of poles positioning area:160m2positioning area:256m2 lighting area:800m2lighting area:1056m2Generationsby eachnumber ofpoles5010Total timedemanded forsolution212hours4hours•MWF230/150A/47.5—MHN-TD150W;•SWF23570SY—SON-T70W;•MRP822POS.16/19—HPI/T250W;•SRC612—HPI/T400W.The poles used were those of3m,5m,10m and15m-high.The implemented application run on a PC with3Gb DDR2of RAM and Intel Core2Quad Q8200processor.It took about212hours to reach the solution we can see in Table I. To designing our hypothetical street problem,there are at least5variables to be exchanged in searching efforts(meaning a1-pole solution).This will increase by3for each new pole. So,a3-pole solution will require from GA implementation to search inside an11-dimensional solution space.The number of individuals and generations adopted were really insufficient to reach a near-by optimum solution.So,the algorithm moves on,adding a new pole to the probable solution.In this way,a3-pole solution should be expected for our problem,instead the6-pole solution reached by GA.Figure9a presents one of the early generations andfigure9b the results after300generations(50for each new pole added to solution). As we can see,the best solution tended to be found among the first randomly chosen individual generation.Genetic operators are insufficient to alter this behavior,as we can see in Mean Fitness curve(figure9),due to the small range of generations presented.Another experiment was done,with a handball block chosen the same one used by Baade[13].This time,we used a database with14sets of luminaire&lamps and4different highs for poles.The run of10generations for each number of poles exceeded4hours of processing.Because of this huge dimensionality,the GA parameters used were not enough for its conversion to an acceptable solution,as expected.To emphasize the necessity of faster processing alternatives, a processing time comparison of reported experiments can be found at Table I.It is necessary to improve thefitness function calculation and speed up the processing to enable larger number of individuals and generations.IV.C ONCLUSIONA Genetic algorithm shows itself as a powerful tool to general design optimization.One of its main issues is to(a)One of the early generations(b)Result after300generationsFig.9:Experiment with a parking lot street:10m wide,80m long,sidewalk with1m of service width and2m passage width carefully project afitness function that covers all aspects of the problem and enables thefinding of a near-optimum solution. Another main issue is the correct representation of a solution, here represented by the chromosome.Whereas this initial work revealed possibilities of GA to handle lighting design problems,there are many improvements to be putational costs reduction for these high dimensionality types of problem is our main quest,and is still under development.Already previewed solutions include the use of faster computers on processing phase,and the determination of better heuristics to cover all the nuances of lighting design problem,including complexities inherent to required calculations.For future works are proposed:•Implement new genetic operators or reformulate used ones,to increase the ability to search within the solution space of high dimensionality.•Understand the feasibility of other programming lan-guages to enable this application for a high performance computer cluster.•Feasibility study for incorporation of this application to the one developed for AutoCAD that uses Information Visualization technology.R EFERENCES[1] A.L.Silva and A.Cardoso and mounier Jr,“Using GeometricConstraints to Support Virtual Electrical Installation”,in VIII Symposium on Virtual Reality,Belem,2006.[2]V.A.Moreira,Iluminac¸˜a o El´e trica,S˜a o Paulo,Brazil:Edgar Bl¨u cherLtda,1999.[3]J.Leng,“Scientific examples of Virtual Reality and visualization ap-plications”,in UKHEC–UK High Performance Computing,pp.1–13, 2001.[4]T.M.Buriol and M.F.Miranda and S.Scheer and G.S.Tows and D.F.Zandon´a,“Processamento e Visualizac¸˜a o de Campos de Iluminˆa nciasUtilizando VRML e Integrac¸˜a o com CAD3D”,in Anais do Simp´o sio Brasileiro de Realidade Virtual,v.1,pp.157–169,2006.[5]M.Mitchell,An Introduction to Genetic Algorithms,5th edition,London,England:MIT Press,1999,ISBN0262631857.[6]H.Xu and Y.Ding,“Optimizing Method for Analog Circuit DesignUsing Adaptive Immune Genetic Algorithm”,in Fourth International Conference on Frontier of Computer Science and Technology,pp.359–363,2009.[7]R.Ganjehmarzy and M.Davoody,“Optimization of Circular RingMicrostrip Antenna Using Genetic Algorithm”,in Communication Net-works and Services Research Conference,pp.222–227,2008.[8] D.L.Marinoski and F.S.Westphal and mbert,“Desenvolvimentode um algoritmo de c´a lculo luminot´e cnico para ambientes internos atrav´e s do m´e todo ponto-a-ponto”,in Anais do ENCAC-COTEDI,pp.1066–1073,2003.[9]K.Deb and A.Pratap and S.Agarwal and T.Meyarivan,“A fast andelitist multiobjective genetic algorithm:NSGA-II”,in IEEE Transactions on Evolutionary Computation,v.6,n.2,pp.182–197,apr2002. [10] C.A.C.Coello,“Evolutionary multi-objective optimization:a historicalview of thefield”,in IEEE Computational Intelligence Magazine,v.1, n.1,pp.28–36,feb2006.[11] E.R.Tufte,The Visual Display of Quantitative Information,2nd edition,Graphic Press,2001,ISBN0961392142.[12]G.F.M.Lima et al.,“Projetos de Iluminac¸˜a o por Quantizac¸˜a o deIluminˆa ncias em Ambientes3D”,in Simp´o sio Brasileiro de Sistemas El´e tricos,2010.[13]W.Baade,“Normas para ensaios de instalac¸˜o es el´e tricas de baixatens˜a o”,in Eletricidade Moderna,n.428,pp.40–46,nov2009.。

Keysight TechnologiesMeasuring ACLR Performancein LTE TransmittersApplication NoteIntroductionAs wireless service providers push for more bandwidth to deliver IP services to more users, LTE hasemerged as a next-generation cellular technology with great potential to enhance current deployments of 3GPP networks and to enable signiicant new service opportunities. However, LTE’s complex, evolved archi-tecture introduces new challenges in designing and testing network and user equipment. The commercial success of LTE will depend in part on the ability of all devices to work as speciied. One of the particular challenges at the air interface will be power management during signal transmission.In a digital communication system such as LTE, the power that leaks from a transmitted signal into adjacent channels can interfere with transmissions in the neighboring channels and impair system performance. The adjacent channel leakage-power ratio (ACLR) test veriies that system transmitters are performing within speciied limits. This critical yet complex transmitter test can be made quickly and accurately using modern signal analyzers such as the Keysight Technologies, Inc. X-Series (PXA/MXA/EXA) signal analyzers with LTE measurement software and signal generators such as the Keysight MXG signal generator with LTEsignal creation software.Challenges of LTE transmitter designLTE product development is underway, and RF engineers are tackling the many design and measurement challenges this complex technology presents. LTE requires support for six channel bandwidths (from 1.4 to 20 MHz), different transmission schemes for the downlink and the uplink (OFDMA and SC-FDMA), two transmission modes (FDD and TDD), and multiple antenna techniques (MIMO spatial multiplexing, diversity, beamsteering). As a result of LTE’s lexible transmission schemes, the physical channel coniguration has a large impact on RF performance—much greater than in current CDMA-based systems. With performance targets set exceptionally high for LTE, engineers have to make careful design tradeoffs to cover each criti-cal part of the radio transmitter chain.One important aspect of transmitter design is the need to minimize unwanted emissions. Because LTE will be deployed in the same frequency bands as W-CDMA and other legacy cellular technologies, the 3GPP speciications regulate emissions to minimize interference and ensure compatibility between the different radio systems. The primary concern is control of spurious emissions, which can occur at any frequency. In this respect LTE is similar to other radio systems. However, new challenges arise at the band edges, where the transmitted signal must comply with rigorous power leakage requirements. With LTE supporting chan-nel bandwidths up to 20 MHz, and with many bands too narrow to support more than a few channels, a large proportion of the LTE channels will be at the edge of the band.Controlling transmitter performance at the edge of the band requires a design with iltering to attenuate out-of-band emissions without affecting in-channel performance. Factors such as cost, power eficiencies, physical size, and location in the transmitter block diagram are also important considerations. Ultimately the LTE transmitter must meet all speciied limits for unwanted emissions, including limits on the amount of power that leaks into adjacent channels, as deined by ACLR.ACLR test requirements ACLR is a key transmitter characteristic included in the LTE RF transmitterconformance tests (Table 1). These tests verify that minimum requirements arebeing met in the base station (eNB) and user equipment (UE). Most of the LTEconformance tests for out-of-band emissions are similar in scope and purposeto those for W-CDMA and should look familiar. However, while W-CDMA speci-fies a root-raised cosine (RRC) filter for making transmitter measurements, noequivalent filter is defined for LTE. Thus different filter implementations canbe used for LTE transmitter testing to optimize either in-channel performance,resulting in improved error vector magnitude, or out-of-channel performance,resulting in better adjacent channel power characteristics.Table 1. Conformance tests for RF transmitters (from 3GPP TS 36.141 [1] and 36.521-1 [2]) Array ArrayGiven the extensive number of complex transmitter configurations that canbe used to test transmitter performance, LTE specifies a series of downlinksignal configurations known as E-UTRA test models (E-TM) for testing theeNB. These test models are grouped into three classes: E-TM1, E-TM2, andE-TM3. The first and third classes are further subdivided into E-TM1.1, E-TM1.2,E-TM3.1, E-TM3.2, and E-TM3.3 (Table 2). Note that the “E” in E-UTRA stands for“enhanced” and designates LTE UMTS terrestrial radio access, whereas UTRAwithout the “E” refers to W-CDMA.ACLR test requirements (continued)For UE testing, transmitter tests are carried out using the reference measurement channels (RMC) specified for eNB receiver testing. The ACLR requirement for the UE is not as stringent as it is for the eNB, so our focus will be on the latter.Table 2. E-UTRA test models (from 3GPP TS 36.141 [1])The 3GPP specifications for LTE define ACLR as the ratio of the filtered mean power centered on the assigned channel frequency to the filtered mean power centered on an adjacent channel frequency. Minimum ACLR conformance requirements for the eNB are given for two scenarios: for adjacent E-UTRA (LTE) channel carriers of the same bandwidth, E-UTRAACLR1, and for the UTRA (W-CDMA) adjacent and alternate channel carriers, UTRAACLR1and UTRAACLR2, respectively.Different limits and measurement filters are specified for E-UTRA and UTRA adja-cent channels, and are provided for both paired spectrum (FDD) operation and unpaired spectrum (TDD) operation. The E-UTRA channels are measured using a square measurement filter, while UTRA channels are measured using an RRC filter with a roll-off factor of = 0.22 and a bandwidth equal to the chip rate, whichis 3.84 MHz in the example of paired spectrum operation shown in Figure 1.Figure 1. Measurement filters forACLR measurements ACLR limits definedfor adjacent E-UTRA carriers ACLR limits definedfor adjacentUTRA carriersACLR test requirements (continued)ACLR test requirements for the eNB including paired and unpaired spectrum operation are summarized in Table 3. As of the September 2009 specification release, the ACLR test cases for the UE were not fully complete. However, the UE test procedure is essentially the same as that used for the base station.Table 3. ACLR base station conformance test requirements (from 3GPP TS 36.141 6.2 [1]). Note that the specification defines the minimum requirement plus the test tolerance (TT)). * Relative limits are 44.2 dB = 45 dB min requirement + 0.8 dB TT. Both Absolute and Relative limitsare provided. Whichever is less stringent is to be used for the conformance requirement. Sophisticated signal evaluation tools are available for making complex LTE measure-ments quickly and accurately. Power measurements including ACLR generally are made using a spectrum or signal analyzer, and the required test signals are built using a signal generator. In the following examples, Keysight’s PC-based Signal Studio application connected to an MXG signal generator is used to build the standards-compliant E-TM signal required for transmitter testing. The output signal is connected to the RF input of an Keysight X-Series signal analyzer running N9080A LTE measure-ment application, which is used for signal analysis. This equipment setup follows the simple block diagram provided in the 3GPP LTE specifications (Figure 2). Although the measurement process described here is for FDD operation, the process for TDD operation is similar.According to the specifications, the carrier frequency must be set within a frequency band supported by the base station under test, and ACLR must be measured for frequency offsets on both sides of the channel frequency, as specified for paired or unpaired spectrum operation (T able 3). The test is performed first using a transmitted signal of type E-TM1.1, in which all of the PDSCH resource blocks have the same power, and then again using E-TM1.2, in which power boosting and deboosting are used. The E-TM1.2 configuration is useful because it simulates multiple users whose devices are operating at different power levels. This scenario results in a higher crest factor, which makes it more difficult to amplify the signal without creating additional,unwanted spectral content—i.e., ACLR.Spectrum Bands Adjacent Carrier Limits (Min req + TT) *PairedSpectrumCategory AE-UTRA 44.2 dB or -13 dBm/MHzUTRA 44.2 dB or -13 dBm/MHz PairedSpectrumCategory BE-UTRA 44.2 dB or -15 dBm/MHzUTRA 44.2 dB or -15 dBm/MHzUnpairedspectrumCategory AE-UTRA (LTE) 44.2 dB or -13 dBm/MHz1.28 Mcps UTRA 44.2 dB or -13 dBm/MHz3.84 Mcps UTRA 44.2 dB or -13 dBm/MHz7.82 Mcps UTRA 44.2 dB or -13 dBm/MHzUnpairedspectrumCategory BE-UTRA (LTE) 44.2 dB or -15 dBm/MHz1.28 Mcps UTRA 44.2 dB or -15 dBm/MHz3.84 Mcps UTRA 44.2 dB or -15 dBm/MHz7.82 Mcps UTRA 44.2 dB or -15 dBm/MHzSetting up the ACLR testFigure 2. Measurement equipment setup (3GPP TS 36.141 [1] Annex I, Figure I.1-1)Setting up the ACLR test (continued)In this ACLR measurement example, Signal Studio is set up to generate a standards-compliant E-TM1.2 test signal. The frequency is set to 2.11 GHz, a frequency that is in several of the major downlink operating bands specified for LTE. The output signal amplitude—an important consideration in determining ACLR performance—is set to -10 dBm. A 5 MHz channel bandwidth is selected from the range that extends from 1.4 to 20 MHz.Figure 3 shows the eNB setup with Transport Channel selected. A graph of the resource allocation blocks for the test signal appears at the bottom. The Y axis indicates frequency or resource blocks, while the X axis indicates slots or time. The different colors correspond to different channels, with the white areas representing Channel 1 and the pink areas Channel 2. Both are downlink shared channels, of interest in this measurement. The other colors represent synchroni-zation channels, reference signals, etc.Figure 3. Resource allocation blocks (at bottom) for E-TM1.2 test signalSelecting Channel 1 shows the output power level to be at -4.3 dB, so the channel power has been deboosted. The output power of Channel 2 has been boosted and is set at 3 dB (Figure 4). A complex array of power boosting and deboosting options can be set for the different resource blocks from resource block allocation graph. The resulting composite signal will have a higher peak-to-average ratio than a single channel in which all blocks are at the same power level. Amplifying a boosted signal such as this can be difficult, as noted earlier. Without sufficient back-off in the power amplifier, clipping may result.Figure 4. Boosted output power in Channel 2Setting up the ACLR test(continued)The test signal is now generated using the Signal Studio software. Because Signal Studio is PC-based, it can be run from the PC-based X-Series signal ana-lyzer. The waveform in this case is created on the desktop of the signal analyzer and then downloaded to the signal generator via LAN or GPIB. The RF output of the signal generator is connected to the RF input of the signal analyzer, where the ACLR performance is measured using swept spectrum analysis. In this example, the signal analyzer is in LTE mode with a center frequency of 2.11 GHz and the ACP measurement selected. At this point it is possible to make a quick, one-button ACLR measurement according to the LTE standard by recalling the appropriate parameters and test limits from a list of available choices in the LTE application. These choices include options for paired and unpaired spectrum, Category A or Category B limits (as defined in ITU-R SM.329), and type of carrier in the adjacent and alternate channels—E-UTRA (LTE), UTRA (W-CDMA), or TD-SCDMA.Recall that for FDD operation, LTE defines two different methods of making ACLR measurements. In Figure 5, the upper graph shows the case in which E-UTRA (LTE) is used at the center and offset frequencies. The lower graph shows LTE at the center frequency and UTRA (W-CDMA) at the adjacent and alternate offsets.Figure 5. Two specified methods of ACLR measurementIn Figure 6 the measurement result shows the E-UTRA adjacent and alternate offset channels. For this measurement a 5 MHz carrier is selected; however, the measure-ment noise bandwidth is 4.515 MHz, because the downlink contains 301 subcarriers. The first offset (A) is at 5 MHz, with an integration bandwidth of 4.515 MHz. The second offset (B) is at 10 MHz with the same integration bandwidth.Figure 6. ACLR measurement result using Keysight X-Series analyzer before optimization This one-button measurement gives a very quick, usable ACLR measurementaccording to the LTE standard. While the result in Figure 6 of about -62 dBc is good, the analyzer settings can be optimized to get even better performance. Four ways to further improve the measurement results are (1) optimize the signal level at the mixer; (2) change the resolution bandwidth filter; (3) turn on noise correction; and (4) use a different measurement methodology called filtered integration bandwidth.To optimize the signal level at the input mixer, the attenuator is adjusted for minimal clipping. The X-Series signal analyzer will automatically select an attenuation value based on the current measured signal value. This automated technique provides a good starting point for achieving optimal measurement range. Signal analyzers such as the X-Series, which have both electronic and mechanical attenuators, can use the two in combination to optimize performance. In such cases the mechanical attenuator can be adjusted slightly to get even better results of about 1 or 2 dB.Next the resolution bandwidth can be lowered by pressing the bandwidth filter key. Note that sweep time increases as the resolution bandwidth is lowered. For example, with the MXA signal analyzer, sweep time at 30 kHz is 676.3 ms. At a lower 10 kHz RBW, the sweep time is about 6 seconds. The slower sweep time reduces variance in the measurement, but reduces measurement speed.Another step is to turn on noise correction. The analyzer then takes one sweep to measure its internal noise floor at the current center frequency, and in subse-quent sweeps subtracts that internal noise floor from the measurement result. This technique substantially improves ACLR, in some cases by up to 5 dB.Setting up the ACLR test(continued)Optimizing theanalyzer settingsOptimizing the analyzer settings (continued)Changing the measurement method is a fourth way to optimize the analyzer. In this case the default measurement method (integration bandwidth or IBW) is changed to the filtered IBW method. Filtered IBW uses a sharp, steep cutoff filter. This technique does degrade the absolute accuracy of the power measurement result, but it does not degrade the relative power accuracy, and ACLR is a relative power measurement. Therefore, filtered IBW does not degrade the ACLR result. Using these techniques in combination, an Keysight’s X-series analyzer can optimize the ACLR measurement automatically for performance versus speed via the analyzer’s embedded LTE application. The result for a typical ACLR mea-surement is improved by up to 10 dB or more. Figure 7 shows an 11 dB ACLR improvement after optimization (compared to Figure 6) using the embedded LTE application. For measurement scenarios requiring the maximum performance, the analyzer settings can be further adjusted.Figure 7. ACLR measurement result using Keysight X-series analyzer after optimization Standards-compliant spectrum measurements such as ACLR are invaluable for RF engineers developing the next generation radio systems. With LTE, however, these measurements are complicated by factors such as variations in the band-width of adjacent channels, choice of transmission filter, and interaction of RF variables between channels of different bandwidth and different susceptibility to interference. A practical solution is to use a spectrum or signal analyzer with a standards-specific measurement application. This combination can reduce error in complex measurements, automatically configuring limit tables and specifiedtest setups and ensuring measurement repeatability.ConclusionAcronyms3GPP 3rd Generation Partnership ProjectACLR Adjacent channel leakage-power ratioA-MPR Additional maximum power reductionCDMA Code division multiple accesseNB Evolved node BE-TM E-UTRA test modelE-UTRA Evolved universal terrestrial radio accessEVM Error vector magnitudeFDD Frequency division duplexGPIB General purpose interface busIBW Integration bandwidthLAN Local area networkLTE Long term evolutionMPR Maximum power reductionMIMO Multiple input multiple outputOFDMA Orthogonal frequency division multiple accessPRACH Physical random access channelQAM Quadrature amplitude modulationQPSK Quadrature phase-shift keyingRB Resource blockRBW Resolution bandwidthRE Resource elementRF Radio frequencyRRC Root-raised cosineSC-FDMA Single carrier frequency division multiple accessSRS Sounding reference signalTDD Time division duplexTD-SCDMA Time domain synchronous code division multiple accessUE User equipmentUMTS Universal mobile telecommunications systemUTRA Universal terrestrial radio accessW-CDMA Wideband code division multiple access11ReferencesMore Information[1] 3GPP TS 36.141 V8.4.0 (2009-09) Base Station (BS) Conformance Testing [2] 3GPP TS 36.521-1 V8.3.1 (2009-09) User Equipment (UE) Conformance Specification; Radio Transmission and Reception Part 1: Conformance T estingFor more information about the 3GPP, visit the 3GPP home page /3GPP specifications home page/specs/specs.htm 3GPP Series 36 (LTE) specifications/ftp/Specs/archive/36_seriesFor more information about Keysight design and test products for LTE visit /find/lteKeysight LTE application notes and technical overviews:3GPP Long T erm Evolution: System Overview, Product Development, and Test Challenges: 5989-8139EN LTE Component T est: 5990-5149ENMIMO in LTE Operation and Measurement—Excerpts on LTE T est: 5990-4760EN MIMO Performance and Condition Number in LTE T est: 5990-4759ENN9080A & N9082A LTE Modulation Analysis Measurement Application Technical Overview: 5989-6537ENLearn more about LTE and its measurements in the new book written by 30 LTE experts:LTE and the Evolution to 4G Wireless Design and Measurement Challenges/find/ltebookFor more information on KeysightTechnologies’ products, applications or services, please contact your local Keysight office. The complete list is available at:/find/contactus Americas Canada (877) 894 4414Brazil 55 11 3351 7010Mexico001 800 254 2440United States (800) 829 4444Asia Paciic Australia 1 800 629 485China800 810 0189Hong Kong 800 938 693India 1 800 112 929Japan 0120 (421) 345Korea 080 769 0800Malaysia 1 800 888 848Singapore 1 800 375 8100Taiwan0800 047 866Other AP Countries (65) 6375 8100Europe & Middle East Austria 0800 001122Belgium 0800 58580Finland 0800 523252France 0805 980333Germany 0800 6270999Ireland 1800 832700Israel 1 809 343051Italy800 599100Luxembourg +32 800 58580Netherlands 0800 0233200Russia 8800 5009286Spain 0800 000154Sweden 0200 882255Switzerland0800 805353Opt. 1 (DE)Opt. 2 (FR)Opt. 3 (IT)United Kingdom0800 0260637For other unlisted countries:/find/contactus(BP-07-10-14)This information is subject to change without notice.© Keysight Technologies, 2009-2014Published in USA, July 31, 20145990-5089EN12 | Keysight | Measuring ACLR Performance in LTE Transmitters - Application NoteLAN eXtensions for Instruments puts the power of Ethernet and the Web inside your test systems. Keysight is a founding member of the LXI consortium.Keysight Assurance Plans/find/AssurancePlansUp to five years of protection and no budgetary surprises to ensure your instruments are operating to specification so you can rely on accurate measurements.Keysight Channel Partners/find/channelpartnersGet the best of both worlds: Keysight’s measurement expertise and product breadth, combined with channel partner convenience./find/lte。

第14卷第6期2023年12月有色金属科学与工程Nonferrous Metals Science and EngineeringVol.14，No.6Dec. 2023加强炼镁传热效率的研究进展郭军华1， 丁天然1， 李培艳1， 孙逸翔1， 刘洁1， 钟素娟1， 张廷安*2（1.郑州机械研究所有限公司新型钎焊材料与技术国家重点实验室， 郑州 450000；2.东北大学冶金学院， 沈阳 110819）摘要：随着轻量化需要日益迫切，金属镁及其合金由于具有质量轻、比强度和比刚度高等特性，应用越来越广泛，镁行业的发展也愈发受人关注。

皮江法是国内炼镁的主要生产工艺，但是随着绿色低碳发展理念的推行，该炼镁工艺在生产过程中传热效率低、还原周期长、能耗高和排放大等缺点突显，一直制约着炼镁行业的发展。

经过多年的研究，学者们在提高镁冶炼传热效率，降低还原温度，缩短还原周期等方面取得一系列成果。

本文主要从还原剂、工艺条件、传热装置3个方面详细综述了提升炼镁传热效率的研究进展，并对未来炼镁技术发展提出了建议和思路，仅供参考。

关键词：镁冶炼；传热效率；还原剂；传热装置；优化工艺中图分类号：TF822 文献标志码：AResearch progress in strengthening the heat transfer efficiencyof magnesium smeltingGUO Junhua 1, DING Tianran 1, LI Peiyan 1, SUN Yixiang 1, LIU Jie 1, ZHONG Sujuan 1, ZHANG Ting ’an *2（1. State Key Laboratory of Advanced Brazing Filler Metals & Technology ， Zhengzhou Research Institute of Mechanical EngineeringCo.， Ltd.， Zhengzhou 450000， China ； 2. School of Metallurgy ， Northeastern University ， Shenyang 110819， China ）Abstract: With the increasing need for lightweight materials, magnesium and its alloys have been widely used because of their light quality, high specific strength and specific stiffness, and the development of the magnesium industry has attracted increasing attention. The Pidgeon process is the main production process of magnesium smelting in China. However, with the implementation of the green and low-carbon development concept, the process has many shortcomings, such as low heat transfer efficiency, long reduction cycle, high energy consumption and large emissions, which has been restricting the development of the magnesium smelting industry. After years of research, scholars have made a series of achievements in improving the heat transfer efficiency of magnesium smelting, reducing reduction temperature, shortening the reduction cycle, etc. In this paper, the research progress in improving the heat transfer efficiency of magnesium smelting was reviewed in detail from three aspects including reductant, process conditions and heat transfer device, and suggestions and ideas on the existing magnesium smelting technology were put forward for reference only.Keywords: magnesium smelting ； heat transfer efficiency ； reducing agent ； heat transfer device ； optimization process收稿日期：2022-11-15；修回日期：2022-12-24基金项目：国家自然科学基金辽宁联合基金资助项目（U1508217）通信作者：张廷安（1960—　），教授，主要从事有色金属冶炼、新工艺的开发、固废处理等方面的研究。

中南林2020级PM第二周作业C4-542.Examiner’s report March/June 2021A manufacturing company uses machine C, which is operational for five hours a day to manufacture four products: W, X, Y and Z. Factory costs are $150,000 per day. The company uses throughput accounting and its objective is to maximise profits. Information relating to these products is as follows:Product Production rate per machine hour （units） Selling price per unit （$） Material cost per unit （$） Conversion cost per unit （$）W 200 350 120 40X 500 190 95 25Y 400 270 160 20Z 350 215 75 35If the company is not able to increase the availability of machine C's operational hours, what is the production ranking of product Y? [单选题] *A FirstB SecondC ThirdD Forth(正确答案)答案解析：The correct answer is D Forth As the products are all produced in the same factory the cost per machine hour will be the same across all the products so they can be ranked on their throughput return per machine hour （otherwise they should be ranked ion their throughput accounting ratio）.The products would be ranked in order: Z, X, W and Y. Product Y would be ranked forth （last）.43.Examiner’s report June 2017. Which of the following statements is NOT consistent with the theory of constraints? [单选题] *A There is no inventory of work in progress or finished good held(正确答案)B Raw materials are converted into sales as quickly as possibleC Operations prior to the bottleneck operate at the same level as the bottleneckD Conversion costs and investment costs are kept to a minimum答案解析： Holding no inventory of work in progress or finished goods is NOT consistent with the theory of constraints as the theory of constraints specifies that a small amount of buffer inventory should be maintained prior to the bottleneck activity so that the bottleneck never has to be slowed down or delayed. The other three statements are all consistent with the theory of constraints.44.Examiner’s report September 2016. A manufacturing company uses three processes to make its two products, X and Y. The time available on the three processes is reduced because of the need for preventative maintenance and rest breaks. ……Which of the following will improve throughput? [单选题] *A Increasing the efficiency of the maintenance routine for Process 2(正确答案)B Increasing the demand for both productsC Reducing the time taken for rest breaks on Process 3D Reducing the time product X requires for Process 1答案解析：This question tested candidates’ knowledge of throughput accounting, and specifically how throughput can be improved. It is also a good example of a question where it’s important not to rush – all of the possible answers could improve throughput, depending on what our bottleneck （limiting factor） is – Process 1,2 or 3. If none of the processes are limited, then increasing demand would improve throughput.Once we have realised this, we need to find out which （if any） process is limiting us. This is a common calculation in F5 but worth going over. We need to calculate, for each process, how many hours are required to meet maximum demand – 10 units of X and 16 of Y:As the table shows, Process 2 is the bottleneck – we cannot produce 10 units of X and 16 of Y in the 22 hours available. Therefore increasing the efficiency of the maintenance routine will increase the amount of units we can produce, and therefore improve throughput. Any of the other options will mean we are still limited by Process 2, and therefore will have no effect.45.Past year paper December 2016.A company decides which of three mutually exclusive products to make in its factory on the basis of maximizing the company's throughput accounting ratio.……Which of the following actions would improve the company's existing throughput accounting ratio? [单选题] *A Increase the selling price of product Z by 10%(正确答案)B Increase the selling price of product Y by 10%C Reduce the material cost of product Z by 5%D Reduce the material cost of product Y by 5%答案解析：As all three products are mutually exclusive the company would choose to make X as it has the highest throughput return per hour of $2.Of the four possible options only increasing the selling price of product Z by 10% would give a higher throughput return per hour of $2.40.46.Specimen exam September 2016.A company manufactures a product which requires four hours per unit of machine time. Machine time is a bottleneck resource as there are only ten machines which are available for 12 hours per day, five days per week. The product has a selling price of $130 per unit, direct material costs of $50 per unit, labour costs of $40 per unit and factory overhead costs of $20 per unit. These costs are based on weekly production and sales of 150 units.What is the throughput accounting ratio? [单选题] *A 1·33(正确答案)B 2·00C 0·75D 0·31答案解析：Return per factory hour = （$130 – $50）/4 hours = $20Factory costs per hour = （$20 + $40）/4 = $15TPAR = $20/$15 = 1·3347.Past year paper June 2015.X Co uses a throughput accounting system. Details of product A, per unit, are as follows:……What is the return per hour for product A? [单选题] *A $40B $2,400(正确答案)C $30D $1,800答案解析：$320 – $80/（6/60） = $2,40048.Past year paper June 2015.The following statements have been made in relation to the concepts outlined in throughput accounting:（i） Inventory levels should be kept to a minimum.（ii） All machines within a factory should be 100% efficient, with no idle timeWhich of the above statements is/are correct? [单选题] *A （i） only(正确答案)B （ii） onlyC Both （i） and （ii）D Neither （i） nor （ii）答案解析：The first statement is correct as throughput accounting discourages production for inventory purposes and is often used in a just in time environment. The second statement is incorrect as in throughput accounting it is the bottleneck resource which should be 100% efficient which actually may mean unused capacity elsewhere. 49.On Monday, in addition to the baking ovens, Sweet Treats Bakery has the following process resources available:……Which of the three processes, if any, is a bottleneck activity? [单选题] *A WeighingB Mixing(正确答案)C BakingD There is no bottleneck答案解析：The bottleneck is the mixing process as 188 minutes are required to meet maximum demand but there are only 180 minutes available.Note: Four batches of brownies need to be made in order to have sufficient cakes to meet maximum demand as the cakes must be made in their batch sizes.50.On Wednesday, the mixing process is identified as the bottleneck process. On this day, only 120 minutes in the mixing process are available.Assuming that Sweet Treats Bakery wants to maximise profit, what is the optimal production plan for Wednesday? [单选题] *A 80 brownies, 30 muffins and 100 cupcakes(正确答案)B 0 brownies, 90 muffins and 100 cupcakesC 120 brownies, 0 muffins and 100 cupcakesD 40 brownies, 60 muffins and 100 cupcakes51.Sweet Treats Bakery has done a detailed review of its products, costs and processes.Which TWO of the following statements will improve the throughput accounting ratio? *A The cafe customer wants to negotiate a loyalty discountB A bulk discount on flour and sugar is available from suppliers(正确答案)C There is additional demand for the cupcakes in the marketD The rent of the premises has been reduced for the next year(正确答案)答案解析：Reduction in rent and discounts on materials will reduce costs and will improve the TPAR.Giving a customer a loyalty discount will reduce sales revenue and as a result the TPAR. Demand for cupcakes can increase but it will not impact the TPAR as demand is not the restriction.52.On Friday, due to a local food festival at the weekend, Sweet Treats Bakery is considering increasing its production of cupcakes. These cupcakes can be sold at the festival at the existing selling price.The company has unlimited capacity for weighing and mixing on Friday but its existing three ovens are already fully utilised, therefore in order to supply cupcakes to the festival, Sweet Treats Bakery will need to hire another identical oven at a cost of $45 for the day.How much will profit increase by if the company hires the new oven and produces as many cupcakes as possible? [单选题] *A $50.00B $140.00C $95.00(正确答案)D $31.00答案解析：Each oven has a capacity of eight hours and each cupcake batch takes two hours so four extra batches can be made.Extra throughput = four batches x $35 = $140.Less the hire costs will result in an additional profit of $9553.In a previous week, the weighing process was the bottleneck and the resulting throughput accounting ratio （TPAR） for the bakery was 1.45.Which of the following statements about the TPAR for the previous week are true?（1）The bakery’s operating costs exceeded the total throughput contribution generated from its three products.（2）Less idle time in the mixing department would have improved the TPAR.（3）Improved efficiency during the weighing process would have improved the TPAR. [单选题] *A 1 and 2B 3 only(正确答案)C 2 and 3D 1 only答案解析：As the TPAR exceeds 1 then the throughput contribution exceeds operating costs so Statement 1 is false.Less idle time on a non-bottleneck process would notimprove the TPAR so Statement 2 is false.Improving efficiency during the weighing process would improve the TPAR as any actions to improve throughput on a bottleneck will improve the TPAR so Statement 3 is true.54.What is the annual capacity of the bottleneck activity? [单选题] *A 2,400 1,600B 4,800 4,800C 7,200 4,800(正确答案)D 9,600 9,600答案解析：Total salon hours = 8 x 6 x 50 = 2,400 each year.There are three senior stylists, therefore total hours available = 7,200.Based on the time taken for each activity, they can perform 7,200 cuts （7,200 hours/1 hour per cut） or 4,800 treatments （7,200 hours/1·5 hours per treatment）.55.The salon has calculated the cost per hour to be $42·56.What is the throughput accounting ratio （TPAR） for both services? [单选题] *A 1·37 1·58(正确答案)B 1·41 2·38C 1·37 1·61D 1·41 2·41答案解析：CutsReturn per hour = （Selling price – materials）/time taken on the bottleneck = （60 – 1·50）/1 = 58·50TPAR = Return per hour/cost per hour =58·50/42·56 = 1·37 （to two decimal places）TreatmentsReturn per hour = （Selling price – materials）/time taken on the bottleneck = （110 – 8·90）/1·5 = 67·40TPAR = Return per hour/cost per hour = 67·40/42·56 = 1·58 （to two decimal places）56.Which of the following activities could the salon use to improve the TPAR?（1）Increase the time spent by the bottleneck activity on each service. （2） Identify ways to reduce the material costs for the services. （3） Increase the level of inventory to prevent stock-outs.（4） Increase the productivity of the stage prior to the bottleneck. （5）Improve the control of the salon’s total operating expenses. （6） Apply an increase to the selling price of the services. [单选题] *A （1）, （2） and （4）C （2）, （5） and （6）(正确答案)D （1）, （4） and （6）答案解析：The factors which are included in the TPAR are selling price, material costs, operating expenses and bottleneck time. Increasing the selling price and reducing costs will improve the TPAR.Increasing the time which each service takes on the bottleneck （the senior stylists’ time） will only reduce the number of services they can provide, so this will not improve throughput.Throughput accounting does not advocate the buildingof inventory as it is often used in a just-in-time environment and there is no point increasing the activity prior to the bottleneck as it will just create a build-up of work-in-progress. Neither of these will improve the rate of throughput through the process.57.What would be the effect on the bottleneck if the salon employed another senior stylist? [单选题] *A The senior stylists’ time will be a bottleneck for cuts onlyB The senior stylists’ time will be a bottleneck for treatments only(正确答案)C The senior stylists’ time will remain the bottleneck for both cuts and treatmentsD There will no longer be a bottleneck答案解析：The existing capacity for each activity is: Cut TreatmentAssistants 48,000 16,000Senior stylists 7,200 4,800Junior stylists 8,000 9,600If another senior stylist is employed, this will mean that their available hours will be （4 x 2,400） = 9,600.Thiswill give them capacity to now do 9,600 cuts （9,600 hours/1 hour per cut） and 6,400 treatments （9,600 hours/1·5 hours per treatment）.As a result, the senior stylists willstill be the bottleneck activity for treatments but for cuts the bottleneck will now be the junior stylists as they can only do 8,000 cuts compared to the senior stylists of 9,600. 58.Which of the following statements regarding the theory of constraints are correct?（1）It focuses on identifying stages of congestion in a process when production arrivesmore quickly than the next stage can handle.（2）It is based on the concept that organisations manage three key factors – throughput, operating expenses and inventory.（3）It uses a sequence of focusing steps to overcome a single bottleneck, at which point the improvement process is complete.（4）It can be applied to the management of all limiting factors, both internal and external, which can affect an organisation. [单选题] *A （1） and （2） only(正确答案)C （2）, （3） and （4）D （1）, （3） and （4）答案解析：The theory of constraints is focused on identifying restrictions in a process and how to manage that restriction （commonly termed a bottleneck）.It is based on the concept of managing throughput, operating expenses and inventory.It does use a series of focusing steps but it is not complete once the bottleneck has been overcome. In fact it is an ongoing process of improvement, as once the bottleneck has been elevated it is probable that another bottleneck will appear and the process willcontinue. It cannot be applied to all limiting factors as some, particularly those external to the organisation, may be out of the organisation’s control.59.Examiner’s report Sept/Dec 2021.A company's actual production figures for a batch of products are as follows: kg $Material 2,000 10,000Labour and overhead 26,000 2,000 36,000Normal loss 10% -200 1,800 36,000Abnormal loss -100 -2,000Good output 1,700 34,000In terms of environmental cost categorisations, how would the normal and abnormal losses be described? [单选题] *A Normal loss = Potentially hidden costs Abnormal loss = Conventional costs(正确答案)B Normal loss = Potentially hidden costs Abnormal loss = Contingent costsC Normal loss = Conventional costs Abnormal loss = Contingent costsD Normal loss = Conventional costs Abnormal loss = Contingent costs答案解析：The correct answer is A Normal loss = Potentially hidden costs, Abnormal loss = Conventional costs Performance management candidates should be aware that there are different environmental cost categorisations. One set of definitions provided by the US Environmental Protection Agency made a distinction depending on how an organisation intended to use the information. They identified four types of costs: conventional costs: raw material and energy costs having environmental relevance potentially hidden costs: costs captured by accounting systems but then losing their identity in ‘general overheads’contingent costs: costs to be incurred at a future date –for example, clean-up costs image and relationship costs: costs that, by their nature, are intangible, for example, the costs of preparing environmental reports. In this example, normal loss costs are spread over good units of production thus losing their identity and being lost and potentially hidden. Abnormal losses are reported separately in results so are like conventional costs of material, energy etc. The numbers given in the question are for illustration only and are not required to answer the question.60.Examiner’s report December 2020.Which TWO of the following activities are environmental INTERNAL failure costs? A Quality control inspections to monitorpollution levels in water leaving a production process B Water purification treatment to clean waste water before it leaves the factory C Fitting of carbon filters to machine processes to reduce carbon emissions D Power usage measuring system to monitor energy consumption within the factory E Payment of fines for breaching environmental regulations in the industry F Insulation of heating pipes in the factory to reduce heat loss G Public relations costs to remedy reputational damage caused by accidental river pollution H Capturing and recycling of waste exhaust gases to generate energy [填空题] *_________________________________(答案：BH)答案解析：Environmental costs can be categorised as prevention costs, detection cost, internal failure costs and external failure costs. This question asks for the identification of the two environmental internal failure costs. These are costs relating to an environmental failure, for example pollution or wastage but where the failure has been identified and dealt with within the organisation before it manages to affect the external environment. Water purification treatment to clean waste water before it leaves the factory and capturing and recycling of waste exhaust gases to generate energy would be examples of environmental internal failure costs. They key here is that waste water has been produced which is a failure cost, but it has been cleaned before leaving the factory, making it an internal failure cost. Likewise, waste exhaust gases have been produced, but these have been captured and recycled within the factory. Environmental prevention costs relate to activities or measure which aim to avoid the pollution or wastage occurring. Insulation of heating pipes in the factory to reduce heat loss and fitting of carbon filters to machine processes to reduce carbon emissions would come under this heading. Environmental detection costs relate to the costs incurred to test the levels of emissions and wastage to ensure that the organisation is being compliant with internal standards and external regulations. Quality control inspections to monitor pollution levels in water leaving a production process and power usage measuring system to monitor energy consumption within the factory would come under this heading. External failure costs relate to pollution which has affected the outside environment.Payment of fines for breaching environmental regulations in the industry and public relations costs to remedy reputational damage caused by accidental river pollution would be examples of environmental external failure costs. A number of candidates did not answer the question which was asked and instead gave the answer for the two environmental prevention costs, or the two environmental detection costs or the two environmental external failure costs. Environmental management accounting is an important area and candidates should ensure that they are familiar with the different categories of environmental costs. They should also read the requirements of the question carefully.61.Examiner’s report September 2019.A company grows different types of tea leaves and blends them together. The tea leaves are picked by hand because too many leaves were damaged and wasted when picking machines were used. The tea leaves are then dried andprocessed and any waste produced is recycled. The company regularly tests its tea for contaminants, in line with food safety legislation. It recently identified that one tea plantation contained high levels of contaminated soil after the use of a new pesticide. Identify, by selecting the relevant box in the table below, the environmental cost classification of each of the following costs.（分别用1 2 3 4 四个数字代表下面四个环境成本分类 1 PREVENTION 2 DETECTION 3 INTERNAL FAILURE 4 EXTERNAL FAILURE），请问按照顺序填写下面四个成本对应的成本分类——Cost incurred to clean contaminated soil，Staff cost for picking leaves by hand，Recycling waste cost，Cost incurred for testing tea for contaminants （答案示例：1234） [填空题] *_________________________________(答案：4132)答案解析：The cost incurred to clean contaminated soil is an external failure cost. This is because this cost has arisen as a result of discharging contaminants into the external environment. The staff cost for picking leaves by hand is a prevention cost. The picking machines damage too many leaves which causes waste, therefore handpicking leaves prevents this wastage. The recycling waste cost is an internal failure cost because it is waste which is created by the business but is dealt with by the company so that it is not released into the environment. The cost incurred for testing tea for contaminants is a detection cost as testing the tea ensures that the company is compliant with legislation.62.Examiner’s report September 2017.Which of the following should be categorised as environmental failure costs by an airline company?（1） Compensation payments to residents living close to airports for noise pollution caused by their aircraft.（2） Air pollution due to the airl ine’s carbon emissions from their aircraft engines.（3） Penalties paid by the airline to the government for breaching environmental regulations. [单选题] *A 2 onlyB 1, 2 and 3(正确答案)C 1 and 3 onlyD 2 and 3 only答案解析： Environmental failure costs are costs incurred as a result of environmental issues being created either internally or outside the company. These can be financial or societal costs. Compensation, penalties and air pollution are all environmental failurecosts. This is a good example of why covering the whole syllabus is important. Environmental management accounting is often an area which is overlooked.63.Past year paper December 2016.Flow cost accounting is a technique which can be used to account for environmental costs. Inputs and outputs are measured through each individual process of production.Which of the following is NOT a category used within flow cost accounting? [单选题] *A Material flowsB System flowsC Delivery and disposal flowsD Waste flows(正确答案)答案解析：Waste flows are not a category used within flow cost accounting, however the other three categories are.64.Past year paper September 2016.Different management accounting techniques can be used to account for environmental costs. One of these techniques involves analysing costs under three distinct categories: material, system, and delivery and disposal.What is this technique known as? [单选题] *A Activity-based costingB Life-cycle costingC Input-output analysisD Flow cost accounting(正确答案)答案解析：Under a system of flow cost accounting material flows are divided into three categories – material, system, and delivery and disposal65.Specimen exam September 2016.Which of the following statements regarding environmental cost accounting are true?（1）The majority of environmental costs are already captured within a typical organisation’s accounting system. The difficulty lies in identifying them.（2）Input/output analysis divides material flows within an organisation into three categories: material flows; system flows; and delivery and disposal flows.（3）One of the cost categories used in environmental activity-based costing is environment-driven costs which is used for costs which can be directly traced to a cost centre.（4）Environmental life-cycle costing enables environmental costs fromthe design stage of the product right through to decommissioning at the end of its life to be considered. [单选题] *A （1）, （2） and （4）B （1） and （4） only(正确答案)C （2）, （3） and （4）D （2） and （3） only答案解析：Most organisations do collect data about environmental costs but find it difficult to split them out and categorise them effectively.Life-cycle costing does allow the organisation to collect information about a product’s environmental costs throughout its life cycle.The technique which divides material flows into three categories is material flow cost accounting, not input/output analysis.ABC does categorise some costs as environment-driven costs, however, these are costs which are normally hidden within total overheads in a conventional costing system. It is environment-related costs which can be allocated directly to a cost centre.66.Past year paper June 2015.When activity-based costing is used for environmental accounting, which statement is correct for environment-related costs and environment-driven costs? [单选题] *A Environment-related costs can be attributed to joint cost centres and environment-driven costs cannot be(正确答案)B Environment-driven costs can be attributed to joint cost centres and environment-related costs cannot beC Both environment-related costs and environment-driven costs can be attributed to joint cost centresD Neither environment-related costs nor environment-driven costs can be attributed to joint cost centres答案解析：This is the correct option as environment-driven costs are allocated to general overheads, not joint cost centres.67.Past year paper December 2014.The following statements have been made about environmental cost accounting:（1）The majority of environmental costs are already captured within a typical organisation’s accounting system. The difficulty lies in identifying them.（2）Input/output analysis divides material flows within an organisation into three categories: material flows; system flows; and delivery and disposal flows.Which of the above statements is/are true? [单选题] *A 1 only(正确答案)B 2 onlyC Neither 1 nor 2D Both 1 and 268.Specimen Exam December 2014.The following are types of management accounting techniques:（i） Flow cost accounting.（ii） Input/output analysis.（iii） Life-cycle costing.（iv） Activity based costing.Which of the above techniques could be used by a company to account for its environmental costs? [单选题] *A （i） onlyB （i） and （ii） onlyC （i）, （ii） and （iii） onlyD All of the above(正确答案)。

THE ACCOUNTING REVIEWV ol.82,No.32007pp.759–796The Effects of Contracting,Litigation, Regulation,and Tax Costs on Conditional and Unconditional Conservatism:Cross-Sectional Evidence at the Firm LevelXinrong QiangUniversity of WyomingABSTRACT:Prior studies propose four explanations for accounting conservatism,andrecent studies classify conservatism into two forms:conditional and unconditional con-servatism.This paper examines whether each proposed explanation applies to con-ditional conservatism,unconditional conservatism,or both.Thefindings are as follows:(1)contracting induces conditional conservatism;(2)litigation induces both forms;(3)regulation induces unconditional conservatism;and(4)taxation induces unconditionalconservatism.Thesefindings indicate that the two forms of conservatism play distinctroles in contracting,regulation,and taxation,as well as a common role in litigation.They also play an interrelated role,as suggested by thefinding that unconditionalconservatism reduces conditional conservatism.The combined evidence implies thatbecause the two forms meet distinct needs but are negatively interrelated,it is nec-essary to trade them off.Keywords:conditional conservatism;unconditional conservatism;contracting;taxation.Data Availability:All data are available from public sources.I.INTRODUCTIONR esearchers have proposed four explanations for accounting conservatism:contract-ing,litigation,regulation,and taxation(Watts2003).The contracting explanation suggests that conservatism counteractsfirms’incentives for aggressive accounting and,hence,lowers potential losses to investors who use accounting numbers in contracts. To induce conservatism for contracting purposes,the investors can impose onfirms certain costs(hereafter,contracting costs).The litigation explanation points out thatfirms also face asymmetric litigation costs that induce conservatism.Under the regulation hypothesis,ac-counting standard-setters and regulators prefer conservatism and induce it by imposingThis paper is based on my dissertation at the State University of New York at Buffalo.I appreciate the helpful comments and suggestions from my dissertation committee—Jerry Han(former Chair,now deceased),Michael Rozeff(Chair),Samuel Tiras(Chair),and Susan Hamlen—as well as those from two anonymous referees,Dan Dhaliwal(editor),Sudipta Basu(the conference discussant).The paper also benefited from the comments of Gary Fleischman,Frederic Sterbenz,Steven Rock,Penne Ainsworth,Linda Kidwell,Phil Shane,John Jacob,and other workshop participants at the University of Hong Kong,State University of New York at Buffalo,University of Wyoming,and University of Colorado,and the2005AAA Annual Meeting.Editor’s note:This paper was accepted by Dan Dhaliwal.Submitted October2005Accepted September2006759760QiangThe Accounting Review,May 2007regulation costs on firms.To lower the contracting,litigation,and regulation costs,firmstend to report conservatively.Finally,the taxation hypothesis posits that firms also reportconservatively to defer tax costs.Most studies of these explanations focus on conditional conservatism.This form ofconservatism (e.g.,impairment accounting)requires a higher degree of verification for goodnews than for bad news after difficult-to-verify news occurs (Basu 1997).Watts (2003)points out that conditional conservatism biases equity book value and earnings downward.In comparison,unconditional conservatism (e.g.,excessive depreciation)also addressesdifficult-to-verify information and biases the numbers downward;however,this form isapplied before difficult-to-verify news occurs.Thus,it immunizes accounting systemsagainst future bad news.1This study explores the distinct,common,and interrelated roles of the two forms ofconservatism.2This is important for two reasons.First,recent studies note that some pro-posed explanations do not apply to both forms.Basu (2005)suggests that taxation is likelyto generate unconditional conservatism.Ball and Shivakumar (2005)argue that contractinggenerates conditional conservatism,which is partially supported by Ball et al.’s (2005)finding on debt contracting.Thus,while the two forms may play a common role (e.g.,inlitigation),they are also likely to play distinct roles (e.g.,in contracting and taxation).Evidence of their distinct and common roles can help identify the desirable form(s)to meeta specific need.Second,Beaver and Ryan (2005)demonstrate that unconditional conservatism (e.g.,excessive depreciation)lowers book values,therefore reducing subsequent conditional con-servatism (e.g.,impairment write-downs).This role of the unconditional form in reducingthe conditional form suggests that the two forms are negatively interrelated.Evidence ofthe interrelated role of the two forms,taken together with the evidence of their distinctroles,highlights a necessary trade-off between them.3For example,firms may sacrificecontracting efficiency (via conditional conservatism)for tax reduction (via unconditionalconservatism).In addition,controlling for this negative interrelation can help to properlyidentify the effect of a given factor on conditional conservatism;otherwise,the effect isunderestimated (overestimated)if the factor encourages (discourages)unconditionalconservatism.The four explanations of the two forms of conservatism and the interrelation of the twoforms are unified as follows.Regulation and taxation provide incentives to record expensesand losses as early as possible,without requiring a particular form of conservatism toachieve this.Because unconditional conservatism has some advantages over conditionalconservatism (e.g.,recording losses earlier),these two factors tend to induce unconditionalconservatism.As this form can fully preempt conditional conservatism—for instance,ex-pensing R&D eliminates any future write-down—the two factors do not induce conditionalconservatism.Like regulation and taxation,litigation sometimes simply provides incentivesto record expenses and losses as early as possible,and thus induces unconditional conser-vatism.However,litigation sometimes induces conditional conservatism because some po-tential plaintiffs particularly demand this form for contracting purposes.Contracting onlygenerates conditional conservatism because unconditional conservatism does not provide1I thank an anonymous referee for the insight that ‘‘[u]nconditional conservatism immunizes the accounting system against future bad news.’’2Accounting conservatism as defined in this study includes conditional and unconditional conservatism but ex-cludes recording no rents (e.g.,historical cost accounting)and understating easy-to-verify numbers (e.g.,delaying easy-to-verify gains or subtracting easy-to-verify income).See Section IV for a discussion of these exclusions.3I thank the anonymous referee again for this insight.Explanations for Conditional and Unconditional Conservatism 761The Accounting Review,May 2007new information and adds noise to the payoffs to contracting parties.In sum,the hypothesesare as follows:(1)contracting induces conditional conservatism;(2)litigation induces bothforms;(3)regulation induces unconditional conservatism;and (4)taxation induces the un-conditional form.Conditional (unconditional)conservatism is measured by downward biases in equitybook value and earnings that result from conditionally (unconditionally)conservative ac-counting.4In univariate tests,unconditional conservatism is regressed on each hypothesizedfactor.In addition,conditional conservatism is regressed on each factor before and aftercontrolling for unconditional conservatism.Consistent with unconditional conservatism pre-empting conditional conservatism,the coefficient on unconditional conservatism is alwayssignificantly negative.The coefficient on a factor sometimes changes substantially after thecontrol,suggesting a need for the control.In multivariate tests,the preemption effect iscontrolled for and all factors are incorporated simultaneously.The findings are as follows:(1)Firms bearing higher equity and debt contracting costschoose greater conditional conservatism.(2)Firms facing larger litigation costs choosegreater conditional and unconditional conservatism (with the results for auditor litigationcosts being weak).(3)Firms bearing higher accounting regulation costs choose greaterunconditional conservatism.(4)Firms facing a larger reduction in tax costs choosegreater unconditional conservatism.These findings indicate that the two forms of conser-vatism not only play a common role in avoiding litigation,but also play distinct roles.Theconditional form improves contracting efficiency,while the unconditional form helps pre-vent regulation and defer taxes.Further,a significantly negative association between thetwo forms of conservatism suggests that they play an interrelated role.The combined evi-dence implies that because the two forms meet distinct needs but are negatively interrelated,it is necessary to trade them off.This paper contributes to our understanding of conservatism in two ways.First,itprovides new evidence of the distinct,common,and interrelated roles of the two forms ofconservatism.This evidence helps identify (1)which form(s)should be used to meet aspecific need,(2)whether a trade-off between the two forms is necessary,and (3)whichfactors should be considered in the trade-off.This evidence also allows researchers toproperly assess the effect of a factor on conditional conservatism because unconditionalconservatism is controlled for.Second,by simultaneously investigating multiple conservatism factors rather than ex-amining different factors in isolation,this paper disentangles effects of the correlated factorsand,hence,provides evidence that enables researchers to accurately gauge the effect ofeach factor.5Conducting firm-level tests further helps to properly identify these effects.Ithas the advantage that it is easier to control for confounding factors at the firm level,alleviating omitted variable and endogeneity problems that are inherent in country-level andtime-series tests (Sloan 2001;Holthausen 2003).Another advantage is that substantial4See Section IV for a discussion of potential measurement error introduced by recording no rents and understating easy-to-verify numbers.5Ball et al.(2003)note that properties of reported numbers are determined by firm incentives,which depend on many factors such as investor demand,regulation,and taxation.Bushman and Piotroski (2006)stress the im-portance of incorporating multiple factors in obtaining a complete understanding of the property in question.Holthausen (2003)suggests that examining multiple factors helps us properly gauge the effects of individual factors.762QiangThe Accounting Review,May 2007variation in firm incentives and the considerable discretion available to firms allow forpowerful tests at the firm level.6The remainder of the paper is organized as follows.Section II motivates the researchquestion with prior research.Section III develops hypotheses.Section IV describes researchdesign.Section V presents results.Section VI concludes.II.PRIOR RESEARCHThe extant literature provides ample evidence of the various conservatism explanations.Researchers document greater conservatism among firms in common-law countries (Ball etal.2000),public firms (Ball and Shivakumar 2005),firms facing more severe conflicts overdividend policy (Ahmed et al.2002),and firms monitored by more independent boards(Beekes et al.2004).All of this evidence is consistent with the contracting hypothesis.Several studies provide evidence of the litigation explanation.Huijgen and Lubberink(2005)find greater conservatism among U.K.firms that are cross-listed in the U.S.(a morelitigious country),while Chandra et al.(2004)show greater conservatism,but not higherlitigation risk,among high-tech firms.Greater conservatism is also observed during periodsof higher auditor liability (Basu 1997),during quarters of audited reports (Basu et al.2001b),and among firms hiring big auditing firms (Basu et al.2001a).Consistent with theregulation explanation,Sivakumar and Waymire (2003)show that enforceable accountingrules induced conservatism among early 20th century railroads.Two recent studies examinemultiple explanations at the country level.Ball et al.(2003)find that four East Asiancountries with high-quality standards show low levels of conservatism,indicating that fac-tors other than standards also affect conservatism.Bushman and Piotroski (2006)show thatlegal and political institutions at the economy level interact with contracting and litigationfactors at the country level to influence conservatism.While the existing evidence provides significant insights,it is limited in two ways.First,because most of the above studies focus on conditional conservatism,the evidenceonly allows researchers to assess the effect of a factor on this form of conservatism.Butthe factor may not affect unconditional conservatism in the same way.Thus,the evidencedoes not allow researchers to understand the distinct/common roles of the two forms ofconservatism.Without such an understanding,it is difficult to determine which form(s)should be used to meet a specific need.Furthermore,if the two forms meet distinct needsbut are negatively interrelated,then firms face a trade-off between them.Without an un-derstanding of the distinct and interrelated roles,it is difficult to determine whether a trade-off is necessary and which factors should be considered in the trade-off.In addition,withthe interrelation of the two forms omitted,even the effect of a factor on the conditionalform cannot be properly identified.Second,because most of the above studies focus on different factors in isolation,theevidence does not allow researchers to accurately assess the effect of each factor.Several6For example,although firms within a country have the same regulatory system,firms of different sizes bear different levels of regulation risk and thus face different amounts of expected regulation costs.Similarly,firms of different levels of market volatility bear different levels of litigation risk and thus face different amounts of expected litigation costs.Hence,there is substantial variation in firm incentives for conservatism.In addition,firms have discretion in exercising conditional conservatism (e.g.,selecting the timing and amount of write-downs)and applying unconditional conservatism (e.g.,implementing rules and choosing methods,estimates,and assumptions),as elaborated in Givoly et al.(2006)and indicated by accounting choice and earnings man-agement studies.Explanations for Conditional and Unconditional Conservatism 763The Accounting Review,May 2007factors likely exist or change jointly (e.g.,co-existing contracting and taxation or co-changing litigation and regulation).Thus,it is unclear which factors cause the existence ofconservatism in a given period,such as that before regulation and litigation regimes(Holthausen and Watts 2001),or cause an inter-temporal change in conservatism,such asits increase during periods of increased auditor liability (Basu 1997).Countries or firmslikely differ in many aspects (e.g.,governance,litigation,and regulation).Hence,it isunclear which factors cause a cross-sectional difference in conservatism,such as that be-tween common-law and code-law countries (Ball et al.2000)or that between U.S.-listedand non-U.S.-listed firms (Huijgen and Lubberink 2005).Overcoming these two limitations motivates this paper.To identify not only the effectof a factor on conditional conservatism,but also its effect on unconditional conservatism,this paper examines both forms of conservatism,after controlling for the preemption ofconditional conservatism by unconditional conservatism.To accurately assess the effect ofindividual factors on conservatism,this paper investigates multiple factors simultaneously.III.HYPOTHESESContractingTo mitigate agency problems (Jensen and Meckling 1976),equity and debt holderscontract with firms based on accounting numbers.Firms thus have incentives to bias equitybook value and earnings upward.This tendency poses potential losses to the investors.Watts (2003)argues that accounting conservatism counteracts this tendency and biases thenumbers downward,thereby reducing the expected losses to investors who have asymmetricloss functions.Thus,conservatism improves contracting efficiency.Ball and Shivakumar (2005)specify that conditional conservatism improves contractingefficiency.This form curbs negative net present value (NPV)projects,thereby improvingthe efficiency of equity contracting,and it quickly triggers debt covenant violations,therebyenhancing the efficiency of debt contracting.On the other hand,this form forgoes positiveNPV projects (Guay and Verrecchia 2006)and retains unnecessary capital within a firm(Watts 2003),thereby reducing contracting efficiency.7Watts (2003)resolves this conflictby pointing out that,for equity (debt)holders with asymmetric loss functions,the benefitsof avoiding negative NPV projects (preventing excessive dividends)outweigh the costs offorgoing positive NPV projects (retaining unnecessary capital),resulting in a net increasein contracting efficiency.Unconditional conservatism,which lowers equity book value and earnings as well,canalso curb negative NPV projects and quickly trigger debt covenant violations.However,Basu (2005)notes that this form does not utilize new information,and Ball and Shivakumar(2005)argue that this form adds noise to payoffs to contracting parties and thus couldreduce contracting efficiency.Overall,contracting is expected to encourage the conditional7Guay and Verrecchia’s (2006)argument that conditional conservatism forgoes positive NPV projects is the flip side of the argument that it curbs negative NPV projects.Under conservative accounting,a positive NPV project may cause near-term earnings to be negative,and future earnings may reflect the benefits of the project after the tenure of a manager.This can lower the manager’s compensation,causing him to forgo the project.Similarly,Watts’(2003)argument that conditional conservatism retains unnecessary capital is the flip side of the argument that it quickly triggers debt covenants violation and,hence,prevents excessive dividends.When the return on equity provided by a firm is lower than the return that shareholders can earn elsewhere,paying dividends is of the best interests of shareholders.However,conservative accounting understates equity book value and earnings;the understated numbers,when used in debt covenants that constrain dividends,reduce firms’flexibility in paying dividends.Consequently,conservative firms may not be allowed to pay dividends when it is efficient to do so.764QiangThe Accounting Review,May 2007form only.Investors can induce this form by imposing costs on firms via governancemechanisms.8LitigationLitigation also provides incentives for conservatism,as suggested by Kellogg’s (1984)evidence that firms and auditors are far more likely to be sued for overstatements than forunderstatements.Firms thus bear costs associated with litigation against them.Further,auditors can pass on their litigation costs to firms by being less accommodating to aggres-sive accounting,by increasing auditing fees (Pratt and Stice 1994),by issuing unfavorableaudit opinions,or by terminating relations with risky clients (Krishnan and Krishnan 1997).Hence,firms also bear litigation costs that auditors pass on to them.Firms can minimize these expected litigation costs by recognizing bad news as earlyas possible,because early recognition of bad news can reduce the probability that classaction lawsuits will be certified and can shorten the length of the class period when thoselawsuits are certified.Unconditional conservatism is preferred to achieve the early recog-nition of bad news because it records losses earlier than conditional conservatism does.However,if some potential plaintiffs particularly demand conditional conservatism in con-tracting,litigation also induces this form.In sum,litigation is expected to generate bothconditional and unconditional conservatism.RegulationTo avoid blame from constituents,accounting standard-setters and regulators (hereafter,the regulators)have an incentive to respond to constituent demand.Accordingly,if theirconstituents demand conservatism,then the regulators tend to induce conservatism (Watts2003).Furthermore,the form(s)of conservatism that the regulators induce would dependon (1)which form(s)their constituents demand,(2)whether the regulators understand theconstituent demand,and (3)which form(s)the regulators prefer.In general,constituents probably prefer the unconditional form because the conditionalform generates big negative shocks (e.g.,large write-downs),which are unlikely to bedesirable.Similarly,the regulators probably prefer the unconditional form,because relativeto smooth value decreases recognized before bad news occurs,large negative shocks rec-ognized after bad news occurs are more likely to cause the regulators problems.However,some constituents tend to favor conditional conservatism for contracting purposes.Ball andShivakumar (2005)present evidence of the historical ambivalence in accounting standards,and suggest that the ambivalence could reflect the regulators’confusion of the two formsof conservatism.Thus,although these constituents favor conditional conservatism,the reg-ulators may not well understand this need and thus may not favor this form.Collectively,the regulators tend to induce unconditional conservatism.They can induce this form byimposing regulation costs on firms (e.g.,costs due to restatements and SEC investigations).Given that firms capture the regulators’intention to induce unconditional conservatism,they tend to adopt this form to prevent regulatory intervention.In addition,when explicitor implicit rate of return regulation is in use,the unconditional form is more likely toprevent regulatory intervention because this form results in fewer large negative shocks inincome and,hence,a lower probability of rate violation.In contrast,it is not apparent that8Specifically,equity holders can impose costs such as potential reduced compensation,slowed promotion,and shortened tenure via board monitoring.Similarly,debt holders can impose costs such as potential immediate repayment,additional covenants,increased collateral,and raised interest rates (Gopalakrishnan and Parkash 1995)via debt monitoring such as having a trustee,hiring bond-rating agencies,and setting tight covenants (Dichev and Skinner 2002).Explanations for Conditional and Unconditional Conservatism 765The Accounting Review,May 2007conditional conservatism can prevent regulatory intervention.Instead,this form results inlarge negative shocks in income and,thus,may even trigger regulatory intervention thatwould not otherwise occur.Overall,regulation is expected to induce unconditional conser-vatism only.TaxationShackelford and Shevlin (2001)summarize two book-tax links.First,tax-minimizingstrategies often result in lower book income due to book-tax conformity.Second,becausefirms are concerned with large book-tax differences,they choose to lower book incomewhen reducing taxable income.Because of these links,taxation tends to induce conservativeaccounting.Basu (2005)points out that firms’choices between the two forms of conser-vatism in tax planning result from cost-benefit trade-offs.He also presents historical evi-dence that taxation induces the unconditional form.I extend his point as follows.From the perspective of the first link,the unconditional form is more likely to defertaxes.Tax codes often allow extra expenses due to unconditional conservatism (e.g.,extraCOGS due to LIFO)to be deductible;however,losses due to conditional conservatism arebarely allowed to be deductible,perhaps because such losses are based on unrealized de-creases in market value rather than realized losses from transactions.From the perspectiveof the second link,it is likely that firms prefer to use the unconditional form to reducebook income.Whereas both forms can lower book income,the unconditional form (1)recognizes losses earlier (i.e.,before news occurs),(2)is easier to control (i.e.,independentof news),(3)costs less to implement (e.g.,no impairment test),and (4)results in smootherearnings (e.g.,fewer shocks).Collectively,taxation is expected to induce the unconditionalform only.Hypothesis TestsTo test whether unconditional conservatism preempts conditional conservatism andwhether this relationship affects hypothesis tests,I regress conditional conservatism on eachfactor before and after controlling for unconditional conservatism.Unconditional conser-vatism is also regressed on each factor.The cross-sectional regressions are as follows:c CSV ϭ␣ϩ␣Cost ϩε;j 01j j c u CSV ϭ␣ϩ␣Cost ϩ␣CSV ϩε;andj 01j 2j j u CSV ϭ␤ϩ␤Cost ϩ␦(1)j 01j jwhere:ϭc CSV j a conditional conservatism measure for firm j ;ϭu CSV j an unconditional conservatism measure for firm j ;Cost j ϭone of the following proxies for factors inducing conservatism for firm j :CtrCost Eq ϭa proxy for equity contracting costs;j CtrCost Dt ϭa proxy for debt contracting costs;j LitiCost Fm ϭa proxy for firm litigation costs;j LitiCost Aud ϭa proxy for auditor litigation costs;j ReguCost ϭa proxy for regulation costs;andj TaxCost ϭa proxy for tax cost reduction.j766QiangThe Accounting Review,May 2007The expected sign of ␣1is positive when Cost is CtrCost Eq ,CtrCost Dt ,LitiCostFm ,or LitiCost Aud .The expected sign of ␤1is positive when Cost is LitiCost Fm ,LitiCost Aud ,ReguCost ,or TaxCost .The expected sign of ␣2is negative because uncon-ditional conservatism is expected to preempt conditional conservatism.The following cross-sectional multivariate regressions are used to test the incremental effect of each factor:c CSV ϭ␣ϩ␣CtrCost Eq ϩ␣CtrCost Dt ϩ␣LitiCost Fm j 01j 2j 3j (ϩ)(ϩ)(ϩ)ϩ␣LitiCost Aud ϩ␣ReguCost ϩ␣TaxCost 4j 5j 6j (ϩ)u ϩ␣CSV ϩε;and 7j j (Ϫ)u CSV ϭ␤ϩ␤CtrCost Eq ϩ␤CtrCost Dt ϩ␤LitiCost Fm j 01j 2j 3j (ϩ)ϩ␤LitiCost Aud ϩ␤ReguCost ϩ␤TaxCost ϩ␦(2)4j 5j 6j j (ϩ)(ϩ)(ϩ)where all variables are as defined in Equation (1).IV .RESEARCH DESIGNSample and Descriptive StatisticsThe original sample consists of all Compustat 2004industry and research firms in theperiod of 1982–2002,including an estimation period of 1982–87,a sample period of 1988–1999,and a sensitivity test period of 2000–02.9Among these firms,884firms meet thedata requirements for book-value-based conservatism measures and 1,500firms meet thosefor accruals-based conservatism measures.10Additional data requirements for all explana-tory variables result in a final sample of 633firms over the sample period of 1988–1999.Panel A of Table 1presents descriptive statistics for key accounting and market variables.The statistics show that the final sample generally consists of well-established firms,indi-cating the presence of survivorship bias due to requirements for substantial data over a longperiod.To address survivorship bias,I use three larger samples over shorter sample periods:719firms over 1990–99,825firms over 1992–99,and 1,050firms over 1994–99.In ad-dition,I use two samples,614firms over 1988–2001and 720firms over 1990–2001,toaddress endogeneity.Proxies for Factors Inducing ConservatismPanel B of Table 1reports descriptive statistics for proxies for factors inducing con-servatism.Because an expected cost equals the probability of incurring a cost times theamount of the cost,expected costs imposed for contracting purposes increase with theprobability that investors detect aggressive reporting.This probability is likely to be highwhen governance is strong,as suggested by the evidence that governance mechanismsconstrain opportunistic reporting,including both the extreme tail of frauds (Dechow et al.1996)and more subtle accruals management (Klein 2002).Therefore,stronger equity ordebt governance proxies for higher expected costs imposed for equity or debt contracting9The sample period is constrained by my limited access to some databases,such as Compact D (used to estimate CtrCost Eq )and SDC (see footnote 29).10I exclude observations with a negative equity book value and those with the highest or lowest 1percent value of any variable necessary to estimate a conservatism measure.。

SPSS词汇(中英文对照) Absolute deviation, 绝对离差Absolute number, 绝对数Absolute residuals, 绝对残差Acceleration array, 加速度立体阵Acceleration in an arbitrary direction, 任意方向上的加速度Acceleration normal, 法向加速度Acceleration space dimension, 加速度空间的维数Acceleration tangential, 切向加速度Acceleration vector, 加速度向量Acceptable hypothesis, 可接受假设Accumulation, 累积Accuracy, 准确度Actual frequency, 实际频数Adaptive estimator, 自适应估计量Addition, 相加Addition theorem, 加法定理Additivity, 可加性Adjusted rate, 调整率Adjusted value, 校正值Admissible error, 容许误差Aggregation, 聚集性Alternative hypothesis, 备择假设Among groups, 组间Amounts, 总量Analysis of correlation, 相关分析Analysis of covariance, 协方差分析Analysis of regression, 回归分析Analysis of time series, 时间序列分析Analysis of variance, 方差分析Angular transformation, 角转换ANOVA （analysis of variance）, 方差分析ANOVA Models, 方差分析模型Arcing, 弧/弧旋Arcsine transformation, 反正弦变换Area under the curve, 曲线面积AREG , 评估从一个时间点到下一个时间点回归相关时的误差ARIMA, 季节和非季节性单变量模型的极大似然估计Arithmetic grid paper, 算术格纸Arithmetic mean, 算术平均数Arrhenius relation, 艾恩尼斯关系Assessing fit, 拟合的评估Associative laws, 结合律Asymmetric distribution, 非对称分布Asymptotic bias, 渐近偏倚Asymptotic efficiency, 渐近效率Asymptotic variance, 渐近方差Attributable risk, 归因危险度Attribute data, 属性资料Attribution, 属性Autocorrelation, 自相关Autocorrelation of residuals, 残差的自相关Average, 平均数Average confidence interval length, 平均置信区间长度Average growth rate, 平均增长率Bar chart, 条形图Bar graph, 条形图Base period, 基期Bayes' theorem , Bayes定理Bell-shaped curve, 钟形曲线Bernoulli distribution, 伯努力分布Best-trim estimator, 最好切尾估计量Bias, 偏性Binary logistic regression, 二元逻辑斯蒂回归Binomial distribution, 二项分布Bisquare, 双平方Bivariate Correlate, 二变量相关Bivariate normal distribution, 双变量正态分布Bivariate normal population, 双变量正态总体Biweight interval, 双权区间Biweight M-estimator, 双权M估计量Block, 区组/配伍组BMDP(Biomedical computer programs), BMDP统计软件包Boxplots, 箱线图/箱尾图Breakdown bound, 崩溃界/崩溃点Canonical correlation, 典型相关Caption, 纵标目Case-control study, 病例对照研究Categorical variable, 分类变量Catenary, 悬链线Cauchy distribution, 柯西分布Cause-and-effect relationship, 因果关系Cell, 单元Censoring, 终检Center of symmetry, 对称中心Centering and scaling, 中心化和定标Central tendency, 集中趋势Central value, 中心值CHAID -χ2 Automatic Interaction Detector, 卡方自动交互检测Chance, 机遇Chance error, 随机误差Chance variable, 随机变量Characteristic equation, 特征方程Characteristic root, 特征根Characteristic vector, 特征向量Chebshev criterion of fit, 拟合的切比雪夫准则Chernoff faces, 切尔诺夫脸谱图Chi-square test, 卡方检验/χ2检验Choleskey decomposition, 乔洛斯基分解Circle chart, 圆图Class interval, 组距Class mid-value, 组中值Class upper limit, 组上限Classified variable, 分类变量Cluster analysis, 聚类分析Cluster sampling, 整群抽样Code, 代码Coded data, 编码数据Coding, 编码Coefficient of contingency, 列联系数Coefficient of determination, 决定系数Coefficient of multiple correlation, 多重相关系数Coefficient of partial correlation, 偏相关系数Coefficient of production-moment correlation, 积差相关系数Coefficient of rank correlation, 等级相关系数Coefficient of regression, 回归系数Coefficient of skewness, 偏度系数Coefficient of variation, 变异系数Cohort study, 队列研究Column, 列Column effect, 列效应Column factor, 列因素Combination pool, 合并Combinative table, 组合表Common factor, 共性因子Common regression coefficient, 公共回归系数Common value, 共同值Common variance, 公共方差Common variation, 公共变异Communality variance, 共性方差Comparability, 可比性Comparison of bathes, 批比较Comparison value, 比较值Compartment model, 分部模型Compassion, 伸缩Complement of an event, 补事件Complete association, 完全正相关Complete dissociation, 完全不相关Complete statistics, 完备统计量Completely randomized design, 完全随机化设计Composite event, 联合事件Composite events, 复合事件Concavity, 凹性Conditional expectation, 条件期望Conditional likelihood, 条件似然Conditional probability, 条件概率Conditionally linear, 依条件线性Confidence interval, 置信区间Confidence limit, 置信限Confidence lower limit, 置信下限Confidence upper limit, 置信上限Confirmatory Factor Analysis , 验证性因子分析Confirmatory research, 证实性实验研究Confounding factor, 混杂因素Conjoint, 联合分析Consistency, 相合性Consistency check, 一致性检验Consistent asymptotically normal estimate, 相合渐近正态估计Consistent estimate, 相合估计Constrained nonlinear regression, 受约束非线性回归Constraint, 约束Contaminated distribution, 污染分布Contaminated Gausssian, 污染高斯分布Contaminated normal distribution, 污染正态分布Contamination, 污染Contamination model, 污染模型Contingency table, 列联表Contour, 边界线Contribution rate, 贡献率Control, 对照Controlled experiments, 对照实验Conventional depth, 常规深度Convolution, 卷积Corrected factor, 校正因子Corrected mean, 校正均值Correction coefficient, 校正系数Correctness, 正确性Correlation coefficient, 相关系数Correlation index, 相关指数Correspondence, 对应Counting, 计数Counts, 计数/频数Covariance, 协方差Covariant, 共变Cox Regression, Cox回归Criteria for fitting, 拟合准则Criteria of least squares, 最小二乘准则Critical ratio, 临界比Critical region, 拒绝域Critical value, 临界值Cross-over design, 交叉设计Cross-section analysis, 横断面分析Cross-section survey, 横断面调查Crosstabs , 交叉表Cross-tabulation table, 复合表Cube root, 立方根Cumulative distribution function, 分布函数Cumulative probability, 累计概率Curvature, 曲率/弯曲Curvature, 曲率Curve fit , 曲线拟和Curve fitting, 曲线拟合Curvilinear regression, 曲线回归Curvilinear relation, 曲线关系Cut-and-try method, 尝试法Cycle, 周期Cyclist, 周期性D test, D检验Data acquisition, 资料收集Data bank, 数据库Data capacity, 数据容量Data deficiencies, 数据缺乏Data handling, 数据处理Data manipulation, 数据处理Data processing, 数据处理Data reduction, 数据缩减Data set, 数据集Data sources, 数据来源Data transformation, 数据变换Data validity, 数据有效性Data-in, 数据输入Data-out, 数据输出Dead time, 停滞期Degree of freedom, 自由度Degree of precision, 精密度Degree of reliability, 可靠性程度Degression, 递减Density function, 密度函数Density of data points, 数据点的密度Dependent variable, 应变量/依变量/因变量Dependent variable, 因变量Depth, 深度Derivative matrix, 导数矩阵Derivative-free methods, 无导数方法Design, 设计Determinacy, 确定性Determinant, 行列式Determinant, 决定因素Deviation, 离差Deviation from average, 离均差Diagnostic plot, 诊断图Dichotomous variable, 二分变量Differential equation, 微分方程Direct standardization, 直接标准化法Discrete variable, 离散型变量DISCRIMINANT, 判断Discriminant analysis, 判别分析Discriminant coefficient, 判别系数Discriminant function, 判别值Dispersion, 散布/分散度Disproportional, 不成比例的Disproportionate sub-class numbers, 不成比例次级组含量Distribution free, 分布无关性/免分布Distribution shape, 分布形状Distribution-free method, 任意分布法Distributive laws, 分配律Disturbance, 随机扰动项Dose response curve, 剂量反应曲线Double blind method, 双盲法Double blind trial, 双盲试验Double exponential distribution, 双指数分布Double logarithmic, 双对数Downward rank, 降秩Dual-space plot, 对偶空间图DUD, 无导数方法Duncan's new multiple range method, 新复极差法/Duncan新法Effect, 实验效应Eigenvalue, 特征值Eigenvector, 特征向量Ellipse, 椭圆Empirical distribution, 经验分布Empirical probability, 经验概率单位Enumeration data, 计数资料Equal sun-class number, 相等次级组含量Equally likely, 等可能Equivariance, 同变性Error, 误差/错误Error of estimate, 估计误差Error type I, 第一类错误Error type II, 第二类错误Estimand, 被估量Estimated error mean squares, 估计误差均方Estimated error sum of squares, 估计误差平方和Euclidean distance, 欧式距离Event, 事件Event, 事件Exceptional data point, 异常数据点Expectation plane, 期望平面Expectation surface, 期望曲面Expected values, 期望值Experiment, 实验Experimental sampling, 试验抽样Experimental unit, 试验单位Explanatory variable, 说明变量Exploratory data analysis, 探索性数据分析Explore Summarize, 探索-摘要Exponential curve, 指数曲线Exponential growth, 指数式增长EXSMOOTH, 指数平滑方法Extended fit, 扩充拟合Extra parameter, 附加参数Extrapolation, 外推法Extreme observation, 末端观测值Extremes, 极端值/极值F distribution, F分布F test, F检验Factor, 因素/因子Factor analysis, 因子分析Factor Analysis, 因子分析Factor score, 因子得分Factorial, 阶乘Factorial design, 析因试验设计False negative, 假阴性False negative error, 假阴性错误Family of distributions, 分布族Family of estimators, 估计量族Fanning, 扇面Fatality rate, 病死率Field investigation, 现场调查Field survey, 现场调查Finite population, 有限总体Finite-sample, 有限样本First derivative, 一阶导数First principal component, 第一主成分First quartile, 第一四分位数Fisher information, 费雪信息量Fitted value, 拟合值Fitting a curve, 曲线拟合Fixed base, 定基Fluctuation, 随机起伏Forecast, 预测Four fold table, 四格表Fourth, 四分点Fraction blow, 左侧比率Fractional error, 相对误差Frequency, 频率Frequency polygon, 频数多边图Frontier point, 界限点Function relationship, 泛函关系Gamma distribution, 伽玛分布Gauss increment, 高斯增量Gaussian distribution, 高斯分布/正态分布Gauss-Newton increment, 高斯-牛顿增量General census, 全面普查GENLOG (Generalized liner models), 广义线性模型Geometric mean, 几何平均数Gini's mean difference, 基尼均差GLM (General liner models), 一般线性模型Goodness of fit, 拟和优度/配合度Gradient of determinant, 行列式的梯度Graeco-Latin square, 希腊拉丁方Grand mean, 总均值Gross errors, 重大错误Gross-error sensitivity, 大错敏感度Group averages, 分组平均Grouped data, 分组资料Guessed mean, 假定平均数Half-life, 半衰期Hampel M-estimators, 汉佩尔M估计量Happenstance, 偶然事件Harmonic mean, 调和均数Hazard function, 风险均数Hazard rate, 风险率Heading, 标目Heavy-tailed distribution, 重尾分布Hessian array, 海森立体阵Heterogeneity, 不同质Heterogeneity of variance, 方差不齐Hierarchical classification, 组内分组Hierarchical clustering method, 系统聚类法High-leverage point, 高杠杆率点HILOGLINEAR, 多维列联表的层次对数线性模型Hinge, 折叶点Histogram, 直方图Historical cohort study, 历史性队列研究Holes, 空洞HOMALS, 多重响应分析Homogeneity of variance, 方差齐性Homogeneity test, 齐性检验Huber M-estimators, 休伯M估计量Hyperbola, 双曲线Hypothesis testing, 假设检验Hypothetical universe, 假设总体Impossible event, 不可能事件Independence, 独立性Independent variable, 自变量Index, 指标/指数Indirect standardization, 间接标准化法Individual, 个体Inference band, 推断带Infinite population, 无限总体Infinitely great, 无穷大Infinitely small, 无穷小Influence curve, 影响曲线Information capacity, 信息容量Initial condition, 初始条件Initial estimate, 初始估计值Initial level, 最初水平Interaction, 交互作用Interaction terms, 交互作用项Intercept, 截距Interpolation, 内插法Interquartile range, 四分位距Interval estimation, 区间估计Intervals of equal probability, 等概率区间Intrinsic curvature, 固有曲率Invariance, 不变性Inverse matrix, 逆矩阵Inverse probability, 逆概率Inverse sine transformation, 反正弦变换Iteration, 迭代Jacobian determinant, 雅可比行列式Joint distribution function, 分布函数Joint probability, 联合概率Joint probability distribution, 联合概率分布K means method, 逐步聚类法Kaplan-Meier, 评估事件的时间长度Kaplan-Merier chart, Kaplan-Merier图Kendall's rank correlation, Kendall等级相关Kinetic, 动力学Kolmogorov-Smirnove test, 柯尔莫哥洛夫-斯米尔诺夫检验Kruskal and Wallis test, Kruskal及Wallis检验/多样本的秩和检验/H检验Kurtosis, 峰度Lack of fit, 失拟Ladder of powers, 幂阶梯Lag, 滞后Large sample, 大样本Large sample test, 大样本检验Latin square, 拉丁方Latin square design, 拉丁方设计Leakage, 泄漏Least favorable configuration, 最不利构形Least favorable distribution, 最不利分布Least significant difference, 最小显著差法Least square method, 最小二乘法Least-absolute-residuals estimates, 最小绝对残差估计Least-absolute-residuals fit, 最小绝对残差拟合Least-absolute-residuals line, 最小绝对残差线Legend, 图例L-estimator, L估计量L-estimator of location, 位置L估计量L-estimator of scale, 尺度L估计量Level, 水平Life expectance, 预期期望寿命Life table, 寿命表Life table method, 生命表法Light-tailed distribution, 轻尾分布Likelihood function, 似然函数Likelihood ratio, 似然比line graph, 线图Linear correlation, 直线相关Linear equation, 线性方程Linear programming, 线性规划Linear regression, 直线回归Linear Regression, 线性回归Linear trend, 线性趋势Loading, 载荷Location and scale equivariance, 位置尺度同变性Location equivariance, 位置同变性Location invariance, 位置不变性Location scale family, 位置尺度族Log rank test, 时序检验Logarithmic curve, 对数曲线Logarithmic normal distribution, 对数正态分布Logarithmic scale, 对数尺度Logarithmic transformation, 对数变换Logic check, 逻辑检查Logistic distribution, 逻辑斯特分布Logit transformation, Logit转换LOGLINEAR, 多维列联表通用模型Lognormal distribution, 对数正态分布Lost function, 损失函数Low correlation, 低度相关Lower limit, 下限Lowest-attained variance, 最小可达方差LSD, 最小显著差法的简称Lurking variable, 潜在变量Main effect, 主效应Major heading, 主辞标目Marginal density function, 边缘密度函数Marginal probability, 边缘概率Marginal probability distribution, 边缘概率分布Matched data, 配对资料Matched distribution, 匹配过分布Matching of distribution, 分布的匹配Matching of transformation, 变换的匹配Mathematical expectation, 数学期望Mathematical model, 数学模型Maximum L-estimator, 极大极小L 估计量Maximum likelihood method, 最大似然法Mean, 均数Mean squares between groups, 组间均方Mean squares within group, 组内均方Means (Compare means), 均值-均值比较Median, 中位数Median effective dose, 半数效量Median lethal dose, 半数致死量Median polish, 中位数平滑Median test, 中位数检验Minimal sufficient statistic, 最小充分统计量Minimum distance estimation, 最小距离估计Minimum effective dose, 最小有效量Minimum lethal dose, 最小致死量Minimum variance estimator, 最小方差估计量MINITAB, 统计软件包Minor heading, 宾词标目Missing data, 缺失值Model specification, 模型的确定Modeling Statistics , 模型统计Models for outliers, 离群值模型Modifying the model, 模型的修正Modulus of continuity, 连续性模Morbidity, 发病率Most favorable configuration, 最有利构形Multidimensional Scaling (ASCAL), 多维尺度/多维标度Multinomial Logistic Regression , 多项逻辑斯蒂回归Multiple comparison, 多重比较Multiple correlation , 复相关Multiple covariance, 多元协方差Multiple linear regression, 多元线性回归Multiple response , 多重选项Multiple solutions, 多解Multiplication theorem, 乘法定理Multiresponse, 多元响应Multi-stage sampling, 多阶段抽样Multivariate T distribution, 多元T分布Mutual exclusive, 互不相容Mutual independence, 互相独立Natural boundary, 自然边界Natural dead, 自然死亡Natural zero, 自然零Negative correlation, 负相关Negative linear correlation, 负线性相关Negatively skewed, 负偏Newman-Keuls method, q检验NK method, q检验No statistical significance, 无统计意义Nominal variable, 名义变量Nonconstancy of variability, 变异的非定常性Nonlinear regression, 非线性相关Nonparametric statistics, 非参数统计Nonparametric test, 非参数检验Nonparametric tests, 非参数检验Normal deviate, 正态离差Normal distribution, 正态分布Normal equation, 正规方程组Normal ranges, 正常范围Normal value, 正常值Nuisance parameter, 多余参数/讨厌参数Null hypothesis, 无效假设Numerical variable, 数值变量Objective function, 目标函数Observation unit, 观察单位Observed value, 观察值One sided test, 单侧检验One-way analysis of variance, 单因素方差分析Oneway ANOVA , 单因素方差分析Open sequential trial, 开放型序贯设计Optrim, 优切尾Optrim efficiency, 优切尾效率Order statistics, 顺序统计量Ordered categories, 有序分类Ordinal logistic regression , 序数逻辑斯蒂回归Ordinal variable, 有序变量Orthogonal basis, 正交基Orthogonal design, 正交试验设计Orthogonality conditions, 正交条件ORTHOPLAN, 正交设计Outlier cutoffs, 离群值截断点Outliers, 极端值OVERALS , 多组变量的非线性正规相关Overshoot, 迭代过度Paired design, 配对设计Paired sample, 配对样本Pairwise slopes, 成对斜率Parabola, 抛物线Parallel tests, 平行试验Parameter, 参数Parametric statistics, 参数统计Parametric test, 参数检验Partial correlation, 偏相关Partial regression, 偏回归Partial sorting, 偏排序Partials residuals, 偏残差Pattern, 模式Pearson curves, 皮尔逊曲线Peeling, 退层Percent bar graph, 百分条形图Percentage, 百分比Percentile, 百分位数Percentile curves, 百分位曲线Periodicity, 周期性Permutation, 排列P-estimator, P估计量Pie graph, 饼图Pitman estimator, 皮特曼估计量Pivot, 枢轴量Planar, 平坦Planar assumption, 平面的假设PLANCARDS, 生成试验的计划卡Point estimation, 点估计Poisson distribution, 泊松分布Polishing, 平滑Polled standard deviation, 合并标准差Polled variance, 合并方差Polygon, 多边图Polynomial, 多项式Polynomial curve, 多项式曲线Population, 总体Population attributable risk, 人群归因危险度Positive correlation, 正相关Positively skewed, 正偏Posterior distribution, 后验分布Power of a test, 检验效能Precision, 精密度Predicted value, 预测值Preliminary analysis, 预备性分析Principal component analysis, 主成分分析Prior distribution, 先验分布Prior probability, 先验概率Probabilistic model, 概率模型probability, 概率Probability density, 概率密度Product moment, 乘积矩/协方差Profile trace, 截面迹图Proportion, 比/构成比Proportion allocation in stratified random sampling, 按比例分层随机抽样Proportionate, 成比例Proportionate sub-class numbers, 成比例次级组含量Prospective study, 前瞻性调查Proximities, 亲近性Pseudo F test, 近似F检验Pseudo model, 近似模型Pseudosigma, 伪标准差Purposive sampling, 有目的抽样QR decomposition, QR分解Quadratic approximation, 二次近似Qualitative classification, 属性分类Qualitative method, 定性方法Quantile-quantile plot, 分位数-分位数图/Q-Q图Quantitative analysis, 定量分析Quartile, 四分位数Quick Cluster, 快速聚类Radix sort, 基数排序Random allocation, 随机化分组Random blocks design, 随机区组设计Random event, 随机事件Randomization, 随机化Range, 极差/全距Rank correlation, 等级相关Rank sum test, 秩和检验Rank test, 秩检验Ranked data, 等级资料Rate, 比率Ratio, 比例Raw data, 原始资料Raw residual, 原始残差Rayleigh's test, 雷氏检验Rayleigh's Z, 雷氏Z值Reciprocal, 倒数Reciprocal transformation, 倒数变换Recording, 记录Redescending estimators, 回降估计量Reducing dimensions, 降维Re-expression, 重新表达Reference set, 标准组Region of acceptance, 接受域Regression coefficient, 回归系数Regression sum of square, 回归平方和Rejection point, 拒绝点Relative dispersion, 相对离散度Relative number, 相对数Reliability, 可靠性Reparametrization, 重新设置参数Replication, 重复Report Summaries, 报告摘要Residual sum of square, 剩余平方和Resistance, 耐抗性Resistant line, 耐抗线Resistant technique, 耐抗技术R-estimator of location, 位置R估计量R-estimator of scale, 尺度R估计量Retrospective study, 回顾性调查Ridge trace, 岭迹Ridit analysis, Ridit分析Rotation, 旋转Rounding, 舍入Row, 行Row effects, 行效应Row factor, 行因素RXC table, RXC表Sample, 样本Sample regression coefficient, 样本回归系数Sample size, 样本量Sample standard deviation, 样本标准差Sampling error, 抽样误差SAS(Statistical analysis system ), SAS统计软件包Scale, 尺度/量表Scatter diagram, 散点图Schematic plot, 示意图/简图Score test, 计分检验Screening, 筛检SEASON, 季节分析Second derivative, 二阶导数Second principal component, 第二主成分SEM (Structural equation modeling), 结构化方程模型Semi-logarithmic graph, 半对数图Semi-logarithmic paper, 半对数格纸Sensitivity curve, 敏感度曲线Sequential analysis, 贯序分析Sequential data set, 顺序数据集Sequential design, 贯序设计Sequential method, 贯序法Sequential test, 贯序检验法Serial tests, 系列试验Short-cut method, 简捷法Sigmoid curve, S形曲线Sign function, 正负号函数Sign test, 符号检验Signed rank, 符号秩Significance test, 显著性检验Significant figure, 有效数字Simple cluster sampling, 简单整群抽样Simple correlation, 简单相关Simple random sampling, 简单随机抽样Simple regression, 简单回归simple table, 简单表Sine estimator, 正弦估计量Single-valued estimate, 单值估计Singular matrix, 奇异矩阵Skewed distribution, 偏斜分布Skewness, 偏度Slash distribution, 斜线分布Slope, 斜率Smirnov test, 斯米尔诺夫检验Source of variation, 变异来源Spearman rank correlation, 斯皮尔曼等级相关Specific factor, 特殊因子Specific factor variance, 特殊因子方差Spectra , 频谱Spherical distribution, 球型正态分布Spread, 展布SPSS(Statistical package for the social science), SPSS统计软件包Spurious correlation, 假性相关Square root transformation, 平方根变换Stabilizing variance, 稳定方差Standard deviation, 标准差Standard error, 标准误Standard error of difference, 差别的标准误Standard error of estimate, 标准估计误差Standard error of rate, 率的标准误Standard normal distribution, 标准正态分布Standardization, 标准化Starting value, 起始值Statistic, 统计量Statistical control, 统计控制Statistical graph, 统计图Statistical inference, 统计推断Statistical table, 统计表Steepest descent, 最速下降法Stem and leaf display, 茎叶图Step factor, 步长因子Stepwise regression, 逐步回归Storage, 存Strata, 层（复数）Stratified sampling, 分层抽样Stratified sampling, 分层抽样Strength, 强度Stringency, 严密性Structural relationship, 结构关系Studentized residual, 学生化残差/t化残差Sub-class numbers, 次级组含量Subdividing, 分割Sufficient statistic, 充分统计量Sum of products, 积和Sum of squares, 离差平方和Sum of squares about regression, 回归平方和Sum of squares between groups, 组间平方和Sum of squares of partial regression, 偏回归平方和Sure event, 必然事件Survey, 调查Survival, 生存分析Survival rate, 生存率Suspended root gram, 悬吊根图Symmetry, 对称Systematic error, 系统误差Systematic sampling, 系统抽样Tags, 标签Tail area, 尾部面积Tail length, 尾长Tail weight, 尾重Tangent line, 切线Target distribution, 目标分布Taylor series, 泰勒级数Tendency of dispersion, 离散趋势Testing of hypotheses, 假设检验Theoretical frequency, 理论频数Time series, 时间序列Tolerance interval, 容忍区间Tolerance lower limit, 容忍下限Tolerance upper limit, 容忍上限Torsion, 扰率Total sum of square, 总平方和Total variation, 总变异Transformation, 转换Treatment, 处理Trend, 趋势Trend of percentage, 百分比趋势Trial, 试验Trial and error method, 试错法Tuning constant, 细调常数Two sided test, 双向检验Two-stage least squares, 二阶最小平方Two-stage sampling, 二阶段抽样Two-tailed test, 双侧检验Two-way analysis of variance, 双因素方差分析Two-way table, 双向表Type I error, 一类错误/α错误Type II error, 二类错误/β错误UMVU, 方差一致最小无偏估计简称Unbiased estimate, 无偏估计Unconstrained nonlinear regression , 无约束非线性回归Unequal subclass number, 不等次级组含量Ungrouped data, 不分组资料Uniform coordinate, 均匀坐标Uniform distribution, 均匀分布Uniformly minimum variance unbiased estimate, 方差一致最小无偏估计Unit, 单元Unordered categories, 无序分类Upper limit, 上限Upward rank, 升秩Vague concept, 模糊概念Validity, 有效性VARCOMP (Variance component estimation), 方差元素估计Variability, 变异性Variable, 变量Variance, 方差Variation, 变异Varimax orthogonal rotation, 方差最大正交旋转Volume of distribution, 容积W test, W检验Weibull distribution, 威布尔分布Weight, 权数Weighted Chi-square test, 加权卡方检验/Cochran检验Weighted linear regression method, 加权直线回归Weighted mean, 加权平均数Weighted mean square, 加权平均方差Weighted sum of square, 加权平方和Weighting coefficient, 权重系数Weighting method, 加权法W-estimation, W估计量W-estimation of location, 位置W估计量Width, 宽度Wilcoxon paired test, 威斯康星配对法/配对符号秩和检验Wild point, 野点/狂点Wild value, 野值/狂值Winsorized mean, 缩尾均值Withdraw, 失访Youden's index, 尤登指数Z test, Z检验Zero correlation, 零相关Z-transformation, Z变换。

学校代码：10004密级：公开北京交通大学硕士专业学位论文OLED发光层对器件视角特性影响的研究Research on the Influence of OLED Light-emitting Layer on the Viewing Angle Characteristics of the Device作者姓名：马子杰学号：18126212导师姓名：徐叙瑢徐征职称：院士教授工程硕士专业领域：光学工程学位级别：硕士北京交通大学2020年6月致谢时间飞逝，我的研究生生活即将结束了，在研究生期间，我的老师、公司主管、同事、同学、父母都给与了我非常大的帮助，在此我要向他们表达衷心的谢意。

首先我要感谢我的老师徐征教授，徐老师是发光与显示行业的资深专家，是我在从事发光与显示行业的领路人，在我的日常学习和工作中对我进行了耐心的指导。

徐老师不仅传授给我们所必须的理论知识和严谨认真的工作态度，更教会了我们为人处事之道，给我们提供了非常好的锻炼成长的机会，对我们后续的工作起到了非常大的帮助，在此衷心的感谢徐征老师对我的指导和帮助。

我还要感谢徐叙瑢院士，徐先生是中国发光学的奠基人之一，徐先生务实求真的治学态度、杰出的科研水平、虚怀若谷的高尚品格都是我们学习的榜样。

此外，还要感谢课题组的赵谡玲教授、宋丹丹教授、乔泊教授、冀国瑞老师在学习生活中给予我的帮助。

感谢维信诺科技股份有限公司OLED器件部给与了我为期一年的实习机会，可以让我在这里磨炼自己。

感谢公司周小康博士对我的指导和帮助，作为联合培养的公司里的导师，周小康博士在学习和工作上给与了我极大的帮助。

他做事严谨，知识面广，为人随和，幽默风趣。

无论是在工作还是学习上都是我要学习的典范，能跟随他学习也是我的一笔财富。

在此衷心的感谢周小康博士对我的指导。

感谢我的同事秦齐齐、李晓聪、娄振花在实验中给了我极大的帮助，以及工作和论文方面提出的许多宝贵的意见，感谢我的父母，给与了我一个安定愉快的学习和生活环境，感谢1807班全体同学。