搜索
本科毕业设计（论文）基于RBF神经网络的短期负荷预测学院自动化学院专业电气工程及其自动化___（电力系统自动化方向）年级班别2007级（3）班学号3107001208学生姓名郭祝帆指导教师彭显刚2011年5月基于RBF神经网络的短期负荷预测郭祝帆自动化学院I摘要电力系统负荷预测的水平已成为衡量电力系统运行管理现代化的标志之一。精确的短期负荷预测，对电力系统的生产安排、经济调度和安全分析都起着十分重要的作用，也直接影响着电力企业的经济效益。因此，短期负荷预测结果成为制定电力市场交易计划的重要依据，这就对短期负荷预测提出了更高的要求。由于常规算法不能较好地反映气象条件等外界因素对负荷的影响，而近年来人工神经网络法等智能算法具有高度的非线性映射能力，可以较好地考虑气象条件等因素对电网负荷的影响，所以本文采用了基于RBF(RadialBasisFunction)神经网络的电力系统短期负荷预测方法。该模型训练速度快，收敛性好，而且可以大大地减少隐含层神经元的数目，有效地提高了预测精度。本文在分析了目前短期电力负荷预测的现状及各种预测方法、预测模型的基础上，根据电力负荷特性的变化规律，通过对河源地区的历史负荷数据分析，考虑了日期类型、温度、天气状况等影响负荷预测的因素，结合神经网络的预测算法，建立RBF神经网络的短期负荷预测数学模型，并在此基础上，利用面向对象的编程方法实现短期负荷预测程序。本文讨论了影响负荷的各种因素，在输入变量中考虑临近日负荷特点，以及各种气象因素，对输入负荷值进行归一化处理，对温度、天气和日期等因素进行了量化处理。利用河源地区的历史负荷数据比较未含天气因素的神经网络和具有天气因素的神经网络的预测效果，根据本文所介绍的方法编程，其结果表明预测精度是符合要求的，从而说明了该方法的可行性和实用性。关键词：短期负荷预测，RBF神经网络，编程IIAbstractThelevelofloadforecastingisoneofthemeasuresofmodernizationofPowersystemmanagement.Accurateshort-termloadforecastingplaysanimportantroleforplanning,economicalschedulingandsecurityanalysisinproduction,whichdirectlyinfluencestheprofitoftheelectricutilityenterprises.Therefore,short-termloadforecastingreseultbecomeimportancebasisofdrawinguptheelectricpowermarketbargainplan.Sotheseputshort-termloadforecastingforwardahigherrequest.Thenormalcalculatewaycannotreflectgoodlyweatherconditionandotheroutsidefactorstotheinfluenceforloadforecasting.Inrecentyears,theartificialneuralnetworkmethodetchaveheightnonlineartoreflecttheabilityofshoot,canreflectgoodlytheweatherfactoretc.Sothispaperpresentsashort-termloadforecastmethodbasedonRBF(RadialBasicFunction)neuralnetworkforpowersystem.Thismodelspeedsrapidly,improvesconvergencepropertyintrainingprocessandthenumberofneuronsinthehiddenlayercanbesignificantlydecreased.Sotheforecastingaccuracycanbeincreasedeffectively.Thistextanalyzethepresentconditionandvariousmethodsandmathematicsmodeloftheshort-termloadforecasting.Accordingtotheruleofchangeofloadcharacteristic,theRBFmodelsfortheshort-termloadforecastingareproposedbycombiningtheartificialneuralnetworksandelectricloadcharacteristicsonHeYuanPowerMarkets,aftercalculatingthefactorssuchasdatetype,temperature,weatherstatusetcwhichinfluencingtheloadforecasting.Basedonthemodels,theloadforecastingsoftwarehasprogrammedbyObjectOrientedmethod.Thisthesisanalyzeseverykindoffactorwhichimpactsload.Initsinputfeatures,theloadcharacteristicofneatdayseverykindofweatherfactorsthatconsidered.Thenweunifytheinputvariables,quantifythetemperature,weatheranddateetc.TheforecastingaccuracyofneuralnetworksmodelsincludingclimatefactorsandnothosefactorsiscomparedbytheloaddatafromHeYuan.Thetestingresultsillustratethattheforecastingaccuracyissatisfactory,accordinglyitshowsthevalidityandpracticabilityofthemethod.Keywords:Neuralnetwork,RBF,Short-termloadforecasting目录1绪论........................................................................................................................................11．1课题研究的背景........................................................................................................11．2国内外负荷预测的研究现状....................................................................................21．3本课题研究的意义....................................................................................................51．4本课题的主要工作....................................................................................................62电力负荷预测概述................................................................................................................72.1负荷预测的概念和原理..............................................................................................72.1.1负荷预测的概念................................................................................................72.1.2负荷预测的基本原理........................................................................................72.2电力负荷预测的分类..................................................................................................82.3负荷预测的步骤........................................................................................................102.4电力负荷的特性分析................................................................................................112.4.1负荷的周期性..................................................................................................112.4.2负荷的随机性..................................................................................................122.4.3负荷的影响因素分析......................................................................................122.5影响负荷预测的因素及误差分析............................................................................142.5.1影响负荷预测的主要因素..............................................................................142.5.2负荷预测的误差分析......................................................................................142.6本章小结....................................................................................................................163RBF神经网络及其结构分析...............................................................................................173.1人工神经网络的基本概念......................................