欢迎来到天天文库
浏览记录
ID:39402614
大小:273.50 KB
页数:77页
时间:2019-07-02
《基于支持向量机的静态电压稳定评估》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、西南交通大学硕士学位论文基于支持向量机的静态电压稳定评估姓名:赵万明申请学位级别:硕士专业:电力系统及其自动化指导教师:黄彦全20081201西南交通大学硕士研究生学位论文第1I页AbstractVoltagecollapsehasbecomeoneofthemostimportantproblemswhichhavethreatenedtheoperationsafetyofelectricpowersystems.Itisnecessarytoevaluatethedistancebetweentheoperationstateandthevoltagecriticalpoint
2、inordertoescapefromthevoltagecollapse.Bymeansofcalculatingthecriticalpoint,theloadingmargintovoltagecollapsecanbedetermined.Butatthecriticalpoint,theJacobianmatrixofconventionalpowerflowequationsbecomessingular.ThecontinuationpowerflowmethodofgettingthecriticalpointbytracingthePVcurvehasbeenapp
3、liedtoovercomethisdifficulty.ThecalculationspeedofthismethodiSslowforpowersystemswithhighdimension,SOitisdifficulttorealizereal-timevoltagestabilityassessment.Theapplicationofafasterandmorereliableevaluationtechniqueisveryimportanttoshortentheevaluationtime.Supportvectormachine(SVM),amethodbase
4、donstatisticslearningtheory,isamachinelearningalgorithmofthenewera.Itequalstosolvingaquadraticprogrammingproblemintheprincipleofminimumstructuralrisk.Thisalgorithmisfeaturedwithstrongforecastingability,globaloptimizationandfastspeedofapproaching.etc.HenceamethodofmodelconstructionwhichbasedOilS
5、VMispresentedforthepowersystemstaticvoltagestabilityassessment,andaSVMmodelforvoltagestabilityassessmentisestablishedinthisdissertation.Basingontheoperationstate,thecriticalpointcanbeestimatedbythemodelbeingtrainedaccordingtothetestresults.ThismethodtakesfulladvantageofSVM’Sabilitytosolvethepro
6、blemswithhighdimension,nonlinearandsmallsample.Hence,withquickerassessmentspeedandhigherforecastprecision,bettergeneralizationabilityisguaranteed.ComparedwithANNmodel,itcanbeseenthatSVMmodelhashigherprecision.Inmisdissertation,thefeaturesofthedatasetaleextractedbyusingprinciplecomponentanalysis
7、toreducetheinputdimension.Thentheprinciplecomponentcontainingtheinformationofsampledataissenttosupportvectorm.achinefortraining.Thisproposalcombinesthefeatureextractionabilityof西南交通大学硕士研究生学位论文第1II页principlecomponentanalysistogethe
此文档下载收益归作者所有