欢迎来到天天文库
浏览记录
ID:34876874
大小:2.15 MB
页数:61页
时间:2019-03-13
《基于线性神经网络的配电网谐波检测技术研究》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、摘要随着一些新技术在电力系统中的广泛应用,谐波问题也变得愈加严重。这不仅对电力系统的安全和经济运行造成严重危害,而且还可能会影响到电力用户的正常工作,给用户带来不必要的经济损失。因此,采取有效的谐波检测措施有着相当重要的现实意义。本文先对现有谐波检测方法进行对比分析,而后针对配电网的谐波特点,采用基于BP神经网络和线性神经网络相结合的方法对配电谐波进行精确检测。课题主要思路是先利用BP神经网络原理及方法构造出基波检测模型,运用最速下降算法对配电网中的基波分量进行精确的检测。以检测出的基波为参数依据,再采用线性神经网络方法对配电网中各类谐波进行检测。由于整次谐波和非整次谐波各自
2、特点不同,所以本文有针对性的分别构造线性神经网络整次谐波和非整次谐波的检测模型,并详细的推导了相关算法的实现过程。最后本文对各个谐波检测阶段分别在无噪和有噪的环境下进行了仿真实验和结果分析。实验结果表明,本文所采用的结合BP神经网络和线性神经网络的配电网谐波检测方法具有较高的检测精度,较传统加汉宁窗FFT算法高出2~3个数量级,且表现出了很好的抗噪能力。关键词:配电网;谐波检测;神经网络;LMS算法;加汉宁窗FFTIAbstractWithsomeofthenewtechnologyiswidelyusedinpowersystems,harmonicproblemsbeco
3、memoreserious.Thisnotonlycausesseriousharmtothesecurityandeconomicoperationofthepowersystem,butalsomayaffectthenormaloperationofpowerusers,givingusersunnecessaryeconomiclosses.Therefore,totakeeffectivemeasuresharmonicdetectionhasaveryimportantpracticalsignificance.Inthispaper,thecurrentharm
4、onicdetectionmethodwereanalyzedfirst,andthenfortheharmoniccharacteristicsofthedistributionnetwork,theuseofBPneuralnetworkandlinearneuralnetworkbasedonacombinationofmethodstoaccuratelydetectharmonicdistribution.ThemainideaistousethesubjectofBPneuralnetworktoconstructthefundamentalprinciplesa
5、ndmethodsofdetectionmodel,usingthesteepestdescentalgorithmforthefundamentalcomponentofthedistributionnetworkforaccuratetesting.Todetectthefundamentalbasisforthearguments,andthenusinglinearneuralnetworkmethodfordistributionnetworkinallkindsofharmonicdetection.Becausethewholeharmonicandnon-in
6、tegerharmonicstheirdifferentcharacteristics,sothistargetedlinearneuralnetworksareconstructedandthewholeentireharmonicharmonicdetectionmodelnon,andadetailedderivationoftheimplementationprocessrelatedalgorithms.Thelastarticleintheabsenceofnoise,respectively,ofeachharmonicdetectionphaseandnois
7、yenvironmentforthesimulationandanalysisofresults.Experimentalresultsshowthat,withaharmonicdetectionmethodcombinesBPneuralnetworkandlinearneuralnetworkusedinthispaperhashigherdetectionaccuracythanthetraditionalFFTalgorithmHanningwindowup2to3ordersofmagnit
此文档下载收益归作者所有