資源描述:
《基于PSO—RBF監(jiān)測預(yù)測模型的電力電子電路-論文.pdf》由會員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在行業(yè)資料-天天文庫。
1、第15卷第2期上海應(yīng)用技術(shù)學(xué)院學(xué)報(bào)(自然科學(xué)版)V0l_15NO.22Ol5年6月JOURNALOFSHANGHAIINSTITUTEOFTECHNOLOGY(NATURALSCIENCE)Jun.2O15文章編號:1671—7333(2015)02—0162—05DOI:10.3969/j.issn.1671-7333.2015.02.013基于PSO—RBF監(jiān)測預(yù)測模型的電力電子電路王紳宇,陳丹江,葉銀忠(1.上海應(yīng)用技術(shù)學(xué)院電氣與電子工程學(xué)院,上海201418;2.浙江萬里學(xué)院電子信息學(xué)院,浙江寧波315100)摘要:針對現(xiàn)有電力電子
2、電路故障狀態(tài)預(yù)測技術(shù)的不足,提出將電路特征性能參數(shù)與粒子群算法(PSO)優(yōu)化的徑向基函數(shù)(RBF)神經(jīng)網(wǎng)絡(luò)相結(jié)合,對電力電子電路進(jìn)行故障狀態(tài)監(jiān)測預(yù)測.以電源電路中Buck電路為例,選擇電路輸出電壓作為監(jiān)測信號,提取輸出電壓平均值及紋波電壓值作為電路特征性能參數(shù),并利用改進(jìn)后的RBF神經(jīng)網(wǎng)絡(luò)實(shí)現(xiàn)狀態(tài)預(yù)測.結(jié)果表明,利用PSO改進(jìn)后的RBF神經(jīng)網(wǎng)絡(luò)對電路輸出平均電壓和紋波電壓的預(yù)測比單純RBF神經(jīng)網(wǎng)絡(luò)預(yù)測的結(jié)果更加精準(zhǔn),能夠跟蹤電源電路狀態(tài)特征性能參數(shù)的變化趨勢,有效實(shí)現(xiàn)電力電子電路狀態(tài)監(jiān)測和預(yù)測.關(guān)鍵詞:故障狀態(tài)預(yù)測;RBF神經(jīng)網(wǎng)絡(luò);粒子群算
3、法;電力電子電路中圖分類號:TP206.3文獻(xiàn)標(biāo)志碼:AFaultConditionMonitoringPredictionTechniqueofPowerElectroniCCircuitsBasedonPSO—RBFNeuralNetworkWANGShenyu,CHENDanjiang,YEYinzhong(1.SchoolofElectricalandElectronicEngineering,ShanghaiInstituteofTechnology,Shanghai201418,China;2.SchoolofElectroni
4、cInformation,ZhejiangWanliUniversity,Ningbo315100,Zhejiang,China)Abstract:Aimingattheissueoffaultconditionmonitoringpredictiontechniqueofpowerelectroniccircuits,amethodbasedoncharacteristicparameterdataandparticleswarmoptimization(PSO)radialbasisfunction(RBF)neuralnetworkf
5、orthefaultconditionmonitoringpredictionofpowerelectroniccircuitswasproposed.TheBuckconvertercircuitwastakenasanexample,thentheaveragevoltagewasextractedascharacteristicparameters,thefaultpredictionofpowerelectroniccircuitswasachieved.Theoutputvoltagewasselectedasmonitoring
6、signal,thentheaveragevoltagewasextractedascharacteristicparameters.PSO—RBFneuralnetworkwasusedtopredicttheBuckconvertercircuit.TheexperimentalresultsshowedthatthePSO-RBFneuralnetworkwasmoreaccurateinpredictingthanthatoftheonlyRBFneuralnetwork.Thenewmethodcouldtracethechara
7、cteristicparameterstrendandcouldbeeffectivelyappliedinfaultconditionmonitoringpredictionofpowerelectroniccircuits.Keywords:conditionmonitoringprediction;RBFneuralnetwork;particleswarmoptimization(PSO);powerelectroniccircuits收稿日期:2015-01—10基金項(xiàng)目:國家自然科學(xué)基金資助項(xiàng)目(61374132)第一作者:王紳
8、宇(1988一),男,碩士生,主要研究方向?yàn)楣收显\斷與容錯控制.E—mail:Wangshenyu00@163.com通信作者:葉銀忠(1964一),男,教授,博士生導(dǎo)師,主要研