資源描述:
《基于遺傳算法優(yōu)化BP神經(jīng)網(wǎng)絡(luò)接觸電阻預(yù)測(cè)》由會(huì)員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、基于遺傳算法優(yōu)化BP神經(jīng)網(wǎng)絡(luò)的接觸電阻預(yù)測(cè)孫海峰,沈穎,王亞楠(華北電力大學(xué)電氣與電子工程學(xué)院,河北保定071003)摘要:接觸電阻是反應(yīng)導(dǎo)體間電接觸性能的重要參數(shù),在實(shí)際的工程中往往采用經(jīng)驗(yàn)公式對(duì)接觸電阻進(jìn)行計(jì)算,精度難以滿足要求。為解決這一問題,將遺傳算法(GA)與BP神經(jīng)網(wǎng)絡(luò)相結(jié)合對(duì)接觸電阻進(jìn)行預(yù)測(cè)。通過實(shí)驗(yàn)得到數(shù)據(jù),分別利用遺傳算法優(yōu)化BP神經(jīng)網(wǎng)絡(luò)、BP神經(jīng)網(wǎng)絡(luò)以及回歸分析模型進(jìn)行訓(xùn)練和測(cè)試,并將各算法所得誤差進(jìn)行對(duì)比。誤差對(duì)比結(jié)果表明:遺傳算法優(yōu)化BP神經(jīng)網(wǎng)絡(luò)的收斂效果優(yōu)于其他兩種算法,且
2、遺傳算法優(yōu)化BP神經(jīng)網(wǎng)絡(luò)所得接觸電阻模型的相對(duì)誤差平均值比BP神經(jīng)網(wǎng)絡(luò)減少了4.01%,比回歸分析減少了4.72%,且預(yù)測(cè)效果較穩(wěn)定。利用遺傳算法與BP神經(jīng)網(wǎng)絡(luò)相結(jié)合的接觸電阻預(yù)測(cè)模型較單獨(dú)使用BP神經(jīng)網(wǎng)絡(luò)預(yù)測(cè)模型具有更好的非線性擬合能力和更高的預(yù)測(cè)精度。關(guān)鍵詞:電接觸;接觸電阻;遺傳算法;BP神經(jīng)網(wǎng)絡(luò);回歸分析中圖分類號(hào):TM769文獻(xiàn)標(biāo)識(shí)碼:A文章編號(hào):1001-1390(2019)00-0000-00PredictionofcontactresistancebasedonoptimizedBP
3、neuralnetworkofgeneticalgorithmtooptimizeBPneuralnetworkSunHaifeng,ShenYing,WangYanan((SchoolofElectricalandElectronicalElectronicEngineering,NorthChinaElectricPowerUniversity,Baoding071003,Hebei,China))Abstract:Thecontactresistanceisanimportantparamet
4、eroftheelectricalcontactperformancebetweenthereactionconductors.Inpractice,theempiricalformulaisoftenusedtocalculatethecontactresistance,whichisdifficulttomeettheaccuracyrequirements.InorderTotosolvethisproblem,thegeneticalgorithm(GA)combinedwithBPneur
5、alnetworkisadoptedtopredictthecontactresistance.ThedataareobtainedThroughthroughexperiment,,thedataareobtained,andtheoptimizedBPneuralnetworkofgeneticalgorithmoptimizedBPneuralnetwork,theBPneuralnetworkandregressionanalysismodelarerespectivelyusedfortr
6、ainingandtesting,andtheerrorsobtainedbyeachalgorithmarecompared.TheresultsoferrorcomparisonshowthatgeneticalgorithmoptimizestheconvergenceeffectofBPneuralnetworkbetterthantheothertwoalgorithms,andtheaveragerelativeerrorofthecontactresistancemodelobtain
7、edbygeneticalgorithmoptimizationBPneuralnetworkisreducedby4.01%comparedwithBPneuralnetwork,whichislowerthantheregressionanalysis4.72%,andtheforecastingeffectismorestable.ThecontactresistancepredictionmodelusinggeneticalgorithmandBPneuralnetworkhasbette
8、rnonlinearfittingabilityandhigherpredictionaccuracythantheBPneuralnetworkpredictionmodelalone.Keywords:electricalcontact,contactresistance,geneticalgorithm,BPneuralnetwork,regressionanalysis0引言在電力系統(tǒng)、自動(dòng)控制系統(tǒng)和信息傳遞系統(tǒng)等領(lǐng)域廣泛存在著電接觸,如閥廳金具的連接處、繼電