基于改进BP神经网络的充电模块故障状态识别.pdfVIP

基于改进BP神经网络的充电模块故障状态识别

摘要：为提高充电桩运行质量，研究了基于改进BP神经网络的充电模块故障状态识别方法。利用充电模块振

动信号包络峭度检测充电模块振动信号包络线的尖峰度，确定充电模块振动信号起始点，通过KS检验方法

提取充电模块故障特征。选取BP神经网络作为充电模块故障状态识别方法，利用遗传算法的寻优性能对BP

神经网络进行改进，将所提取充电模块故障特征输入到改进的BP神经网络之中，以获得充电模块故障状态

识别结果。实验结果表明，该方法可精准识别充电模块故障状态，改善充电桩接入配电网的网侧电流谐波含

量过高问题。

关键词：充电模块；神经网络；故障状态识别

Abstract:Inordertoimprovetheeffectofchargingmodulefaultontheoperationqualityofchargingpile,amethod

ofchargingmodulefaultstateidentificationbasedonimprovedBPneuralnetworkisstudied.Thekurtosisofthevi-

brationsignalenvelopeofthechargingmoduleisusedtodetectthespikeofthevibrationsignalenvelopeofthe

chargingmodule,determinethestartingpointofthevibrationsignalofthechargingmoduleandextractthefault

characteristicsofthechargingmodulethroughKStestmethod.BPneuralnetworkisselectedasthechargingmodule

faultstaterecognitionmethodandtheBPneuralnetworkisimprovedbyusingtheoptimizationperformanceofgenet-

icalgorithm.TheextractedchargingmodulefaultfeaturesareinputintotheimprovedBPneuralnetworktoobtainthe

chargingmodulefaultstaterecognitionresults.Theexperimentalresultsshowthatthemethodcanaccuratelyidentify

thefaultstateofthechargingmoduleandimprovetheproblemofhighharmoniccontentofthegridsidecurrent

whenthechargingpileisconnectedtothedistributionnetwork.

Keywords:chargingmodule;neuralnetwork;faultstateidentification

FoundationProject:SupportedbyStateGridBeijingInstituteofElectricalScienceandTechnology2021High-power

ChargingEquipmentDetectionCapabilityImprovement(No.67022320003F)

1引言充电模块故障等。直流充电桩可能发生的故障中

充电模块故障发生率极高，主要原因是充电模块

风能、太阳能等能源广泛应用于发电领域中，

工作电压过高，组成充电模块的电力电子功率器

提升了电动汽车