基于改进BP神经网络的充电模块故障状态识别

Fault State Identification of Charging Module Based on Improved BP Neural Network

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。

In order to improve the effect of charging module fault on the operation quality of charging pile,a method of charging module fault state identification based on improved BP neural network is studied.The kurtosis of the vibration signal envelope of the charging module is used to detect the spike of the vibration signal envelope of the charging module,determine the starting point of the vibration signal of the charging module and extract the fault characteristics of the charging module through KS test method.BP neural network is selected as the charging module fault state recognition method and the BP neural network is improved by using the optimization performance of genetic algorithm.The extracted charging module fault features are input into the improved BP neural network to obtain the charging module fault state recognition results.The experimental results show that the method can accurately identify the fault state of the charging module and improve the problem of high harmonic content of the grid side current when the charging pile is connected to the distribution network.