基于时频域特征分析和ML-NN的故障电弧检测与选线

Fault arc detection and line selection based on time-frequency domain feature analysis and ML-NN

针对低压配电系统方式复杂、负载种类繁多、串联故障电弧的检测难度越来越大的问题,提出了1种基于时频域特征分析和多标签神经网络(ML-NN)分类的串联故障电弧检测与选线的方法.该方法通过采集多回路负载的不同支路发生电弧时的干路电流,对其时域采取统计的方法对故障电流的方差、均值、偏度和峰度进行分析,对其频域采用小波变换的方法得到其故障电流的小波系数特征.将时频域特征作为神经网络的输入进行训练,同时采用反向传播方法来训练模型,实现故障电弧检测和故障选线.经过实验验证,故障电弧检测和选线的准确度分别达到了97.57%、99%.

Aiming at the problems that series fault arc detection is more and more difficult in low-voltage distribu⁃tion system with complex modes and various loads,a method of series fault arc detection and line selection based on time-frequency domain feature analysis and multi-label neural network(ML-NN)classification was proposed.In this method,the variance,mean value,skewness and kurtosis of the fault current are analyzed by statistical meth⁃od in time domain,and the wavelet transform method is used to get the wavelet coefficient characteristics of the fault current in frequency domain.The time-frequency domain features are used as the input of neural network for train⁃ing,and the back propagation method is used to train the model to realize fault arc detection and fault line selec⁃tion.The accuracy of fault arc detection and line selection reaches 97.57%and 99%respectively.