基于GWO-SVM模型的智能电气故障检测与识别

Intelligent Electrical Fault Detection and Recognition Based on GWO-SVM Model

针对常见的分类算法在电气故障诊断中分类准确度不高的问题,提出了一种灰狼优化支持向量机(GWO-SVM)模型来提高电气故障诊断的识别率。首先采集了现实生活中最常见的线性和非线性家用电器白炽灯和微波炉在正常工作和发生电弧故障时的波形信号;其次对其进行了频域特征提取;最后使用灰狼优化算法对支持向量机进行优化,并与未优化SVM和BP神经网络进行了对比。结果表明,GWO-SVM模型的正确率达到了90%,优于对比算法。

To address the issue of low accuracy in traditional classification algorithms for distinguishing electrical faults,a Grey Wolf Optimization-Support Vector Machine(GWO-SVM)model is proposed to improve electrical fault diagnosis accuracy.Firstly,waveform signals of commonly encountered linear and nonlinear household electrical appliances such as incandescent lamps and microwave ovens in normal operation and during arc faults are collected from real-life scenarios.Secondly,frequency domain features are extracted from these signals.Finally,the GWO algorithm optimizes the Support Vector Machine,and the performance of GWO-SVM is compared with unoptimized SVM and BP neural networks.The GWO-SVM model achieves an accuracy of 90%.