基于小波变换的交流系统串联电弧故障诊断

Series Arc Fault Diagnosis Based on Wavelet Transform in AC System

为实现对用电系统低压用户端中串联电弧故障的准确诊断,根据交流系统中低压串联电弧故障特性,通过自主搭建的电弧故障模拟实验平台及不同负载下的串联电弧故障模拟实验,本文提出一种基于小波变换的串联电弧故障诊断方法。该方法首先采用极大极小原理对信号进行降噪处理,并结合小波变换模极大值对信号进行多分辨分析;将三阶Daubechies小波基函数提取出的各频段细节信号模极大值作为网络输入的特征向量,利用基于阻尼最小二乘法改进的多层前馈(Back Propagation,BP)神经网络构建特征向量与电弧故障之间的映射关系进行故障诊断分类。测试结果表明,该方法可有效实现交流系统中串联电弧故障的诊断分类。

For diagnosing the series arc fault in the power system accurately, according to the characteristic of series arc fault in AC system, this paper presents a method of series arc fault analysis and diagnosis based on the wavelet transform using the experimental platform of arc fault and arcing fault experiment in different loads. Combined with the wavelet transform module maxima and multi-resolution analysis, the db3 wavelet function is used to extract the feature vector in each frequency band of both the normal and fault signals which are denoised by the minimax principle. With the improved BP neural network based on the Levenberg-Marquarat, the fault is diagnosed by the mapping relationship between the feature vector and arc fault. The result of test shows that the proposed method can realize the diagnosis and classification of the series arc fault in AC system effectively.