基于小波变换和奇异值分解的串联电弧故障检测方法

发布时间：2018-06-21 02:00

本文选题：电弧故障 + 小波变换　；参考：《电工技术学报》2017年17期

【摘要】：根据线路中电流信号的变化来检测电弧故障,小波变换是一种常用的检测方法,但是单纯利用小波变换对于正常情况和电弧故障的区分并不明显,而且其结果存在很大的冗余。针对这一问题,提出了采用一种基于小波变换和奇异值分解的串联电弧故障检测的方法。利用电弧模拟发生装置产生串联故障电弧,采集在多种负载下线路正常工作和发生串联电弧故障时的电流。首先对采集的电流信号进行离散小波变换,得到离散小波系数序列,构造特征矩阵;然后对特征矩阵进行奇异值分解,并定义电流信号的特征参数,利用特征参数作为串联电弧故障检测的依据。试验结果表明:正常情况和电弧故障下的特征参数区分明显且没有交叉,易于确定阈值,利用该方法进行串联电弧故障检测的准确率较高,且大大压缩了小波变换结果的冗余性。
[Abstract]:Wavelet transform is a common detection method to detect arc fault according to the variation of current signal in the circuit, but the difference between normal condition and arc fault by wavelet transform is not obvious, and the result is very redundant. To solve this problem, a method of series arc fault detection based on wavelet transform and singular value decomposition is proposed. The series fault arc is generated by using the arc simulation generator to collect the current of the line working normally and when the series arc fault occurs under various loads. Firstly, discrete wavelet transform is carried out on the collected current signal, and the discrete wavelet coefficient sequence is obtained to construct the characteristic matrix, and then the singular value decomposition of the characteristic matrix is carried out, and the characteristic parameters of the current signal are defined. The characteristic parameters are used as the basis for fault detection of series arc. The experimental results show that the characteristic parameters of normal condition and arc fault are distinguished obviously and have no crossover, and the threshold is easy to be determined. The accuracy of detecting series arc fault by using this method is high. The redundancy of wavelet transform is greatly reduced.
【作者单位】：中国矿业大学(北京)机电与信息工程学院;
【基金】：国家自然科学基金项目资助(51577187)
【分类号】：TM501.2

【相似文献】