基于小波奇异熵和支持向量机的配电网单相接地选线新方法

A Novel Approach to Fault Line Selection in Distribution Networks Based on Wavelet Singularity Entropy and Supporting Vector Machine

针对目前小电流系统接地系统选线方法存在的问题,提出了一种新的故障选线方法。对各条线路故障后一个周波内的零序电流作小波分析,并且将由系数构成的矩阵作奇异分解并得到对应的小波奇异熵(wavelet singular entropy,WSE),利用网格搜索法寻优,得到最优的非负惩罚因子参数C和核函数参数g。采用优化的支持向量机(support vector machine,SVM)对各条线路的WSE进行分类,即选出故障线路。仿真结果表明提出的方法不受噪声干扰、电弧重燃、采样不同步、补偿度不同等干扰因素对故障选线结果的影响,可以准确、可靠地实现故障选线。

In view of the problems existing in the selection method of the small current grounding system, this paper pro- poses a new fault line selection method. Zero sequence current within a period after fault of each line is analyzed by wavelet, the matrix consisting of wavelet coefficients is decomposed by singular value decomposition, and then the singular entropy can be obtained. The optimum parameters of non-negative penalty factor C and kernel function parameter g can be got through grid search method. The wavelet singular entropy is classified by optimized SVM, which can select the fault line accurately. The simulation resuhs verify that the proposed approach is accurate and reliable for the fault line selection.