基于小波熵权和支持向量机的高压输电线路故障分类方法

High Voltage Transmission Line Fault Classification Based on Entropy Weight of Wavelet and Support Vector Machines

高压输电线路故障类型的正确识别是进行故障定位和事故分析的前提。为此,作者提出一种分层的故障类型识别方法,首先根据线路故障时三相电流小波熵权分布曲线相互间距离的差异、距离之和进行故障的初步归类,构造表征不同故障类别的样本,然后采用支持向量机算法对样本进行训练,得到识别不同故障类型的最优分类面。仿真结果表明:该方法识别速度快,克服了常规线性分类方法的局限性,且故障识别精度不受系统运行方式、过渡电阻以及故障位置的影响,具有较强的通用性和实用性。

Correct identification of faults occurred in high voltage transmission line is the presupposition of fault location and fault analysis. For this reason, the authors propose a layered fault type recognition method, in which at first according to the differences of mutual distances among entropy weight distribution curves of three-phase currents during the fault of transmission line and the sum of distances the fault type is preliminarily classified; then by means of support vector machine （SVM） technique the samples are trained to obtain optimal hyperplane for the recognition of different faults. Simulation results show that the recognition speed of the proposed method is rapid and the proposed method overcomes the limitations of conventional linear classification methods; meanwhile, the fault recognition accuracy is not affected by power system operation modes, transition resistance and fault position, thus the proposed method is versatile and practicable.