基于支持向量机的雾霾绝缘子泄漏电流分析

Analysis of Insulator Leakage Current under Fog and Haze Based on Support Vector Machine

雾霾天气中,绝缘子处于极为不利的运行状态,污闪事故发生的可能性大为增加。为研究雾霾天气对绝缘子运行造成的影响,在高压实验室中模拟了燃煤和汽车尾气导致的雾霾环境,测量了在不同雾霾成分作用下绝缘子表面的泄漏电流。对实验中采集到的泄漏电流信号利用小波阈值去噪法进行去噪处理,对不同的阈值选择方法所得到的去噪效果的优劣进行了比较,得到了最佳的阈值选择法。利用小波包分解和信息熵分析方法提取了泄漏电流信号的特征向量,并通过特征向量建立了基于支持向量机理论的分类模型。测试证明,该模型能够准确地区分不同运行环境下的绝缘子泄漏电流。

Under fog and haze weather, the insulator is operating in extremely unfavorable state, the possibility of flashover accident is greatly increased. In order to study the effect of haze weather on the operation of the insulator, haze environment caused by coal combustion and automobile exhaust was simulated in high pressme laboratory, and leakage current in insulator surface under different haze components was measured. The leakage current signal collected in experiment was denoised by wavelet threshold denoising method, and the optimal threshold selection method was obtained by comparing and different threshold selection method threshold denoising effect. The feature vector was extracted by wavelet packet decomposition and information entropy analysis method, and the classification model based on support vector machine theory was established by using feature vector. After testing, the model can accurately classify insulator leakage current under different operating conditions.