计及雷击情况的基于PDT-SVM暂降源辨识方法研究

PDT-SVM-based sag source identification considering lightning strike

对雷击故障暂降的特点进行了分析,归纳了短路故障,雷击,变压器投切,感应电机启动四种暂降类型电压有效值波形的特点,引入5个暂降电压特征指标,建立了暂降源类型辨识特征矩阵。然后采用基于粒子群聚类优化最优决策树SVM分类器对四种暂降类型进行辨识。分类器的训练与测试数据均来自电网实测暂降电压数据,与工程实际密切贴合。算例分析结果验证了算法的有效性和准确性。

This paper analyzes the characteristics of the type of lightning strike sag, and summarizes the characteristics of short-circuit fault, lightning strike, transformer switching, induction motor starting four kinds of voltage sag’s RMS waveforms, and introduces five sag voltage characteristics indicators to establish the sag source feature matrix. Then, the optimal decision tree SVM classifier based on particle swarm clustering optimization is used to identify the 4 types of sag. The training and test data of the classifier are derived from the actual data of the power grid. The identification algorithm is closely matched with the engineering practice. The analysis results verify the effectiveness and accuracy of the algorithm.