复杂电力网异常输电线路自适应识别研究分析

Adaptive Identification of Abnormal Transmission Lines in Complex Power Network

为了减少人力甄别复杂输电线路异常的工作量,实现自动化巡检,研究了复杂电路网络异常线路自适应识别的问题。当前输电线路故障识别方法采用经验模态分解获得的特征向量通常具有模糊性,难以准确观测输电线路的运行状态,导致最终输电线路故障判别结果不准确、耗时长。提出基于经验模态分解的异常输电线路自适应识别方法。对输电线路信号进行经验模态分解,获得多个真实有效的线路信号的本征模态分量,结合相关性分析方法过滤出含有关键信息的经验模态分量进行奇异值分解,将分解得到的奇异值矩阵作为异常输电线路特征分量。采用动态柯西蜂群算法对故障分类器的参数寻优,构建蜂群优化的故障分类器模型,结合实际工况下的特征数据训练该分类器模型并进行故障自适应识别。实验结果表明,通过采用蜂群优化算法优化分类器模型显著提高了输电线路故障识别精度和速度。

At present,the method of recognizing fault in electric transmission line is difficult to accurately observe the operating state of electric transmission line,which results in inaccurate and time-consuming recognition of final fault diagnosis of transmission line.Therefore,a method for adaptive recognition of abnormal electric transmission line based on empirical mode decomposition was put forward.At first,the empirical mode decomposition was per- formed on transmission line signal to obtain the intrinsic mode components of many real and effective line signals. Combined with correlation analysis method,the empirical modal components containing key information were filtered for singular value decomposition.Then,the singularity matrix obtained by decomposition was used as the characteristic component of abnormal electric transmission line.Moreover,the dynamic Cauchy bee colony algorithm was used to optimize the parameter of fault classifier and the fault classifier model was built based on bee colony optimization. Finally,the classifier model was trained based on the characteristic data of actual operating condition and the fault was adaptively recognized.Experimental results prove that the bee colony optimization algorithm significantly improves the accuracy and speed of transmission line fault recognition in optimizing the classifier model.