基于改进K-means聚类的变压器局部放电定位诊断方法

Transformer Partial Discharge Fault Location and Diagnosis Based on Improved K-means Clustering Algorithm

针对现有的变压器局部放电故障定位诊断研究未能充分有效考虑时间延迟误差对定位点结果的影响,提出了一种基于改进K-means聚类算法的变压器局部放电故障定位诊断方法。首先,通过消去非线性定位方程的二阶项,实现非线性定位方程的线性化处理;然后,通过使用多个声发射传感器定位PD故障点初始值,并在考虑时间延迟误差的情况下通过冗余时间量测计算多组PD故障定位点;最后,通过使用改进K-means聚类算法,利用聚类密度和凝聚度参数确定聚类中心和聚类个数,并进行聚类优化处理以确定最佳的PD定位点位置。仿真实验验证了所提方法的有效性。

Targeting the problem that the existing research on transformer partial discharge fault location and diagnosis doesn’t consider the influence of time delay error on the location point,the paper proposes a method of the transformer partial discharge fault location and diagnosis based on the improved K-means clustering algorithm.Firstly,the second-order term of a nonlinear positioning equation is eliminated to realize the linearization of the equation.Then the initial value of the partial discharge fault point is located using multiple acoustic emission sensors,and multiple groups of the fault location points are calculated with redundant time measurement by considering the time delay error.Finally by using the improved K-means clustering algorithm,the cluster center and cluster number are determined by using the cluster density and cohesion parameters,and the optimal partial discharge location point is determined via clustering optimization.The simulation experiments verify the effectiveness of the proposed method.