高压断路器声纹及振动熵特征联合的GWO-KFCM故障诊断  

GWO-SVM Fault Diagnosis of High Voltage Circuit Breaker Based on Sound Texture and Vibration Entropy Characteristics

作　　者：董耀 陈云 樊万昌 刘会兰[2] 常珂 DONG Yao;CHEN Yun;FAN Wanchang;LIU Huilan;CHANG Ke(Haixi Power Supply Company of State Grid Qinghai Electric Power Company,Qinghai Golmud 817000,China;North China Electric Power University,Hebei Baoding 071003,China)

机构地区：[1]国网青海省电力公司海西供电公司,青海格尔木817000 [2]华北电力大学,河北保定071003

出　　处：《高压电器》2023年第3期53-60,共8页High Voltage Apparatus

基　　金：国网青海省电力公司科技项目资助(5228061900AL)。

摘　　要：针对复杂环境下高压断路器故障诊断算法的准确率和泛化性问题,提出一种声纹及振动熵特征联合的GWO-KFCM故障诊断算方法。首先,对声音信号进行广义S变换,提取反应声纹的盒维数、方向度和对比度纹理特征;对振动信号进行变分模态分解(VMD),计算信号的排列熵。最后,构造联合特征向量送入模糊核C—均值聚类(KFCM)学习训练,利用灰狼优化(GWO)算法优化KFCM初始聚类中心,对训练样本进行预分类后输入SVM,辨识操动机构运行状态。结果表明,声纹及振动熵特征联合的GWO-KFCM故障诊断方法充分利用声振信号互补优势,对实验样本总体诊断准确率达到了100%,并且有较好的泛化能力。In view of accuracy and generalization on fault diagnosis algorithm for high-voltage circuit breakers,a kind of GWO-KFCM fault diagnosis algorithm combining sound texture and vibration entropy characteristics is proposed in this paper.Firstly,the generalized S transform is performed on the sound signal to extract the box dimension,directionality and contrast texture characteristics of the sound texture.The vibration signal is subjected to variational modal decomposition(VMD)to calculate the permutation entropy of the signal.Finally,the joint feature vector is constructed and sent to the fuzzy kernel C-mean clustering(KFCM)learning training,and the gray wolf optimization(GWO)algorithm is used to optimize the KFCM initial clustering center.The training sample is input to SVM after pre-classification to identify operation condition of the operating mechanism.The results show that the GWOKFCM fault diagnosis method combining sound texture and vibration entropy features makes full use of the complementary advantages of acoustic and vibration signals,and the overall diagnostic accuracy of the experimental samples reaches 100%,and it has good generalization ability.

关 键 词：断路器 声纹 振动熵 GWO KFCM 广义S变换 VMD 

分 类 号：TM561[电气工程—电器]