基于LMD和DE-PNN的高压断路器机械故障识别方法  

A METHOD FOR IDENTIFYING MECHANICAL FAULTS OF HVCB BASED ON LMD AND DE-PNN

作　　者：陈佳豪 吴浩 胡潇涛 顾小平 宋弘 Chen Jiahao;Wu Hao;Hu Xiaotao;Gu Xiaoping;Song Hong(School of Automation and Information Engineering,Sichuan University of Science&Engineering,Zigong 643000,Sichuan,China;Artificial Intelligence Key Laboratory of Sichuan Province,Zigong 643000,Sichuan,China)

机构地区：[1]四川轻化工大学自动化与信息工程学院,四川自贡643000 [2]人工智能四川省重点实验室,四川自贡643000

出　　处：《计算机应用与软件》2023年第6期57-62,165,共7页Computer Applications and Software

基　　金：四川省科技厅项目(2019YJ0477,2018GZDZX0043);企业信息化与物联网测控技术四川省高校重点实验室项目(2018WZY01);人工智能四川省重点实验室项目(2019RYY01);四川理工学院四川省院士(专家)工作站项目(2018YSGZZ04);国家电网有限公司科技项目(521997180016)。

摘　　要：为了克服基于声音信号的断路器机械故障难以诊断的问题,提出一种基于局域均值分解(LMD)和差分进化算法(DE)优化后的概率神经网络(PNN)的断路器故障诊断方法。通过模拟故障发生条件采集传动机构卡涩、基座螺丝松动、合闸弹簧储能不足、正常合闸四种状态的声音数据,对采集到的数据进行LMD分解并利用皮尔逊相关系数法进行信号重构,计算重构信号的分段能量熵构成故障诊断特征向量。利用差分进化算法优化的概率神经网络对训练集进行训练,将测试集输入模型进行测试,实现断路器机械故障诊断。实验结果表明,基于LMD-DE-PNN的高压断路器机械故障诊断方法相较于传统断路器故障诊断方法能快速有效地识别断路器机械故障。To overcome the difficulty of diagnosing mechanical faults of circuit breakers based on sound signals,this paper proposes a circuit breaker fault diagnosis method based on probabilistic neural network(PNN)optimized by local mean decomposition(LMD)and differential evolution(DE)algorithm.Sound data of four states of transmission mechanism jamming,loosening of base screws,insufficient closing spring energy storage,and normal closing were obtained by simulating fault occurrence conditions.LMD decomposition of the collected data was performed,Pearson correlation coefficient method was used to signal reconstruction,and we calculate the segmented energy entropy of the reconstructed signal to form a fault diagnosis feature vector.PNN optimized by DE algorithm was used to train the training set,and the test set was input to the model for testing to realize the mechanical fault diagnosis of the circuit breaker.Experimental results show that the HVCB mechanical fault diagnosis method based on LMD-DE-PNN can quickly and effectively identify the mechanical fault of the circuit breaker compared with the traditional fault diagnosis method of circuit breaker.

关 键 词：机械故障诊断 声音信号 局域均值分解 差分进化优化算法 概率神经网络 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置] TP3[自动化与计算机技术—控制科学与工程]