基于ARO和TWSVM的高压断路器故障诊断方法  

Fault diagnosis method for high voltage circuit breaker based on ARO and TWSVM

作　　者：赵岩 徐天 王梓毅 Zhao Yan;Xu Tian;Wang Ziyi(School of Electrical&Control Engineering,Heilongjiang University of Science&Technology,Harbin 150022,China)

机构地区：[1]黑龙江科技大学电气与控制工程学院,哈尔滨150022

出　　处：《黑龙江科技大学学报》2025年第2期329-336,共8页Journal of Heilongjiang University of Science And Technology

基　　金：黑龙江省省属高等学校基本科研业务费项目(2021-KYYWF-1476)。

摘　　要：为了提高孪生支持向量机对高压断路器故障状态的诊断性能,提出一种采用人工兔算法改进孪生支持向量机参数的方法。采用小波包改进阈值法对高压断路器的振动信号去噪,通过集合经验模态分解提取若干阶本征模态函数分量,选取前7阶可表征振动信号主要信息的本征模态函数分量,计算其样本熵作为输入诊断模型的特征量,利用人工兔算法优化孪生支持向量机的核函数参数和惩罚参数,诊断高压断路器的故障状态。结果表明,文中提出的故障诊断模型准确率达到97.5%,相比于孪生支持向量机和灰狼优化孪生支持向量机等模型,训练时间最短,取得了较好的诊断效果。This paper describes a study designed to improve the diagnosis performance of the twin support vector machine on the fault state of high voltage circuit breaker and proposes an artificial rabbit algorithm to improve the parameters of the twin support vector machine.The study is enabled by using the wavelet packet improvement threshold method to denoise the vibration signals of the high voltage circuit breaker,extracting a certain number of order intrinsic mode function components by the ensemble empirical modal decomposition,selecting first seven order intrinsic mode function components characterizing the main information of the vibration signals to calculate the feature quantity of the diagnostic model of the sample entropy,optimizing the kernel function parameter and the penalty parameter of the twin support vector machine by the artificial rabbit algorithm;and diagnosing the fault state of the high voltage circuit breaker.The experimental results show that the accuracy rate of the fault diagnosis model proposed in this paper reaches 97.5%,and compared with the twin support vector machine,grey wolf optimized twin support vector machine and other models,its training time is shortest,verifying the better diagnosis results.

关 键 词：高压断路器 故障诊断 振动信号 样本熵 人工兔算法 孪生支持向量机 

分 类 号：TM713[电气工程—电力系统及自动化] TP181[自动化与计算机技术—控制理论与控制工程]