基于特征选择和ICOA-LSSVM的变压器故障诊断  

Transformer Fault Diagnosis Based on Feature Selection and ICOA-LSSVM

作　　者：向小民[1] 盛刘宇 刘谦 刘闯 XIANG Xiaomin;SHENG Liuyu;LIU Qian;LIU Chuang(College of Electrical and New Energy,Three Gorges University,Yichang 443000;Jingzhou Power Supply Company,State Grid Hubei Electric Power Co.,Ltd.,Jingzhou 434000;Jingmen Power Supply Company,State Grid Hubei Electric Power Co.,Ltd.,Jingmen 448000)

机构地区：[1]三峡大学电气与新能源学院,宜昌443000 [2]国网湖北省电力有限公司荆州供电公司,荆州434000 [3]国网湖北省电力有限公司荆门供电公司,荆门448000

出　　处：《电气工程学报》2024年第4期397-406,共10页Journal of Electrical Engineering

基　　金：国家自然科学基金资助项目(61876097)。

摘　　要：为提高变压器故障诊断的准确率,提出一种基于特征选择和改进黑猩猩算法(Improved chimp optimization algorithm,ICOA)优化最小二乘支持向量机(Least squares support vector machine,LSSVM)的变压器故障诊断方法。采用F-score和信息增益两种方法对故障特征进行筛选,根据特征选择结果确定变压器故障诊断模型的输入量。采用ICOA算法对LSSVM的惩罚因子和核参数进行优化,建立了基于特征选择和ICOA-LSSVM的变压器故障诊断模型。采用实际变压器故障数据进行算例分析,并与其他变压器故障诊断方法进行对比,结果表明,考虑特征选择的ICOA-LSSVM模型诊断结果的正确率高达95.83%,高于其他方法,验证了所提变压器故障诊断方法的正确性和优越性。In order to improve the accuracy of transformer fault diagnosis,a transformer fault diagnosis method based on feature selection and improved chimp optimization algorithm(ICOA)optimized least squares support vector machine(LSSVM)is proposed.Two methods,F-score and information gain,are used to screen the fault features,and the input of transformer fault diagnosis model is determined according to the feature selection results.Using ICOA algorithm to optimize the penalty factor and kernel parameters of LSSVM,a transformer fault diagnosis model based on feature selection and ICOA-LSSVM is established.The actual transformer fault data are used for example analysis,and compared with other transformer fault diagnosis methods.The results show that the accuracy of the diagnosis results of ICOA-LSSVM model considering feature selection is as high as 95.83%,higher than other methods,which verifies the correctness and superiority of the transformer fault diagnosis method proposed.

关 键 词：变压器 故障诊断 改进黑猩猩算法 最小二乘支持向量机 特征选择 

分 类 号：TM41[电气工程—电器]