基于IGWO-SVM的接触网绝缘子污秽识别研究  

Research on Pollution Identification of Catenary Insulator Based on Improved Gray Wolf Optimization Algorithm

作　　者：李致远 王致诚 刘洪 LI Zhiyuan;WANG Zhicheng;LIU Hong(Liuzhou Railway Vocational Technical College,Guangxi Liuzhou,545616,China)

机构地区：[1]柳州铁道职业技术学院,广西柳州545616

出　　处：《广西电力》2023年第1期38-42,74,共6页Guangxi Electric Power

基　　金：甘肃省科技计划基金项目《电气化铁路接触网绝缘子清洗机器人研制与产业化》(18CX6JA022);2022年度广西高校中青年教师科研基础能力提升项目《电气化铁路接触网绝缘子污秽识别定位技术研究》(2022KY1403);柳州铁道职业技术学院科技立项《基于自适应大领域搜索的堆垛机调度策略研究》(柳铁校发【2021】13号2021-KJB04)。

摘　　要：针对接触网绝缘子污秽等级识别,采用SVM作为分类器。为提高识别精度,改进灰狼算法的搜索策略,利用混沌映射规则对种群进行初始化,基于动态权重策略对灰狼个体位置更新,收敛因子采用非线性策略。将改进算法应用于SVM的核参数和惩罚因子寻优,求出优化参数构建识别模型,并对比了其他几种智能算法的识别状况。实验结果表明,本文所提方法有更高的污秽等级识别精度,可以为工程实际应用提供一定的借鉴。Support vector machine is used as the classifier of catenary insulator pollution level recognition model.In order to improve the recognition accuracy of the model,the search strategy of Gray Wolf algorithm is improved.The population is initialized by chaotic mapping rules,and the individual position of gray wolf is updated based on dynamic weight strategy.The convergence factor of nonlinear strategy is adopted.The improved algorithm is applied to the kernel parameters and penalty factor finding of SVM to find the optimized parameters to construct the recognition model,and the recognition status of several other intelligent algorithms is compared.The experimental results show that the method mentioned in this article has higher dirty level recognition accuracy,which can provide a certain reference for the actual application of the project.

关 键 词：污秽等级识别模型 绝缘子 改进灰狼算法 支持向量机 

分 类 号：U225.43[交通运输工程—道路与铁道工程] TP391.41[自动化与计算机技术—计算机应用技术]