基于加权灰靶与LPSO-BP算法的交流接触器运行状态划分与识别  

Classification and Identification of AC Contactor Operational States Based on Weighted Grey Target and LPSO-BP Algorithm

作　　者：周厚霖 邢朝健 许静[1] 刘树鑫 明欣 刘丙泽 ZHOU Houlin;XING Chaojian;XU Jing;LIU Shuxin;MING Xin;LIU Bingze(Key Laboratory of Special Electric Machines and High Voltage Apparatus in the Ministry of Education,Shenyang University of Technology,Shenyang 110870,China)

机构地区：[1]教育部特种电机与高压电器重点实验室(沈阳工业大学),辽宁沈阳110870

出　　处：《电器与能效管理技术》2025年第2期6-14,共9页Electrical & Energy Management Technology

基　　金：国家自然科学基金项目(51977132);辽宁省科技重大专项(2020JH1/10100012);辽宁省科技厅“揭榜挂帅”科技攻关专项(2022JH1/10800015)。

摘　　要：针对目前交流接触器运行状态的分界不明显、划分依据不充分以及训练算法过早收敛的问题,提出一种基于加权灰靶和莱维飞行粒子群优化(LPSO)算法的交流接触器运行状态划分与识别算法模型。首先,对原始数据进行阈值去噪和特征选择。其次,通过层次分析法(AHP)加权改进灰靶算法,得到交流接触器全寿命开断过程的灰靶值,并将其作为运行状态划分的数据依据,具体划分为良好、一般、高危3种状态。最后,利用LPSO优化BP神经网络(LPSO-BP)算法进行归类预测,并与其他算法进行比较,验证了所提算法具有更好的稳定性和准确性。In response to the problems of unclear boundaries,insufficient division criteria,and premature convergence of training algorithms,an algorithm model for dividing and identifying the operational states of AC contactors based on weighted grey target and Levy Flight Particle Swarm Optimization(LPSO)algorithm is proposed.Firstly,the threshold denoising and feature selection are performed on the raw data.Secondly,by using the Analytic Hierarchy Process(AHP)to weight and improve the grey target algorithm,the grey target value of the entire life cycle of the AC contactor is obtained as the data basis for dividing the operational state into three states:good,general,and high-risk.Finally,the BP neural network optimized by LPSO(LPSO-BP)algorithm is used for classification prediction,and compared with other algorithms to verify that the proposed algorithm has better stability and accuracy.

关 键 词：交流接触器 状态识别 数据降噪 最大互信息系数 莱维飞行 

分 类 号：TM572.2[电气工程—电器]