基于优化支持向量机的用户窃电行为识别方法  

User Electricity Theft Behavior Recognition Method Based on Optimized Support Vector Machine

作　　者：谢伟涛 XIE Weitao(Guangdong Power Grid Co.,Ltd.Guangzhou Yuexiu Power Supply Bureau,Guangzhou 510000,China)

机构地区：[1]广东电网有限责任公司广州越秀供电局,广东广州510000

出　　处：《安徽电气工程职业技术学院学报》2025年第1期20-27,共8页Journal of Anhui Electrical Engineering Professional Technique College

摘　　要：为了实现对用户窃电行为的精准识别,文章设计了一种基于金枪鱼群优化(tuna swarm optimization,TSO)算法优化支持向量机(support vector machine,SVM)的用户窃电行为识别方法。采用TSO算法对SVM的罚系数和核系数进行寻优搜索,以提升SVM的性能,在此基础上构建了TSO-SVM模型。利用TSO-SVM模型对用户窃电行为进行识别,并将识别结果与其他模型进行对比。对比分析结果表明,TSO-SVM模型识别结果的平均准确率高达99.25%,相比其他模型具有更高的精度和更好的稳定性,验证了基于优化SVM的用户窃电行为识别方法的有效性。In order to accurately recognize the electricity theft behavior of users,this paper designs a recognition method based on tuna swarm optimization(TSO)algorithm to optimize the support vector machine(SVM).The TSO algorithm is used to search for the optimal penalty coefficient and kernel coefficient of SVM,in order to improve its performance.Based on this,a TSO-SVM model is constructed.The TSO-SVM model is used to recognize the electricity theft behavior of users,and the recognition results are compared with other models.The comparative analysis shows that the average accuracy of the recognition results of the TSO-SVM model is as high as 99.25%,which has higher accuracy and better stability than other models,verifying the effectiveness of the electricity theft behavior recognition method proposed in this paper.

关 键 词：窃电行为 识别 支持向量机 金枪鱼群优化算法 

分 类 号：TM73[电气工程—电力系统及自动化]