摘要
为改善电力行业日益严重的窃电现状,对用户窃电嫌疑因子进行分析,提出了一种基于改进的遗传优化神经网络的评估模型。传统的BP神经网络算法存在一定的局限性,学习过程收敛速度慢,容错能力差,学习过程容易陷入局部极小值。针对BP神经网络存在的缺陷,利用遗传算法进行性能优化,有效地避免了陷入局部最优的问题。通过实例分析表明,提出的改进模型对窃电嫌疑因子的判断准确率达到88%以上,证明文中建立的评估模型针对窃电问题提供了一种切实可行的方案。
In order to improve the situation about the increasingly serious power theft in the electric power industry,the suspicion coefficient of electricity stealing is analyzed,and an improved evaluation model based on GA-BP neural network is introduced in this paper. Several limitations exist in traditional BP neural network,it takes a long time to converge and is fragile to errors and easy to get stuck in local minima. For the weakness of BP neural network algorithm,GA algorithm is used to optimize algorithm performance to effectively avoid the problem of local optimality. In this paper,some typical companies are selected to verify the improved evaluation model,and the accuracy of suspicion coefficient of power theft reached 88%. The evaluation model studied in this paper provides a feasible idea for the power theft problem.
作者
王庆宁
张东辉
孙香德
沈杨
许湘莲
Wang Qingning;Zhang Donghui;Sun Xiangde;Shen Yang;Xu Xianglian(Wuhan University of Technology,Wuhan 430070,China;Wuhan Nari Limited Company of State Grid Electric Power Research Institute,Wuhan 430074,China)
出处
《电测与仪表》
北大核心
2018年第11期35-40,共6页
Electrical Measurement & Instrumentation
关键词
窃电嫌疑因子
指标评价体系
BP神经网络
遗传算法
suspicion coefficient of power theft
indictor evaluation system
BP neural network
genetic algorithm