基于神经网络算法的反间歇性窃电行为监测方法  

Anti-intermittent Electricity Stealing Monitoring Method Based on Neural Network Algorithm

作　　者：黄根 王大成 张辉 叶晟 莫雨阳 Huang Gen;Wang Dacheng;Zhang Hui;Ye Cheng;Mo Yuyang(Shanghai Qingpu Electric Power Company of State Grid,Shanghai 201799,China)

机构地区：[1]国网上海青浦供电公司,上海201799

出　　处：《兵工自动化》2025年第1期38-42,共5页Ordnance Industry Automation

基　　金：国网上海市电力公司科技项目(B30934210001)。

摘　　要：针对当前反间歇性窃电行为监测方法准确性差用户用电数据识别平均精度较低等问题,提出基于神经网络算法的反间歇性窃电行为监测方法。构建基础窃电分析模型;使用中值滤波器剔除采集到的无用数据,完成用电数据采集及预处理;应用神经网络逆传播算法优化窃电行为监测神经网络模型,并设定间歇性窃电行为识别函数,实现对间歇性窃电行为的监测;构建实验环节,通过F_(1)值与平均精度2种指标分析应用效果。实验结果表明:该方法使数据分析能力得到提升,能提高反间歇性窃电行为监测准确性。In order to solve the problems of low average accuracy of power consumption data identification and poor accuracy of anti-intermittent electricity stealing behavior monitoring,an anti-intermittent electricity stealing behavior monitoring method based on neural network algorithm is proposed.Constructing a basic electricity-stealing analysis model,using a median filter to eliminate acquired useless data,completing the acquisition and preprocessing of power consumption data,applying a neural network back propagation algorithm to optimize an electricity-stealing behavior monitoring neural network model,setting an intermittent electricity-stealing behavior identification function,and realizing the monitoring of the intermittent electricity-stealing behavior;The experimental link was constructed,and the application effect of this method was analyzed by F_(1) value and average accuracy.The experimental results show that this method improves the ability of data analysis,and further improves the accuracy of anti-intermittent electricity theft monitoring.

关 键 词：神经网络算法 反间歇性窃电行为监测 用电信息采集系统 高压电能采集 反窃电技术 线损计算 

分 类 号：TM933[电气工程—电力电子与电力传动]