首页|基于特征挖掘的低压台区窃电行为识别研究

基于特征挖掘的低压台区窃电行为识别研究

扫码查看
原文链接
  • 万方数据
  • 维普
为保障低压台区用电安全,提出一种基于特征挖掘的低压台区窃电行为识别方法.首先,采用ZigBee网络模型采集低压台区用户用电数据,并从中提取用电量突变指标、用电数据差别指标以及用电量变化指标等用户异常用电特征.然后,对提取到的特征量进行不确定度量、修正以及归一化处理,得到具有代表性的特征量.接着,采用主成分分析方法对特征量实施降维处理,以达到提高分类器效率的目的.最后,将降维后的特征作为支持向量机分类器的输入信息,通过与三类指标进行匹配,判断是否存在窃电行为.试验结果表明,该方法能够有效地识别低压台区的窃电行为,识别准确率达到 93.4%.该方法可以为防范和打击窃电行为提供有效的技术支持.
Research on Identification of Electric Stealing Behavior in Low Voltage Substation Based on Feature Mining
To protect the security of electricity consumption in low voltage substation,a feature mining-based identification method of electric stealing behaveior in low voltage substation is proposed.Firstly,the ZigBee network model is used to collect the power consumption data of the users in the low voltage substation,and extract the abnormal power consumption features of the users such as the indicators of sudden change in power consumption,the indicators of difference in power consumption data,and the indicators of change in power consumption,etc.Then,the extracted features are subjected to uncertainty measurement,correction and normalization to obtain representative features.Then,the principal component analysis method is used to reduce the dimensionality of the features to improve the efficiency of the classifier.Finally,the dimensionality reduced features are used as the input information of the support vector machine classifier and are matched with the three types of indicators to determine whether there is electric stealing behavior.The experimental results show that the method can effectively identify electric stealing behavior in low-voltage substation,and the identification accuracy rate reaches 93.4%.The method can provide effective technical support for preventing and combating electric stealing behavior.

Electricity consumption featureFeature miningElectric stealing behaviorNormalization processPrincipal component analysisDimensionality reduction processSupport vector machine

邓瑞麒、黄国政、黄亮浩、郑广勇、李永乐

展开 >

广东电网有限责任公司江门供电局,广东 江门 529000

用电特征 特征挖掘 窃电行为 归一化处理 主成分分析 降维处理 支持向量机

中国南方电网有限责任公司科技基金资助项目

GDKJXM20220767

2024

10.16086/j.cnki.issn1000-0380.2023070112

自动化仪表
中国仪器仪表学会 上海工业自动化仪表研究院

自动化仪表

CSTPCD
影响因子：0.655
ISSN：1000-0380
年,卷(期)：2024.45(10)