摘要
当前电力用户行为特征分类方法对于离散数据的处理能力较差,导致客户服务支撑效果依旧较差。针对此问题,设计基于数据挖掘的电力用户行为特征分类方法。使用LOF算法对离散数据与标准数据之间的距离进行测算,对原始电力数据进行处理,使用主元分析法设定电力用户行为数据观测变量,结合决策树技术构建电力用户特征分类模型,完成行为特征分类。实验结果表明,分类结果更精准,平均电网设备故障发生率为4.06%,用户窃电管控率最高达到87.43%,可有效支撑电力营销服务多个领域,用户服务效果较好。
At present,the power user behavior feature classification method has poor processing ability for discrete data,which leads to poor customer service support effect.In order to solve this problem,a classification method based on data mining is designed.LOF al-gorithm is used to measure the distance between the discrete data and the standard data,and the original power data is processed.Principal component analysis is used to set the observation variables of power user behavior data.Combined with decision tree technology,the power user feature classification model is constructed to complete the behavior feature classification.The experi-mental results show that the classification results are more accurate.The average grid equipment failure rate is 4.06%,and the us-er stealing control rate is up to 87.43%.It can effectively support multiple areas of power marketing services,and the user service effect is better.
作者
费飞
翁利国
寿挺
霍凯龙
FEI Fei;WENG Li-guo;SHOU Ting;HUO Kai-long(State Grid Zhejiang Hangzhou Xiaoshan Electric Power Company,Hangzhou 311200 China)
出处
《自动化技术与应用》
2023年第11期101-104,117,共5页
Techniques of Automation and Applications
关键词
数据挖掘技术
窃电行为
决策树
离散数据
data mining technology
electricity stealing behavior
decision tree
discrete data