基于数据挖掘技术的电价执行在线稽查模型

Online Inspection Model for Electricity Price Implementation Based on Data Mining Technology

针对目前供电企业营销稽查主要依靠人工巡检等被动方法而无法及时发现电价执行异常用户的现状,构建了基于数据挖掘技术的电价执行在线稽查模型。该模型以计量营销等海量用电数据为研究对象,首先利用Kmeans聚类算法构建典型用电轨迹模块,用以识别客户的典型用电模式;其次,利用马氏距离判别算法构建电价异常辨别模块,用以辨别电价执行异常用户;模型的输出为电价执行异常嫌疑用户,可为电力稽查人员提供稽查的范围及依据。利用该方法对中国南方某区域电网进行电价执行营销稽查,结果表明其能快速诊断电价执行异常用户,具有良好的实用性和可行性。下一步的研究重点是采用分布式计算方法来提高计算速度,以及通过调整判别阈值、增加异常判别方法来进行辅助稽查。

In allusion to the situation that marketing inspection of power supply enterprises mostly depends on some passive methods such as manual inspection, which may cause a problem of being unable to discover abnormal electricity price imple- mentation customers, this paper introduces construction of online inspection model for electricity price implementation based on data mining technology. Taking mass power data such as measurement marketing as research objects, this model firstly u- ses K-means clustering algorithm to construct electricity tracking module for identifying typical electricity mode. Secondly, it uses Mahalanobis distance discriminant analysis algorithm to establish abnormal electricity price distinguish module for identi- fying abnormal electricity price implementation customers. Outputs of the model arc regarded as suspected abnormal custom- ers, which may provide inspection range and basis for power inspectors. This method is used for marketing inspection on e- lectricity price implementation in some Chinese southern region and results indicate that it is able to rapidly diagnose abnor- mal electricity price implementation customers which means good practicability and feasibility of this method. Next research key points are using distributed computing methods to improve computing speed and carry on assistant inspection by using ad- justing distinguishing thresholds and increasing distinguishing method for identifying abnormalities.