基于用电信息采集系统用户负荷特性聚类分析

基于用电信息采集系统的用户负荷数据聚类分析,是获得典型负荷曲线和按负荷特性完成用户分类的重要手段。K均值聚类算法(K-means)是目前应用较多的电力负荷分类算法,但K-means算法最大问题在于无法自动获取最优聚类数目。对此,文章提出了一种基于聚类结果评价指标及分类复杂程度确定聚类数目的方法,得到的聚类数目可作为K-means的初始输入。该方法可以有效降低K-means分类算法中人工参与程度,并能获得较优的聚类结果。文章末尾通过实际算例分析验证了所提分类方法的正确性。

Cluster analysis of load characteristic based on electricity consumption information acquisition system

Clustering analysis of load characteristic based on user load data of Consumption information acquisition sys-tem, is an important means for user classification. K-means clustering algorithm (K-means) is the the generally used load classification algorithm at present. K-means algorithm, however cannot ensure the optimal clustering number auto-matically. Aiming at this problem, the article puts forward a method to determine the clustering number based on cluster-ing result evaluation index and classification complexity. The method can reduce manual handling works in K-means classification process, and can obtain better clustering results. At the end of the article through the actual example analy-sis demonstrate the validity of the proposed classification method.