引入改进模糊C均值聚类的负荷数据辨识及修复方法

高级量测体系的建设促使大量用电负荷数据增加了可观性,但由于通信等原因,量测数据中存在不良数据。文中提出一种引入改进模糊C均值(FCM)聚类算法的负荷数据辨识及修复方法,该方法利用快速爬山技术,对标准FCM聚类算法中聚类数目难以预先确定、初始聚类中心随机选取等缺点进行改进,实现用电负荷数据的精准聚类。在此基础上提取可行域矩阵及特征曲线,实现对新量测数据的辨识及修正。最后采用某地实际负荷测量数据进行分析,并通过与基于标准FCM聚类算法的对比,验证了该方法的快速性、高效性及其应用前景。

Load Data Identification and Correction Method with Improved Fuzzy C-means Clustering Algorithm

With the development of advanced metering infrastructure(AMI), the observability of power utility information is promoted, but lots of bad data exist in the vast amounts of measurement data due to communication and other reasons. A load data identification and correction method with an improved fuzzy C-means(FCM)clustering algorithm is proposed. With the rapid mountaineering technique, the proposed method has overcome the shortcomings of the existing standard FCM clustering algorithm, which is unable to determine the number of clusters in advance and the random choice of the initial cluster centers. Based on the load data clustering obtained, the feasible region matrix and the load characteristic curve can be calculated for identification and correction of the new load data. Finally, a case of actual load measuring data is analyzed and compared with similar cases treated by the standard FCM clustering algorithm, the proposed method proved fast and efficient.