改进FCM算法及其在电力负荷坏数据处理的应用  

Application of Improved FCM Algorithm in Outlier Processing of Power Load

作　　者：蒋雯倩[1] 李欣然[1] 钱军[1] 

机构地区：[1]湖南大学电气与信息工程学院,长沙410082

出　　处：《电力系统及其自动化学报》2011年第5期1-5,共5页Proceedings of the CSU-EPSA

基　　金：国家自然科学基金资助项目(50977023)

摘　　要：变电站日负荷曲线含有丰富的综合负荷构成特性信息,可以用于负荷特性的分类与综合,但必须对原始生数据进行坏数据的辨识与调整。在深入分析已有方法以及负荷建模对日负荷曲线分类与综合要求的基础上,提出一种基于拉格朗日(Lagrange)插值方法和模糊聚类原理的改进的模糊C均值聚类FCM(fuzzy C-means)算法应用于变电站日负荷曲线的坏数据辨识与调整。首先运用内维尔(Neville)算法对缺失数据补全;然后采用改进FCM算法对日负荷曲线进行聚类,产生各类的特征曲线,利用负荷曲线的横向相似性辨识负荷坏数据;最后利用特征曲线进行坏数据调整。实例分析取得了良好效果。Substation daily load curve has the rich information on the integrated load structure features, which can be used to the classification and synthesis of load characteristics. However, the identification and justification of outlier should be done for the original data. In this paper, based on the analysis of existing methods and requirements of load modeling to the classification and synthesis of daily load curve, an algorithm based on Lagrange interpolation method and improved fuzzy C-means（FCM） algorithm of fuzzy clustering principle is pro- posed to the identification and justification of outlier in the substation daily load curve. Firstly, the Neville algorithm is used to complete the missing data. Then the improved FCM algorithm is applied to cluster the daily load curve to produce various characteristic curves. So the outlier can be identified by using the horizontal similarity of load curves. Finally, the outlier can be adjusted by using the characteristic curves. The case analysis has proved the good results.

关 键 词：负荷特性 坏数据辨识 内维尔算法 改进的模糊C均值算法 日负荷曲线 

分 类 号：TM772[电气工程—电力系统及自动化]