基于数理统计和模糊聚类的变电站负荷特性判别比较  

Load characteristics discrimination analysis based on mathematical statistics and fuzzy clustering

作　　者：李培强[1] 李欣然[1] 罗安[1] 

机构地区：[1]湖南大学电气与信息工程学院,湖南长沙410082

出　　处：《中国电力》2011年第1期15-18,共4页Electric Power

基　　金：国家自然科学基金资助项目(06071007);中国博士后科学基金资助项目(20100471211);湖南省自然科学基金资助项目(10JJ9023)

摘　　要：变电站的综合负荷判别分析是负荷模型走向实用化的重要手段之一。随着电网规模扩大和新(扩)建变电站的数目的日益增加,电网必须动态调整综合变电站的负荷特性分类结果。针对不断新增的和用户发生变化的变电站的负荷特征数据,提出了基于数理统计和模糊聚类的2种判别方法,前者采用基于马氏距离的判别准则,后者采用模糊C均值为判别依据。通过某实际变电站已知的负荷特性数据分别进行判别聚类,论证两者的各自优缺点,指出前者适用于变电站个数较少的情形,计算简单方便;后者适合于大量样本变电站的聚类判别,精度更高。2种方法为变电站的综合负荷特性的准确判别提供了高效便捷的判别方法。The load characteristics discrimination is an important task in load modeling. The substation characteristics clustering must be adjusted dynamically with the power grid expansion and substation growth. Two algorithms were proposed to discriminate loads by using Mahalanobis distance in mathematical statistics and fuzzy C means clustering. Based on the testing results on known characteristics of actual substation load data, the advantages and disadvantages of both algorithms are discussed. The mathematical statistics method has simple calculation and suitable when the substation number is small. The fuzzy clustering method yields more accurate results and is suitable for the case with large number of substations. The two algorithms are accurate and effective for substation load characteristics identification.

关 键 词：综合负荷 数理统计 模糊聚类 特性判别 

分 类 号：TM63[电气工程—电力系统及自动化] TM714