基于模糊聚类的电力系统负荷特性分析-《郑州大学学报(工学版)》

文章信息/Info

Author(s):: LI Canbing; QU Fang; WANG Xiaoning; etc; 1．School of Electrical Engineering，Zhengzhou University，Zhengzhou 450001，China；2．Anyang Electrical Power Company，Anyang 455000，China

Keywords:: power system; load characteristics; classification and synthesis; fuzzy clustering

摘要:: 以变电站负荷构成比例作为基本特征向量,提出采用自适应矢量量化(AVQ)神经网络模糊聚类方法对电力负荷特性进行分类与综合.将AVQ神经网络模糊聚类方法和模糊C均值(FCM)聚妾方法进行了对比研究.通过对福建电网44个变电站进行聚类分析,得出两种聚类方法不仅具有良好的聚类综合能力,同时能够通过优化理论获得聚类中心矩阵,与其他聚类方法相比有明显优势;两者相比,AVQ神经网络模糊聚类算法本身机时小,处理速度更快,而且结果更为合理.

Abstract:: The Adaptive Vector Quantization(AVQ)neuronic network fuzzy clustering means are used toclassify and synthesize load characteristics in this paper．The composition proportion of substation load is usedas the characteristic vector．Two classification approaches are studied，including AVQ neuronic network fuzzyclustering means and Fuzzy C Means(FCM)．In case study，a case with 44 substations of Fujian province isstudied using these two means．Both of them could obtain the clustering center and aggregate load characteristics．They have a clear advantage over other clustering means．Compared with FCM，AVQ fuzzy clusteringmeans process large amounts of data more quickly．And its clustered results are more reasonable．