一种基于小波变换的模糊聚类算法及其应用  

Wavelet transform-based fuzzy clustering algorithm and its application

作　　者：杨健维[1] 何正友[1] 赵静[1] 张海平[1] 

机构地区：[1]西南交通大学电气工程学院,成都610031

出　　处：《大连海事大学学报》2008年第3期67-71,共5页Journal of Dalian Maritime University

基　　金：国家自然科学基金资助项目(50407009);四川省杰出青年基金资助项目(06ZQ026-012);教育部优秀新世纪人才支持计划项目(NCET-06-0799)

摘　　要：为准确对输电线路故障性质、故障相等进行识别,提出一种基于小波能量比值的模糊C-均值聚类(FCM)算法,并研究了该算法在输电线路的永久性和瞬时性故障识别中应用的可行性.结合小波分析的时频分析能力、小波能量比值的特征提取能力和FCM的模式识别能力,建立一实际500kV输电线路的PSCAD模型,对单相故障产生时的暂态电流进行了聚类分析和识别.仿真结果表明:基于小波能量比值的FCM算法能较好地识别故障相与非故障相,且算法收敛速度快,识别结果准确.A fuzzy C-means clustering algorithm （ FCM ） based on wavelet energy ratio was developed to identify different types of faults such as fault property and fault phase effectively for transmission lines. The feasibility of applying fuzzy C-Means clustering algorithm to fault recognition for permanence and instantaneity for transmission line was studied. Time-frequency analysis ability of wavelet analysis, feature extraction ability of wavelet energy ratio and pattern recognition ability of fuzzy C-means clustering were also considered. A PSCAD model of 500 kV transmission line was established, and a sample of transient current under the condition of single phase fault was analyzed and identified by using FCM algorithm. Simulation results show that the proposed algorithm can recognize fault-phase and non-fault- phase well, and it has fast convergence speed and good recognition results.

关 键 词：电力系统 故障诊断 小波变换 模糊C-均值聚类 特征提取 

分 类 号：TM711[电气工程—电力系统及自动化]