中文摘要：

提出了一种基于两步聚类法的电力系统受扰机群同调性分析方法。首先通过BIRCH算法处理广域量测系统提供的受扰机群电角速度数据．构建受扰机群的聚类特征（clusterinlg feature，CF）树来实现预聚类；然后采用层次聚类法处理预聚类结果并自动确定聚类数量来实现正式聚类：最后根据聚类结果进行受扰机群的同调性分析。通过对新英格兰10机39节点系统的算例仿真，对比不同的同调机组识别方法，验证了该办法的有效性以及能够实现自动确定同调机组数的特点。

英文摘要：

Based on TwoStep Cluster, a new method to analyze the coherency of disturbed generator groups is proposed. Firstly, the generator rotor speed gathered from the Wide Area Measurement System （WAMS） is disposed to build CF tree and produce sub-clusters by dint of BIRCH algorithm method; then through hierarchical clustering, the sub-clusters are clustered into the specified number of clusters that is automatically determined; finally the clustering result is used to analyze the coherency of the disturbed generator groups. The effectiveness of applying the proposed method to identify the coherent generators, as well as the feature to determine the clustering numbers automatically, is verified by simulation of New England 10-machine 39-bus system and by comparing the results of different methods of identification of coherent generators.