A Graph Cut Approach for Coherency Identification with Wind-Integrated Power System

This paper presented a graph cut approach to detect coherent groups. By calculating the cosine similarity, an undirected graph and normalized Laplacian matrix are created. The graph cut for the formed undirected graph was carried out by using spectral clustering, and the coherent groups were separated by inverse mapping the solved subgraphs into the power system. The proposed method is applied to the modified IEEE39 node system for analysis and verification.