Objectives The existing methods are criticized for their single-perspective approach, disregard for the local and global characteristics of the nodes, and insufficient consideration of the load carried by the nodes.

Methods The proposed method identifies the inherent brittleness of the ship power system as the root cause of the interruption and chain collapse. To establish the brittleness indexes, the method first considers node degree, eigenvector centrality, node cohesiveness, closeness centrality, and brittleness electrical betweenness from a brittleness perspective, based on complex network theory. Then, the game combination assignment method is proposed to determine the optimal combination weights of intuitionistic fuzzy hierarchical analysis and entropy weight method with Nash equilibrium as the coordination objective. Combine five indexes through comprehensive assignment to obtain a comprehensive brittleness index. Finally, the proposed brittle source identification method is validated using a typical four-station annular ship grid.

Results The method identifies four levels of brittle sources based on the network efficiency of different brittleness indexes, and the results demonstrate its ability to identify brittle sources more comprehensively and reasonably.

Conclusions This research provides effective guidance for the prevention and control of interlocking faults in ship power grids.