The synchronous stability of cluster networks composed of distributed power generation multi-cell microgrids is studied. Based on three commonly used models of power network, namely star, tree and ring network, the internal relation between network topology and network characteristics is deeply studied, and the factors affecting system synchronization stability are analyzed. Firstly, based on the complex network graph theory and the Kuramoto oscillator model, the cluster network model was constructed. By observing the changes of the relevant parameters of the cluster network, the influence of the degree of subnet interconnection on the synchronization performance of the whole network was analyzed. The results show that the decentralized connection mode has better synchronization performance than the centralized connection mode. The closer the connections among the subnets of the cluster network, the better the synchronization performance of the cluster network. The results show that the increase of the connection strength is beneficial to the stability of the whole network, but has more complex influence on the synchronization stability of the sub-network.