Abstract: In the process of exploring complex networks, the identification of critical nodes and the detection of community structure are widely concerned, which is helpful for people to better understand and use the structural characteristics of complex networks and reveal the relationships between nodes. In this paper, Coulomb force centrality (CFC) is proposed and applied to label propagation algorithm (LPA) to design community detection algorithm CFCLPA, which eliminates the randomness in LPA and has a high community structure detection performance. A series of tests and comparisons were made between real networks and LFR benchmark networks. The experimental results show that the proposed algorithm has better community detection performance.