Abstract: In order to effectively prevent the collapse of diversion tunnels, this paper proposes a method to identify the key causes of the collapse of diversion tunnels based on complex network and grey relational analysis. Firstly, the historical accident is analyzed deeply, and the set of risk factors is summarized systematically from four dimensions. Then, based on the risk factors and their influence relationship, the network model of collapse accident causes is constructed, and the importance of nodes is analyzed with multiple indexes and dimensions. In view of the fact that a single feature may not be able to fully represent the influence of nodes in the network, this study introduced grey relational analysis to effectively integrate multiple feature indicators of nodes, and then determined the importance of each risk factor, and finally put forward suggestions from the perspective of engineering practice.