Analysis of Coal Mine Safety Accident Risk Intervention Propagation Based on Agent Heterogeneity: Considering Intervention Time and Risk Preference

ABSTRACT Complex nonlinear relationships exist among coal mine accident risks, which provide favorable conditions for risk propagation. Insufficient identification and prevention of these risks can easily trigger chain reactions among risk factors and eventually lead to accidents. To accurately locate risk intervention nodes, this study took coal mine accident cases that occurred in China from 2020 to 2024 as the research object. By integrating association rule mining and complex network methods, a risk interaction network for coal mine accidents was constructed, and the interaction strengths among risks were quantified. Furthermore, factors such as risk concealment, personnel risk preference, risk propagation intensity, and risk intervention timing were introduced to improve the traditional SEIR risk propagation model. The entire process of coal mine accident risk propagation was then simulated. The main findings are as follows: (1) For every 5 s that risk intervention is advanced, the spread of risk propagation can be reduced by 23.7%, and the risk network can return to a stable state earlier. (2) Personnel risk preference significantly affects the peak value of risk propagation and the time required to return to a stable state, with risk‐averse preferences showing a more significant effect on suppressing risk propagation. (3) The risk intervention effect of regulatory and supervision departments and coal mining enterprises is significantly better than that of ordinary operational personnel. (4) Multi‐agent collaborative intervention can achieve an optimal level of risk recovery; however, collaborative intervention is not always superior to single‐agent intervention. Based on these findings, it is recommended to improve the risk awareness and assessment ability of coal mine workers, optimize the allocation of risk intervention resources, and formulate differentiated and dynamic risk intervention measures. Moreover, when implementing collaborative risk intervention, key nodes and appropriate intervention intensity should be selected to avoid weakening the intervention effect due to resource dispersion. The coal mine accident risk propagation and intervention model constructed in this study deepens the understanding of risk propagation mechanisms and helps move the accident risk threshold forward.