Aiming at the problem of fuzziness of influencing factors and lack of dynamic interpretation ability in mine fire risk assessment under the action of multi-factor coupling, a mine fire risk assessment index system was constructed based on four dimensions of human, machine, environment and management. The combination ordered weighted averaging(C-OWA) operator was used to weaken the influence of subjective factors and establish the interaction matrix to identify the key risk factors. The matrix elements were integrated into the fuzzy cognitive map inference model to predict the evolution trend of fire risk. Finally, the risk evaluation level is determined by the combination evaluation coefficient, and the proposed method is applied to a mine instance to verify. The results show that the fire risk of the mine is a general risk, which is consistent with the actual situation, which verifies the rationality and feasibility of the method. At the same time, the method can not only carry out the coupling analysis of risk factors, but also realize the quantitative prediction of risk trend, which has certain advantages.