Abstract: In order to solve the limitations and applicability of traditional safety assessment methods in coal mine safety risk assessment, and the analysis process was easily affected by subjective factors, the HAZOP-LOPA coal mine risk assessment method based on Bayesian network (BN) was proposed. By analyzing the advantages and disadvantages of hazard and operability analysis (HAZOP) and layers of protection analysis (LOPA), qualitative analysis and semi-quantitative analysis were combined. The BN model was inserted into each protective layer in LOPA, and the failure probability of the independent protective layer was calculated by using "GeNie" software. The Bayesian model was constructed to complete the analysis and calculation of the BN model. This method was applied to coal mine gas explosion accidents. The results show that the two independent protective layers of gas sensor and optimized ventilation system can reduce the risk.