In order to accurately predict the rock burst level,five factors of rock burst occurrence were selected from three aspects of lithology,stress,and surrounding rock as the prediction evaluation index,and the importance of evaluation index was calculated by using the improved entropy weight formula.Combined with the multi-dimensional Indian huts model in the catastrophe progression method,the data of rock burst samples after dimensionless processing were normalized to obtain the catastrophe sequence value,and the rock burst interval was divided.Finally,this method is applied to the rock burst prediction of Jiangbian hydropower station tunnel.The results show that the method is objective,accurate,and consistent with the actual engineering situation and has certain advantages over other aspects.