In the analysis of rock burst criterion prediction at home and abroad, the prediction standards of rock burst are selected including the conditions of mechanics、integrity、energy and brittle. The concept of relative membership degree on the rock burst prediction was introduced. The weight of standards and fuzzy matrix of relative membership degree are calculated. Uncertainty in rock burst prediction is described and compared according to the information entropy. Generalized weighted distance is also defined to characterize the differences in rock burst based on the maximum entropy principle, the establishment of a rock burst prediction fuzzy optimization model. The results from the application to practical example and comparisons with other methods are fairly good. Finally, the prediction model is applied in Putaoshan tunnel and the predictions are consistent with the actual rock burst.