In high terrestrial stress regions, rockburst is a major geological disaster significantly influencing the underground engineering construction. How to carry out the efficient and accurate rock burst prediction remains to be solved. In order to comprehensively consider the objective information of index data and the important role of subjective judgment and decision-making in rockburst prediction, the improved analytic hierarchy process and the CRITIC method based on index correlation are used to obtain the subjective and objective weights of each index, respectively. Fusion is carried out to obtain the comprehensive weight according to the principle of minimum discriminative information. The original cloud model and the classification interval of the predictive index are revised to make up for the excessive sensitivity of original cloud model to the average of the grade interval. A hierarchical comprehensive cloud model of each index is generated through a cloud algorithm. Finally, the reliability and validity of the model are verified by rockburst examples. The model in this paper is compared with the entropy cloud model, the CRITIC cloud model and the set pair analysis multi-dimensional cloud model. The results show that the model can not only describe the various uncertainties of interval value indicators, but also quickly and effectively determine the rockburst strength level.