Abstract: In order to reasonably determine the weights of index in the evaluation of the possibility and classification of rockburst, a combination weight method is proposed. Based on the theory of information entropy, the original formula is modified. Compared with the original formula, the modified formula is suitable under any conditions, which is more reasonable in theory. Then, Euclidean distance function is introduced to make the difference between the subjective and objective weight same as the difference between the favorable coefficients; and then the combination weight can be calculated. On the basis, the evaluation indexes are chosen in the analysis according to the causes of rockburst and its characteristics. Combining with the optimized grey clustering method, a grey evaluation model based on combination weight (GEM-CW) for predicting of rockburst proneness is established. Based on some deep rock projects at home and abroad, the GEM-CW model is adopted to predict the possibility and classification of rockburst. Compared with the fuzzy synthetic evaluation method, the attribute synthetic evaluation method, the unascertained measurement model and the matter-elements method, the prediction results of the GEM-CW model in the paper are close to the practical records, so as to prove that the proposed model is effective and available. Therefore, the proposed method provides a practical way to accurately predict the possibility and classification of rockburst in deep underground engineering.

Key words: rockburst proneness, combination weight (GEM-CW), grey evaluation model, Euclidean distance function, entropy