Prediction of rock burst intensity based on multi-source evidence weight and error-eliminating theory.

Abstract

Rock burst is the main geological hazard in deep underground engineering. For the prediction of the intensity of rock burst, a model for prediction of rock burst intensity on the basis of multi-source evidence weight and error-eliminating theory was established. Four indexes including the ratio of rock's compressive-tensile strength [Formula: see text], the stress coefficient of rock [Formula: see text], the elastic energy index of rock Wet, and integrality coefficient Kv were chosen as the prediction variables of rock burst; the index weights are calculated by different weighting methods and fused with evidence theory to determine the final weight of each index. According to the theory of error-eliminating, taking "no rock burst" (I in classification standards of rock burst intensity) as the objective and using the error function to process 18 sets of typical rock burst data and the weight of evidence fusion as the normalized index limit loss value, a model for prediction of rock burst intensity was built. It is verified by the actual situation and three other models. Finally, the model has been applied to rock burst prediction of Zhongnanshan tunnel ventilation shaft. The results show that evidence theory fuses multi-source index weights and improves the method of determining index weights. The index value is processed by error-eliminating theory, and the limit value problem of index value normalization is optimized. The predicted results of the proposed model are consistent with the situation of Zhongnanshan tunnel. It improves the objectivity of the rock burst prediction process and provides a research idea for rock burst intensity prediction index.