Abstract: In order to evaluate the rockburst intensity grade accurately and ensure safety of excavation engineering, a classification model based on MIV-MA-KELM was built. Firstly,the mechanism of rockburst occurrence was used to analyze the factors influencing the intensity grade and further determine the main evaluation indicators. Then,the MA was introduced to carry out an optimum selection of Kernel-based KELM parameters, and KELM was used to fit the nonlinear mapping relationship between the influencing factors and rockburst intensity. Finally, the MIV was used to calculate the weight of each indicator at 20% mediation rate, and indicators having a low weight were removed and the others were fed back to MA-KELM model for retraining and testing. According to the 68 sets of data from the Bayu tunnel, a series of comparative experiments were designed and carried out, and the model was applied to the prediction of rockburst intensity grade of Qinling Mountains tunnel. Experimental results show that the prediction results conform with the reality, and that the MIV-MA-KELM model can construct the indicators system more reasonably, and avoid local optimal solution, contributing to improving the classification accuracy.

Key words: rockburst intensity grade, memetic algorithm(MA), mean impact value(MIV), kernel-based extreme learning machine(KELM), MIV mediation rate