Rockburst prediction study based on AVOA-XGBoost model

To extract mineral resources safely and efficiently，it is necessary to study rockburst prediction． Therefore，an AVOA-XGBoost model was proposed for rockburst intensity prediction． Based on the initial selection of six evaluation indicators，326 rockburst cases were collected，and the Boruta algorithm and synthetic minority over-sampling technique (SMOTE) were used to filter features and solve the class imbalance problem． The pre-processed dataset was divided into the training set (80%) and the testing set (20%) by stratified sampling for training and testing the model respectively．Based on the results，it can be easily shown that African vultures optimization algorithm (AVOA) can efficiently determine the hyperparameters of the extreme gradient boosting (XGBoost) algorithm． Compared to existing intelligent models，the model shows excellent accuracy，and it has the Kappa coefficient of 0.92，and it shows better convergence speed than the single XGBoost． The important analysis of the features shows that the elastic energy index of rocks contributed the most to the model．Finally，the model was applied to the Sanshandao gold mine project case to verify the effectiveness and applicability of AVOA-XGBoost in rockburst prediction. © 2023 Huazhong University of Science and Technology. All rights reserved.