Computational Intelligence Model for Estimating Intensity of Blast-Induced Ground Vibration in a Mine Based on Imperialist Competitive and Extreme Gradient Boosting Algorithms

Abstract

In this paper, we developed a novel hybrid model ICA–XGBoost for estimating blast-produced ground vibration in a mine based on extreme gradient boosting (XGBoost) and imperialist competitive algorithm (ICA). For comparison, we used another hybrid model combining particle swarm optimization and XGBoost [i.e., particle swarm optimization (PSO)–XGBoost] as well as other models, namely classical XGBoost, artificial neural network (ANN), gradient boosting machine (GBM), and support vector regression (SVR). We compared these techniques using 136 blasting events data gathered at an open-pit coal mine in Vietnam. The models’ performance evaluation criteria were the determination coefficient (R2), root-mean-square error, mean absolute error, ranking, and color intensity. Based on the results, our ICA–XGBoost model is the most robust in predicting blast-produced ground vibration. The PSO–XGBoost model provided a slightly poorer performance. The classical XGBoost model showed a lower performance than the hybrid models (i.e., ICA–XGBoost and PSO–XGBoost). The SVR and ANN models gave average performances, whereas the GBM model yielded the worst performance. The results also reveal that the maximum explosive charge capacity, the elevation between blast sites and monitoring points, and the monitoring distance are the most critical variables that should be used in predicting the intensity of blast-induced ground vibration in a mine.