Novel Soft Computing Model for Predicting Blast-Induced Ground Vibration in Open-Pit Mines Based on Particle Swarm Optimization and XGBoost

Blasting is a useful technique for rocks fragmentation in open-pit mines, underground mines, as well as for civil engineering work. However, the negative impacts of blasting, especially ground vibration, on the surrounding environment are significant. Ground vibration spreads to rocky environment and is characterized by peak particle velocity (PPV). At high PPV intensity, structures can be damaged and cause instability of slope. Therefore, accurately predict PPV is needed to protect the structures and slope stability. In this research, a novel intelligent approach for predicting blast-induced PPV was developed. The particle swarm optimization (PSO) and extreme gradient boosting machine (XGBoost) were applied to obtain the goal, called the PSO-XGBoost model. Accordingly, the PSO algorithm was used for optimization of hyper-parameters of XGBoost. A variety of empirical models were also considered and applied for comparison of the proposed PSO-XGBoost model. Accuracy criteria including mean absolute error, determination coefficient (R2), variance account for, and root-mean-square error were used for the assessment of models. For this study, 175 blasting operations were analyzed. The results showed that the proposed PSO-XGBoost emerged as the most reliable model. In contrast, the empirical models yielded worst performances.