Prediction of Vibration Velocity Generated in Mine Blasting Using Support Vector Regression Improved by Optimization Algorithms

Ground vibration generated from blasting is a detrimental side effect of the use of explosives to break the rock mass in mines. Therefore, accurately predicting ground vibration is a practical need, especially for safety issues. This research proposes hybrid artificial intelligence schemes for predicting ground vibration. The approaches are based on support vector regression (SVR) optimized with firefly algorithm (FFA), genetic algorithm (GA), and particle swarm optimization (PSO). Additionally, a hybrid FFA and artificial neural network (ANN) model and several well-known empirical models were also employed in this study. In the predictive modeling process, 90 sets of data, collected from two quarry mines in Iran, divided into two datasets, namely a training dataset and a testing dataset, were used. After model development, to provide an objective assessment of the predictive model performances, their results were compared based on several well-known and popular statistical criteria. FFA-SVR exhibits much more efficiency and reliability than PSO-SVR, GA-SVR, FFA-ANN models in terms of ground vibration prediction, indicating the superiority of FFA over PSO and GA in the SVR training.