Optimization of blasting parameters for an underground mine through prediction of blasting vibration

Drilling and blasting remains the primary method of rock fragmentation in metal mining. However, blasting vibration can adversely affect the stability of the rock. Therefore, prediction of blasting vibration is essential in the mining industry. This paper proposes a combination of principal component analysis (PCA) and support vector machine (SVM) model to predict blasting vibration. Here, PCA was used to simplify the inputs of the SVM. Relative location of the monitoring point to blasting source, total charge, maximum charge per delay, number of delays, burden, spacing, height, and horizontal distance were used as inputs of the combination model (PCA-SVM), while peak particle velocity was set as output. The PCA-SVM model was successfully employed to adjust blasting parameters of the No. 21 stope in Hongtoushan Copper Mine. Two blasting data sets were used to compare the capability of the PCA-SVM model with conventional predictors. The results prove the superiority of the PCA-SVM model in estimating blasting vibration.