Estimation of blast-induced ground vibration through a soft computing framework

Blasting is considered as the most common technique for rock breakage in surface mines, tunneling and large infrastructural development projects. Ground vibration, as one of the most adverse effects induced by blasting operations, can cause substantial damage to structures in the nearby environment. Therefore, there is a need to make accurate predictions of blast-induced ground vibration for minimizing the environmental effects. The present paper explores the possibility of using the genetic algorithm (GA) to create a new predictive model for estimating blast-induced ground vibration at Bakhtiari dam region, Iran. In this regard, two form of equations, i.e., power and linear forms, were developed by GA. Maximum charge weight per delay (W) and distance between monitoring station and blasting point (D), as the most influential parameters on the ground vibration, were used as the input or independent variables for modeling. Also, the peak particle velocity (PPV) parameter, as a descriptor for evaluating blast-induced ground vibration, was considered as the output or dependent variables for modeling. In total, 85 blasting events were considered and the D, W and PPV parameters were precisely measured. The selected GA forms were then compared with the several empirical prediction models. Finally, it was found that the GA power form (with root-mean-square error (RMSE) 0.45 and coefficient of multiple determination (R (2)) of 0.92) was more acceptable model for predicting PPV than the GA linear form and the empirical prediction models.