Development of a model to predict peak particle velocity in a blasting operation

Ground vibrations arising from rock blasting is one of the fundamental problems in the mining industry, and predicting it plays an important role in the minimization of environmental complaints. To evaluate and calculate the blast-induced ground vibration by incorporating blast design and rock strength, artificial neural networks (ANN) and dimensional analysis techniques were used. First a three-layer, feed-forward back-propagation neural network having nine input parameters, twenty-five hidden neurons and one output parameter was trained using 116 experimental and monitored blast records from one of the most important copper mines in Iran. Seventeen new blast datasets were used for the validation of the peak particle velocity (PPV) by ANN. In the second step, a new formula was developed applying dimensional analysis on results obtained from the sensitivity analysis of the ANN consequences. Results from the calculated formula were compared based on correlation coefficient and root mean square error (RMSE) between monitored and predicted values of PPV. In addition to providing the best prediction of vibration, the new formula has the greatest correlation coefficient and the lowest RMSE, 74.5% and 3.49, respectively.