An artificial neural network-based mathematical model for the prediction of blast-induced ground vibration in granite quarries in Ibadan, Oyo State, Nigeria
Abstract
Blast-induced ground vibration is one of the most severe and complex environmental problems associated with blasting operation. The scaled-distance approach is the common method of estimating the magnitude of the blast-induced ground vibration. However, the prediction of this approach is inaccurate as evident in the literature. Therefore, this study proposed an artificial neural network model for the prediction of blasting operations in five granite quarries in Ibadan, Oyo State, Nigeria. The distance from the measuring station to the blasting point (D) and a charge per delay (Q) were the two input parameters into the model while the peak particle velocity (PPV) was the targeted output. 100 datasets were used in developing the model. The datasets were divided into training, testing, and validation. The ANN model was trained using backpropagation algorithm with the Levenberg-Marquardt training function. The weights and biases obtained from the trained ANN architecture were extracted and transformed into a simple mathematical equation for the computation of PPV. The obtained results from the ANN model was compared with the prediction of multilinear regression (MLR). The coefficient of determination (R2) of the proposed ANN model is 0.988 while that of the MLR model is 0.738. The mean absolute percentage error (MAPE), root-mean-squared error (RMSE), and variance accounted for (VAF) were also used to further evaluate the performance of the models. The MAPE, RMSE, and VAF of the ANN model are 7.14, 2.90, and 98.74 while that of the MLR model is 40.90, 13.35, and 73.76. Therefore, the proposed ANN model can give a reasonable prediction of the PPV.
Description
Scientific African
Keywords
Blasting operation, Ground vibration, Peak particle velocity, Scaled-distance approach, Neural network
Citation
10.1016/j.sciaf.2020.e00413