Application of Metaheuristic Techniques for the Prediction and Optimization of Blast Induced Ground Vibration

Abstract

Mining is considered to be the primary industry. The industrial and technological infrastructure of the modern world is one way or another dependent upon this sector. With the advancement of many alternate exploitation technologies and methods, blasting is an inseparable part of mining ever since its first implementation though it has many disadvantages. Despite its ease of use and being economic it adversely affects the surrounding structures and environment. A major challenging parameter is to reduce the ground vibration induced from blasting so as to minimize the damage to the surrounding structures and environment. Therefore, the blasting parameters need to be optimized in order to reduce the ground vibration. Before the optimization, a correlating objective function has to be established so as to ensure the most accurate prediction model. Although there have been many implementations of ordinary backpropagation networks for prediction of peak particle velocity (PPV), still not a single one can be recommended because of their underwhelming accuracy. Therefore, many of the hybrid neural networks have been implemented in this study and compared with existing ordinary neural networks, regression methods and also various empirical predictors. The best performing networks are coalesced to create an ensemble network and used in the final prediction to compensate for the biases of a single neural network. The ensemble network using firefly algorithm based artificial neural network (FA-ANN), grey wolf optimization (GWO-ANN) and teaching and learning based optimization (TLBO-ANN) is used as the objective function for the optimization of the blasting parameters. The firefly algorithm has been used for optimization of the blasting parameters and it resulted in a considerable reduction in the ground vibration. The ensemble network is also used to create a graphical user interface (GUI) for the end user to easily predict and display the prediction when the input parameters are given. Another GUI has been built for the optimization which displays the optimized parameters from the given upper and lower bounds of input parameters. This GUIs can help the field technicians to estimate the PPV and optimize their blast design to minimize the ground vibration before conducting the blasting.