Prediction of Rock Size Distribution in Mine Bench Blasting Using a Novel Ant Colony Optimization-Based Boosted Regression Tree Technique

Abstract

In this paper, we used artificial intelligence (AI) techniques to investigate the relation between the rock size distribution (RSD) and blasting parameters for rock fragmentation in quarries. Moreover, the ant colony optimization (ACO)–boosted regression tree (BRT) model, which is a novel AI technique for predicting RSD using the blasting parameters, is proposed based on the ACO and BRT algorithms. For predicting RSD, three well-developed models, namely the particle swarm optimization–adaptive neuro-fuzzy inference system (PSO–ANFIS), firefly algorithm (FFA)–ANFIS, and FFA–artificial neural network, were applied to the same dataset. Additionally, four benchmark AI techniques, i.e., support vector machine, k-nearest neighbors, principal component regression, and Gaussian process, and a conventional approach, i.e., the Kuz–Ram model, were employed for considering and predicting RSD. Using an image processing technique, the Split-Desktop software package was used to analyze the RSD of 136 blasting events at a quarry in Vietnam. Results were used as inputs, such as powder factor, explosive charge per delay, bench height, stemming length, and burden, and outputs, i.e., RSD, in this study. The novel scoring and color-intensity methods were used for visualizing several statistical criteria, including the correlation coefficient, mean absolute error, and root-mean-square error, to evaluate the model performance. Results indicate that the proposed ACO–BRT hybrid model yields higher RSD predictive accuracy than that obtained using any other model. The proposed model seems to be promising for optimizing the blasting parameters to increase the production efficiency while reducing the production costs.