Artificial Intelligence Based Smart Blasting Using ICA Optimized Neural Network for Oversize Prediction in a Small Scale Dolomite Quarry in Nigeria

Abstract The integration of artificial intelligence (AI) into blasting operations has demonstrated significant improvements in precision, safety, and operational efficiency. By leveraging AI algorithms to optimize blast design, monitor field conditions, and analyze fragmentation data, more informed decision-making is achieved. In small-scale mining, safe and efficient blasting practices are critical to protecting workers, maximizing resource extraction, and minimizing environmental impacts. This study first reviews the applications, advantages, and limitations of various AI techniques used in predicting blast performance and environmental effects, with specific attention to the overlooked impact of multicollinearity and the absence of explicit mathematical expressions in many soft computing models. In the experimental section, an imperialist competitive algorithm (ICA)-optimized artificial neural network (ANN) model is developed to predict the percentage of oversized material produced by small-scale blasting. Field data, including blast parameters and rock strength, were collected from a dolomite quarry in Akoko, Edo State, Nigeria. Fragmentation analysis was conducted using WipFrag 4.0 software across 48 blast rounds, using the primary crusher gape as the decision threshold. Input parameter selection was guided by multicollinearity analysis to ensure robust modeling. Evaluation metrics such as root-mean-square error (RMSE), correlation coefficient (R²), mean absolute percentage error (MAPE), variance accounted for (VAF), Nash–Sutcliffe efficiency (NSE), and performance index (PI) confirmed that the ICA-ANN model significantly outperformed the standard ANN. While the conventional ANN model underestimated oversize by 14.7%, the ICA-ANN achieved a lower prediction error of 2.7%. The proposed model offers a practical and accurate tool for predicting oversized fragmentation in small-scale rock engineering scenarios, contributing to improved blasting efficiency and sustainability in the mining sector.