Prediction of Blast-Induced Ground Vibration at a Limestone Quarry: An Artificial Intelligence Approach

Ground vibration is one of the most unfavourable environmental effects blasting activities, which can cause serious damage to neighboring homes and structures. As a result, effective forecasting their severity critical controlling reducing recurrence. There are several conventional predictor equations available proposed by different researchers but them based on only two parameters, i.e., explosive charge used per delay distance between blast face monitoring point. It well-known fact that results influenced number design such as burden, spacing, powder factor, etc. these not being considered in any predictors due they show high error predicting vibrations. Nowadays, artificial intelligence has been widely engineering. Thus, three approaches, namely Gaussian process regression (GPR), extreme learning machine (ELM) backpropagation neural network (BPNN) were this study estimate ground caused Shree Cement Ras Limestone Mine India. To achieve aim, 101 datasets with average depth, distance, weight, stemming length input parameters collected from mine site. For comparison purposes, simple multivariate analysis (MVRA) model well as, nonparametric regression-based technique known adaptive splines (MARS) was also constructed using same datasets. This serves foundational for GPR, BPNN, ELM, MARS MVRA ascertain respective predictive performances. Eighty-one (81) representing 80% total construct train various models while 20 data samples (20%) utilized evaluating capabilities developed models. Using testing datasets, major indicators performance, mean squared (MSE), variance accounted (VAF), correlation coefficient (R) determination (R2) compared statistical evaluators performance. revealed GPR exhibited superior capability MARS, ELM MVRA. The showed highest VAF, R R2 values 99.1728%, 0.9985 0.9971 respectively lowest MSE 0.0903. engineer employ an appropriate method blast-induced vibration.