Prediction of Blast-Induced Ground Vibration in Open-Pit Mines Using a New Technique Based on Imperialist Competitive Algorithm and M5Rules

In this paper, blast-induced ground vibration (BIGV) was considered as the primary objective, and a new artificial intelligence system was proposed to predict BIGV with high accuracy based on the M5Rules and imperialist competitive algorithm (ICA), called ICA-M5Rules technique. Accordingly, the ICA was considered to optimize the M5Rules based on the rules of the M5 model, as well as the prune and smooth procedures. To evaluate the effectiveness of the proposed ICA-M5Rules technique, random forest (RF), classical M5Rules, and support vector machine (SVM) were developed as the benchmark techniques to compare with the proposed ICA-M5Rules technique. Besides, two existing empirical equations were each used to develop models based on the experimental datasets to estimate BIGV for comparison with the proposed ICA-M5Rules model. A case study of a quarry mine in Vietnam was adopted for the developed (ICA-M5Rules, M5Rules, RF, SVM, empirical) models based on 125 blasting events. Mean absolute error, root-mean-squared error, and determination of coefficient (R-2) were applied and computed to evaluate the accuracy, as well as the performance of the developed models. The findings indicated that the proposed ICA-M5Rules model provided highest accuracy. The primary objective was appropriately addressed based on the obtained results of the developed models. As well established, the proposed ICA-M5Rules model was introduced as a new system to predict BIGV in open-pit mines accurately.