Prediction of blast-induced ground vibration in dolomitic marble quarry using Z-number information and fuzzy cognitive map based neural network models

Blast-induced ground vibration (BIGV) is one of the detrimental environmental consequences of blasting operations in mining and civil engineering. Hence, accurate prediction of BIGV is highly imperative. Therefore, different novel artificial intelligence (AI) methods such as Bayesian regularized neural network (BRNN), Bayesian regularized causality-weighted neural network (BRCWNN) and Z- number-based Bayesian regularized causality-weighted neural network (Z-BRCWNN) are proposed in this study for the reliable prediction of BIGV in a dolomitic marble quarry using the obtained field data. The outcome of the proposed models is subjected to rigorous statistical analyses. The outcome of analyses revealed that the Z-BRCWNN model outperformed the other models with 70%, 82% and 82% threshold statistic values evaluated at the 5%, 10% and 15% confidence levels for the testing phase and 63%, 91% and 91% threshold values for the validation phase evaluated at the same levels as above. The sensitivity analysis conducted revealed that DI has the highest influence on BIGV. • The BIGV affects the structural integrity and cause court cases • Novel models are proposed for accurate prediction of BIGV • Z-BRCWNN outperformed other models in BIGV prediction