فهرست نشریات علمی دسترسی آزاد ایران

Blast-induced ground vibration poses significant environmental and safety challenges in mining operations. Traditional predictive models relying on Peak Particle Velocity (PPV) face limitations due to the confounding effect of distance, a non-controllable variable. This study introduces a novel integrated framework for predicting and optimizing blast vibrations through four key contributions. First, we propose the Vibration Power Index (VPI = PPV × D^α), a location-independent metric derived from seismic attenuation laws with an empirically determined site-specific coefficient. Second, to address data scarcity, we implement SMOTER (Synthetic Minority Over-sampling Technique for Regression) for enhanced dataset augmentation. Third, we develop a robust Artificial Neural Network (ANN) model for VPI prediction, which is subsequently integrated with an enhanced Hybrid Firefly Algorithm (HFA) featuring chaotic initialization and adaptive parameters for global optimization. Finally, a closed-loop methodology from data preprocessing to optimization was established. Applied to 77 blast records from the Asbcheran mine, our ANN achieved superior performance (R²=0.97, RMSE=4.33), while the HFA identified an optimal pattern reducing mean VPI by 28%. This framework provides a practical tool for sustainable blast design optimization.