Dynamic evaluation of blast-induced slope stability in open-pit iron mines using deterministic and machine learning approaches

Blast-induced ground vibrations significantly influence the dynamic stability of open-pit slopes, with slope response governed by blast design parameters, vibration characteristics, slope geometry, and rock mass quality. This study presents an integrated framework for evaluating blast-induced slope stability using deterministic and machine learning approaches that combine field monitoring, numerical modelling, and analytical assessment. Dynamic slope performance was evaluated through an integrated methodology based on Newmark’s sliding-block analysis and validated using field-monitored vibration data. To characterize vibration attenuation, the Sadovsky empirical model was first calibrated using site-specific blasting records. Subsequently, a novel Geological Strength Index (GSI)-based peak particle velocity (PPV) prediction model was developed and optimized using a hybrid Differential Evolution–Trust Region Reflective (DE–TRF) algorithm. The proposed model achieved high predictive accuracy (R 2 = 0.846, RMSE = 4.23 mm/s), demonstrating the effectiveness of incorporating rock mass quality into vibration prediction. Results indicate that higher-GSI rock masses transmit stress waves more efficiently, producing relatively higher PPV values at equivalent distances. However, owing to their greater strength and structural competence, these rock masses exhibited superior dynamic stability compared with lower-GSI fractured rock masses, which showed greater susceptibility to instability despite lower vibration amplitudes. Numerical and empirical analyses confirmed stable slope performance under the investigated blasting conditions. The maximum Newmark displacement of 50 mm occurred at a PPV of 17.4 mm/s, a peak particle acceleration (PPA) of 0.2 g, and a scaled distance of 11.24 m/kg 1 ᐟ 2 . Parametric analysis further identified an optimal slope configuration comprising a slope height of 65 m and a slope angle of 30°, yielding a factor of safety of 1.195 under blast loading. The proposed framework integrates empirical, analytical, numerical, and machine learning techniques to support vibration-controlled blast design and enhance slope safety and operational efficiency in open-pit iron mines.