Efficient calibration of discrete element meso-parameters is essential for reliable rock fragmentation modeling. This study focuses on green sandstone, combining uniaxial compression tests with PFC3D simulations to establish an XGBoost–stepwise regression framework for macro–meso-parameter calibration of the parallel bond model. XGBoost was used to identify the dominant meso-parameters governing peak strength, elastic modulus, and Poisson’s ratio, and stepwise regression was applied to construct explicit nonlinear mapping equations. Peak strength is mainly controlled by shear strength τcp and normal strength σcp, while elastic modulus and Poisson’s ratio are primarily influenced by bond modulus Ecp and stiffness ratio kp*. Introducing quadratic and interaction terms improved model fit, with adjusted R2 increasing by 27.5% and 11.2%, respectively. The calibrated parameters reproduced laboratory mechanical indices with errors of 0.09%–3.745% and showed good agreement with the observed shear–brittle failure pattern. Based on the calibrated model, a representative impact-fragmentation simulation further revealed staged conversion of input energy into fracture-related energy during crack initiation, propagation, and through-failure. The proposed framework improves the efficiency and interpretability of PBM parameter calibration and supports DEM-based analysis of rock fragmentation and energy evolution.
Chao et al. (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: