Calibration of DEM macro and micro parameters via XGBoost method

Abstract

To solve the complicated macro and micro parameter calibration problem in the discrete element method (DEM) simulation of rock mechanics, macro parameter prediction and micro parameter inversion model are established based on the XGBoost model. Firstly, a parameter database for the uniaxial compressive test in DEM has been established by literature research and numerical simulation. The critical parameters in the uniaxial compressive test have been chosen with data and theoretical analysis. The influence of the number of selected parameters on accuracy has also been discussed. Then, the prediction model and inversion model have been established which can complete the calibration quickly. The two models were tested by test samples, and the accuracy of the model can generally reach more than 90%. This research has great significance for improving the efficiency of discrete element modeling.