Research on evaluation model of rock failure integrity under complex geological conditions in karst area

Blasting lumpiness prediction is one of the most important research contents in engineering blasting. Although the traditional KUZ-RAM model is widely used, it often overestimates the size of blasting. Therefore, the KUZ-RAM model was updated or corrected in this paper by simplifying the difficult problem of statistical burst fragmentation in LS-DYNA. Based on the theory of area measurement method, the fitting mechanism of machine learning is used to study the lumpiness of simulation results. The updated KUZ-RAM model adds a coefficient of 0.623 to the original equation of average lumpiness x_m. The linear coefficient R² between the predicted results and the field blasting results increases from −1.99 to 0.97, which significantly improves the prediction of blasting lumpiness.