When analyzing the physical and mechanical properties of rocks using the discrete element method, it is necessary to calibrate the micro-scale parameters in order to ensure that the macroscopic parameters in numerical simulations meet the analysis requirements. A particle flow model based on uniaxial compression and Brazilian splitting tests was used to simulate physical experiments, and the Plackett-Burman design (PB design) method was employed to screen out the micro-scale parameters that have the most significant influence on the macroscopic mechanical parameters. Linear equations describing the relationship between the micro-scale parameters and the macroscopic mechanical parameters were obtained. Then, based on these equations, the central composite design (CCD design) method was used to determine the responsiveness of the micro-scale parameters to the macroscopic mechanical parameters, as well as the non-linear relationship between the micro-scale parameters and the macroscopic parameters. Linear and non-linear equations were simultaneously solved to address the multi-objective optimization problem. Finally, the numerical simulation results were validated by comparing them with indoor experimental results. The results show that and in parallel bonding have the most significant impact on the uniaxial compressive strength and tensile strength of rocks, and exhibit a positive correlation. Ē_c and have the most significant influence on the elastic modulus of rocks, where Ē_c shows a positive correlation with the elastic modulus, while shows a negative correlation. Moreover, has the most significant impact on the Poisson's ratio of rocks and exhibits a positive correlation.