Ball mill plays a key role in the beneficiation process, and its performance directly affects the efficiency and economic benefits of subsequent separation. In order to address issue such as high cost and low efficiency in the performance optimization method of ball mill, a two-way coupling method based on coarse-grained CFD-DEM is proposed to study the influence of steel ball grade pairing grinding efficiency. This method uses a coarse-grained model to represent a certain number of particles in the original system by larger particles in the coarse-grained system. By simulating different ball gradation schemes, the specific energy consumption of different ore particle sizes and the total power consumption of the ball mill are analyzed. The results show that the total power consumption of recommended scheme 1 [m(?80):m(?60):m(?30)=40:25:35] is 7.5% lower and the dissipation power is 5.2% lower than that of the field scheme [m(?100)=100] , and the cumulative specific power is the highest and the crushing effect is the best at crushing [8,3] mm and [3.0,0.5] mm ore particles. The application of the coarse-grained CFD-DEM two-way coupling method effectively reduces the computational overhead of large-scale particle system, improves the computational efficiency of ball mill grinding process simulation, and provides more efficient practical guidance for the optimal design of ball mill.