To study and optimize the tracked chassis performance of electric shovel performance. First, the sand-gravel pavement model is built based on the discrete element method (DEM), and the virtual prototype model of the electric shovel is established. On this basis, the impacts of the pre-tension force, the road-wheel spacing, the sprocket speed, and the grouser height on the tracked chassis performance are explored by the coupling method of DEM and multi-body dynamics (MBD), and the performance prediction model of tracked chassis is obtained through Kriging interpolation method. Finally, taking ground pressure distribution, power, and track tension as the optimization goal, multi-objective optimization based on genetic algorithm (GA) and performance prediction model of tracked chassis is implemented, and the corresponding optimization values are 21,028 N, 1.10 rad/s, −12.97 mm, and 9.50 mm. The feasibility of the optimization results is then confirmed via comparison with results of DEM-MBD simulation, the results show that the power and tractive force are reduced to varying degrees, and the ground pressure variation coefficient and track tension are increased.