In order to solve the defects of insufficient filling, folding, too large forming load and poor surface forming quality of a steering knuckle after forming, an optimization strategy based on Kriging model, multi-objective genetic algorithm and numerical simulation technology was proposed. The punch draft angle, the fillet radius of long sheep horn side for punch head and the fillet radius of connection between rod and flange were selected as the design variables, and the unfilled distance at the end of rod and the maximum forming load of final forging were selected as the optimization objectives. Firstly, the experimental scheme was obtained by orthogonal experimental design method. Secondly, each group of experiments was simulated by Deform-3D, and the response data between design variables and optimization objectives was obtained. Thirdly, the mapping relationship between design variables and optimization objectives was approximated by Kriging model, and the approximate model was globally optimized by multi-objective genetic algorithm to obtain the frontier optimal solution set. Finally, the optimized parameters were simulated and verified in production. The results show that the simulation results and production results prove the effectiveness of the optimization strategy. Therefore, the products with good forming effect are obtained by the optimized mold parameters, and the material utilization rate is increased from 75% to 85%.