In order to improve comprehensive steering performance, the approach based on simulated annealing modified particle swarm algorithm (SA-PSO) for parameters optimization of rotary valve was proposed. Models of vehicle dynamics, steering system, high-speed road feel and energy consumption were built. The quantitative indexes of low-speed steering portability, high-speed road feel, energy consumption of rotary valve were derived. Optimization problem with minimal root of quadratic sum of the indexes of steering portability, road feel and energy consumption were defined as objective function, and the scope of the parameters as constraints was put forward. The fitness function of particle swarm optimization algorithm (PSO) modified by simulated annealing theory was constructed. The global optimal solution of rotary valve parameters was obtained by using the improved PSO algorithm. SA-PSO and PSO optimization results were presented, which indicated that the improved PSO provided stronger global convergence and faster convergence speed. The correctness of optimization method was verified through contrastive analysis of the steering performance above-mentioned under optimized parameters and other two sets of parameters. Finally, the experiments of lemniscate steering, high-speed steering and energy consumption were carried out. The results demonstrate that comprehensive steering performance is optimal with optimized parameters of rotary valve.