During the progress of oil filling in the vehicle, the deviation between actual pressure and ideal pressure always affects the shift quality. For enhancing the shift quality, the model for simulation of the vehicle shift system was established, which included fuel supply unit, electromagnetic reversing valve, pressure regulating valve and clutch. The oil filling progress was simulated and analyzed in AMESim. Taking the oil filling pressure as optimization target, the optimized function was obtained on the basis of the error integral criterion. In pressure regulating valve, the key parameters affecting the oil filling pressure were obtained by the Pareto diagram as follows: the opening of the valve, the spring preloading force and the diameter of damping hole. Nonlinear programming by quadratic Lagrangian and genetic algorithms were respectively applied to optimize parameters above. According to the comparison between the optimized simulation oil filling curve and the ideal oil filling curve, the genetic algorithm was defined as the optimal algorithm, and the optimal structural parameters of the pressure regulating valve were determined. Based on the variable controlling approach, the experimental schemes were designed. The results of the experiments showed that the stability of the optimal shift system was improved to some extent, and the optimal oil filling process was more close to the ideal oil filling process, providing a reliable reference for improving the shift quality.