Abstract: In order to achieve the stable walking and optimal feet forces distribution for quadruped robots in a dynamic gait, a control framework based on the fusion of the full-scale virtual model and the dynamic model is proposed. The distribution of the virtual forces at the torso center and the feet forces in the support phase is transformed into a quadratic optimization problem. The optimal distribution of feet forces is achieved in real time through Gurobi library. The dynamical feedforward and the virtual model control methods are integrated in the swing phase to achieve smooth trajectory tracking. Through dynamic simulation software Webots, the method is compared with the basic position impedance control method on the quadruped robot platform. The results show that the method can reduce the impact force by about 30% in the ground interaction, and the stability of the robot movement is effectively improved.