In order to obtain the normal gait for the prosthesis-carrier with the change of external environment and gait, we designed a model of dynamic ankle prosthesis and control system and introduced the strategy of centralpatterngenerator (CPG) about the moving trail of dynamic ankle prosthesis. The dynamic parts, which are incorporated in the model of dynamic ankle prosthesis, provide power in order to have anthropic function and character. The tool of Matlab/simulink was used to simulate the strategy. The simulation results showed that the strategy of CPG learning control in this study was effective and could track the reference trail rapidly and fit the moving trail of a person's normal limb. It can make the prosthetic timely regulation and action, enhance the prosthetic intelligence. It has important practical value for intelligent prosthesis development based on this analysis of technology.