The biological nervous system has always been one of the inspirations of robot controller research. People try to control external equipments by using bio-inspired control models, so that the equipments can achieve bio-like motion effect. Therefore, a closed-loop brain-machine interface(BMI) model based on the discharge activity of cerebral cortical neurons is used as a controller to control the movement of the three-joint artificial prosthesis to the target position. And under a given target distance, a simplified model of the three-joint prosthesis is established, and a loss function of time and kinetic energy also is proposed to describe the loss generated by the prosthetic motion. Finally, a Bayesian iterative algorithm is used to obtain the optimal motion of the prosthesis, so that the loss function value is minimized when the target grabbing action is completed, and the motion control of the prosthesis is optimized.