Abstract:At present,the trajectory of mechanical arm joint is vulnerable to the interference of the external environment, which leads to unstable trajectories and serious dithering phenomenon,and it cant satisfy the requirement of trajectory tracking task well.so the movement diagram model of mechanical arm double joint was created by using radial basis function (RBF) neural network adaptive control method to track the trajectory of mechanical arm joint.The error generated by mechanical arm trajectory was analyzed by designing the mechanical arm joint adaptive neural network controller,and lyapunov function was used to prove the stability and convergence of controller.Combined with specific examples,the trajectory tracking error of the double joint of the mechanical arm was simulated by Matlab software.At the same time, the simulation error of fuzzy PID control was compared and analyzed.The simulation curves show that the error generated by trajectory tracking is smaller and the vibration amplitude of input torque is relatively small by using RBF neural network adaptive control method for mechanical arm joints. Therefore, Mechanical arm joint end by using RBF neural network adaptive controller can reduce the tracking error of trajectory and improve the vibration phenomenon.