On-line control of manipulator joint based on RBFNN

The on-line control strategy of Manipulator joint is proposed by using RBF (Radial basis function) neural network of characteristics with high accuracy, high-speed learning. Firstly, (RBF) neural network is applied to on-line control as a identifier, here the gradient descent method is used and improved. The center, width and weight of the network are regulated in order to reach the smallest error function.. Then Jacobi matrix is obtained according to the calculated value. At last three parameters of controller are obtained by Jacobi matrix, thus obtain the purpose of on-line control. The simulation results of dynamics model have verified the proposed control scheme possesses characteristics of high-speed adjustment, and high-precision steady-state error and strong self-adaptive capability. This paper also proves mathematically the stability of the control system under the existence of disturbances and modeling errors.