Self-Adaptive PID Control Strategy Based on RBF Neural Network for Robot Manipulator

Abstract

To the strong nonlinearity and parameter uncertainty in robot manipulator control, a novel self-adaptive PID controller based on RBF neural network on-line identification for the robot manipulator is proposed in this paper, which has solved the weak adaptive ability and poor robustness of the conventional PID control. The control scheme designed in this paper is realized by two kinds of neural networks, where the self-adaptive single neuron network is implemented to tune the parameters of the PID controller. Another RBF neural network is built to identify the robot manipulator on-line, simultaneously get the Jacobian information for the controller. This paper compares the proposed control strategy with the conventional PID control mainly from the following four respects: tracking performance without and with ambient disturbance, varying frequency and amplitude of the desired input signal and parameter variation of the robot manipulator. Simulation results have shown that the proposed approach presents a fast and high-precise tracking ability.