Trajectory Tracking Control Based on RBF Neural Network Learning Control

Chengyu Han, Yiming Fei, Zixian Zhao, Jiangang Li

Abstract: In this paper, a radial basis function neural network (RBFNN) learning control scheme is proposed to improve the trajectory tracking performance of a 3-DOF robot manipulator based on deterministic learning theory, which explains the parameter convergence phenomenon in the adaptive neural network control process. A new kernel function is proposed to replace the original Gaussian kernel function in the network, such that the learning speed and accuracy can be improved. In order to make more efficient use of network nodes, this paper proposes a new node distribution strategy. Based on the improved scheme, the tracking accuracy of the 3-DOF manipulator is improved, and the convergence speed of the network is improved.