Adaptive Neural Network Control for a Class of Nonlinear Systems with Unknown Control Direction
Chenliang Wang,Lei Guo,Changyun Wen,Qinglei Hu,Jianzhong Qiao
DOI: https://doi.org/10.1109/tsmc.2018.2861003
2018-01-01
IEEE Transactions on Systems Man and Cybernetics Systems
Abstract:In this paper, a novel adaptive neural network (NN) control scheme is proposed for a class of nonlinear systems with unknown control direction. By introducing some differentiable functions and high-order Lyapunov functions, the obstacle caused by unknown control direction in NN control is successfully circumvented and all closed-loop signals are shown to be uniformly bounded up to infinite time. Meanwhile, by introducing an error transformation technique, it is rigorously proved that the argument of the unknown nonlinearities remains within a compact set which can be explicitly calculated a priori , making the NN approximation always valid. Moreover, with the aid of a bound estimation approach, we effectively compress the impact of approximation errors and external disturbances and steer the tracking error into a predefined small residual set. Simulation results illustrate the effectiveness of the proposed scheme.