Neural-Adaptive Dynamic Surface Control for Nonlinear Unmodeled Systems

Abstract

Aiming at a type of strict feedback system with uncertain nonlinearity, we designed a new controller to solve a constraint on input and unmodeled dynamics in the system. We used backstepping technology to design the controller. Simultaneously, in order to avoid the “dimension explosion” problem due to the introduction of the virtual controllers, we applied the dynamic surface control (DSC) technique when we designed the virtual controller for each subsystem. For the nonlinear input problem of the controller, we introduce a continuously differentiable hyperbolic tangent function. We designed a new dynamic signal limited by the controller's input signal and to solve the difficulty of the unmodeled dynamics. Finally, all closed-loop signals in this paper proved to be semi-global uniformly, ultimately bounded (SGUUB), and a simulation experiment using Matlab software is shown in this paper.