Adaptive event-triggered control of switched nonlinear systems with unknown input dead zones and unmodeled dynamics

Abstract

In this article, the problem of adaptive neural event-triggered control (ETC) based on command filter and dynamic surface control (DSC) is discussed for a class of nonstrict-feedback switched nonlinear systems with unknown asymmetric dead zones and unmodeled dynamics. The unknown nonlinear continuous functions are approximated by radial basis function neural networks (RBFNNs). Unmodeled dynamics can be handled efficiently by applying distinct dynamic signals for distinct subsystems. The input dead zone is linearized to facilitate controller design and stability analysis. The explosion of complexity can be avoided by using command-filtered backstepping technology. A new event-triggered strategy with no triggering and multiple triggering is designed for each switching interval. Furthermore, by selecting the initial values of the next subsystem at the time of switching and using DSC, the stability of a single switching interval is linked to the stability of all switching intervals. By theoretical analysis, all signals in the adaptive system are proved to be semi-global uniform ultimate bounded (SGUUB) under arbitrary switching. Meanwhile, the Zeno behavior is removed. Simulation results verify that the proposed control approach is feasible.