Event-triggered adaptive NN control for MIMO switched nonlinear systems with non-ISpS unmodeled dynamics

This paper investigates the problem of event-triggered adaptive neural network (NN) control for multi-input multi-output (MIMO) switched nonlinear systems with output and state constraints and noninput-to-state practically stable (ISpS) unmodeled dynamics. A nonlinear mapping is firstly utilized to deal with output and state constraints. Also, by developing a new switching signal with persistent dwell-time (PDT) and a switching dependent dynamic signal, the difficulty caused by some non-ISpS unmodeled dynamics is overcome. Then, a type of switching event-triggering mechanisms (ETMs) and event-triggered adaptive NN controllers of subsystems are designed, which handle the issue of asynchronous switching without requiring any known restriction on maximum asynchronous time. A piecewise constant introduced into this ETM effectively ensures a strict positive lower bound of interevent times. Zeno behavior is thus ruled out. Finally, by proposing a novel class of switching signals with reset PDT, it is ensured that all output and state constrains are never violated and all signals of the switched closed-loop system are semi-global uniform ultimate boundedness (SGUUB). A two inverted pendulum system and a numerical example are provided for illustrating the applicability and validity of the proposed method. (C) 2022 The Franklin Institute. Published by Elsevier Ltd. All rights reserved.

Automated summary

This paper investigates the problem of event-triggered adaptive neural network control for multi-input multi-output switched nonlinear systems with output and state constraints and noninput-to-state practically stable unmodeled dynamics. The paper proposes a nonlinear mapping to handle constraints, overcomes the difficulty caused by some non-ISpS unmodeled dynamics with a new switching signal, and designs event-triggering mechanisms and adaptive neural network controllers to deal with asynchronous switching without any known restriction on maximum asynchronous time.