Dynamic Learning-Based Optimal Sliding Mode Control for Fuzzy Singularly Perturbed Systems With FDI Attacks and Communication Constraints

Abstract

This article explores the sliding mode control (SMC) issue for Takagi–Sugeno (T–S) fuzzy model-based singularly perturbed systems (SPSs) with bandwidth-limited and cyberattack-threatened communication. First, to ease the communication constraints on system performance, a novel dynamic event-triggering mechanism (DETM) is designed to reduce the transmission of redundant data adaptively; moreover, considering that the network bandwidth is now generally divided into multiple channels, a multichannel-oriented weighted try-once-discard (MWTOD) protocol is proposed to realize collision-free data transmission over multiple communication channels at event-triggering instants. Then, focusing on false data injection (FDI) attacks, which are a type of commonly encountered security threat, a secure observer-assisted sliding mode controller with undetermined gain matrices is presented. Subsequently, by constructing an augmented T–S fuzzy SPS model, the sufficient conditions for the stability with guaranteed $H_{\infty }$ performance of the system and the reachability of the sliding surface are analyzed, which is accompanied by the derivation of the observer and controller gains. To improve the control performance, a dynamic learning-based adaptive particle swarm optimization (APSO) algorithm is further devised with the aim to minimize the sliding domain. Simulations are finally conducted to verify the effectiveness of the proposed SMC strategy.