Neural-Network Approximation-Based Adaptive Periodic Event-Triggered Output-Feedback Control of Switched Nonlinear Systems

• Published in: IEEE transactions on cybernetics Vol. 51; no. 8; pp. 4011 - 4020
• Main Authors: Li, Shi, Ahn, Choon Ki, Guo, Jian, Xiang, Zhengrong
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.08.2021; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Adaptive control; Adaptive neural-network (NN) control; Adaptive sampling; Adaptive systems; Artificial neural networks; Continuously stirred tank reactors; Feedback control; Mathematical analysis; Neural networks; Nonlinear control; Nonlinear systems; Observers; Output feedback; periodic event-triggered control (PETC); State observers; switched nonlinear system (SNS); Switched systems; Switches
• ISSN: 2168-2267; 2168-2275; 2168-2275
• DOI: 10.1109/TCYB.2020.3022270

This study considers an adaptive neural-network (NN) periodic event-triggered control (PETC) problem for switched nonlinear systems (SNSs). In the system, only the system output is available at sampling instants. A novel adaptive law and a state observer are constructed by using only the sampled system output. A new output-feedback adaptive NN PETC strategy is developed to reduce the usage of communication resources; it includes a controller that only uses event-sampling information and an event-triggering mechanism (ETM) that is only intermittently monitored at sampling instants. The proposed adaptive NN PETC strategy does not need restrictions on nonlinear functions reported in some previous studies. It is proven that all states of the closed-loop system (CLS) are semiglobally uniformly ultimately bounded (SGUUB) under arbitrary switchings by choosing an allowable sampling period. Finally, the proposed scheme is applied to a continuous stirred tank reactor (CSTR) system and a numerical example to verify its effectiveness.