Decentralized Event-Triggered Online Adaptive Control of Unknown Large-Scale Systems Over Wireless Communication Networks

• Published in: IEEE transaction on neural networks and learning systems Vol. 31; no. 11; pp. 4907 - 4919
• Main Authors: Su, Hanguang, Zhang, Huaguang, Liang, Xiaodong, Liu, Chong
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.11.2020; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Adaptive control; Adaptive critic design; Communication networks; Communications networks; Control methods; Control systems; Decentralized control; decentralized event-triggered control (DETC); Heuristic algorithms; Interconnected systems; large-scale system (LSS); Large-scale systems; Neural networks; Nonlinear dynamical systems; Optimal control; reinforcement learning; Subsystems; System dynamics; Wireless communication; Wireless communications; Wireless networks; wireless sensor-actuator network (WSAN)
• ISSN: 2162-237X; 2162-2388; 2162-2388
• DOI: 10.1109/TNNLS.2019.2959005

In this article, a novel online decentralized event-triggered control scheme is proposed for a class of nonlinear interconnected large-scale systems subject to unknown internal system dynamics and interconnected terms. First, by designing a neural network-based identifier, the unknown internal dynamics of the interconnected systems is reconstructed. Then, the adaptive critic design method is used to learn the approximate optimal control policies in the context of event-triggered mechanism. Specifically, the event-based control processes of different subsystems are independent, asynchronous, and decentralized. That is, the decentralized event-triggering conditions and the controllers only rely on the local state information of the corresponding subsystems, which avoids the transmissions of the state information between the subsystems over the wireless communication networks. Then, with the help of Lyapunov's theorem, the states of the developed closed-loop control system and the critic weight estimation errors are proved to be uniformly ultimately bounded. Finally, the effectiveness and applicability of the event-based control method are verified by an illustrative numerical example and a practical example.