Data-Based Adaptive Event-Triggered Transfer Stabilization for Nonlinear Networked Systems

This paper investigates the adaptive event-triggered data-driven control problem for a class of unknown nonlinear discrete networked systems. To address this problem, a stochastic configuration network-based algorithm is developed to construct a candidate mapping set within the modeling-valid domain. Subsequently, an adaptive event-triggered control protocol is proposed, and a closed-loop mapping set is obtained. Then, by leveraging the ideas of transfer stabilization and the S-lemma, a data-driven stability criterion for nonlinear discrete networked systems is derived. The stability criterion solely relies on the data of the controlled system and is independent of both the system model and the data model. Based on this stability criterion, the control gain and triggering matrix of the controlled system can be obtained. Additionally, the effectiveness and practicality of the proposed method are validated through a numerical example and a complex memristive Hopfield neural network circuit. Note to Practitioners—The increasing complexity of systems leads to higher data density, which in turn imposes significant computational demands, particularly when system models are unavailable, making efficient computation a challenging task. Neural network-based modeling techniques provide powerful tools for reconstructing system models. However, the high-dimensional weight matrices introduced during modeling, along with the need to address additional modeling errors, often intensify computational complexity. This paper investigates adaptive event-triggered data-driven control for unknown nonlinear discrete networked systems, presenting an enhanced modeling algorithm based on stochastic configuration neural networks to avoid high-dimensional weight matrices, along with an adaptive event-triggered transfer stabilization strategy that accounts for modeling errors. The proposed approach is validated through its application to a complex memristive Hopfield neural network circuit, demonstrating its efficiency, practicality, and effectiveness in tackling real-world challenges.