Dynamic Event-Triggered Optimized Backstepping Control for Strict-Feedback Nonlinear Systems With Communication Constraints

In this work, a dynamic event-triggered optimized backstepping control approach based on neural networks is proposed to address communication limitations in strict-feedback nonlinear systems. Neural networks are employed to approximate the unknown system uncertainties, with their parameters updated via a gradient descent algorithm. To enhance the communication efficiency of both control inputs and neural network parameters, a dynamic event-triggering scheme is devised. This mechanism adaptively modifies its threshold parameters in real time, depending on the system’s tracking performance. Additionally, a disturbance observer is incorporated to mitigate the influence of external perturbations. By formulating a barrier-type performance index for subsystem optimization and integrating observer–actor–critic structures, both virtual and actual optimal controllers are derived within an inversion-based control framework. Lyapunov theory is utilized to demonstrate that all signals in the closed-loop system are uniformly ultimately bounded. Finally, the effectiveness and practicality of the control strategy are verified by numerical simulations and an application case of an electromechanical system. Note to Practitioners—This paper addresses the practical challenge of controlling strict-feedback nonlinear systems in real-world environments where communication bandwidth is limited and external disturbances are prevalent. In scenarios such as robotic manipulators, aerospace propulsion, and power electronics, frequent data exchange between sensors, controllers, and actuators can lead to network congestion, increased energy consumption, and degraded real-time responsiveness. This work proposes a novel control framework that integrates dynamic event-triggered mechanisms, disturbance observers, and actor–critic neural networks within an optimized backstepping structure. By co-optimizing the triggering of control commands and neural network weight updates, the method significantly reduces communication overhead without sacrificing tracking accuracy or system stability. Furthermore, the disturbance observer ensures robust performance under unknown perturbations. This technique offers a practical and scalable solution for networked control systems in engineering applications that demand high reliability and communication efficiency.