Zero-sum game-based dynamic self-triggered sliding mode control for unknown nonlinear systems with asymmetric input...

Area of Science:

• Control Theory
• Nonlinear Systems
• Adaptive Dynamic Programming

Background:

• Controlling unknown nonlinear systems with asymmetric input constraints presents significant challenges.
• Existing methods often struggle with real-time adaptation and computational efficiency.

Purpose of the Study:

• To develop a robust control strategy for unknown continuous-time nonlinear systems with asymmetric input constraints.
• To address the zero-sum game-based dynamic self-triggered sliding mode control problem.

Main Methods:

• Formulation of a novel nonquadratic value function to convert H-infinity control into an unconstrained zero-sum game.
• Utilizing recurrent neural networks for data-driven reconstruction of unknown system dynamics.
• Design of a dynamic self-triggered mechanism for adaptive determination of triggering instants.

Main Results:

• A streamlined single-critic neural architecture is proposed to solve the Hamilton-Jacobi-Isaacs equation, eliminating actor-network dependency.
• Lyapunov-based analysis guarantees uniform ultimate boundedness of all closed-loop signals.
• The control strategy demonstrated excellent dynamic response characteristics in simulations.

Conclusions:

• The proposed adaptive dynamic programming method effectively addresses the control of unknown nonlinear systems with asymmetric input constraints.
• The dynamic self-triggered sliding mode control strategy ensures system stability and optimal performance.
• Validated on robotic arm and microgrid systems, showcasing practical applicability and robustness.