Event-triggered adaptive dynamic programming for decentralized tracking control of input constrained unknown...

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• Nonlinear Dynamics

Background:

• Decentralized tracking control (DTC) is crucial for complex interconnected systems.
• Unknown dynamics and input constraints pose significant challenges in DTC.
• Existing methods often require continuous data exchange, leading to high resource consumption.

Purpose of the Study:

• To develop an event-triggered adaptive dynamic programming approach for DTC in input-constrained, unknown nonlinear interconnected systems.
• To reconstruct system dynamics using local information and observer-based techniques.
• To optimize control policies while minimizing computational and communication overhead.

Main Methods:

• A neural-network-based local observer is designed to reconstruct subsystem dynamics.
• A nonquadratic value function transforms the DTC problem into an optimal control problem.
• An observer-critic architecture with experience replay tunes the critic neural network weights for policy optimization.

Main Results:

• The proposed method successfully implements decentralized tracking control for unknown nonlinear systems.
• The event-triggering mechanism significantly reduces computational and communication load.
• Lyapunov's direct method confirms the stability of the closed-loop system under the proposed control policy.

Conclusions:

• The event-triggered adaptive dynamic programming scheme provides an effective solution for DTC in complex nonlinear systems.
• The approach conserves computational resources and communication bandwidth.
• The demonstrated stability guarantees the reliability of the proposed control strategy.