An integrated dynamic event-driven ADP framework for distributed resource allocation in nonlinear multi-agent systems

Area of Science:

• Control Theory
• Artificial Intelligence
• Systems Engineering

Background:

• Distributed resource allocation is crucial for multi-agent systems (MASs).
• Existing methods often struggle with nonlinear dynamics and real-time control demands.
• Optimizing resource allocation requires balancing performance and computational efficiency.

Purpose of the Study:

• To develop an integrated adaptive dynamic programming (ADP) framework for second-order nonlinear MASs.
• To address the distributed resource allocation problem with enhanced efficiency and accuracy.
• To design a dynamic event-triggered mechanism for optimized control input updates.

Main Methods:

• Formulation of a coupled performance index based on resource allocation errors.
• Transformation of the resource allocation problem into optimal control policy design.
• Development of a fixed-time single-critic learning law for online Hamilton-Jacobi-Bellman (HJB) equation approximation.
• Implementation of a dynamic event-triggered mechanism for adaptive control input updates.

Main Results:

• The proposed ADP framework effectively solves the distributed resource allocation problem in MASs.
• The dynamic event-triggered mechanism ensures control inputs are updated only when necessary, reducing communication load.
• Theoretical analysis confirms the uniform ultimate boundedness of all closed-loop signals.
• Numerical simulations demonstrate accurate resource allocation and balanced error/control effort.

Conclusions:

• The integrated ADP framework with a dynamic event-triggered mechanism offers a robust solution for resource allocation in nonlinear MASs.
• This approach enhances control performance while optimizing computational resources.
• The study validates the effectiveness and practical applicability of the proposed method.