Dynamic Event-Driven Finite-Horizon Optimal Consensus Control for Constrained Multiagent Systems.

Area of Science:

• Control Theory
• Artificial Intelligence
• Networked Systems

Background:

• Finite-horizon optimal control for multiagent systems presents challenges due to time-varying Hamilton-Jacobi-Bellman equations.
• Input constraints in multiagent systems require sophisticated control strategies.

Purpose of the Study:

• To develop an event-driven finite-horizon optimal consensus control for multiagent systems.
• To address symmetric or asymmetric input constraints effectively.
• To enhance communication resource utilization through a novel dynamic event-triggering mechanism.

Main Methods:

• A single-critic neural network (NN) with a time-varying activation function approximates the optimal control.
• An augmented error vector updates NN weights, minimizing terminal errors.
• An improved learning law relaxes persistence excitation conditions and eliminates initial stability requirements.
• A dynamic event-triggering mechanism (DETM) with dynamic threshold parameters (DTPs) and auxiliary dynamic variables (ADVs) is proposed.

Main Results:

• The proposed method effectively approximates optimal control under input constraints.
• The novel DETM enhances communication efficiency compared to existing methods.
• The learning law accelerates convergence and removes the need for initial stability control.
• Simulation results validate the method's effectiveness and the DETM's superiority.

Conclusions:

• The developed event-driven optimal consensus control is effective for multiagent systems with input constraints.
• The proposed dynamic event-triggering mechanism offers improved communication resource utilization.
• The approach provides a robust and efficient solution for complex control problems in networked systems.