Data-driven output consensus for a class of discrete-time multiagent systems by reinforcement learning techniques

Abstract

The primary objective of this research paper is to examine the challenges associated with optimization and tracking control in discrete-time multiagent systems. This study begins by defining the optimal tracking control issue within the framework of a leader–follower multiagent system. A novel approach based on policy iteration is introduced for calculating the control law and performance index. The paper also includes an analysis of the convergence properties of this technique, ensuring its effectiveness in reaching optimal solutions. To facilitate this, an actor–critic neural network framework is utilized to approximate the control law and iterative performance index function. The proposed approach enables the real-time implementation of the policy iteration method, eliminating the requirement for prior knowledge of the system’s dynamics. The paper concludes with a series of simulation experiments that demonstrate the practicality and efficiency of the developed optimal tracking control strategies.