Data-Driven Optimal Output Cluster Synchronization Control of Heterogeneous Multi-Agent Systems

This paper presents a novel data-driven optimal output cluster synchronization control method for heterogeneous multi-agent systems with disturbances based on adaptive dynamic programming. Traditional cluster synchronization control methods require the information of system matrices, which limit the application scope in the reality. In order to solve this problem, a novel data-driven optimal control method is presented, where the input-state data is utilized without the information of system matrices of state equations, to realize the output cluster synchronization of multi-agent systems. The major contributions are displayed as follows: 1) a novel data-driven optimal output cluster synchronization control method is presented which requires the input-state data of multi-agent systems; 2) a novel distributed adaptive observer is designed which can avoid the effects of negative edge weights between the clusters; 3) the output cluster synchronization control problem is transformed into the output regulation problem, and adaptive dynamic programming method is presented which can realize the disturbance rejection. First, the output cluster synchronization control problem is formulated. Next, a novel optimal output cluster synchronization control method is presented based on distributed adaptive observer and adaptive dynamic programming. Numerical experiment shows the good performance of the presented method. Note to Practitioners—Most of the existing cluster synchronization control methods require the information of system matrices. However, it is hard to obtain the accurate system models in the reality, which limits the application scope of the existing methods. On the other hand, the disturbances in practice make the effective cluster synchronization control of multi-agent systems very challenging. Aiming at the above problems, this paper designs novel data-driven optimal control laws for heterogeneous multi-agent systems with disturbances to realize the output cluster synchronization. A novel distributed adaptive observer is designed to estimate the states and system matrices of leaders. Then, the adaptive dynamic programming method is presented to obtain the optimal control law of each agent based on the estimated information. Comparative experiment is provided to show the good performance of the presented method.