Data-Driven Iterative Learning Cluster Consensus Control for Nonlinear MASs Under Actuator Faults and DoS Attacks

This article explores the resilient fault-tolerant cluster consensus control problem for the repeatable nonlinear multiagent systems under actuator faults and denial-of-service (DoS) attacks. To circumvent unknown agent dynamics and capitalize on prior operation data, the data mapping of agents with fault information is established along the iteration domain by the locally dynamic linearization technique. In the cyber layer, stochastic DoS attacks are assumed to follow the Bernoulli distribution, and a backward attack compensation mechanism is introduced. In the physical layer, an adaptive actuator fault compensation strategy derived from the improved projection algorithm is built. Within this design, a data-driven fault-tolerant and intrusion-tolerant iterative learning control (FTIT-ILC) method is formulated to ensure dual guarantees. The convergence analysis condition of the method is presented by the nature of irreducible sub-stochastic matrices. Finally, experiments verify the FTIT-ILC method.