Fault-Tolerant Consensus of Multi-Agent Systems Subject to Multiple Faults and Random Attacks

This paper explores the consensus control problem of nonlinear multi-agent systems (MASs) under complex cyber-physical threats (CPTs), which encompass sensor/actuator faults, input/output channel noises, and random cyber-attacks. The multiple sensor/actuator faults are uniformly modeled as an exponential type, while random cyber-attacks are characterized by a Markov chain. To enhance the safety and security of MASs under CPTs, the distributed normalized observers are first developed, enabling precise estimations of unknown state and fault information. Subsequently, the distributed fault-tolerant consensus control (FTCC) scheme with a positive reconstruction mechanism is proposed to maintain resilience against attacks, compensation for faults, and robustness to noises in MASs under adverse CPTs. The two notable innovations can be outlined as follows: i) The achievement of FTCC objectives under complex CPTs, demonstrating strong algorithmic transferability in both non-attack and random attack scenarios. ii) The adoption of a double-layer distributed framework in the estimation layer and control layer, balancing computational complexity and efficiency improvements compared to a combination of decentralized and distributed approaches. Simulation results finally confirm the efficacy and feasibility of the proposed FTCC algorithm.