Tracking Consensus of Multi-agent Systems with Varying Number of Agents under Actuator Attacks

This article studies the tracking consensus of multi-agent systems (MASs) with varying number of agents under actuator attacks. Solving such a problem is quite challenging due to the effect of actuator attacks and the switching dynamics of tracking error signals caused by the entry and departure of agents. To facilitate the analysis, it is assumed that the dynamic changes of tracking error and estimation error signals around the switching time instants can be characterized by several inequalities. To solve the tracking consensus problem under consideration, a full-order distributed observer designed based only on the output information of agent is proposed for each following agent to observe its states. Then, a kind of fault-tolerant controllers is skillfully constructed based upon the observer states and the upper bound of actuator attack signals to guarantee the achievement of tracking consensus under actuator attacks. It is demonstrated that the tracking consensus in the closed-loop MASs can be ensured if there is at least one directed path from the leader to each following agent and the dynamics of each agent is controllable. Finally, a numerical simulation is used to demonstrate the effectiveness of the designed controller.