Research On Event-Triggered Predictive Control Of Networked Multi-Agent Systems

Consensus control problem of networked multi-agent systems(MASs)is a basic problem of coordinative control of MASs.Some problems in the system,such as process constraints,uncertainty,external disturbances and actuator failures,bring trouble to the consensus control of multi-agent systems.On the other hand,there are some inevitably problems in networked systems,such as network-induced delay,network attacks,and limited communication bandwidth.Model predictive control(MPC)is a well-known and effective method with the ability to deal with complex constraints.However,the MPC needs to continuously solve optimization problems online,so as to consume a lot of computation resources.Based on the above analysis,an event-triggered scheme is introduced into MPC to solve the consensus problem of MASs with constraints,time-delays,cyber-attacks and disturbances in this paper.The main research contents of the paper are as follows:Firstly,an event-triggered robust model predictive control algorithm is proposed for a class of discrete multi-agent systems with actuator faults.The event-triggered mechanism is used to save communication resources and energy consumption;the designed robust model predictive controller gain can be obtained by solving a constrained optimization problem.Through theoretical analysis,it could be found that the proposed controller can enable multi-agents achieve average consensus in the presence of faults,control input amplitude constraints,and time-delays.Then,an adaptive event-triggered robust model predictive control algorithm is proposed for a class of discrete multi-agent systems with time-varying delays and unknown disturbances under arbitrary network topology.An allocating minimum number of leader nodes algorithm is employed first to allocate the least number of leader nodes in arbitrary network topology,and a virtual leader is introduced to connect the allocating leader nodes so as to construct a network topology including a directed spanning tree.In the design of the consensus control protocol,a new adaptive event-triggering mechanism is introduced to save communication and computing resources under the premise of ensuring system control performance.The controller gain of RMPC and the event-triggering conditions can be obtained by solving an optimization problem constrained by linear matrix inequalities.In addition,by using the eigenvalue decomposition method,the dimension of solving linear matrix inequalities will not increase with the increase of the number of agents.Finally,an event-triggered robust model predictive control algorithm is proposed for a class of discrete multi-agent systems under hybrid cyber-attack.The hybrid attacks,including deception attacks on actuators and denial-of-service attacks on communication networks,could destroy the control data and affect information transmission so as to endanger the functionality of the closed-loop system.The proposed controller gain can be obtained by solving the optimization problem constrained by linear matrix inequalities,meanwhile,an event-triggered scheme with a maximum trigger interval is introduced to save computing and communication resources.It could be proved that the H∞performance of the closed-loop system is guaranteed.