Distributed Time-Triggered And Event-Triggered Optimization Control For Multi-Agent Systems

In the past decade,the research on distributed optimization control for multi-agent systems has attracted a large number of scholars,it's the reason that this field has important significance in practical systems,including computer network,power systems and so on.However,most of the existing works of distributed optimization control are considering the continuous-time control algorithms.It's worth noting that the communication network bandwidth among agents and energy resources are restrained,so there are more and more scholars who gradually began to consider design the discrete-time control algorithms to provide more theoretical basis for practical applications.This paper mainly studies the distributed optimization control of second-order multi-agent system.The main work is as follows.1.Firstly,the distributed time-triggered optimal control algorithms for the secondorder multi-agent system with external disturbance are proposed.The time-triggered strategy is the effective tool to reduce the communication frequency among agents and the computation frequency of gradient information.The destination of distributed optimization control is to minimize the system performance index function,which is the sum of the local cost function of each agent.It is worth noting that the local cost function here can only be obtained by the corresponding agent.A series of time-triggered distributed optimization control protocols are proposed in this part,where the agents can gradually converge to the optimal state.The instant of communication among agents and the gradient information sampling times are determined by a fixed period.Finally,the theoretical proof and the numerical simulation are provided to validate the effectiveness of the conclusion.2.Secondly,the distributed event-triggered optimal control algorithms for the second-order multi-agent system with external disturbance are proposed.The eventtriggered strategy is used to reduce the communication frequency among agents and the computation frequency of gradient information.A series of event-triggered distributed optimization control protocols are proposed in this part,the instant of communication among agents and the gradient information sampling times are determined by two different event conditions.What's more,the agents can gradually converge to the optimal state.Then the theoretical proof is given,which indicates that the trigger interval of two different event conditions has a lower bound strictly greater than zero,it means that Zeno phenomenon will not occur by the designed event-triggered algorithms.Further,the numerical simulation is provided to verify the validity of the conclusion.