Research On Event-Triggered Consensus Algorithm For Multi-Agent Systems

In recent years,the coordination control problem of multi-agent systems(MASs)has received wide attentions from the experts and scholars in different fields,such as computer,communication,control and so on.It has played a huge role in the practical applications,because it can make the large and complex systems into small,coordinate communication with each other and easy to control.The related theoretical research has been applied to robot formation,unmanned air vehicles,underwater vehicles,etc.Among the research directions in MASs,the consensus problem is a fundamental but most important one.However,in practical occasion,the MASs usually relies on the digital platform for data processing,communication and task execution.In today's increasingly complex device structure and limited network bandwidth,the event-triggered mechanism is obviously more effective than the traditional time-triggered control methods.In this thesis,some event-triggered control protocols are studied on the background of the consensus problems for MASs.The actuator failure,input saturation and unavailable state are considered respectively,taking the general linear MASs as a model.Based on the existing research results,an adaptive event-triggered mechanism is proposed and the leaderless and leader-following consensus problems of MASs are studied under the above model.Moreover,the corresponding event-triggered consensus control protocols are designed at the same time.The main work is as follows:Chapter 3 investigates the general linear MASs model with actuator failure and the corresponding event-triggered control protocol is proposed.This chapter mainly improves the event-triggered mechanism on the basis of existing results.A novel adaptive event-triggered function is designed,which the trigger threshold depends on the time and system states at the same time.It not only overcomes the defect that the state-dependent threshold diminished as the number of agents increases,but also overcomes the defect of the time-dependent threshold cannot be adjusted according to the agents' initial states.In addition,when the global information is unknown and multiplicative fault occurs in the system,the corresponding distributed adaptive laws are designed by using the local interactive information between agents.What's more,some adaptive parameters are also introduced into the trigger function to improve its self-regulation ability.Not only the leaderless consensus is achieved,but also ensure the system will not happen Zeno-behavior.In Chapter 4,the general linear MASs subject to actuator failures and input saturation are considered.Different from the Chapter 3,the case of leader-following consensus with input saturation is studied,which is more suitable for practical application.By using low gain feedback technique,a completely distributed adaptive event-triggered fault-tolerant tracking consensus algorithm design step is given.In order to avoid large gain in Chapter 3,more comprehensive coupling gain and multiplicative fault adaptive laws are developed by using?-modify technique.It is proved that all tracking errors eventually convergence to zero and the considered MASs does not occur saturation.In Chapter 5,for general linear MASs with input saturation,the fact that the agent states are not always measurable is further considered.We design a reasonable state observer to observe the states based on the output information.a distributed event-triggered consensus algorithms is proposed for the considered system and the semi-global consensus of system is also effectively solved.Some conclusions are presented at the end of this thesis and the further research direction is discussed.