Data-driven Security Control Of Multi-agent Systems Under Cyber Attack

Benefiting from the progress and mutual influence in the fields of control science,computer science and manufacturing,multi-agent systems have developed rapidly and are widely used in social production.Traditional model-based control methods are gradually unable to meet the practical application requirements of heterogeneous multi-agent systems,which makes the model-free control methods more popular.The multi-agent system relies on the network for data exchange,and the network security problem threatens the stable operation of the system,so it must be paid attention to.Research on how to use limited network and computing resources more efficiently can help accelerate the practical application of multi-agent systems.Under the above background,this paper designs an event-triggered model-free adaptive control scheme for the multi-agent cooperative control problem under false data injection attack.The main research contents of this paper are as follows.First,the distributed model-free adaptive cooperative control of single-input singleoutput heterogeneous nonlinear multi-agent systems under false data injection attack is studied.Based on the single-system compact form dynamic linearization technology,a new distributed compact form dynamic linearization technology is proposed for multi-agent systems.Considering the impact of false data injection attacks,a distributed model-free adaptive safety control algorithm is proposed to achieve cooperative control of the multiagent.The designed controller does not need the information of the network topology and only uses the input and output data of the system,which makes the algorithm more suitable for the actual system.Finally,numerical examples are used to verify that the proposed method can achieve consensus control,tracking control and containment control.Then,an event-triggered mechanism is introduced to the distributed model-free adaptive control algorithm to reduce the occupation of network communication resources.The designed event-triggering mechanism is less conservative than existing works.At the same time,considering the impact of false data injection attacks,an event-triggered based distributed model-free adaptive safety control algorithm is proposes to achieve the cooperative control of multi-agent.Finally,it is verified that the proposed method can achieve consensus control using numerical examples and examples with practical background.The final chapter summarizes the research results of the full text and discusses possible future research directions.