Event-triggered Control For Multi-Agent Systems Under Network Attacks And Ship Applications

In a multi-agent system,communication networks play an important role in connecting agents.They allow agents to exchange information,share knowledge,coordinate actions,and achieve collaborative work within the system.However,the introduction of communication networks has led to increasingly prominent issues of network security and resource limitations.Therefore,studying event-triggered control of multi-agent systems under network attacks is of great theoretical significance and engineering value to ensure the safe and reliable operation of the system.In this thesis,the design and analysis of event-triggered multi-agent systems under network attacks are considered,and the theoretical methods are extended to the application of multiple autonomous surface ships.The main research work is as follows:First,this thesis investigates the event-triggered control problem in multi-agent systems based on simplified adaptive dynamic programming.Based on the state information of the agent itself and its neighbors,the local neighborhood consistent error is obtained,the local performance index function of each subsystem is defined,and the event-triggered HamiltonJacobi-Bellman equation is constructed.The single network adaptive dynamic programming architecture is used to solve the event-triggered Hamilton-Jacobi-Bellman equation and obtain the optimal event-triggered control strategy.The experience delay technique is introduced to update the neural network weights,thereby relaxing the requirement of persistence of excitation conditions.The Lyapunov functional is used to prove that all signals in the closedloop system are ultimately uniformly bounded.Next,this thesis investigates the optimal event-triggered control problem in multi-agent systems under false data injection attacks.By designing a secure pre-selector and using sorting and median operators to obtain sensor data that has not been attacked.Meanwhile,in order to solve the problem of unmeasurable internal states of the system,a state observer is designed to estimate the unmeasurable states of the system using the obtained sensor data.Based on this,the local neighborhood consistent error and local performance index function are obtained,and the Hamilton-Jacobi-Bellman equation is established.The event-triggered mechanism is introduced to transform the optimal event-triggered control problem into the optimal control problem.The single-network adaptive dynamic programming method is used to solve the optimal event-triggered control strategy,which avoids the explicit calculation of the value function and policy function,simplifies the calculation process,and reduces the computational complexity.Then,this thesis investigates the event-triggered control problem with input quantization in multi-autonomous surface vessels under denial-of-service attacks.The structure combining graph theory and leader-following method is used to decouple the multi-autonomous surface vessel system through coordinate transformation.The output of the hysteresis quantizer is nonlinearly decomposed by utilizing sector-bounded property to eliminate quantization error.A relative threshold event-triggering mechanism is designed,and an event-triggered control strategy based on reliable attack detection mechanism is proposed according to the logical relationship between the output voltage level signals of physical components,which can effectively reduce the adverse effects of denial-of-service attacks while reducing the control input update frequency.Finally,through Lyapunov analysis,the proposed control protocol ensures that all closed-loop signals remain bounded under attacks.Finally,this thesis investigates the dynamic event-triggered control problem with unknown disturbances in multi-autonomous surface vessels under denial-of-service attacks.Based on the structure of graph theory and leader-following method,the multi-autonomous surface vessels are decoupled by coordinate transformation.An interference observer with auxiliary variables is designed to estimate the unknown disturbances,and the estimation and compensation of unknown disturbances can be achieved by adding equal compensation in the virtual controller.At the same time,a dynamic event-triggered control strategy based on a reliable attack detection mechanism is proposed by combining the triggering error and dynamic adjustment rules and according to the logical relationship between the output voltage level signals of physical components.Unlike the relative threshold event-triggering mechanism,the designed dynamic event-triggering mechanism does not need to be fixed all the time,and the threshold parameter can be dynamically adjusted according to the dynamic law,which further reduces the unnecessary transmission between the controller and the actuator and improves the resource efficiency.