Research On Optimal Consensus Control For Multi-agent System With Unknown Dynamic Based On Event-triggered Method

Due to the extensive application of the consensus control of multi-agent system in the engineering field,more and more researchers pay attention to it.As a hot research direction,optimal consensus control problems of multi-agent system also attract lots of scholars.In practical application,it is difficult to obtain the complete model of multiagent system because of the unknowability and complexity of multi-agent system.Therefore,the research on the optimal consistency of multi-agent systems with unknown dynamic models has important practical value.However,a large number of data transmission and computation is a great burden for agents with limited volume and computation.Therefore,how to ensure consistency and reduce data transmission between agents is a challenge faced by many scholars.Event-triggered control mechanism,especially the edge-based event triggering control,provides an effective way to reduce the frequent communication between agents.In this paper,the general event-triggered control mechanism and edge-based event triggered control mechanism are used to solve the optimal consensus problem of discrete-time and continuous time multi-agent systems with unknown dynamics.The main research contents of this paper are as follows:1.Research on event-triggered optimal consensus control problem of discrete-time multi-agent system.For the optimal consensus problem of multi-agent system with unknown dynamic model,the optimal control is generally obtained by solving Hamilton-Jacobi-Bellman(HJB)equation to solve the optimal consensus problem.Combining reinforcement learning with control theory,a novel ?-policy-iterative algorithm is proposed to approximate the solution of HJB equation.The HJB equation is used as the event-triggered control conditions.The neural network will be trained and the system will be updated when the conditions are violated.The simulation results show that the convergence speed of the results is faster than that of the traditional algorithm.Compared with no event triggering,the system does not update continuously,which reduces the computational load while achieving consensus.2.Research on edge-event-triggered optimal consensus control problem of multiagent systems.By using the neighbor-based event triggering method,the agent needs to obtain the information of all its neighbor agents at event-triggered instant.This inevitably produces a lot of unnecessary information,resulting in increased computing load and data redundancy.Different from the neighbor-based event-triggered control method which starts from the agent,the edge event triggering method starts from the edges between agents.In addition,the information of edges is employed to construct the HJB equation and design the edge-event-triggered conditions.When the edge event is triggered,the agent only gets the information of the agent connected through this edge.The effectiveness of the edge-event-triggered control method is proved by simulation experiments.