| The multi-agent systems are distributed systems composed of agents with communication,computing,and decision-making capabilities,which can realize tasks that cannot be accomplished by a single huge complex system.This pattern has been widely applied to in military,production and life fields,etc.The existing results of the multi-agent system control theory rely on the model information of the system,however,with the increasing complexity and demand of the production process,the modeling process of the actual control system is increasingly difficult.On the other hand,with the use of advanced sensing technology and data storage technology,a large amount of online and offline operation data in the multi-agent system performing production tasks is easy to be collected,transmitted and stored.Then,in the presence of the system model is difficult to establish,only using I/O data to achieve the cooperative control of the multi-agent systems has important theoretical significance and application value.Therefore,this paper studies a data-driven distributed cooperative control method for nonlinear multi-agent systems with unknown models.The main innovative work is summarized as follows:(1)Consider the consensus problem of nonlinear multi-agent systems with cooperative-competitive interactions,a distributed bipartite model-free adaptive consensus control protocol is proposed.The tracking error of the agent under the structural balanced signed directed graph is defined,and the sufficient condition to ensure the asymptotic convergence of the tracking error is obtained.Secondly,considering the asymmetry of the agent’s output in practice,an asymmetrical bipartite consensus control objective is defined.By introducing an asymmetric influence factor,an asymmetric bipartite model-free adaptive consensus control protocol is designed,and the stability result of agents converging to different modulus values and different symbols is given.(2)Consider the formation problem of multi-input multi-output nonlinear multi-agent systems with the cooperative-competitive interactions,a distributed bipartite formation control protocol is proposed.First,an equivalent data model is established through a compact dynamic linearization method.By using the matrix decoupling,the disc theorem and the binomial theorem,the formation control problem is studied under constant desired output and time-varying desired output,and sufficient conditions for asymptotic and bounded convergence of formation tracking errors are respectively given.Finally,the simulation example is verified that the proposed protocol can realize the symmetrical bipartite formation of the multiple-input multiple-output agents with fixed and time-varying formations,respectively.(3)Consider the consensus problem of nonlinear multi-agent systems,an event-triggered distributed model-free adaptive control protocol is proposed.By using Lyapunov stability theory,an event-triggered function is first designed,then a sufficient condition that ensures the asymptotic convergence of the tracking error is given.The design method also extends to solve the consensus problem of the multi-agent system with cooperative-competitive interactions.The results show that when the in-degree is not less than the out-degree for each agent,the distributed data-driven protocol can ensure the asymptotic convergence of the system with reducing the communication of the agents.(4)Consider the consensus problem of nonlinear repetitive multi-agent systems within a finite time,an event-triggered model-free adaptive iterative learning control protocol based on offline data is proposed.According to the defined consistent norm and matrix factorization method,the bounded convergence of multi-agent systems with weakly connected directed graph is given along the iterative domain.Meanwhile,the problem of asymmetric bipartite consensus for multi-agent systems under weakly connected signed directed graph is studied,and an asymmetric bipartite model-free adaptive iterative learning consensus protocol is designed,which realizes the high-precision tracking of the multi-agent systems with desired time-varying trajectory under the weak connectivity condition.(5)Consider the multi-wheeled robots that performing inspection tasks in the complex environment of the mine,the problem of consensus trajectory tracking control is studied,and distributed data-driven motion control protocols are proposed.Based on the dynamic linearization method,the dynamic linearization data model between the robot’s position information output and the velocities(linear velocity and angular velocity)input data is established.Firstly,consider the data loss phenomenon in the communication process of robots,a model-free adaptive motion control protocol based on incomplete distributed position data is proposed,and a sufficient condition for bounded convergence of multi-robot systems in the mathematical expectation sense is given.Secondly,an event-triggered control protocol based on robot position information is designed,and the bounded convergence of the system is obtained while reducing the number of communications between robots.The results show that the proposed data-driven method can well complete the trajectory consensus control task of the wheeled mobile robots for the inspection task,and the feasibility of the formation control based on the design of consensus trajectory tracking protocol is further verified by simulation. |
