The Research Of Multi-agent Iterative Learning Formation Control Based On Wireless Communication

Multi-agent system can accomplish some parallel and complicated tasks more efficiently than multiple agents alone.The Iterative Learning Control(ILC)can realize the formation control with high precision for multi-agent systems(MAS)with repetitive motion,such as multiple satellite systems that can perform tasks around the Earth and multiple unmanned vehicles and so on.When using iterative learning control to realize the formation of multi-agent system,it is necessary to transmit the control signal through wireless communication.However,two main issues,including communication delay and packet dropouts,occur when signals are exchanged among adjacent agents.As a result,the convergence of iterative learning control is interfered by these issues.In this thesis,the high-precision formation control of multi-agent systems based on iterative learning control by wireless communication are studied.The main contributions of this thesis are as follows:Firstly,when adjacent agents send signal with one-step random time delay for the formation of multi-agent systems without reference trajectories,the system transfer matrix is derived and the influence of random one-step time delay for system convergence speed is analyzed.An iterative learning control method is proposed to implement formation based on adjacent agents.The simulation results demonstrate that the output error of system can converge in the method.Then,when adjacent agents send signal with arbitrary random time delay for the formation tracking of multi-agent systems with reference trajectories,the iterative learning control model in the frequency domain is established,and the first-order and second-order iterative learning laws are proposed respectively.Using the generalized Nyquist criterion and the Gershgorin disk theorem,the convergence condition of system error is obtained.The results show that the formation tracking error of multi-agent systems converges to a certain value using the first-order iterative learning control,and the formation tracking error of multi-agent systems converges to zero using the second-order iterative learning control.Finally,the iterative learning control is used to realize the formation tracking of multi-agent system when adjacent agents send signal with communication packet dropout for the multi-agent systems with reference trajectories.Similarly,an iterative learning control model in the frequency domain is established.The method using previous iterative data to compensate the data lost in this iteration is proposed.Simulation results demonstrate that the system error can converge to zero using the proposed compensation method regardless of the probability of packet loss.