Prescribed Performance Consensus Control And Optimization For Nonlinear Multi-Agent Systems

In recent years,multi-agent systems are widely utilized in many military and industrial fields such as satellite,UAV,robot and traffic scheduling,the distributed cooperative control technology of multi-agent system has received extensive attention.The consensus control is the basic of distributed cooperative control,in which the control objective is driving the key states to achieve the consensus by exchanging information with neighbor agents.Nowadays,the research on consensus control focuses on designing methods to realize the consensus and ensure the stability.However,in practical engineering systems,it is required that not only guarantee the stability,but also ensure some key performances to satisfy requirements or optimize some performances.Therefore,based on the summary of previous work,the consensus control problems with prescribed performance requirements and optimal consensus control problems are deeply investigated under Leader-Follower framework in this paper.The results about the consensus control are further enriched.The main contributions are summarized as follows.(1)The consensus control problem with prescribed performance requirements is investigated for a class of nonlinear multi-agent systems with quantized control input connected by directed communication topology graph.Considering nonlinear multi-agents systems with strict-feedback and nonstrict-feedback nonlinearities,neural state observer are designed to estimated unavailable states.The output-feedback consensus tracking control protocol is designed based on prescribed performance control strategy and dynamic surface control methods.By using Lyapunov stability theory,all signals of the closed-loop system are proved to be ultimately uniformly bounded,the consensus is achieved,transient and steady performances satisfy given requirements.(2)The event-triggered prescribed performance consensus control problem is investigated for a class of nonlinear multi-agent systems with input saturation connected by directed communication topology graphs.Firstly,neural networks are utilized to approximate unknown nonlinear functions and neural state observers are designed to estimated unavailable states.Secondly,based on prescribed performance control method,dynamic surface control method,prescribed performance consensus control protocols are designed by introducing relative threshold conditions and switching threshold conditions.The auxiliary signal is constructed to solve the control constraints.By using Lyapunov stability theory,all signals of the closed-loop system are proved to be ultimately uniformly bounded,the consensus is achieved,transient and steady performances satisfy given requirements,and Zeno behavior is avoided.(3)The event-triggered optimal consensus control problem is investigated for nonlinear multiagent systems with input saturation based on Bellman optimality principle and adaptive dynamic programming methods.A class of non-quadratic index is introduced to handle control constraints.Based on event-triggered sampling mechanism,under the single network framework,the optimal control protocol is designed by neural networks.Only when the triggering conditions were satisfied,the controller will sample the signals of sensors and update the weight estimation vectors and control laws,the computation and communication costs are effectively reduced.All signals of the closed-loop systems are guaranteed ultimately uniformly bounded and Zeno behavior is avoided.(4)The robust optimal consensus control protocol is investigated for nonlinear multi-agent systems consisted of uncertain multiple Euler-Lagrange systems with external disturbance connected by directed communication topology graphs.The distributed state estimators and event-triggered mechanism are designed by utilizing exchanged states and the local adaptive controller is designed.Based on differential game theory and adaptive dynamic programming,robust optimal consensus control protocol is derived by using neural network to approximate the Nash solution under the single network framework.The requirement of real-time communication is avoided,computation and communication costs are reduced effectively,meanwhile undirected communication topology graphs,global knowledge of graph is avoided and robustness is improved.All signals of the closed-loop systems are guaranteed ultimately uniformly bounded and Zeno behavior is avoided.(5)The formation control problem with prescribed performance is investigated for uncertain nonlinear multi-agent systems consisted of multiple wheeled nonholonomic mobile robots in the presence of unknown skidding and slipping effects.Based on prescribed performance control strategy and dynamic surface control method,formation control protocol is designed by using neural networks to estimate and compensate nonlinear functions with unknown skidding and slipping terms.Moreover,hyperbolic tangent functions are used to approximate control constraints.Based on Lyapunov stability theory,all signals of the closed-loop systems are guaranteed ultimately uniformly bounded,the formation position errors satisfy given transient and steady performance requirements.The adaptation to environment for wheeled nonholonomic mobile robots is improved.By using Lyapunov stability theory,all signals of the closed-loop system are proved to be ultimately uniformly bounded,the transient and steady performances of distributed formation errors satisfy given requirements.