Adaptive Neural Control For Nonlinear Multi-agent Systems With Complicated Constraints

As an important branch of distributed artificial intelligence,multi-agent systems increase capabilities by virtue of local communication and interaction among agents,so that they can complete complex tasks that cannot be achieved by a single system,and have higher flexibility,reliability and versatility.At present,the applications of multi-agent systems have been extended to various fields,such as multi-unmanned aerial vehicles systems for collaborative operations,reconnaissance,vessels formation cruising and collaborative operation of multiple manipulators,etc.In these practical applications of multi-agent systems,they are usually limited by various constraints caused by the operating environments,performance requirements and their own physical devices.Once these constraints are transgressed,it will lead to a sharp decline in performance,system instability and even cause serious safety accidents.Therefore,it is of great significance to investigate the cooperative control method for multi-agent systems subject to constraint conditions.This paper focuses on the following studies of nonlinear multi-agent systems with different constraint conditions:1.Under the conditions of full state constraints and prescribed performance,the consensus control problem of non-strict feedback nonlinear multi-agent systems is considered.An adaptive neural controller is constructed based on a fixed threshold event-triggered mechanism,and the number of controller updates is reduced.In order to achieve the desired performance requirements,an equivalent transformation based on a performance function is performed on the tracking error,so that the tracking error can converge to the ideal range within a specified time.Aiming at the problem of full state constraints,the barrier Lyapunov function is introduced into the controller design process,and the system states are guaranteed to be confined within the constraint range.2.Under the conditions of asymmetric time-varying state constraints and input saturation,the consensus control strategy for non-strict feedback nonlinear multi-agent systems with unknown control direction is researched.By designing a distributed sliding-mode estimator,each follower can obtain the estimate of leader's trajectory and track it directly.Then,the asymmetric time-varying barrier Lyapunov function is introduced to design an adaptive neural controller to solve the problem of asymmetric time-varying state constraints.The mean-value theorem and Nussbaum function are employed to deal with the problems of input saturation and unknown control direction,simultaneously.In addition,the least parameter method is used to select the adaptive parameters,and a compensation function is also presented,so that only one adaptive law needs to be designed for each agent,which greatly reduces the computational complexity.3.Under the condition of deferred asymmetric time-varying state constraints,the consensus control method for nonlinear multi-agent systems is investigated,and an event-triggered adaptive neural control strategy is developed.A shifting function is utilized to transform the error variables such that the initial tracking condition can be totally unknown,and the constrained control can be realized at a specified time instant.Meanwhile,the deferred asymmetric time-varying full state constraints are addressed by a class of asymmetric barrier Lyapunov function.By integrating a relative threshold event-triggered mechanism into the adaptive neural controller,the communication burden is released and the ideal control performance is ensured,simultaneously.