Neural Network Adaptive Control For Uncertain Nonlinear Multi-agent Systems With State Constraints

In recent years,the multi-agent systems have been widely used in various fields due to the demand of actual production and life.Therefore,the application of intelligent control in multi-agent systems have been widely concerned by scholars.At the same time,due to the production process,implementation mechanism and safety considerations,constraints often exist in the actual system,and once the constraints are violated,the work efficiency may be reduced or even the safety of people's lives and property may be threatened.Therefore,it is very important to study multi-agent systems with full state constraint.This paper will carry out research from the following three aspects:(1)An adaptive neural network distributed control scheme is proposed for nonlinear non-affine multi-agent systems with unknown control direction and full state constraints.Firstly,the mean value theorem is used to transform the non-affine multi-agent systems into affine systems,in which the properties of neural networks are used to approximate the unknown items in the systems.Secondly,the original systems with state constraints are transformed into equivalent unconstrained systems by the method of function transformation in one-to-one mapping,and the transformed systems have the same properties as the original systems,so as to ensure that the state of each agent does not violate the predetermined constraint boundary.Thirdly,based on the backstepping design,the adaptive distributed controller is designed,and the Nussbaum function is used to solve the unknown control direction of the systems.Meanwhile,all signals of the closed-loop systems are bounded by using the stability theory of Lyapunov functions.Finally,a simulation example is given to verify the effectiveness of the proposed method.(2)An adaptive neural network fault-tolerant distributed control method for nonlinear non-affine multi-agent systems with actuator failure and input saturation is proposed.Firstly,by using the mean value theorem to transform the non-affine multi-agent systems,an equivalent affine model can be obtained.Secondly,based on integrating Barrier Lyapunov function and backstepping recursion,an adaptive fault-tolerant distributed controller is designed,in which the unknown functions in the system are approximated by neural network method.At the same time,the integration Barrier Lyapunov function is selected to make the state variables of the system not violate the constraint bounds.Finally,a simulation example is given to verify the effectiveness of the proposed method.(3)An adaptive distributed control strategy is proposed for nonlinear multi-agent systems under the constraints of directed graph and state.In this paper,the integration Barrier Lyapunov function(i BLF)is introduced to overcome the conservative limitation of the barrier Lyapunov function with error variables,relax the feasibility conditions,and simultaneously solve the coupling term and state constraints of the communication error between agents.An adaptive distributed controller was designed based on i BLF and backstepping method,and i BLF was differentiated by means of the integral mean value theorem.At the same time,the properties of neural network are used to approximate the unknown terms,and the stability of the system is proved by Lyapunov stability theory.This scheme can not only ensure that the output of all the followers meet the output trajectory of the leader,but also make the state variables not violate the constraint bounds,and all the closed-loop signals are bounded.Finally,a simulation example is given to verify the effectiveness of the proposed method.