Research On Multi-agent Consensus Control Method Based On Adaptive Dynamic Programming

With the rapid development of computing technology,communication technology and control technology,distributed collaborative control of multi-agent systems has been paid close attention in the academic community.Consensus control is a very common control problem in multi-agent cooperative control.Its purpose is to make all agents in the same communication network reach an agreement and achieve the same goal.At present,the common control technologies in multi-agent cooperative control include proportional integration control,fuzzy control,model predictive control,etc.these methods only consider the stability of the system,and can not achieve the minimization of energy consumption.Adaptive dynamic programming solves the optimal control of the system on the basis of ensuring the stability of the system,and the performance index function is significantly improved compared with the traditional control methods.Adaptive dynamic programming is a multi-stage decision-making method for solving optimal control in automation field.The algorithm uses neural network to approximate the mathematical model of nonlinear complex system,which provides theoretical support for intelligent upgrading of many industries.Adaptive dynamic programming has the advantages of low energy consumption,good control effect and strong computing power for solving the complex nonlinear problem of multi-agent consensus control.In this paper,adaptive dynamic programming algorithm is used to solve the problem of multi-agent consensus control:(1)For a class of multi-agent systems with input constraints,a consistency control scheme based on adaptive dynamic programming algorithm is proposed.In this paper,the saturated multi-agent control problem is transformed into an optimization problem by introducing a non quadratic functional into the performance index function.The Hamilton-Jacobi-Bellman equation can be constructed by the performance index function.The strategy iterative algorithm is introduced to solve the equation,and the stability of the iterative algorithm is analyzed,which makes the algorithm feasible.In order to fit the local performance index function,an on-line neural network controller is designed,and the corresponding Hamilton-Jacobi-Bellman equation is expressed.The adaptive updating law of weights is obtained by gradient descent method,which makes the algorithm run smoothly.(2)An optimal consensus control algorithm is designed for multi-agent systems with external disturbances.Firstly,the disturbance term is added to the dynamic system equation of the follower,and then the quadratic function of the disturbance term is added to its utility function.In order to solve the uniform optimal control under the influence of disturbance,the influence of control input and disturbance term should be considered at the same time.Based on the theory of nonzero sum differential game,Nash equilibrium is realized.Finally,in the simulation phase,the comparison with the experimental results in ideal state is added to show the implementation effect of the algorithm in this chapter.(3)A distributed consensus optimal control scheme is designed for multi-agent systems with external disturbances and input constraints.The performance index functions of multi-agent systems with input constraints and external disturbances are defined by combining their respective performance index functions in optimal control.The Lyapunov function is designed and the stability condition of the system is obtained.In the final simulation experiment,an external disturbance is added under the premise of ensuring the stability of the system.At the same time,considering the constraints of the control input,the stable simulation results are obtained.The experimental results verify the effectiveness of the algorithm.