Decentralized Guaranteed Cost Optimal Control Of Reconfigurable Manipulators Based On Adaptive Dynamic Programming

With the development of modern science and technology intelligence,such as,the reconfigurable manipulators are automatic assembly,flexible and environment adaptable intelligent equipment which modules can be combined according to the different task requirements,it will be to apply for deep-space and deep-sea exploration,nuclear industry and other high-risk industries,intelligent entertainment factory.However,reconfigurable manipulators should consider energy consumption and control precision problems to ensure fulfill the task that strong coupling and nonlinear uncertainty.Therefore,developing the easier controller structure and more optimized consumption of energy is important and profound significant.Firstly,this paper discusses the background and significance of selected topic,and analyzes the domestic and abroad status and progress of reconfigurable manipulators,reconfigurable manipulators control method and adaptive dynamic programming(ADP)theory respectively,determines the research target of this article at last.Secondly,two kinds of dynamic models are built for the reconfigurable manipulators.One is the traditional dynamics model based on Newton-Euler iterative algorithm,and coupling cross-linking uncertainties is represented a set of subsystems associated.Another is the dynamic model of the harmonic drive based on joint torque measurement information,which simplifies the traditional dynamic model greatly.Thirdly,a decentralized Guaranteed Cost Optimal Control(GCOC)based on ADP as traditional reconfigurable manipulators system dynamic model is presented.According to the trajectory tracking problems,we simplify the controller structure and improve the control precision of the system,combined with the optimal control theory,built guaranteed cost upper bound performance index function of matching configuration and trajectory,establish performance index function with guaranteed cost upper bound function and build a single network to solve optimal feedback tracking control law of the Hamilton Jacobi Bellman(HJB)equation,its stability analysis and simulation experiment verify the effectiveness.Then,an energy guaranteed cost optimal control method based on ADP is presented for decentralized control problem of coupling crosslinking uncertainty reconfigurable manipulators with joint torque measurement information,according to the control precision and energy consumption of the performance index function to build HJB equation,using the policy iteration(PI)method to solve the HJB equation and optimal control strategy.The asymptotic stability of closed-loop system is proved by Lyapunov theory.Compared the algorithm of this section and traditional RBF neural network algorithm,the results demonstrate the effectiveness of this section algorithm in the simulation experiment.Finally,the paper summarizes the work and prospects of the future.