Adaptive Dynamic Programming Based Fault Tolerant Control For Reconfigurable Manipulators

With the development of modern science and technology,and the arrangements for "13th Five-Year" national science and technology innovation plan,many intelligent mechanical equipments,such as,the modular and flexible reconfigurable manipulateors,which can adapt the external environments and operational tasks,will be vigorously applied to deep space exploration,intelligent factories,nuclear industries and many other applications.However,actuators and other components of reconfigurable manipulators will break down inevitably when they work in extremely complex environments for a long time.If we don't deal with it on time,it may lead to serious damage.Therefore,fault diagnosis and tolerant control is desperately required to solve such problems to maintain the reliable performance.Meanwhile,how to improve the control precision of mechanical reconfigurable manipulator system,simplify the structure of the controller and optimize the energy consumption are also necessary.It will not only to provide strong technical support for manipulateors with large-scale complex structure in the future,but also great significance for automation in the direction of the optimization and healthy developments.In this thesis,based on the geometric formula of the rigid body,the dynamic model of the reconfigurable manipulator system is established by using the Newton-Euler iterative algorithm.On this basis,the dynamic model of the fault system is given.For trajectory tracking problem,in order to improve the control precision of the system,optimize the energy consumption,and simplify the controller structure,combine with the optimal control theory,an optimal tracking control method based on critic-only policy iteration(CoPI)algorithm for reconfigurable manipulators is proposed.The optimal tracking control law is calculated by CoPI algorithm,and the global tracking control law which is combined with nominal control law,ensures the tracking error of the system is asymptotically convergent to zero.Then,considering that the reconfigurable manipulator may need to increase or decrease modules for different task requirements,decentralized control is more suitable for reconfigurable manipulator.Meanwhile,in order to fulfill the idea of "optimization",a decentralized control method based on ADP algorithm observer critic structure for reconfigurable manipulators is proposed.The difficulty of decentralized control lies in how to handle the coupled interconnection term.In this chapter,we combine alternative ideas with observer techniques,by using the neural network to identify the dynamic model of subsystem which is replaced.By using CoPI algorithm to design the local feedback controller,utilize the adaptive robust term offsets the global approximation errors of the system,and combine with the designed local nominal controller to ensure the system trajectory tracking error asymptotically converge to zero.Finally,for the reconfigurable manipulator system with actuator failures,based on ADP theory and fault observer,we present a novel fault tolerant control scheme of ADP algorithm with an improved performance index function.By using the actuator fault observation value from an adaptive fault observer which is utilized to design an improved performance index function.Thus,the fault tolerant control problem can be transformed into an optimal control problem.The fault tolerant controller consists of two parts: the nominal controller and the optimal feedback controller.The nominal controller is designed to ensure the tracking performance of the system stabilize to zero,and the optimal feedback controller is designed to make the tracking error converges to a steady-state in an optimal manner.The simulation experiments verify the effectiveness of the designed observer and controller.