Research On Fault Observer Based Active Fault Tolerant Control Method For Reconfigurable Manipulator

The progress of manipulator technology is continuing to extend the application of manipulator in more aspects and wider range as human beings hope that more complex tasks can be completed by the manipulator. Although the traditional manipulator can easily complete many different tasks by reprogramming, it is constrained to the limited ranges due to its mechanical structure. The reconfigurable manipulator has been researched and developed in such condition and its joints and connecting rod modules can be changed according to the specific tasks to form the shape meeting the requirements of the tasks. There are a lot of things can be studied on the new structure of the reconfigurable manipulator, such as the building of Kinematics and dynamics models, the designing of model controller as well as the fault detection and fault tolerant control in the failure of sub-systems of the reconfigurable manipulator, etc.The rapid development of modern industry pushes the requirements of tasks higher and higher and some tasks cannot be accomplished directly by the human beings, such as the deep space exploration and submarine detection, etc. human beings have been replaced by the reconfigurable manipulators to accomplish those highly dangerous tasks. However, the high risks of tasks, the increasingly complex control system and the long-time operation of reconfigurable manipulator in the extreme environment will unavoidably cause failures in the actuation and sensing mechanisms of the manipulator. The failure of elements in such system will lead to the malfunction of the system and even injuries or property damage. Therefore, it is of significance to design a safe and reliable reconfigurable manipulator system that can guarantee the stable operation of the system in the condition there is a failure in the actuation or sensing mechanism of the system.The study in this Article focuses on a kind of reconfigurable manipulator system with failures in the sensing and actuation mechanisms, including the following contents and work:1 Indicate the background and significance of the study by briefing the status quo and contents of the studies on the reconfigurable manipulator at home and abroad.2 Achieve the dynamic module of the subsystem via the geometric formula of dynamic equation of the rigid objects and, based on which, acquire the Newton-Euler equation for the module subsystem through the forward iterations(Generalized velocity and acceleration) and inverse iterations(generalized force) in accordance with the description of manipulator structure in the kinematics. The dynamic module of the multi-failure subsystem is given in the condition that both the sensor and actuator have failures at same time according to the status-space expression of the subsystem’s dynamic module.3 Trace the problem via the trajectory. The decentralized control subsystem is designed for the reconfigurable manipulator based on the adaptive neural network to make compensations for the approximation error of neural network and the subsystem coupling by utilizing the adaptive control in accordance with the nonlinear items of the approximation subsystem of the neural network function. A decentralized control method is put forward for the adaptive neural network based on the extended state observer when the reconfigurable manipulator system is in the measurable condition. The decentralized controller is designed for the adaptive neural network based on the estimated value of extended state observer and the Lyapunov stability theorem, and the method put forward in this article is verified at the end via the simulation research.4 The Luenberger observer and the dynamic output feedback controller are designed on the basis of the Linear Matrix Inequality(LMI) technology and H? performance indexes in response to the failures in the actuator and sensor of the reconfigurable manipulator subsystems. Estimation is conducted for the actuator and sensor with the Luenberger observer due to the advantages that no failure detection and isolation units are needed. Then the active fault tolerant control is conducted for the dynamic output feedback controller based on the observed values of observer. In the end, the stability of the Luenberger and dynamic output feedback controller has been proven and, meanwhile, the simulation text has been conducted for the reconfigurable manipulator subsystem in different structures to verify the effectiveness of the observer and controller.5 The conclusions and the perspectives of future research are presented at the end of this paper.