Neuro-fuzzy Control For Flexible-link Manipulators

As the demand for manufacturing and the space application extended more and more in these years, flexible manipulator has received special attention from many aspects. In structure, the character of flexible manipulator is that its links are light and elastic. Therefore, it exceeds the traditional industrial robot greatly in operating speed, working efficiency and energy consumption (especially the energy consumption used for space launch). However, due to the elastic deflection and vibration of the links, the flexible manipulator becomes an infinite dimension MIMO system with distributed parameters, and is highly nonlinear and strong coupling. Therefore, the model and the control of flexible manipulator are of great significance both in theory and in application.In this paper the model of flexible manipulator is firstly studied.To design and analyze the system easily, reasonable assumption and approximation on the model are used. The closed dynamic equation is derived using Lagrangian approach. In addition, the model is decomposed into the slow and fast subsystems using singular perturbation theory and output re-definition method.With the dynamic model of flexible manipulator, PD control, fuzzy PD control and neural-network control are studied. PD control focuses on the control law design and the stability of the whole system. Fuzzy PD control emphasizes how to adjust the gains of PD control to improve the performance of the system. Besides, the structure of fuzzy controller and the membership functions are dealt with. In the control scheme using neural network,the control of flexible manipulator is separated into angle-following control of slow subsystem andvibration-suppressing control of fast subsystem based on singular perturbation and time decomposition. Moreover, the paper discusses the structure and the algorithm of neural network plus fuzzy PD control of tip vibration. Finally the performance of designed methods is validated by simulation.To compare the advantages and the disadvantages of the above schemes, the flexible-manipulator simulations on PD control, fuzzy PD control and neural-network control are carried out. After comparing the simulation results with each other the author analyzes the causes of different effects. An experiment on the testbed of a two-link flexible manipulator show that the neural-network plus fuzzy PD control with tip vibration has better performance than others (simple PD control and fuzzy PD control). It not only achieves accurate tip positioning but also guarantees the fast response.