| Recently,the motion characteristics and controllability of rigid-flexible coupled robots have received much attention as a key research direction in the basic robotics research field.The rigid-flexible coupled robot,which is composed of partly rigid and partly flexible components and has flexible characteristics,has become a piece of essential automation equipment to promote the gradual transformation of the manufacturing industry to the direction of high precision and human-machine integration.However,its complex coupling structure characteristics also bring non-negligible flexible interference,which seriously affects the motion control accuracy and operation precision of the robot,and easily leads to control system divergence.It has been a very challenging research direction to cope with the rigid-flexible coupling problem in order to achieve stable and accurate control.Rigid-flexible coupled robots are mainly oriented to three basic types of objects: flexible-joint robots,flexible-link robots,and cable-driven robots.The researches show that the rigid-flexible coupling problems in the rigid-flexible coupled robots have mathematical models with very similar characteristics after equivalent processing.Focusing on the rigid-flexible coupling problem in flexible-joint robots is representative and positive for solving rigid-flexible coupling problems in other types of robots.This dissertation on the dynamic characteristics and control problem of flexible-joint robots,involves rigid-flexible decoupling modeling of flexible-joint robots,adaptive control problem under uncertain nonlinearities,bounded control problem with torque saturation constraints,and non-full-state feedback control problem.To solve above problems,some control schemes are proposed in the framework of singular perturbation theory.The main research contents and contributions are as follows.(1)The advantages and limitations of the singular perturbation theory for decoupling modeling of flexible-joint robots are expounded.From different aspects,the technical methods of rigid-flexible decomposition of the overall system of the flexible-joint robots are systematically studied.First,for the flexible-joint robots with joints of weak flexibility,selecting the link angle and the joint elastic torque as slow and fast variables,respectively,the full-order dynamics system model of the flexible-joint robot is decoupled into a slowly reduced subsystem and a fast subsystem to stabilize the boundary layer.Then,by applying the concept of integral manifold within the frame of singular perturbation theory,the fast subsystem of the robot dynamics is reshaped into a typical tracking system to ensure the fast variable converges to zero over time but without considering it in the system model,providing more flexibility on controller design.Finally,by designing a saturated flexible compensator,a rigid-flexible decoupling modeling is carried out for a flexible-joint robot system with joints of arbitrary flexibility,which breaks the limitation of singular perturbation theory.(2)Considering the dynamic uncertainty(and/or external disturbances),a sliding mode tracking controller with chattering suppression of flexible-joint robot is studied.First,a smooth saturation function is applied internally to replace the commonly used non-smooth switching function in the design of smoothsaturation-function-contained reaching law to alleviate the inherent chattering phenomenon of the sliding mode control.Then,neural network based soft computing is designed externally to run online to avoid chattering amplification caused by high value of switching gain.In addition,a comprehensive analysis of the asymptotic stability of the fast and slow subsystems and the overall system is done.Finally,numerical examples verify the superiority of the dynamic performance of the proposed control approach.(3)A generalized saturation adaptive controller based on backstepping control and neural network is proposed for the tracking control problem of flexiblejoint robots with bounded torque inputs.First,a saturated sub-controller by backstepping method is proposed for the slow subsystem,where the projectiontype parameter adaptation and a class of saturation functions are applied to make the torque inputs bounded,and a saturated neural network approximator is involved to simplify the control law and to compensate for the uncertain nonlinearity.Third,a new filtered tracking error of the elastic torque is designed for the fast subsystem to make the boundary layer subside quickly.In addition,explicit but strict stability analysis is given for the system.Finally,the input boundness,anti-interference capacity and dynamic characteristics of the proposed controller are verified by simulation comparisons.(4)Aiming at the problem of state estimation in the feedback control of flexible-joint robot system,the non-full-state feedback control strategies based on the observers are mainly studied.First,to improve the computing efficiency of extended Kalman filter(EKF),an efficient EKF with new states variable,which simplifies computing the Jacobian matrix,is designed to make the closedloop tracking control without angular position or velocity measurements of links.Then,to improve the robustness of EKF,an adaptive EKF with self-tuning parameters,which combines the noise filter of EKF with global convergence of high-gain observer,is applied to make the output feedback control without angular velocity measurements of the link side and the motor side.In addition,from two different aspects,rigorous stability analyses on the closed-loop control system in terms of two singularly perturbed subsystems and also the whole system under both full-state feedback and non-full-state feedback via improving EKFs are given,respectively.Finally,simulation comparisons are provided to verify the effectiveness and superiority of the proposed observers and the proposed non-full-state feedback controller based on the observers. |
PDF Full Text Request