Research On End Vibration Suppression And Control Strategy Of Flexible Joint Manipulator

With the rapid development of domestic robot technology,flexible cooperative robot has been widely used in 3C electronics,medical,automobile parts integration and other industries due to its high load ratio,compact structure and low power consumption.It is inevitable to use harmonic reducer,joint torque sensor and other flexible components in flexible robot,which makes flexible cooperative robot become a nonlinear system with strong rigid-flexible coupling characteristics.Due to the existence of external interference and model precision deviation,excessive trajectory acceleration will lead to excessive trajectory tracking error.When the robot stops working,the residual vibration will be generated due to the existence of flexibility,which makes the robot unable to reach the desired position quickly and stably,and greatly reduces the working efficiency of the robot.Therefore,this thesis takes the flexible robot as the research object,and adopts singular perturbation method to decompose the dynamics of the flexible robot into fast and slow subsystems.The slow subsystems are controlled by dynamic feedforward PD and sliding mode control algorithms respectively,and the control performance is compared and analyzed.Through the combination of RBF neural network and sliding mode control,the system modeling error can be compensated online.Aiming at the problem that sliding mode control technology will introduce high frequency switching of control rate,the sliding mode switching function is redesigned.The simulation model is built with Simulink/Simscape simulation software,and the control effect before and after improvement is analyzed.Firstly,the existing methods and research status of vibration suppression of manipulator are briefly described.For the control object in this thesis,the vibration sources are analyzed and reasonable assumptions are put forward.The newton-Euler method is used to establish the robot dynamics model considering joint flexibility,and the mathematical characteristics of the flexible robot dynamics model are analyzed.Secondly,for the vibration problem of the flexible robot,the rigid-flexible coupling system is divided into fast and slow subsystems by using singular perturbation method.Singular perturbation sliding mode control and singular perturbation feedforward PD control algorithms are designed respectively.Simulation analysis is carried out by taking the 2-d OF flexible robot as an example to compare the experimental results and existing problems.Thirdly,aiming at the problem that sliding mode control requires high model accuracy,a sliding mode control method based on RBF neural network error compensation is designed.The modeling error in control rate is compensated online by RBF neural network,and the adaptive rate of weight updating of neural network is deduced by Lyapunov method.In order to reduce chattering,the switching function is redesigned,and the improved saturation function is used to replace the traditional symbolic function,so as to further improve the control performance of the system.Finally,in order to verify the control effect of the above method,a simulation model of the6-DOF flexible cooperative robot was built by Simulink/Simscape.Based on the asymmetric Sshaped curve in Cartesian space,a path with both arcs and straight lines was designed to verify the tracking effect of the control algorithm on different paths.Simulation results show that compared with the circular path.Straight paths are easier to follow.At the same time,the chattering degree of the manipulator is greatly reduced by the improved neural sliding mode control.Compared with PD control algorithm,the improved neural sliding mode control has smaller overshoot and higher tracking accuracy.