Research On The Key Technology Of Industrial Robot Motion Control Based On Feedforward Dynamics

Intelligence,integration and high-performance movement are the main development trend of industrial robots.In these,the high-performance movement is the important foundation of the machining quality and working efficiency of the robot,which requires the high dynamics response and high robustness of the controller.Although the conventional closed loop controller is simple in structure and easy to realize,it is usually aimed at the linear constant system,but the robot system features with nonlinearity,strong coupling and time-varying,especially in the high-speed motion,the joint torque couples strongly,the inertia changes a lot,the nonlinear effect is significant.Using conventional control method,it is easy to produce large control deviation and the conservative control parameters limit the system gain bandwidth,affect the system dynamic characteristics,and thus cause large hysteresis error.Moreover,large inertia change may cause system oscillation and damage its stability.For this,the key technology of motion control based on robot dynamics are studied to achieve high-performance motion.The main contents are as follows:In view of the dynamic analysis of 6R industrial robot,the overall architecture of dynamic feedforward control is proposed based on the robot dynamic characteristics with nonlinearity,strong coupling and time-varying.Compared with the conventional closed-loop feedback control,the simulation results show that the robot dynamic feedforward control can effectively improve the motion accuracy and dynamic response.It's difficult to obtain the accurate robot dynamic parameters and avoid the uncertainty disturbances in practice.Thus the ideal performance of feedforward control is hard to be obtained.As the active disturbance rejection Control(ADRC)has the advantage of high interference rejection ability,simple implementation and is not restricted to the models controlled,the ADRC dynamic feedforward control strategy based on ADRC has been proposed.To solve the problems of strong couplings of multiple joints,the dynamic characteristic compensation and the unknown disturbance compensation are compensated to decouple the control of multi-joints,and the serial structure and algorithm of dynamic feedforward control based on ADRC have been studied.The simulation results show that the ADRC controller based on feedforward dynamics has the good robustness to uncertain disturbances.Precise kinetic control requires precise kinetic model parameters.To improve the parameter accuracy of the dynamic model,and solve the problem of complex process,large accumulative error and load effect,the method of parameter identification is studied from the following aspects based on the minimum inertia parameter model.In the aspect of identification strategy,the identification strategy of “high unification and low in sequence” is proposed to improve the problem of multi-joints coupling interference during unified identification and the gradual transmission of errors during identification in sequence.In the aspect of load parameter identification,the "No-load and load Difference Identification strategy" is proposed.The measured moment information of the joint under no-load condition is used to replace the body dynamic model with load under the same trajectory,and the linear dynamic model of load is constructed by the difference value,avoiding the calls of body inertia parameters,error introduction and on-line calculation of robot dynamics,thus improve the load inertia identification accuracy and efficiency.In terms of estimation methods,weighted least squares method and genetic algorithm method are combined effectively to improve the accuracy of parameter estimation.To reduce the effect of model uncertainty to the control performance,and improve the motion control performance,and in view of the diversity,uncertainty and inadequate fitting faced by the traditional friction model,i.e.,it is difficult to propose a unified description form to characterize all the friction situations,a BP neural network friction model based on genetic algorithm optimization is proposed,which has higher friction fitting precision and generalization ability than conventional model.In addition,reducing the sampling noise and improving the feedback signal quality are the basis of improving the dynamic characteristic of the closed-loop system and realizing the high-performance motion control.A novel nonlinear tracking differentiator(NTD)with high stability and fast convergence,is proposed to solve the problems of phase lag,low stability and amplitude attenuation faced by traditional TDs.It integrates tracking rapidness,accuracy,and transitional stability,and has high approximation and filtering effects on generalized derivatives of the signal.Based on the research above,according to the overall structure of robot control system,the paper develops motion control experiment platform,designs motion control software structure,realizes the basic functions of robot motion control and dynamic parameter identification,and then studies the dynamic parameter identification and space trajectory tracking experiments.The results show that the ADRC feedforward controller can effectively improve the tracking accuracy,dynamic response and robustness of the robot.