Research On Dynamics Control Technology Of Robot Based On Identification Model

With the development of intelligent manufacturing industry,automated production lines need higher efficiency and security.The traditional position control based on kinematics is located to the servo driver side by the motion controller,and the accuracy of the control completely depends on the performance of the servo driver.At the same time,when human and robot work together,the motion controller based on position control can't respond even when the robot collides accidentally,resulting in the damage of the robot and even endangering the safety of the operator.Robot collision detection technology is an important measure to ensure the safe operation of robot under the condition of unsupervised.In this paper,the dynamics of SCARA robot and six axis robot are studied.Taking SCARA robot as the research object,the dynamic model is established by Lagrange method,and the relationship between the physical properties and the dynamic model of the robot is clarified.In view of the complex friction nonlinear phenomenon of harmonic drive joint of robot,the improved nonlinear friction model and periodic friction model are proposed;in view of the importance of excitation trajectory to parameter identification,the improved genetic algorithm is proposed,and the excitation trajectory in the form of five order Fourier is designed.Experiments show that the friction model can better represent the friction torque of the robot,and the designed excitation trajectory can fully mine the "characteristics" of the model and improve the identification accuracy.Using Newton Euler method to establish the six_axis robot's dynamic model,and the transfer mode of speed and torque of the serial robot is determined.The accuracy of the dynamic model is verified by joint simulation.To solve the problem that six axis robot can't measure friction torque directly,an empirical friction model is proposed,which is close to the nonlinear friction model and can be identified together with the dynamic parameters.In order to reduce the identification accuracy caused by the imbalance of the front and rear axle moment,the moment measurement variance is used as the weight,and the weighted minimum two is used to estimate the dynamic parameters,which improves the overall identification accuracy of the dynamics.In order to reduce the effect of model-based collision detection algorithm,a new observer equivalent to band-pass filter is proposed.By analyzing the frequency distribution of the dynamic model,the cut-off frequency of the high pass filter is determined.The new observer can effectively reduce the disturbance caused by model error and improve the robustness of collision detection algorithm.Aiming at the limitation of three closed-loop control of servo system,the model-based feedforward control is studied.By establishing the single joint DC motor model and the robot dynamics model,the velocity feedforward compensation and the torque feedforward compensation are simulated and tested.The experimental results show that the single joint velocity and moment feedforward reduce the track tracking errors by 73.5% and 95.7%,and the track tracking errors of the first three joints in the whole experiment reduce by 93.3%,92.3% and 81.9% respectively.In this paper,through joint simulation and experiment,the dynamic modeling is verified to improve the reliability of the model;through improving the friction model,the accuracy of model identification is improved;through a new observer based on the generalized momentum method,the acceleration is avoided,the interference of model error is reduced,and the robustness of collision detection is improved;through the feedforward compensation control strategy based on the model It reduces the joint tracking error and improves the control performance of the servo system.