Research On Vibration Suppression Strategy Of Servo-controlled Multi-axis Motion System

Servo-controlled multi-axis motion systems are widely used as the main moving parts in diversified automated production.In order to achieve a variety of different automation tasks in the same multi-axis motion system,different end effectors are designed according to the task requirements and the end effector is installed on the workbench or the end of the multi-axis motion system.Due to the requirement of light weight,the end effector introduces different degrees of flexibility to the system.In addition,the flexibility of the multi-axis motion system transmission system is also inevitable.Therefore,in the process of high-speed operation,large vibration inevitably occurs.In actual production,in order to ensure the smooth running of the trajectory and accurate positioning,generally choose to reduce the speed acceleration of the trajectory at a special point(the position of the sudden change in speed and acceleration)and wait for the natural attenuation of the vibration.but this kind of method will reduce the efficiency of batch production,which does not meet the high-efficiency and high-precision automated production requirements.Therefore,restraining the excessive vibration of the multi-axis motion system when passing through special points during the high-speed operation of the designated trajectory and the residual vibration of the auxiliary stroke high-speed positioning emergency stop section has important research significance for the efficient operation of the multi-axis motion system in practice.This paper mainly focuses on the problem of excessive vibration generated by special points during the high-speed trajectory process of the multi-axis motion system and the vibration of the multi-axis motion system during emergency stop.The main contents of the research are as follows:The control system of the multi-axis motion system in the industry is mostly PD feedback controller.When the specified trajectory is running at high speed,due to the inherent hysteresis of PD feedback control,vibration will inevitably occur at special points in the trajectory due to insufficient actual torque of the motor.Aiming at this problem,this paper establishes the dynamic model of each axis of the multi-axis motion system by Lagrange method,and designs the excitation trajectory optimization method based on Fourier series,and identifies the dynamic model based on the experimental method.Then we analyze the global asymptotic stability of the feedforward + PD control method based on the dynamic model applied to the multi-axis motion system.When the multi-axis motion system is positioned and stopped at high speed,large residual vibrations will be generated due to the flexibility of the system.To solve this problem,this paper studies a post-adaptive input shaping method.The post-adaptive input shaper starts with the residual vibration response of the system,and obtains the optimal shaper coefficient vector based on the recursive least squares method optimization.Due to the unstable noise in the residual vibration response,in order to improve the tracking performance of the post-input shaper,the effect of the forgetting factor on improving the tracking performance in the optimization process is studied,and the adaptive update algorithm of the forgetting factor is further studied,next,the algorithm is embedded in the iterative process of recursive least squares,a post-adaptive input shaper with an adaptive forgetting factor is obtained.In some application scenarios,the multi-axis motion system needs to change the running track frequently.In this case,applying a post-adaptive input shaper to recalculate the shaping trajectory every time the trajectory is changed takes a long time and is inefficient.In response to this problem,this paper studies a post-input shaper based on a multilayer neural network model,designs the structure and related parameters of the multilayer neural network model according to actual needs,and selects a suitable deep learning framework and learning algorithm to train the neural network model,finally,analyze the generalization performance of the trained neural network model.Finally,three typical multi-axis motion systems are selected as the experimental platform.The feedforward vibration suppression algorithm based on the dynamic model proposed in this paper,the post-adaptive input shaper based on recursive least squares,and the post-adaptive input shaper based on multilayer neural model are verified by experiments on three platforms.The experimental results prove the effectiveness and feasibility of the algorithm described in this paper.