Resaerch On Friction Modeling And Advanced Controll Algorithm Of AC Servo System

In our modern times,permanent magnetic synchronous motors(PMSM)have been increasingly used in manufacturing,aerospace and national defense.Although PMSM is compact and light with its simple structure,its performance is greatly compromised by friction torque in practical application.Friction is a complicated physical phenomenon,creating various negative impact on different control systems.Friction is unavoidable and may cause the "flat-top" in the two-way movement of the servo system and the "creeping" in the low-speed movement.This paper focuses on the influence of friction torque on the position tracking of PMSM and it provides two methods of fitting based on study on multiple friction models and on the real engineering scenarios.The first method is based on a simple Coulomb model,in which identification is achieved with two given sine waves of the same amplitude-frequency ratio.The second method is based on the Stribeck model.In this method,the Lagrange's interpolation is applied to the nonlinear low-speed phase,while the least square method is applied to the linear high-speed phase.Both of the methods take full account of the feasibility in real engineering.For a higher degree of precision in the friction model,this paper adopts the genetic algorithm and provides a method to narrow the fitting deviation and to achieve higher degree of fitting with the use of the Stribeck model.Compensating tests are carried on the screw platform.The test has proved a good compensating result.Changes may exist in the friction model in real work conditions,leading to great deviation of the off-line friction identification model.To pursue a better adaptation to environment,the paper provides a self-adapting controller which can achieve self-adjustment as the changing work conditions,and this enables quick restrain when the parameters of the friction model change.Robot servo systems are impacted not only by friction torque but also by the gravitational torque of the arms.This paper describes the separation of the gravitational torque and the friction torque by designing two groups of movement curves.The Lagrange's interpolation and the least square method are used to create a friction model to compensate the robot servo system.A sliding mode disturbance observer is used to identify the gravitational disturbance and friction model deviation of the robot arms and enables feed-forward compensation accordingly.The paper also provides simulation which shows that the compound feed-forward control algorithm can greatly improve the position tracking performance of robot servo systems.