A Research On Adaptive Iterative Learning Control Algorithm Based On A Two-Degree Step-Scan Platform

Feedforward control plays an important role in control system of motion platform with linear motor.It achieves feedforward control signal from priori knowledge and uses this feedforward control signal to compensate error in advance and improve control performance.It has important practical significance and application value to Research and realize feedforward control methods which can achieve high speed and high precision on trajectory tracking control tasks which requires high tracking precision.Traditional feedforward control methods include iterative learning control,model-based feedforward control and iterative feedforward tuning,these methods has different advantages and disadvantages.Iterative learning control is suitable for repetitive tasks to improve control precision effectively,however non repetitive disturbances will cause accumulation of error.Besides,it has poor extrapolation ability.Model-based feedforward control method has high extrapolation ability,but the control performance it achieves is highly dependent on the quality of system model.Iterative feedforward tuning is independent on system and has high extrapolation ability,but the control precision it can achieve is lower than iterative learning control.In this paper,theoretical analysis and experimental verification is carried on feedforward control method based on two-degree linear motor platform which executes step-scan trajectory tracking tasks.Based on control requirements of high precision,high extrapolation ability and independent on system model,a control algorithm which satisfies these requirements is proposed.Experiments is carried to validate practicability and validity of the algorithm.First of all,mathematical model of control is established based on physical meaning of platform.Parameters of the model is then identified by Swept frequency experiment to build system model and provides theoretical basis for simulationmodeling and convergence analysis.Control requirements are analyzed and reference trajectory is designed based on control task in engineering practice.Second,the problem of non repetitive disturbance suppression is studied to improve control precision and prevent accumulation of disturbance.Based on analysis and segmentation of reference trajectory,an adaptive learning gain matrix and a low-pass filter with adaptive cut-off frequency are therefore designed.After all,an adaptive iterative learning control algorithm is proposed to suppress accumulation of non repetitive disturbance and improve tracking precision of the whole trajectory tracking task.Third,iterative feedforward tuning is studied to alleviate performance deterioration of iterative learning control when reference trajectory changes.By introduction of basis function,parameterized feedforward controller is built.Different kinds of basis function and different optimization methods tuning parameters of controller is researched.Advantages and disadvantages of different methods are compared by theoretical and simulational analysis,and then iterative feedforward tuning method suitable for this project is chosen to compensate reference-induced disturbance and improve the problem of performance deterioration.Then,based on control requirements and researches above,an algorithm combined iterative feedforward tuning and iterative learning control is proposed.This algorithm is data-driven,which only needs collection of input and output data and is independent on system model.Algorithm can reach high control precision and has high extrapolation ability,meanwhile alleviates performance deterioration problem of standard iterative learning control when reference trajectory changes.Programme and flow chart of this control algorithm is proposed,as well as theoretical analysis of the algorithm.At last,to validate control algorithms above,an experimental servo system consists of upper computer(PC computer),POWER PMAC control card,AMAC motion driver and NEWPROT IDL225 linear motor motion platform is established.Experiments on iterative feedforward tuning,adaptive iterative learning control and the proposed IFT-ILC method is carried on this experimental platform.By analysis ofcollected experimental data,effectiveness and superiority of the algorithm is validated from the perspective of control accuracy and extrapolation ability,etc.