| In order to improve the surface machining precision of outline, the tracking error of feed axis in NC machining need to be reduced. So a more accurate model for response characteristics of axis motion is needed. The conventional modeling methods include block diagram model in Simulink and electrical and mechanical joint simulation, the joint simulation need complex finite element model, and are difficult to apply to real-time motion control.This topic raised the axis motion response model based on historical data, a linear iteration model is used to describe the relationship among actual velocity in current period, velocity tracking error in last period and actual velocity in last period, the velocity tracking error is the difference between command velocity and actual velocity in the same period, and extract the linear iteration coefficient of linear iteration model from historical data. The linear iteration can predict the axis response in the process of motion control at the same time, and improve the performance of motion control under control process in real-time.The repeatability of response characteristics of a specific axis is a premise for a modeling method based on historical data. Designed repetitive experiments to analyze the repeatability of actual velocity curve under the same control parameters and command speed. A cubical smoothing algorithm with five-point approximation was used to smooth the actual velocity curve, to remove the noise in the collected data and get the actual velocity curve. The results of repetitive experiments indicate that, the difference of axis velocity curve is 0.1% of the programming velocity, so modeling the axis response using historical data is feasible.Researched the factors which can affect the actual velocity of axis and the factors’ influence degree, and confirmed a linear iteration model use the velocity tracking error and actual velocity in last period as basic variable, Extract the linear iteration coefficient in historical data using the least square method. The validation experiments of actual velocity prediction had been completed, the maximal difference between predicted velocity and actual velocity were less than 5% of the programming speed, and the mean one less than 0.5%, the results indicate that the model have high prediction accuracy of velocity.Analysed the influence degree of different axis and different machining tool, and the validation experiments of predictive tracking error had been conducted, the difference between prediction and actual steady error can be 2% of the actual steady error in line rail machine tool. The result indicate that the model have high prediction accuracy of tracking error.Study the feedforward method based on predicted tracking error and conducted a circle machining experiment, the circle radius error of the circle machining experiment reduced 31.5um with feedforward compensation, whose programming speed is 1000mm/min. The results indicate that, the method can increase the circle machining accuracy effectively. |
