Research On Modeling And Compensation Technology Of NC Machine Tool Feed Axis Reverse Error

When the machine tool feed axis moves in the reverse direction,a non-linearly changing following error will occur,which seriously affects the machining accuracy.With the developments of high-speed and high-accuracy,the traditional compensation methods based on gap model and friction force model cannot meet the accuracy requirements.Therefore,this thesis proposes a speed feedforward compensation method based on the model of the nonlinear following error during the reverse period.For statement the sake of the traditional error models cannot describe the non-linear change characteristics of errors,the thesis proposes the concept of "reverse error" to distinguish from static "mechanical clearance",and establishes a machine tool reverse error prediction model,as follows: analyze the causes of non-linear changes in the reverse error based on clearance model and friction Model and operating data,then analysis repeatability of Error;establish a machine tool servo feed system model,and determine 9 input features of error model includes speed and acceleration after correlation analysis,which is based on the LightGBM algorithm,the model outputs a reverse error-time series.Speedfeed-forward is a good method of reverse error compensation,and the feedforward signal is the key to compensation.The feedforward parameters cannot be calculated accurately on basis of manual experience,so the thesis proposes a method for calculating the feedforward parameters based on the SVM algorithm.The eigenvalues are obtained after preprocessing the instruction information and the reverse error of the model prediction,then they are input into the SVM model of the feedforward parameters and the reverse error.The model generates the feedforward parameters t of the reverse error,and then they are input into the parameters to the speed feedforward interface,the compensation of the corresponding reverse error can be completed.Conduct experiments on the BM8-H machine tool,the absolute value of the inverse error model for the x-axis and y-axis prediction deviations averaged 2.77?m and 2.80?m,which was more accurate than the LSTM model;The compensation parameters generated by the predicted reverse error reduce the circle contour error to 2.90?m on average,and the error compensation rate is 74.90% on average.Compared with the existing compensation method,the accuracy of the compensation method described in this article is significantly improved.