Influence of different data processing methods on error prediction of three-axis NC machine tool based on LSTM Neural Network

Aiming at the error prediction algorithm of three-axis NC machine tool based on LSTM neural network, this paper processes the original data through filtering, difference, and expanding data set to obtain different training characteristics, and then establishes the tracking error prediction model of single axis servo system based on neural network. Finally, the effectiveness of different data processing methods is verified on the verification set. Furthermore, experiments show that Several methods mentioned in this paper can reduce the contour error of trajectory, among which the difference and expanding data set methods can effectively reduce the contour error of trajectory.