Machine tool thermal error modeling method based on deep learning

Abstract

The invention discloses a machine tool thermal error modeling method based on deep learning. The method comprises the following steps of obtaining temperature data of a machine tool main shaft and carrying out normalization processing; calculating correlation between each measuring point and thermal errors of a main axis at three directions through a partial correlation coefficient method, and selecting m measuring points with high correlation as temperature critical measuring points; establishing a SAE network and initializing a network parameter, and taking temperature data of the temperature critical measuring points as independent variables and inputting into the SAE network to extract a temperature characteristic of the temperature data; and taking the temperature characteristic as the independent variable, taking corresponding thermal error data as a dependent variable and inputting into a GA-BP neural network to carry out training, and carrying out thermal error prediction. Themethod has advantages that prediction precision is high, robustness is good, a thermal error change trend of the machine tool can be effectively estimated and so on.

Description

technical field [0001] The invention relates to the technical field of numerical control machine tools, in particular to a method for modeling thermal errors of machine tools based on deep learning. Background technique [0002] Among the various error sources of machine tools, thermal errors and geometric errors are the most important errors in machine tool error sources. 40% to 70% of the errors of high-end CNC machine tools are determined by thermal deformation. Thermal error has become the main source of error affecting the machining accuracy of parts. Establishing a model that can accurately describe the thermal deformation of the machine tool is the basis and decisive factor for thermal error compensation. Existing studies usually use multiple linear regression models, finite element models, least squares support vector machines, support vector machines, gray theory and neural network models and other modeling methods. Wang Xiushan and others established a multiple l...

AI Technical Summary

Problems solved by technology

High-dimensional data will produce dimensionality disaster, which will result in good modeling fitting effect but poor prediction effect; at the same time, the model established by traditional methods will show a strong phase effect due to environmental and seasonal changes, especially In the case of small sample size, it is difficult to meet the robustness requirements of the thermal error model, which restricts the implementation of thermal error compensation technology to a certain extent

Images