Developing Geometric Error Compensation Software for Five-Axis CNC Machine Tool on NC Program Based on Artificial Neural Network

Abstract

This paper aims to develop a geometric error compensation software for five-axis CNC machine tool basing on NC program and machine learning artificial neural network. These modules are used in this software such as Human Machine Interface (HMI), data preprocessing, NC rebuild, and checking NC program. To determine error compensation, the inverse error algorithm from ideal points (from inputted NC program) and machine learning based artificial neural network model are calculated. While, artificial neural network model is optimized based on mean squared error cost function. The performance of the prediction model is confronted with experimental data points and various regression analyses. The geometric compensation software is then constructed and appended to this manuscript for a practical application purpose.