An efficient thermal error prediction model using neural networks and key temperature points for gantry machining centers

Abstract

The gantry machining center is popular for various fabrication, such as milling and tapping. However, thermal errors introduced by the rotation of spindle, workpiece processing, and cooling significantly degrade fabrication precision. The objective of this study is to establish an appropriate and efficient thermal error prediction model for the spindle of gantry machining center. The model will then aid real-time compensation for the error. Firstly, this study presents a systematic strategy for selecting key temperature points on the gantry machining center, reducing the number of required sensors. Subsequently, a thermal error model is developed based on the selected key temperature points. The model will be capable of predicting thermal errors in the x- and z-direction. Finally, this work both validates the thermal error model and exhibits real-time compensation capabilities using a real machine.