Application of Multiple Regressions to Thermal Error Compensation Technology – Experiment on Workpiece Spindle of Lathe

Abstract

This study focuses on a type of intelligent machine tool – Thermal Error Compensation Technology. Lathes are used to rotate a cylindrical workpiece against fixed tools. The high speed of rotation and frequent changes to the cutting load can result in heat deformation which will reduce the accuracy of the relative position between the tool and the workpiece, thus reducing machining precision. This study measures lathe temperature and thermal errors of a workpiece spindle. Experimental results are subjected to multiple regression analysis to assess the relationship between the heat source and deformation to produce a three axis thermal error model. Additional items from polynomial, interaction and cyclical series are included in the model. Finally, we choose the best thermal error model based on predictive ability. The resulting model has a standard deviation of the forecast error within 5 μm.