Theory analysis and system identification methods on thermal dynamics characteristics of ballscrews

Abstract

Empirical model of machine tools on thermal error has been widely researched, which can compensate for thermal error to some extent but not suitable for thermal dynamic errors produced by dynamic heat sources. The thermoelastic phenomenon of unidimensional heat transfer of ballscrews influenced by changeable heat sources is analyzed based on the theory of heat transfer. Two methods for system identification (the least square system identification and BP artificial neural network (ANN) system identification) are put forward to establish a dynamic characteristic model of thermal deformation of ballscrews. The model of thermal error of the X axis in a feed system of DM4600 vertical miller is established with a fine identification effect. Comparing the results of the two identification methods, the BP ANN system identification is more precise than the least square system identification.

Keywords

ballscrew / thermal dynamic characteristic / system identification / least square / BP artificial neural network (ANN)