Thermal Errors Simulation and Modeling of Motorized Spindle

Abstract

Thermal errors of motorized spindle are of great importance to affect final machining precision of CNC machine tool. Thermal characteristics simulation analysis of motorized spindle is realized by ANSYS; thermal errors test measurement is completed based on 5-point method; and prediction models of thermal errors are constructed by multiple linear regression (MLR) method, Back Propagation (BP) neural network method and Radial Basis Function (RBF) neural network method respectively. The results of simulation and experiments illustrate that simulation results can represent thermal characteristics of motorized spindle, whose degree of confidence mainly depending on setting of thermal load and boundary conditions properly or not; RBF neural network model has highest prediction precision for thermal errors of motorized spindle based on test data.