Abstract: To distribute the thermal stiffness of a machine tool reasonably and provide the basis for the optimization of thermal stiffness of machine tool parts, an identification method for the key thermal stiffness of a machine tool based on the thermal error neural network prediction model is proposed. Considering the feature that the influences of the thermal stiffness of different parts on the whole machine tool’s thermal stiffness are non-identical, this method proposes the concept of key thermal stiffness. According to experimental data of temperature and thermal errors, a thermal errors prediction model is built by using radial basis function neural network for its high modeling accuracy and strong generalization ability. Unit temperature risings produced by different machine tool parts after they attain thermal equilibrium are taken as the input vectors of the established prediction model, variation values of thermal errors are calculated to identify the key thermal stiffness. On this basis, the principle and implementation steps of this identification method is presented. Application of the method on a viaduct gantry machining center for its key thermal stiffness identification shows that the identification results are consistent with the verification test results.

Key words: Gantry machining center, Identification, Key thermal stiffness, Neural network, Thermal error prediction model