ニューラルネットワークによる工作機械壁面熱伝達率の推定

概要

論文の詳細を見る
It is necessary to know the exact values of heat transfer coefficients on the surfaces of a machine tool for an analysis of thermal deformation of the machine tool. However, it is difficult to measure heat transfer coefficients preciseIy by the conventional methods. Therefore, we have been developing a method to estimate heat transfer coefficients from surface temperature of the surfaces of the machine tool. This article presents a new approach to estimate heat transfer coefficients of the surfaces of the machine tool using a neural network. The presented method was applied to actual experiments for the machine tool. The results show the feasibility for precise estimating of heat transfer coefficients from surface temperature of the machine tool in case of operation conditions of the machine tool are in the range of pre-learning for the neural network. However, if the operation condition of the machine tool are out of the range pre-learning for the neural network, accuracy of the estimation is relatively low. Singular value decomposition with FEM model is thought to be superior to the neural network method, because singular value decomposition is valid for wide range of operation conditions of the machine tool.
一般社団法人日本機械学会の論文
1998-10-25