ニューラルネットワークによる非弾性構成方程式のパラメータ決定法

概要

論文の詳細を見る
This paper describes a new method for determining appropriate parameters of inelastic constitutive equations using the hierarchical neural network. The present method consists of three subprocesses. First, various behaviors of material under cyclic loading or creep conditions are obtained through numerical integration of constitutive equations for many combinations of their constant parameters. Each relationship between the material behaviors and the parameters is here called "learning data". Next, the hierarchical neural network is trained using the learning data. Finally, the trained network is utilized to determine appropriate parameters corresponding to actual material behaviors measured in laboratory tests. To demonstrate its performance, the present method was applied to the parameter determination of Chaboche's viscoplastic model under uniaxial loading and stationary temperature conditions. The experimental data referred to are cyclic hysteresis curves of about 60 cycles and a stress relaxation curve. The results clearly show that the present method is a powerful tool for determining appropriate parameters of inelastic constitutive equations, referring to various material behaviors simultaneously.
一般社団法人日本機械学会の論文
1993-03-25