GA・ニューロによる非弾性構成モデリング(<小特集>ソフトコンピューティング)

概要

論文の詳細を見る
Up to now, a number of models have been proposed and discussed to describe a wide range of inelastic behaviours of materials. The fatal problems of using such explicit constitutive models are however the difficulty of determining parameters in the model and the existence of model errors. In this paper, the author first defines the roles and meanings of genetic algorithms and neural networks in the field of inverse analysis. Genetic algorithms are then applied for the parameter identification of constitutive models, while neural networks are used as implicit constitutive models to eliminate model errors. Having been tested with the actual experimental data, the approaches were found to overcome the problems. Finally, the hybrid explicit/implicit constitutive model is introduced as the ultimate model.
日本シミュレーション学会の論文
1997-09-15