ニューラルネットワークによるき裂を有するはりの逆問題解析 : 解析的手法による学習データの効率的作成

概要

論文の詳細を見る
In this paper, a method for determination of the depth and location of a transverse surface crack in a beam is proposed. The study, which is treated as an inverse problem, was performed using a error-back propagation neural network approach. In the analysis, the crack is simulated by an equivalent rotational spring, connecting the two segments of the beam. Analysis of this approximate model leads to a frequency equation which relates the natural frequencies to the crack location and depth. The equation is then applied to the inverse problem-identification of the crack location and depth from frequency measurement results. First three or four natural frequencies, obtained by solving the non-linear frequency equation, are adopted as training data for the neural network. After the training stage has been completed, the crack parameters are estimated using the output of the trained network. The simulation results indicate that the proposed method is valid and practical.
一般社団法人日本機械学会の論文
1998-01-25