再学習過程を組み込み構築されたニューラルネットワークによるトラス構造の損傷同定

概要

論文の詳細を見る
Damage detection has received much attention for structural health monitoring. In this paper, we develop a damage detection method for truss structures using a neural network with relearning process. The damage of truss structures is characterized as reduction of mass and stiffness of truss member. The input data are composed of the frequencies and mode shapes of damaged truss structure. To verify the validity and effectiveness of the present method, the damage detection is conducted using the input data without measurement errors, and the results of the damage detaction are compared with the exact ones and those by previous damage detection methods. Consequently, it is found that the present method can identify satisfactorily the location and magnitude of damage. Finally, the damage detection is conducted using the input data with measurement errors. It is recognized from the results that the present damage detection method is effective in practical use.
一般社団法人日本機械学会の論文
2008-04-25