ニューラルネットワークと数値シミュレーションを利用した欠陥判別の試み(<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
In Japanese pressurized water reactor plants, all the steam generator tubes are inspected using eddy current testing (ECT) technique in every scheduled inspection. In the defect distinction in ECT inspections of the steam generator tubes, a huge amount of labor is necessary, since human inspectors have to identify all the ECT signals recorded on magnetic tapes. This paper describes a computer-aided defect distinction system, which characterizes defects by combination of neural networks and electromagnetic field analyses. The capabilities of our neural network system and electromagnetic field analysis method are demonstrated. In this paper, we mainly mention about the defect distinction using neural networks and numerical simulation by electro-magnetic field anaylsis.
日本シミュレーション学会の論文
1997-09-15