An artificial neural network is applied by nondestructive inspections in aerospace materials. The use of an artificial neural network is presented for classifying testing data as corresponding to sample materials with defect and without defect. The back-propagation learning for a multi-layer feed-forward neural network is applied to this classification. The trust region method is adopted to the back-propagation learning problem. Results of numerical tests are summarized.
Get full access to this article
View all access options for this article.
References
1.
KojimaF., ICASE report No. 89-26, NASA Langley Research Center, Proc. 5th Symp. on Control of Distributed Parameter Systems, Perpignan, France, July 1989.
2.
BanksH.T. and KojimaF., Quart. Appl. Math.47 (1989) 273.
3.
BanksH.T., KojimaF. and WinfreeW.P.Inverse Problems6 (1990) 897.
4.
KojimaF., International Series of Numerical Mathematics100 (Birkhauser-Verlag, 1991) pp. 207–217.
5.
KojimaF., in: Computation and Control II, Progress in Systems and Control Theory11 (Birkhauser-Verlag, 1991) pp. 237–245.
6.
JohanssonE.M., DowlaF.U. and GoodmanD.M., Technical Report UCRL-JC-104850, Lawrence Liver-more Laboratory (1990).
7.
MoréJ.J.Mathematical Programming: The State of the Art (Springer, New York, 1983).
8.
HeathD.M., WelchC.S. and WinfreeW.P., in: Review of Progress in Quantitative Nondestructive Evaluation5B (Plenum, 1986) pp. 1125–1132.
9.
BanksH.T., KojimaF. and WinfreeW.P.Corrosion profiles identification by PDE model based approach in thermal testing (in preparation).
10.
CarterR.G., Technical Report TR 87-12, Rice University (1987).
