Intelligent evaluation of ultimate strength prediction of composites and fabrics’ extensibility
Authors: Doriana D'Addona, R. Teti, P.N. Koustoumpardis, N.A. Aspragathos
Abstract
An approach to intelligent evaluation of residual flexural strength of glass fiber reinforced plastic composites and to tensile test for the estimation of the fabrics extensibility is proposed.
The glass fiber reinforced plastic composites were subjected to hydrothermal aging cycles in water at 60 °C. Sensor fusion of acoustic emission and load data is carried out through neural networks to obtain a prediction of ultimate flexural strength as early as possible in the loading history of glass fiber reinforced plastic composites.
A robotized system is used to perform the fabrics tensile test and the extensibility of the samples is estimated using a feed-forward neural network while trying to imitate the human expert valuation.
The results show that neural network processing provides an effective method for the monitoring of composite material strength behavior based on acoustic emission detection and analysis and for the estimation of the extensibility of new fabrics.
| Attachment | Size |
|---|---|
| Doriana_Teti_Aspar Iproms08.wmv | 5.91 MB |
Doriana M. D'Addona, Ph.D.
Dept. of Materials and Production Engineering
University of Naples Federico II
Tel.: +39 0817682336
fax: +39 0817682362
Thank you for your comments.
The final NN architecture for the Tensile test data had 6 hidden unit. This number of hidden nodes was dertermined by a procedure for the optimisation of the NN architecture.
The methods used for the AE data have small differences due to the diverse nature of the original data.
Dear Authors,
Thank you for this interesting paper. I have a couple of questions:
1) I could not find which was the final NN architecture for the Tensile test data, the input is 4 but how many hidden units were used?
2)How were these determined? (were these found using the same method as for the AE data?)
Also, there is a small error which you might want to fix so the paper is error-free for the proceedings. In the final paragraphs of section 3, the paper say "Fig. 7" instead of "Fig. 6". (please upload a new version of the paper with this error fixed).
Thank You.
Gonzalo.