Applicability of Adaptive Neural Networks (ANN) in the Extrusion of Aluminum Alloys and in the Prediction of Hardness and Internal Defects

Abstract:

Article Preview

Adaptive Neural Networks (ANN) can be used in the analysis of a complex panorama of interconnected input/output industrial data, even when they present substantial noise. The ANN, despite presenting substantial mathematical complexity associated with non-linear parameterization (which includes transfer equations and corresponding “training”), are largely used under industrial conditions in several engineering areas (such as in steelmaking), with substantial success. This work shows the applicability of the ANN in a specific case related to the analysis of internal defects of extruded aluminum sections (occurring both at the head and at the extrusion tail), and the associated bar hardness as a function of process parameters such as: billet temperature, extrusion ratio, ram speed and billet length. Results were analyzed in terms of the adhesion to an ANN built upon the collected industrial data, as well as the relevance of each variable within the ANN.

Info:

Periodical:

Materials Science Forum (Volumes 638-642)

Main Theme:

Edited by:

T. Chandra, N. Wanderka, W. Reimers , M. Ionescu

Pages:

303-309

DOI:

10.4028/www.scientific.net/MSF.638-642.303

Citation:

R. C. Campana et al., "Applicability of Adaptive Neural Networks (ANN) in the Extrusion of Aluminum Alloys and in the Prediction of Hardness and Internal Defects", Materials Science Forum, Vols. 638-642, pp. 303-309, 2010

Online since:

January 2010

Export:

Price:

$35.00

In order to see related information, you need to Login.

In order to see related information, you need to Login.