Suscripción a Biblioteca: Guest

ISSN Online: 2377-424X

ISBN Print: 0-85295-345-3

International Heat Transfer Conference 10
August, 14-18, 1994, Brighton, UK

THERMALHYDRAULIC CORRELATIONS BY NEURAL NETWORKS

Get access (open in a dialog) DOI: 10.1615/IHTC10.3960
pages 391-396

Sinopsis

In recent years the Artificial Neural Network methodology has received growing attention because of the excellent results achieved in many fields of application. A new area of potential benefit for this methodology seems to be that of correlating experimental data. In this context, we explore a methodology based upon the use of the self-organizing Kohonen's map and the supervised, multilayered, feed forward neural network trained by the error back propagation algorithm. Here the methodology is applied to two-phase pressure drops in horizontal ducts. The problem of correlating in these conditions presents many difficulties revealed when trying to apply some classical correlations. The result seems interesting both for correlating purposes and for singling out groups of data having similar behaviour, identifying at the same time the parameter responsible of this.