Naslov
Application of Neural Networks in Predicting the Steel Properties

Autori
Filetin, Tomislav ; Majetić, Dubravko ; Žmak, Irena

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Proceedings of 10th International DAAAM Symposium / Katalinić Branko - Beč : DAAAM International Vienna, 1999, 155-156

Skup
Intelligent Manufacturing & Automation : Past-Present-Future

Mjesto i datum
Beč, Austrija, 21.10.1999. - 23.10.1999

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
steels; artificial neural network; predicting of properties; Jominy curve; tempering curve

Sažetak
The purpose of this paper is to demonstrate how a neural network is used for predicting steel properties when some of the relevant influence factors and their relations are unknown. An attempt has been made to establish a non-linear static discrete-time neuron model, the so-called Static Elementary Processor (SEP). Based on SEP neurons, a static multi layer forward neural network is proposed to predict the Jominy Hardness curve from chemical composition. The same method has been used in predicting the tempering curve from chemical composition and austenizing temperature of tool steels. The mean error between hardness data fromm catalogue and predicted hardness data, and standard deviation for testing dataset are small and acceptable in both studies.

Izvorni jezik
Engleski

Znanstvena područja
Strojarstvo

POVEZANOST RADA