Artificial neural network-based modeling for preditction of hardness of austempered ductile iron

Savangouder, Ravindra V. (Swinburne University of Technology, Melbourne, Australia) ; Patra, Jagdish C. (Haute école d’ingénierie et de gestion du canton de Vaud (HEIG-VD) HES-SO // Haute Ecole Spécialisée de Suisse Occidentale) ; Bornand, Cédric (Haute école d’ingénierie et de gestion du canton de Vaud (HEIG-VD) HES-SO // Haute Ecole Spécialisée de Suisse Occidentale)

Austempered ductile iron (ADI), because of its attractive properties, for example, high tensile strength along with good ductility is widely used in automotive industries. Such properties of ADI primarily depend on two factors: addition of a delicate proportion of several chemical compositions during the production of ductile cast iron and an isothermal heat treatment process, called austempering process. The chemical compositions, depending on the austempering temperature and its time duration, interact in a complex manner that influences the microstructure of ADI, and determines its hardness and ductility. Vickers hardness number (VHN) is commonly used as a measure of the hardness of a material. In this paper, an artificial neural network (ANN)-based modeling technique is proposed to predict the VHN of ADI by taking experimental data from literature. Extensive simulations showed that the ANN-based model can predict the VHN with a maximum mean absolute error (MAPE) of 0.22%, considering seven chemical compositions, in contrast to 0.71% reported in the recent paper considering only two chemical compositions.


Keywords:
Faculty:
Ingénierie et Architecture
School:
HEIG-VD
Institute:
ReDS - Reconfigurable & embedded Digital Systems
Publisher:
Cham, Springer
Date:
2019-12
Cham
Springer
Pagination:
8 p.
Published in:
International Conference on Neural Information Processing ICONIP 2019: Neural Information Processing
Author of the book:
Gedeon, T.
Wong, K.
Lee, M.
DOI:
ISBN:
978-3-030-36801-2
Appears in Collection:

Note: The status of this file is: restricted


 Record created 2020-01-14, last modified 2020-01-16

Fulltext:
Download fulltext
PDF

Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)