Pregled bibliografske jedinice broj: 913572
From simple empirical methods to artificial neural networks for estimation of cyclic and fatigue material parameters – An overview and outlook
From simple empirical methods to artificial neural networks for estimation of cyclic and fatigue material parameters – An overview and outlook // Abstracts of the 8th International Conference on Structural Engineering and Construction Management 2017
Kandy: University of Peradenya, 2017. str. xxxvi-xxxvi (plenarno, međunarodna recenzija, sažetak, znanstveni)
CROSBI ID: 913572 Za ispravke kontaktirajte CROSBI podršku putem web obrasca
Naslov
From simple empirical methods to artificial neural networks for estimation of cyclic and fatigue material parameters – An overview and outlook
Autori
Basan, Robert
Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, znanstveni
Izvornik
Abstracts of the 8th International Conference on Structural Engineering and Construction Management 2017
/ - Kandy : University of Peradenya, 2017, Xxxvi-xxxvi
ISBN
978-955-589-239-1
Skup
8th International Conference on Structural Engineering and Construction Management 2017
Mjesto i datum
Kandy, Šri Lanka, 07.12.2017. - 09.12.2017
Vrsta sudjelovanja
Plenarno
Vrsta recenzije
Međunarodna recenzija
Ključne riječi
estimation methods ; monotonic properties ; cyclic parameters ; fatigue parameters ; ANN ; materials database
Sažetak
In an attempt to reduce the number of experiments needed and to enable performing of calculations early in product development, empirical methods for estimation of materials' cyclic and fatigue parameters from simple monotonic properties are being developed from mid 1960's to current day. A critical overview and analysis of existing approaches and estimation methods is provided. Their main features and deficiencies including development on limited number of material data, establishment of a direct and independent relationship between monotonic properties and cyclic/fatigue parameters, assignment of constant values to these parameters due to the lack of, or poor correlation among them, disregard of the differences among different material groups, are identified and discussed. Applicability of existing methodology for evaluation of estimation methods is analyzed and some new insights and suggestions are provided. Discussion is complemented with recent developments of artificial neural networks (ANN) for estimation of cyclic and fatigue parameters, own ANN solution developed using MATDAT Materials Properties Database and it's comparison with existing relevant empirical methods. Outlook to future work on previously proposed indirect approach to estimation of cyclic/fatigue parameters and further steps in development of ANN-based solutions are presented.
Izvorni jezik
Engleski
Znanstvena područja
Strojarstvo
