On the Application of Neural Networks Trained with FEM Data for the Identification of Stiffness Parameters of Improved Mechanical Beam Joints

Francisco Badea, JesusAngel Perez, Fikret Can Ozenli and José Luis Olazagoitia ()
Additional contact information
Francisco Badea: Department of Industrial Engineering and Automotive, Nebrija University, Pirineos 55, 28040 Madrid, Spain
JesusAngel Perez: Department of Construction and Manufacturing Engineering, Mechanical Engineering Area, University of Oviedo, 33940 Gijón, Spain
Fikret Can Ozenli: Department of Industy 4.0 Engineering, Ecole Supérieure d’Ingénieurs Léonard de Vinci, La Défense, 92916 Paris, France
José Luis Olazagoitia: Facultad de Diseño y Tecnología, Universidad de Diseño y Tecnología (UDIT), Av. Alfonso XIII, 97, 28016 Madrid, Spain

Mathematics, 2023, vol. 11, issue 15, 1-22

Abstract: Even though beam-type elements are widely adopted in the industry due to their low computational cost and potential time savings when modeling, they present a significant shortcoming given by their own formulation, which makes them incapable of accounting for local joint topology, which has a notable influence on the behavior of these structures. In this scenario, solutions that can mitigate this drawback while still providing improved results with simple models are of special interest. Many research works have focused on joint-specific approaches, as reflected in the literature. This paper introduces a novel generally improved beam model. This model uniquely features 4 nodes, 12 elastic elements, and 1 beam, contrasting starkly with the conventional beam elements that consist of merely 2 nodes and 1 element. This innovative model enhances the adaptability of modeled structures at the joint level. Crucially, it necessitates a methodology for the precise estimation of the elastic elements at the joint level. This article explores the capabilities of artificial neural networks for predicting the stiffness values derived from the calculated displacements at specific points within a complete structure. This research provides a complete analysis of the proposed methodology showing the significant limitations encountered for ANN when predicting finite element methodology (FEM)-derived values. The results and findings obtained in the article serve as a valuable reference paving the way for future studies involving finite element models and artificial neural networks.

Keywords: T-junctions; artificial neural networks; finite elements analysis; beam model improvements (search for similar items in EconPapers)
JEL-codes: C (search for similar items in EconPapers)
Date: 2023
References: View complete reference list from CitEc
Citations:

Downloads: (external link)
https://www.mdpi.com/2227-7390/11/15/3261/pdf (application/pdf)
https://www.mdpi.com/2227-7390/11/15/3261/ (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:gam:jmathe:v:11:y:2023:i:15:p:3261-:d:1201842

Access Statistics for this article

Mathematics is currently edited by Ms. Emma He

More articles in Mathematics from MDPI
Bibliographic data for series maintained by MDPI Indexing Manager ().