This paper presents an improved artificial neural network to predict the damage percentage in the test sample. The main objective is to show that the presence of holes and more generally of notches and other connection gaps lead to a weakening of the structure due to local overstresses, called stress concentrations. It is therefore good to avoid them, as much as possible. When the presence of stress concentrators is inevitable, it is necessary to know the stress concentration factor associated with each geometry, a notion introduced in this problem, in order to dimension the structures. Large holes result in high stress concentration factors. This behavior clearly shows that the presence of holes in a specimen is a place of stress concentration which can lead to the initiation and propagation of cracks. In this study, using an improved artificial neural network (ANN) model, we aim to predict the damage percentage in test samples with higher accuracy and reliability. This improved ANN integrates state-of-the-art algorithms (Arithmetic Optimization Algorithm-AOA, Balancing Composition Motion Optimization-BCMO and Jaya Algorithm), refined training methodologies and an extensive dataset of stress values of different sizes to ensure a more comprehensive and robust understanding of damage prediction, thereby contributing to more accurate assessments of the structural integrity and reliability of tested samples.
Predict Damage Percentage in Test Specimens Using Improved Artificial Neural Network / Oulad Brahim, A.; Capozucca, R.; Magagnini, E.; Khatir, B.; Khatir, A.. - 486:(2024), pp. 105-116. (Intervento presentato al convegno International Conference on Steel and Composite for Engineering Structures, ICSCES 2023 tenutosi a ita nel 2023) [10.1007/978-3-031-57224-1_11].
Predict Damage Percentage in Test Specimens Using Improved Artificial Neural Network
Oulad Brahim A.;Capozucca R.;Magagnini E.;Khatir A.
2024-01-01
Abstract
This paper presents an improved artificial neural network to predict the damage percentage in the test sample. The main objective is to show that the presence of holes and more generally of notches and other connection gaps lead to a weakening of the structure due to local overstresses, called stress concentrations. It is therefore good to avoid them, as much as possible. When the presence of stress concentrators is inevitable, it is necessary to know the stress concentration factor associated with each geometry, a notion introduced in this problem, in order to dimension the structures. Large holes result in high stress concentration factors. This behavior clearly shows that the presence of holes in a specimen is a place of stress concentration which can lead to the initiation and propagation of cracks. In this study, using an improved artificial neural network (ANN) model, we aim to predict the damage percentage in test samples with higher accuracy and reliability. This improved ANN integrates state-of-the-art algorithms (Arithmetic Optimization Algorithm-AOA, Balancing Composition Motion Optimization-BCMO and Jaya Algorithm), refined training methodologies and an extensive dataset of stress values of different sizes to ensure a more comprehensive and robust understanding of damage prediction, thereby contributing to more accurate assessments of the structural integrity and reliability of tested samples.File | Dimensione | Formato | |
---|---|---|---|
Paper-with-Springer-Template-moumin-final19-03-2024.pdf
Open Access dal 01/04/2025
Tipologia:
Documento in post-print (versione successiva alla peer review e accettata per la pubblicazione)
Licenza d'uso:
Tutti i diritti riservati
Dimensione
1.94 MB
Formato
Adobe PDF
|
1.94 MB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.