Communication Dans Un Congrès Année : 2020

Combining Importance Sampling with Machine Learning to Accelerate Structural Reliability Analysis

Résumé

This paper presents an innovative approach combining Importance Sampling (IS) and Machine Learning (ML) to accelerate structural reliability analysis. Reliability analysis is essential for assessing the safety and performance of structures subject to uncertainties and random excitations. However, traditional methods can be computationally expensive, especially when failure events are rare. In this study, we propose a hybrid method that uses Importance Sampling (IS) to generate meaningful samples while integrating machine learning models to predict structural behavior under various loading conditions. This combination reduces the number of required simulations while improving the accuracy of failure probability estimates. Experimental results show that our approach can significantly reduce computational time while maintaining accuracy comparable to conventional methods. This method opens new avenues for reliability analysis, enabling engineers to better assess the risks associated with complex structures.

Fichier principal
Vignette du fichier
Paper1_MEDYNA2020_Paper.pdf (415.25 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-05146779 , version 1 (21-07-2025)

Licence

Identifiants

  • HAL Id : hal-05146779 , version 1

Citer

Alexandre Saidi, Weizhen Hou, Abdelmalek Zine, Mohamed Ichchou. Combining Importance Sampling with Machine Learning to Accelerate Structural Reliability Analysis. Conference MEDYNA 2020 : 3rd Euro-Mediterranean Conference on Structural Dynamics and Vibroacoustics, Italy, AIDAA, Italian Association of Aeronautics and Astronautics (www.aidaa.it), Feb 2020, Naple, Italy. ⟨hal-05146779⟩
233 Consultations
101 Téléchargements

Partager

  • More