0
  • DE
  • EN
  • FR
  • Internationale Datenbank und Galerie für Ingenieurbauwerke

Anzeige

Predictive modelling and latent space exploration of steel profile overstrength factors using multi‐head autoencoder‐regressors

Autor(en): (ETH Zurich, Concrete Structures and Bridge Design & Design++ Zurich Switzerland)
(ETH Zurich, Steel and Composite Structures & Design++ Zurich Switzerland)
(Swiss Data Science Center)
(ETH Zurich, Steel and Composite Structures & Design++ Zurich Switzerland)
Medium: Fachartikel
Sprache(n): Englisch
Veröffentlicht in: ce/papers, , n. 3-4, v. 6
Seite(n): 836-842
DOI: 10.1002/cepa.2587
Abstrakt:

This paper investigates the suitability and interpretability of a data‐driven deep learning algorithm for multi cross sectional overstrength factor prediction. For this purpose, we first compile datasets consisting of experiments from literature on the overstrength factor of circular, rectangular and square hollow sections as well as I‐ and H‐sections. We then propose a novel multi‐head encoder architecture consisting of three input heads (one head per section type represented by respective features), a shared embedding layer as well as a subsequent regression tail for predicting the overstrength factor. By construction, this multi‐head architecture simultaneously allows for (i) the exploration of the nonlinear embedding of different cross‐sectional profiles towards the overstrength factor within the shared layer, and (ii) a forward prediction of the overstrength factor given profile features. Our framework enables for the first time an exploration of cross‐section similarity w.r.t. the overstrength factor across multiple sections and hence provides new domain insights in bearing capacities of steel cross‐sections, a much wider data exploration, since the encoder‐regressor can serve as meta model predictor. We demonstrate the quality of the predictive capabilities of the model and gain new insights of the latent space of different steel sections w.r.t. the overstrength factor. Our proposed method can easily be transferred to other multi‐input problems of Scientific Machine Learning.

Structurae kann Ihnen derzeit diese Veröffentlichung nicht im Volltext zur Verfügung stellen. Der Volltext ist beim Verlag erhältlich über die DOI: 10.1002/cepa.2587.
  • Über diese
    Datenseite
  • Reference-ID
    10766814
  • Veröffentlicht am:
    17.04.2024
  • Geändert am:
    17.04.2024
 
Structurae kooperiert mit
International Association for Bridge and Structural Engineering (IABSE)
e-mosty Magazine
e-BrIM Magazine