Anzeige
Erschienen in:
2021 | OriginalPaper | Buchkapitel
Parallelized Numerical Optimization for Parameter Fitting of an Advanced Elasto-Viscoplastic Model with an Open Source Implementation
verfasst von : Thomas Alexander Vergote, Leung Chun Fai, Chian Siau Chen
Erschienen in: Challenges and Innovations in Geomechanics
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Abstract
To model the elasto-viscoplastic soil response, even in one-dimensional problems, a significant amount of parameters are required. These parameters interact, making it difficult to fit the parameters directly. A model is fit to data by numerical optimization with a number of local and global algorithms. The evaluation showcases that the correct parameters can be found efficiently by optimization, but that there is a significant risk of converging to a local minimum far removed from the correct solution as the dimensionality increases. A randomly repeated local search algorithm is proposed which outperforms the other algorithms. The optimization scheme was applied to find the parameters based on tests with varying stress paths on the same soil by a combined parallelized optimization scheme. The combined optimization allows to find parameters that fit all stress paths as good as possible, serving as a basis for comparison of models.