Adaptive sparse grid based HOPGD: Toward a nonintrusive strategy for constructing space‐time welding computational vademecum

Abstract

Summary Simulation‐based engineering usually needs the construction of computational vademecum to take into account the multiparametric aspect. One example concerns the optimization and inverse identification problems encountered in welding processes. This paper presents a nonintrusive a posteriori strategy for constructing quasi‐optimal space‐time computational vademecum using the higher‐order proper generalized decomposition method. Contrary to conventional tensor decomposition methods, based on full grids (eg, parallel factor analysis/higher‐order singular value decomposition), the proposed method is adapted to sparse grids, which allows an efficient adaptive sampling in the multidimensional parameter space. In addition, a residual‐based accelerator is proposed to accelerate the higher‐order proper generalized decomposition procedure for the optimal aspect of computational vademecum. Based on a simplified welding model, different examples of computational vademecum of dimension up to 6, taking into account both geometry and material parameters, are presented. These vademecums lead to real‐time parametric solutions and can serve as handbook for engineers to deal with optimization, identification, or other problems related to repetitive task.

A multiscale reduced‐order‐model strategy for transient thermoelasticity with variable microstructure

“…Other perspectives concern the case where more parameters are involved. Some solution strategies consist in switching from SVD to higher order SVG (HO‐SVD), 39 higher order proper generalized decomposition (HO‐PGD), 40,41 Candecomp/Parafac (CP), 42 or their gappy versions 43,44 . Other surrogate modelings can be considered, such as Kriging, 45 or even support vector regression (SVR) 46 .…”

Nonintrusive Uncertainty Quantification for automotive crash problems with VPS/Pamcrash

“…The separated character of F (h) means that it is a sum of rank-one terms, being each rank-one term the product of sectional modes (the adjective sectional is used to indicate that the mode depends only on one of the parameters). The algorithm employed to compute the SRS is based on the ideas of the leastsquares PGD approximations described in detail in [3,4,15,14].…”

HiDeNN-PGD: reduced-order hierarchical deep learning neural networks

“…This form or concept is known as CD [29]. The so-called PGD method has adopted this concept for solving PDEs [10,30] and for data learning [14,15,31].…”

“…It has gained increased popularity in recent years. For overcoming the intrusiveness and extending the applicability of the method, the a posteriori data-driven PGD [14][15][16][17] has also been developed more recently. In contrast to a priori PGD, the a posteriori method uses a database to learn the separated functions and thus can be used as regression for constructing reduced order surrogate models.…”