2022
DOI: 10.1002/nme.7178
|View full text |Cite
|
Sign up to set email alerts
|

Data‐physics driven reduced order homogenization

Abstract: A hybrid data‐physics driven reduced‐order homogenization (dpROH) approach aimed at improving the accuracy of the physics‐based reduced order homogenization (pROH), but retain its unique characteristics, such as interpretability and extrapolation, has been developed. The salient feature of the dpROH is that the data generated by a high‐fidelity model based on the first order computational homogenization (i.e., without model reduction) can improve markedly the accuracy of the physic‐based model reduction. The d… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1
1

Citation Types

0
5
0

Year Published

2023
2023
2024
2024

Publication Types

Select...
4
2

Relationship

1
5

Authors

Journals

citations
Cited by 8 publications
(5 citation statements)
references
References 60 publications
0
5
0
Order By: Relevance
“…Data-physics-driven ROH. 52 The solver-free CCH approach proposed in this study pre-computes eigenstrain influence tensors in the spirit of the ROH. Thus, by eliminating the need to solve the unit cell equilibrium equations, solver-free CCH proves remarkably less computationally intensive than CCH.…”
Section: Review Of Cch and Rohmentioning
confidence: 99%
See 3 more Smart Citations
“…Data-physics-driven ROH. 52 The solver-free CCH approach proposed in this study pre-computes eigenstrain influence tensors in the spirit of the ROH. Thus, by eliminating the need to solve the unit cell equilibrium equations, solver-free CCH proves remarkably less computationally intensive than CCH.…”
Section: Review Of Cch and Rohmentioning
confidence: 99%
“…[39][40][41][42][43] A few of the very recent exciting developments in multiscale modeling include mean-field homogenization, 44,45 surrogate modeling of the microscale based on deep learning, 46,47 an adaptive wavelet method for dynamic high-strain damage problems, 48 a self-consistent approach for high strain rate loading of composites, 49 a wavelet-enhanced fast Fourier transform (FFT)-based approach, 50 a ROH approach for polycrystalline microstructures with cracks, 51 and data-physics-driven ROH. 52 Broadly speaking, multiscale methods can be categorized as either resolved-scale methods or upscaling methods. Resolved-scale methods, which include domain decomposition and multigrid methods, directly simulate fine-scale behavior in localized subregions of the problem domain, while employing a coarse-scale model for the remainder of the domain.…”
Section: Introductionmentioning
confidence: 99%
See 2 more Smart Citations
“…GRUs also combine the input gate and forgetting gate of long short-term memory (LSTM) into an update gate, and change the output gate into a reset gate, which makes it have a simpler structure. Thus, it is easier to train and update its hidden status with less computation [44]. However, unidirectional LSTM and GRU neural networks have problems, such as the insufficient utilization of data information.…”
Section: Deep Learning Modelsmentioning
confidence: 99%