Microstructure design using machine learning generated low dimensional and continuous design space

Jung, Jaimyun; Yoon, Jae Ik; Park, Hyung Keun; Jo, Hyeontae; Kim, Hyoung Seop

doi:10.1016/j.mtla.2020.100690

Cited by 42 publications

(14 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…A general process of autoencoders/variational autoencodersbased design [131,264,265] is shown in Figure 21a. Taking microstructure images as the input, the encoder generates parametric latent features as the microstructure design variables.…”

Section: Deep Learning-driven Generative Design and Optimizationmentioning

confidence: 99%

“…Taking microstructure images as the input, the encoder generates parametric latent features as the microstructure design variables. Furthermore, the latent feature-property relationship is established by another machine learning model, such as Gaussian regression for Bayesian optimization, [264] convolutional residual network (ResNet), [131] etc. Microstructure design was done by searching latent feature values that lead to optimal material properties, and then mapping the optimal microstructure design from the latent space to the geometry space by decoding.…”

Section: Deep Learning-driven Generative Design and Optimizationmentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

Guiding the Design of Heterogeneous Electrode Microstructures for Li‐Ion Batteries: Microscopic Imaging, Predictive Modeling, and Machine Learning

Zhu

Finegan

et al. 2021

Advanced Energy Materials

View full text Add to dashboard Cite

Electrochemical and mechanical properties of lithium‐ion battery materials are heavily dependent on their 3D microstructure characteristics. A quantitative understanding of the role played by stochastic microstructures is critical for the prediction of material properties and for guiding synthesis processes. Furthermore, tailoring microstructure morphology is also a viable way of achieving optimal electrochemical and mechanical performances of lithium‐ion cells. To facilitate the establishment of microstructure‐resolved modeling and design methods, a review covering spatially and temporally resolved imaging of microstructure and electrochemical phenomena, microstructure statistical characterization and stochastic reconstruction, microstructure‐resolved modeling for property prediction, and machine learning for microstructure design is presented here. The perspectives on the unresolved challenges and opportunities in applying experimental data, modeling, and machine learning to improve the understanding of materials and identify paths toward enhanced performance of lithium‐ion cells are presented.

show abstract

Section: Deep Learning-driven Generative Design and Optimizationmentioning

confidence: 99%

Section: Deep Learning-driven Generative Design and Optimizationmentioning

confidence: 99%

See 1 more Smart Citation

Guiding the Design of Heterogeneous Electrode Microstructures for Li‐Ion Batteries: Microscopic Imaging, Predictive Modeling, and Machine Learning

Zhu

Finegan

et al. 2021

Advanced Energy Materials

View full text Add to dashboard Cite

show abstract

“…Most importantly, CNN-based models are prone to model overfit due to their large number of tunable parameters. Despite these limitations, recent work has demonstrated the positive impact of data-driven methods on topology optimization (Kollmann et al, 2020;Yilin et al, 2021) and inverse design of microstructures (Jung et al, 2020;Tan et al, 2020).…”

Section: Introductionmentioning

confidence: 99%

Development of a Robust CNN Model for Capturing Microstructure-Property Linkages and Building Property Closures Supporting Material Design

Mann

Kalidindi

2022

Front. Mater.

View full text Add to dashboard Cite

Recent works have demonstrated the viability of convolutional neural networks (CNN) for capturing the highly non-linear microstructure-property linkages in high contrast composite material systems. In this work, we develop a new CNN architecture that utilizes a drastically reduced number of trainable parameters for building these linkages, compared to the benchmarks in current literature. This is accomplished by creating CNN architectures that completely avoid the use of fully connected layers, while using the 2-point spatial correlations of the microstructure as the input to the CNN. In addition to increased robustness (because of the much smaller number of trainable parameters), the CNN models developed in this work facilitate the construction of property closures at very low computational cost. This is because it allows for easy exploration of the space of valid 2-point spatial correlations, which is known to be a convex hull. Consequently, one can generate new sets of valid 2-point spatial correlations from previously available valid sets of 2-point spatial correlations, simply as convex combinations. This work demonstrates the significant benefits of utilizing 2-point spatial correlations as the input to the CNN, in place of the voxelated discrete microstructures used in current benchmarks.

show abstract

“…We note that problem formulations based on Bayesian Optimization [11,12,13]) account for uncertainty in the objective solely due to the imprecision of the surrogate and not due to the aleatoric, stochastic variability of the underlying microstructure. In the context of optimization/design problems in particular, a globally-accurate surrogate would be redundant.…”

Section: Introductionmentioning

confidence: 99%

Self-supervised optimization of random material microstructures in the small-data regime

Rixner¹,

Koutsourelakis²

2021

Preprint

View full text Add to dashboard Cite

While the forward and backward modeling of the process-structure-property chain has received a lot of attention from the materials community, fewer efforts have taken into consideration uncertainties. Those arise from a multitude of sources and their quantification and integration in the inversion process are essential in meeting the materials design objectives. The first contribution of this paper is a flexible, fully probabilistic formulation of such optimization problems that accounts for the uncertainty in the process-structure and structure-property linkages and enables the identification of optimal, high-dimensional, process parameters. We employ a probabilistic, data-driven surrogate for the structure-property link which expedites computations and enables handling of non-differential objectives. We couple this with a novel active learning strategy, i.e. a self-supervised collection of data, which significantly improves accuracy while requiring small amounts of training data. We demonstrate its efficacy in optimizing the mechanical and thermal properties of two-phase, random media but envision its applicability encompasses a wide variety of microstructure-sensitive design problems.

show abstract

Microstructure design using machine learning generated low dimensional and continuous design space

Cited by 42 publications

References 32 publications

Guiding the Design of Heterogeneous Electrode Microstructures for Li‐Ion Batteries: Microscopic Imaging, Predictive Modeling, and Machine Learning

Guiding the Design of Heterogeneous Electrode Microstructures for Li‐Ion Batteries: Microscopic Imaging, Predictive Modeling, and Machine Learning

Development of a Robust CNN Model for Capturing Microstructure-Property Linkages and Building Property Closures Supporting Material Design

Self-supervised optimization of random material microstructures in the small-data regime

Contact Info

Product

Resources

About