Prediction of steel nanohardness by using graph neural networks on surface polycrystallinity maps

Karimi, Kamran; Salmenjoki, Henri; Mulewska, K.; Kurpaska, Ł.; Kosińska, A.; Alava, Mikko J.; Papanikolaou, Stefanos

doi:10.1016/j.scriptamat.2023.115559

Cited by 12 publications

(5 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Indeed, although not yet explored, the integration of information concerning local crystal orientation or metallurgy associated to the hardness values to be estimated is expected to improve the accuracy of the model, also considering the outcomes of some recent literature results. [61] In effect, the metallurgy data that are used in the proposed approach are process data taken on a liquid steel sample and not from the Jominy specimen. On the other hand, this requires additional analytical efforts and costs, which might balance the savings that are achievable with the ML-based solution.…”

Section: Discussionmentioning

confidence: 99%

A Deep Learning‐Based Approach to the Estimation of Jominy Profile of Medium‐Carbon Quench Hardenable Steels

Colla,

Vannucci,

Matino

et al. 2023

steel research int.

View full text Add to dashboard Cite

The possibility to estimate the Jominy profile of steel based on its chemical composition is of utmost importance and high practical relevance for industries, at enables a preliminary assessment of the suitability of a specific steel grade to a particular application or to the requirements of a customer, by saving time and resources as the Jominy end‐quench test is costly and time consuming. More importantly, an estimator can be used in steel grade design, by supporting the investigation of the most suitable chemistry to meet some given specifications. The paper proposes a novel approach to estimate the hardenability profile of medium Carbon quench hardenable steels, which exploits the potential of deep learning to correlate the steel metallurgy to the entire shape of the curve rather than to its single points, by thus being adaptable to a wide range of steel grades while providing very accurate estimates. Moreover, the proposed approach is suitable implement a transfer learning paradigm, as it can exploit the knowledge acquired by training on a specific dataset to adapt the model to different steel grades for which less data or data holding different features are available.This article is protected by copyright. All rights reserved.

show abstract

Section: Discussionmentioning

confidence: 99%

A Deep Learning‐Based Approach to the Estimation of Jominy Profile of Medium‐Carbon Quench Hardenable Steels

Colla,

Vannucci,

Matino

et al. 2023

steel research int.

View full text Add to dashboard Cite

show abstract

“…This approach is adopted in many existing works for learning elastic properties ( e.g. , for alloys 24–27 and polycrystals 28,29 ) and related atomic properties like stress and energy fields. 30 They are, however, limited to derived scalar elastic properties such as bulk modulus and shear modulus, and separate models are built for each derived property.…”

Section: Introductionmentioning

confidence: 99%

“…In a nutshell, state-of-the-art ML models for elastic properties encode compositional information [19][20][21] and/or structural information [20][21][22][23] in a material as feature vectors and then map them to a target using some regression algorithms. This approach is adopted in many existing works for learning elastic properties (e.g., for alloys [24][25][26][27] and polycrystals 28,29 ) and related atomic properties like stress and energy elds. 30 They are, however, limited to derived scalar elastic properties such as bulk modulus and shear modulus, and separate models are built for each derived property.…”

Section: Introductionmentioning

confidence: 99%

An equivariant graph neural network for the elasticity tensors of all seven crystal systems

Wen,

Horton,

Munro

et al. 2024

Digital Discovery

View full text Add to dashboard Cite

show abstract

“…Polycrystals have been studied by ML in several publications 29 – 33 , where experimental data and finite element simulations were used to produce the training data. In contrast, in this work, we study predictability of the deformation process of cube-shaped iron nanopolycrystals by combining strain-controlled molecular dynamics (MD) simulations with ML methods.…”

Section: Introductionmentioning

confidence: 99%

Predicting elastic and plastic properties of small iron polycrystals by machine learning

Mińkowski,

Laurson

2023

Sci Rep

View full text Add to dashboard Cite

Deformation of crystalline materials is an interesting example of complex system behaviour. Small samples typically exhibit a stochastic-like, irregular response to externally applied stresses, manifested as significant sample-to-sample variation in their mechanical properties. In this work we study the predictability of the sample-dependent shear moduli and yield stresses of a large set of small cube-shaped iron polycrystals generated by Voronoi tessellation, by combining molecular dynamics simulations and machine learning. Training a convolutional neural network to infer the mapping between the initial polycrystalline structure of the samples and features of the ensuing stress-strain curves reveals that the shear modulus can be predicted better than the yield stress. We discuss our results in the context of the sensitivity of the system’s response to small perturbations of its initial state.

show abstract

Prediction of steel nanohardness by using graph neural networks on surface polycrystallinity maps

Cited by 12 publications

References 23 publications

A Deep Learning‐Based Approach to the Estimation of Jominy Profile of Medium‐Carbon Quench Hardenable Steels

A Deep Learning‐Based Approach to the Estimation of Jominy Profile of Medium‐Carbon Quench Hardenable Steels

An equivariant graph neural network for the elasticity tensors of all seven crystal systems

Predicting elastic and plastic properties of small iron polycrystals by machine learning

Contact Info

Product

Resources

About