Building a quantitative composition-microstructure-property relationship of dual-phase steels via multimodal data mining

Da, Ren; Wang, Chenchong; Wei, Xiaolu; Lai, Qingquan; Xu, Wei

doi:10.1016/j.actamat.2023.118954

Cited by 14 publications

(4 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, Gradient‐weighted Class Activation Mapping (Grad‐CAM) is applied to visualize important regions, where the value at each position of the input image is computed based on its importance to the prediction results. This algorithm aids in the reverse analysis of core microstructural units, contributing significantly to the target attributes [15] …”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Employing deep learning in non‐parametric inverse visualization of elastic–plastic mechanisms in dual‐phase steels

Han,

Wang,

Zhang

et al. 2024

Materials Genome Engineering Advances

Self Cite

View full text Add to dashboard Cite

Enhancing the interpretability of machine learning methods for predicting material properties is a key, yet complex topic in materials science. This study proposes an interpretable convolutional neural network (CNN) to establish the relationship between the microstructural evolution and mechanical properties of non‐uniform and nonlinear multisystem dual‐phase steel materials and achieve an inverse analysis of the elastic‐plastic mechanism. This study demonstrates that the developed CNN model achieves an accuracy of 94% in predicting the stress‐strain curves of dual‐phase steel microstructures with different compositions and processes, with the mean absolute error not exceeding 50 MPa, representing merely 5.26% of the average tensile strength of dual‐phase steels in the dataset. The reverse visualization results of the CNN model indicate that, during tensile deformation, the grain boundaries maintain deformation coordination within the grains by impeding dislocation slip. This results in a significant stress concentration at the grain boundaries, with stresses at the boundaries being higher than those borne by the martensitic phase and minimal stresses in the ferrite phase. Moreover, compared with traditional crystal plasticity models, the CNN model exhibits a substantial improvement in computational efficiency. This method provides a generic plan for improving the interpretability of machine learning methods for predicting material properties and can be easily applied to other alloy systems.

show abstract

Section: Methodsmentioning

confidence: 99%

“…The features visualized by the CNN confirmed its prediction accuracy. Ren et al [15] . developed a universal deep learning framework that established relationships between composition, microstructure, and properties.…”

Section: Introductionmentioning

confidence: 99%

Employing deep learning in non‐parametric inverse visualization of elastic–plastic mechanisms in dual‐phase steels

Han,

Wang,

Zhang

et al. 2024

Materials Genome Engineering Advances

Self Cite

View full text Add to dashboard Cite

show abstract

“…The realization of cellular automata in the prediction of mechanical properties may be the integration of machine-learning algorithms and data-driven approaches [ 119 , 120 ]. This integration leads to more accurate predictive models by large datasets and optimizing model parameters based on experimental data [ 121 , 122 ]. This research will open a new chapter in the study of mechanical properties and raise our understanding of the nature of mechanical properties of metallic materials to a new level.…”

Section: Future Research Directions Of Ca In Metallic Materials Researchmentioning

confidence: 99%

Review on Cellular Automata for Microstructure Simulation of Metallic Materials

Zhi,

Jiang,

et al. 2024

Materials

View full text Add to dashboard Cite

The cellular automata (CA) method has played an important role in the research and development of metallic materials. CA can interpret the microstructure changes of materials and obtain more abundant, accurate and intuitive information of microstructure evolution than conventional methods. CA can visually represent the process of grain formation, growth, development and change to us in a graphical way, which can assist us in analysis, thinking and solving problems. In the last five years, the application of CA in materials research has been rapidly developed, and CA has begun to occupy an increasingly important position in the simulation research of metallic materials. After introducing the advantages and limitations of CA compared to other widely used simulation methods, the purpose of this paper is to review the recent application progress on the microstructure simulation of metallic materials using CA, such as solidification, recrystallization, phase transformation and carbide precipitation occurring during forming and heat treatment. Specifically, recent research advances on microstructure simulation by CA in the fields of additive manufacturing, welding, asymmetrical rolling, corrosion prevention, etc., are also elaborated in this paper. Furthermore, this paper points out the future work direction of CA simulation in the research of metallic materials, especially in the simulation of the crystal structure, the prediction of mechanical properties, CA simulation software and rule systems, etc. These are expected to attract wide attention of researchers in the field of metallic materials and promote the development of CA in materials research.

show abstract

“…Li et al [12] have proposed a machine learning method guided by physical metallurgy principles to rapidly and accurately predict the thermal deformation mechanisms of antibacterial stainless steels with varying copper (Cu) contents. Ren et al [13] used a deep learning framework to reveal the relationship between material composition microstructure performance and predict the tensile properties of dual phase steel. Wei et al [14] proposed a machine learning model for predicting the rotational bending fatigue strength and high-throughput design of fatigue resistant steel.…”

Section: Introductionmentioning

confidence: 99%

Prediction and Analysis of Hot Rolling Strip Tension Based on Back Propagation Neural Network

Yuan,

Ma,

et al. 2024

Preprint

View full text Add to dashboard Cite

In modern hot strip mill control systems, strip tension control is the core control function, and its performance will be directly reflected in product quality. A strip tension prediction model based on Back Propagation (BP) neural network is proposed. To ensure that the true tension value is obtained, this paper proposes a four-dimensional judgment mode for the contact time between the looper and the strip steel and establishes a data set of tension parameters for hot rolled strip steel. The traditional BP neural network, genetic algorithm optimized BP neural network (GA-BP), and whale algorithm optimized BP neural network (WOA-BP) models were used to predict the strip tension, and their prediction performance was evaluated. The results show that the proposed WOA-BP model has the best prediction effect, with the highest model decision coefficient of 0.9330. At the same time, the contribution rate of each variable to the strip tension was studied, and the results showed that the looper angle and looper roller force had the greatest impact on it, consistent with physical laws. Propose improvement suggestions for the control performance of hot rolled strip tension based on the influence of looper angle on the prediction model.

show abstract

Building a quantitative composition-microstructure-property relationship of dual-phase steels via multimodal data mining

Cited by 14 publications

References 45 publications

Employing deep learning in non‐parametric inverse visualization of elastic–plastic mechanisms in dual‐phase steels

Employing deep learning in non‐parametric inverse visualization of elastic–plastic mechanisms in dual‐phase steels

Review on Cellular Automata for Microstructure Simulation of Metallic Materials

Prediction and Analysis of Hot Rolling Strip Tension Based on Back Propagation Neural Network

Contact Info

Product

Resources

About