.png){kind=icon}
.png){kind=icon}
Return
Evaluation of Micro-Mechanical Behavior for low Carbon-Containing Alumina Refractory assisted by Interpretable Machine Learning: Toward Development of Digital Materials
.png){kind=icon}
PRE
AI
H
Z
J
L
S
Y
X
Y
DOI：10.1016/j.actamat.2026.122331.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Carbon-containing alumina refractories are important materials of several functional metallurgical components, and their service performance is ruled by their micro-scale mechanical behavior. By integrating nano-indentation testing with interpretable machine learning, an experimental-computational workflow comprising probabilistic microstructural clustering and SHapley Additive exPlanations (SHAP)-guided Tabular Prior-Fitting Network (TabPFN) modeling was established. The study quantitatively identifies the micro-scale functional clusters and corresponding unified microstructural states with the proportion and influence score. A correlation between micro- and macro-mechanical behavior is demonstrated. Additionally, the effects of replacement of 10% flake graphite by 1% expanded graphite as well as the impact of thermal shock on the microstructure and properties evolution are clearly interpreted based on the key feature selection. The processing-microstructure-property integrated material model and methodology established in this study bridge micromechanical mechanisms and macroscopic performance, offering an interpretable basis for the development of heterogeneous digital refractory.
Keywords:
micro-mechanical behavior
carbon-containing alumina refractory
interpretable machine learning
microstructure-property relationship
digital materials
Journal
IF:
9.3
Papers:
2.0W
Citations:
12.9W
Organization
Cited Papers