.png){kind=icon}
.png){kind=icon}
Return
Mechanism-guided multi-scale modeling of ESR slag viscosity in CaF2–CaO–Al2O3/SiO2 via microstructural analysis and adaptive multi-source ensemble learning
.png){kind=icon}
PRE
AI
陈
W
Y
A
Z
Y
DOI：10.1007/s42243-026-01880-7.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
A collaborative framework integrating phase diagram digitization, sparse region identification, molecular dynamics simulations, modified Arrhenius equation fitting, and adaptive multi-source ensemble learning is developed for viscosity prediction of CaF2–CaO–Al2O3/SiO=2 slag systems in electroslag remelting. The methodology overcomes challenges posed by high-temperature measurement limitations and sparse compositional coverage by expanding experimental data through systematic phase diagram analysis and supplementing missing values via molecular dynamics simulations in the liquid phase region and empirical extrapolation. An adaptive ensemble strategy combining categorical boosting, eXtreme gradient boosting, and support vector regression achieves test-set performance metrics: 0.0413 for mean squared error, 0.0886 for mean absolute error, and 0.857 for R2 (coefficient of determination), representing a 5.8-fold increase in compositional–temperature space coverage with minimal experimental cost. Microstructural analysis reveals that Al2O3 functions as a “weak network former” with a critical threshold at 20%, while SiO2 acts as a “strong network former” with a polymerization transition at 4 wt.%. The quantitative mapping of bridging oxygen ratios, Qn distribution, and network density to viscosity evolution establishes the composition–structure–property relationship at the atomic scale. A low-viscosity operational window (20% Al2O3–4% SiO2) is proposed for slag system optimization.
Keywords:
Electroslag remelting
Molecular dynamics
Machine learning
Ensemble learning
CaF2–CaO–Al2O3/SiO2
Microstructural analysis
Journal
IF:
3.6
Papers:
3.6K
Citations:
6.1K
Organization
Cited Papers