Pre-Rolling Forming-Quality Prediction in 2219 Aluminum Alloy Ring Rolling Using Physics-Informed Heterogeneous Stacking

Industrial 2219 aluminum alloy ring production requires quality assessment before a prescribed rolling plan is executed. At the investigated site, independent geometric inspection is incomplete and forming-quality feedback is not used for in-process plan adjustment. This study establishes a pre-rolling prediction framework by separating quality supervision derived from historical process signals from information available for a new workpiece. A terminal radial-fluctuation root-mean-square (TRF-RMS) indicator is constructed from the follower-roll terminal response and mapped to three enterprise-defined grades. Physics-informed heterogeneous stacking (PI-HStack) predicts these process-based grades using blank and target geometry, derived geometric relationships, initial temperature, and prescribed rolling time. Engineering descriptors represent the planned transformation; out-of-fold probability fusion combines learners, and class-specific decision adjustment addresses three-grade assignment. Five stratified 80/20 hold-out tests on 2224 industrial records yielded mean accuracy of 0.6769, macro-averaged F1-score (Macro-F1) of 0.6436, and macro-averaged one-vs-rest area under the receiver operating characteristic curve (Macro-AUC) of 0.8316. Basic geometry already provides substantial predictive information, and additional descriptors do not improve all objectives. PI-HStack has the highest reported mean overall metrics, whereas extreme gradient boosting (XGBoost) performs better on poor-grade screening measures. Ranking by the PI-HStack poor-grade score captures 56.28% of poor-grade records within the highest-ranked 20% of test records. The findings support objective-specific feature and model selection for pre-rolling plan review. The evaluation is retrospective and limited to the observed production distribution; independent geometric-tolerance validation and generalization to unseen specifications remain unestablished.

Authors

Institutions

Publication Details

Journal
Symmetry
Published
2026-10-16
DOI
https://doi.org/10.3390/sym18101725
Primary Topic
Fault Detection and Control Systems
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Pre-Rolling Forming-Quality Prediction in 2219 Aluminum Alloy Ring Rolling Using Physics-Informed Heterogeneous Stacking

Xiaojie Jia, Jianping Tan, Ruining Xie, Zehui Li et al.
Symmetry
Fault Detection and Control Systems
article

Pre-Rolling Forming-Quality Prediction in 2219 Aluminum Alloy Ring Rolling Using Physics-Informed Heterogeneous Stacking

Xiaojie Jia, Jianping Tan, Ruining Xie, Zehui Li, Jianqiao Liu, Xiaoxuan Huang, Jie Hou, Liang Zhou, Li Li
article en

Abstract

Industrial 2219 aluminum alloy ring production requires quality assessment before a prescribed rolling plan is executed. At the investigated site, independent geometric inspection is incomplete and forming-quality feedback is not used for in-process plan adjustment. This study establishes a pre-rolling prediction framework by separating quality supervision derived from historical process signals from information available for a new workpiece. A terminal radial-fluctuation root-mean-square (TRF-RMS) indicator is constructed from the follower-roll terminal response and mapped to three enterprise-defined grades. Physics-informed heterogeneous stacking (PI-HStack) predicts these process-based grades using blank and target geometry, derived geometric relationships, initial temperature, and prescribed rolling time. Engineering descriptors represent the planned transformation; out-of-fold probability fusion combines learners, and class-specific decision adjustment addresses three-grade assignment. Five stratified 80/20 hold-out tests on 2224 industrial records yielded mean accuracy of 0.6769, macro-averaged F1-score (Macro-F1) of 0.6436, and macro-averaged one-vs-rest area under the receiver operating characteristic curve (Macro-AUC) of 0.8316. Basic geometry already provides substantial predictive information, and additional descriptors do not improve all objectives. PI-HStack has the highest reported mean overall metrics, whereas extreme gradient boosting (XGBoost) performs better on poor-grade screening measures. Ranking by the PI-HStack poor-grade score captures 56.28% of poor-grade records within the highest-ranked 20% of test records. The findings support objective-specific feature and model selection for pre-rolling plan review. The evaluation is retrospective and limited to the observed production distribution; independent geometric-tolerance validation and generalization to unseen specifications remain unestablished.

SymmetryVol. 18(10)
Central South University (CN), Yueyang Changling Equipment Research Institute (China) (CN), Institute of New Materials (CN)
National Natural Science Foundation of China
Openalex Percentile: Top 17%
Fault Detection and Control Systems
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.