Physics-Informed Machine Learning Prediction of Axial Capacity of Hybrid Double-I-Box Cold-Formed Steel Columns


Date Published : 20 November 2026

Contributors

DEEPAK M S

Postdoctoral Research Fellow, Lincoln University College, Petaling Jaya, Selangor, Malaysia.
Author

Keywords

Cold-formed steel; Hybrid Double-I-Box Column; Axial capacity; Physics-informed machine learning; Bottom-Zone Local Buckling;

Proceeding

Track

General Track

License

Copyright (c) 2026 Sustainable Global Societies Initiative

Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Abstract

Hybrid Double-I-Box Columns (HDIBCs) are built-up cold-formed steel compression members whose axial capacity depends on hybrid material distribution, slenderness and Bottom-Zone Local Buckling (BZLB). This paper reorganises a published experimental–finite-element column database for physics-informed machine-learning screening. Mechanics-based inputs support direct capacity regression, code-resistance correction and response-class prediction. The extraction contains 24 experimental records and 96 ideal finite-element condition records. Missing ideal-section properties restrict the executed five-fold assessment to 24 experiments, grouped into 12 material–section families. The source effective-width resistance gives 16.57% mean absolute percentage error; Ridge and Extra Trees correction models give 2.55% and 1.91%, respectively. Gradient Boosting classifies the two observed responses with 87.50% accuracy and 0.823 macro F1. These results describe internal validation on existing data. Four-class evaluation remains pending because ideal-model inputs are incomplete and condition labels are not identifiable from the specified features. The framework provides axial-capacity and BZLB screening, while source inconsistencies and the small sample limit interpretation. It is not a replacement for code or design verification.

References

No References

Downloads

How to Cite

DEEPAK M S, D. M. S. (2026). Physics-Informed Machine Learning Prediction of Axial Capacity of Hybrid Double-I-Box Cold-Formed Steel Columns. Sustainable Global Societies Initiative, 1(15). https://vectmag.com/sgsi/paper/view/1485