Predicting Fatigue Bond Life of Composite Rebar‐Concrete Structures: A Cross‐Material Transfer Learning Approach and Application
Year: 2026
Type: article
Abstract: Fiber‐reinforced polymer composite rebars offer superior corrosion resistance for marine infrastructure. However, due to data scarcity, the prediction of fatigue bond life for composite rebar‐reinforced concrete (C‐RC) remains a challenge. This study proposes a transfer learning (TL) framework leveraging knowledge from steel rebar to predicting composite rebar fatigue bond life. For comparison, eXtreme Gradient Boosting (XGBoost) and deep neural network (DNN) models were developed using only composite rebar data, additionally with an XGBoost model adopting combined datasets. SHapley Additive exPlanations (SHAP) analysis elucidated similarities and distinctions in fatigue bond mechanisms between steel and composite rebars. Results demonstrate the TL model achieved better performance than the XGBoost and DNN models, with accuracy rising 67.5% above the model with combined dataset. A Streamlit‐based platform was supplementarily established for fatigue bond life prediction, providing a framework for cross‐material fatigue life predictions based on TL techniques, which save cost and time for experiments.
DOI: 10.1111/ffe.70535
Authors: Yiliyaer Tuerxunmaimaiti(Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Writing – original draft), Xuan Zhao(Data curation, Formal analysis, Methodology, Software, Writing – review & editing), Daxu Zhang(Conceptualization, Funding acquisition, Methodology, Supervision, Writing – review & editing), Pei‐Fu Zhang(Methodology, Writing – review & editing), Qi Zhao(Conceptualization, Formal analysis, Methodology, Supervision, Visualization, Writing – review & editing)
Language: English
Subfield: Building and Construction
Field: Engineering
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W7236463900
