Adaptive machine learning framework: Predicting UHPC performance from data to modelling

Ultra-High Performance Concrete (UHPC) is vital for next-generation infrastructure, necessitating complex interaction modeling beyond empirical methods. This study proposes an interpretable machine learning (ML) framework to predict the compressive strength (CS) of UHPC and analyze input variable in...

Full description

Saved in:
Bibliographic Details
Main Authors: Yinzhang He, Shaojie Gao, Yan Li, Yongsheng Guan, Jiupeng Zhang, Dongliang Hu
Format: Article
Language:English
Published: Elsevier 2025-09-01
Series:Results in Engineering
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2590123025027914
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849341034510680064
author Yinzhang He
Shaojie Gao
Yan Li
Yongsheng Guan
Jiupeng Zhang
Dongliang Hu
author_facet Yinzhang He
Shaojie Gao
Yan Li
Yongsheng Guan
Jiupeng Zhang
Dongliang Hu
author_sort Yinzhang He
collection DOAJ
description Ultra-High Performance Concrete (UHPC) is vital for next-generation infrastructure, necessitating complex interaction modeling beyond empirical methods. This study proposes an interpretable machine learning (ML) framework to predict the compressive strength (CS) of UHPC and analyze input variable influences. The framework has several key modules: data preprocessing, feature selection, outlier detection, model training, hyperparameter optimization, and model interpretation. First, a CS dataset of 924 samples with 20 input features was constructed. Outliers were removed using the Isolation Forest algorithm based on binary search trees. Four feature subsets (FSs) were generated via F-score and mutual information score analyses, then used to train Ridge Regression, SVR, RF, GBDT, XGBoost, and LightGBM models. Bayesian optimization fine-tuned each model's hyperparameters to identify the optimal model-FS combination. Finally, SHAP interpreted input feature contributions for the best model. Results showed outlier detection reduced extreme values, improving data distribution. LightGBM demonstrated the most stable performance among all models. As the number of features decreased, model performance initially increased then decreased, peaking at FS_14 (test set: R2 = 0.9677, MAE = 4.4621 MPa, RMSE = 6.9226 MPa), showing the mutual information scoring method can improve the model performance by reducing redundant information. SHAP analysis identified the six most influential features on CS in FS_14: age, silica fume, steel fiber, steel fiber diameter, cement, and polycarboxylate superplasticizer (in order). As CS increases, the relative contributions of constituent materials increase, while the contribution of age decreases. Overall, the proposed framework enhances UHPC prediction accuracy and generalization, advancing engineering application.
format Article
id doaj-art-bc9b91c5ca9a4aeca4b10d4bbfc83b84
institution Kabale University
issn 2590-1230
language English
publishDate 2025-09-01
publisher Elsevier
record_format Article
series Results in Engineering
spelling doaj-art-bc9b91c5ca9a4aeca4b10d4bbfc83b842025-08-20T03:43:44ZengElsevierResults in Engineering2590-12302025-09-012710672410.1016/j.rineng.2025.106724Adaptive machine learning framework: Predicting UHPC performance from data to modellingYinzhang He0Shaojie Gao1Yan Li2Yongsheng Guan3Jiupeng Zhang4Dongliang Hu5School of Highway, Chang’an University, Xi’an 710064, China; The Key Laboratory of lntelligent Construction and Maintenance of CAAC, Xi'an 710064, ChinaSchool of Highway, Chang’an University, Xi’an 710064, China; The Key Laboratory of lntelligent Construction and Maintenance of CAAC, Xi'an 710064, ChinaThe Key Laboratory of lntelligent Construction and Maintenance of CAAC, Xi'an 710064, China; School of Civil Aviation, Northwestern Polytechnical University, Xi'an 710072, ChinaJiangsu Sinoroad Engineering Research Institute Co., Ltd., Nanjing 211805, China; Corresponding authors:School of Highway, Chang’an University, Xi’an 710064, China; The Key Laboratory of lntelligent Construction and Maintenance of CAAC, Xi'an 710064, ChinaSchool of Highway, Chang’an University, Xi’an 710064, China; The Key Laboratory of lntelligent Construction and Maintenance of CAAC, Xi'an 710064, China; Corresponding authors:Ultra-High Performance Concrete (UHPC) is vital for next-generation infrastructure, necessitating complex interaction modeling beyond empirical methods. This study proposes an interpretable machine learning (ML) framework to predict the compressive strength (CS) of UHPC and analyze input variable influences. The framework has several key modules: data preprocessing, feature selection, outlier detection, model training, hyperparameter optimization, and model interpretation. First, a CS dataset of 924 samples with 20 input features was constructed. Outliers were removed using the Isolation Forest algorithm based on binary search trees. Four feature subsets (FSs) were generated via F-score and mutual information score analyses, then used to train Ridge Regression, SVR, RF, GBDT, XGBoost, and LightGBM models. Bayesian optimization fine-tuned each model's hyperparameters to identify the optimal model-FS combination. Finally, SHAP interpreted input feature contributions for the best model. Results showed outlier detection reduced extreme values, improving data distribution. LightGBM demonstrated the most stable performance among all models. As the number of features decreased, model performance initially increased then decreased, peaking at FS_14 (test set: R2 = 0.9677, MAE = 4.4621 MPa, RMSE = 6.9226 MPa), showing the mutual information scoring method can improve the model performance by reducing redundant information. SHAP analysis identified the six most influential features on CS in FS_14: age, silica fume, steel fiber, steel fiber diameter, cement, and polycarboxylate superplasticizer (in order). As CS increases, the relative contributions of constituent materials increase, while the contribution of age decreases. Overall, the proposed framework enhances UHPC prediction accuracy and generalization, advancing engineering application.http://www.sciencedirect.com/science/article/pii/S2590123025027914Ultra-High Performance Concrete (UHPC)Compressive strengthMachine learning (ML)LightGBMSHapley Additional explanation (SHAP)
spellingShingle Yinzhang He
Shaojie Gao
Yan Li
Yongsheng Guan
Jiupeng Zhang
Dongliang Hu
Adaptive machine learning framework: Predicting UHPC performance from data to modelling
Results in Engineering
Ultra-High Performance Concrete (UHPC)
Compressive strength
Machine learning (ML)
LightGBM
SHapley Additional explanation (SHAP)
title Adaptive machine learning framework: Predicting UHPC performance from data to modelling
title_full Adaptive machine learning framework: Predicting UHPC performance from data to modelling
title_fullStr Adaptive machine learning framework: Predicting UHPC performance from data to modelling
title_full_unstemmed Adaptive machine learning framework: Predicting UHPC performance from data to modelling
title_short Adaptive machine learning framework: Predicting UHPC performance from data to modelling
title_sort adaptive machine learning framework predicting uhpc performance from data to modelling
topic Ultra-High Performance Concrete (UHPC)
Compressive strength
Machine learning (ML)
LightGBM
SHapley Additional explanation (SHAP)
url http://www.sciencedirect.com/science/article/pii/S2590123025027914
work_keys_str_mv AT yinzhanghe adaptivemachinelearningframeworkpredictinguhpcperformancefromdatatomodelling
AT shaojiegao adaptivemachinelearningframeworkpredictinguhpcperformancefromdatatomodelling
AT yanli adaptivemachinelearningframeworkpredictinguhpcperformancefromdatatomodelling
AT yongshengguan adaptivemachinelearningframeworkpredictinguhpcperformancefromdatatomodelling
AT jiupengzhang adaptivemachinelearningframeworkpredictinguhpcperformancefromdatatomodelling
AT donglianghu adaptivemachinelearningframeworkpredictinguhpcperformancefromdatatomodelling