Prediction of ultimate strength and strain in FRP wrapped oval shaped concrete columns using machine learning

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: March 28, 2025

The accurate prediction of the strength enhancement ratio ([Formula: see text]) and strain (εcc/εco) in FRP-wrapped elliptical concrete columns is crucial for optimizing structural performance. This study employs machine learning (ML) techniques to enhance accuracy reliability. A dataset 181 samples, derived from experimental studies finite element modeling, was utilized, with a 70:30 train-test split (127 training samples 54 testing samples). Four ML models: Decision Tree (DT), Adaptive Boosting (ADB), Stochastic Gradient (SGB), Extreme (XGB) were trained optimized using Bayesian Optimization refine their hyperparameters improve performance.Results demonstrate that SGB achieved best performance predicting [Formula: text], an R2 0.850, lowest RMSE (0.190), highest generalization capability, making it most reliable model predictions. For (εcc/εco), XGB outperformed other models, achieving 0.779 (2.162), indicating better balance between accuracy, generalization, minimal overfitting. DT ADB exhibited lower predictive performance, higher residual errors capacity. Furthermore, Shapley Additive exPlanations analysis identified FRP thickness-elastic modulus product (tf × Ef) compressive as influential features impacting both ratios. To facilitate real-world applications, interactive graphical user interface developed, enabling engineers input ten parameters obtain real-time

Language: Английский

Concrete compressive strength classification using hybrid machine learning models and interactive GUI

Innovative Infrastructure Solutions, Journal Year: 2025, Volume and Issue: 10(5)

Published: April 28, 2025

Abstract Concrete Compressive Strength (CCS) is a critical parameter in structural engineering, influencing durability, safety, and load-bearing capacity. This study explores the classification of CCS using hybrid Machine Learning (ML) techniques an interactive Graphical User Interface (GUI). Advanced ML algorithms: Random Forest (RF), Adaptive-Boosting (AdaBoost), Extreme-Gradient-Boosting (XGBoost), Light-Gradient Boosting (LightGBM), Categorical-Boosting (CatBoost) were applied to categorize strength into Low, Normal, High classes. The dataset, comprising 1298 samples, was split 80% training 20% testing for evaluation. Hyperparameter tuning Bayesian Optimization with fivefold stratified cross-validation, resulting greatly improved model’s performance. Results showed that LightGBM achieved highest accuracy, scores 0.931 (Low), 0.865 (Normal), 0.935 (High), corresponding area under curve values 0.967, 0.938, 0.981. CatBoost also performed well, particularly Normal classes, while XGBoost slight overfitting class. RF AdaBoost had acceptable performance but struggled boundary cases. To interpret model predictions, SHapley-Additive-exPlanations (SHAP) analysis used. Curing duration cement content most influential factors across all water superplasticizer played secondary roles. Coarse aggregate became more significant High-Strength (HSC). A GUI developed allow practitioners input data receive real-time classifications, bridging gap between machine learning practical applications concrete design.

Language: Английский