A Comparative Study of Data-Driven Models for Shear Strength Prediction of FRP-RC Beam Using Machine Learning Techniques

Nowadays machine learning techniques are effectively used as a means of resolving issues in civil and structural engineering. Accurately evaluating the shear strength of a reinforced concrete beam with fibre reinforced polymer (FRP) is crucial to ensure a secure design and effectively assess its performance. However, the accuracy of the predictions made by current shear models is generally constrained by the use of a limited database and complex parameters. The aim of this study is to create a model based on machine learning techniques that can predict the shear strength of reinforced concrete beams containing fibre reinforced polymer bars, both with and without stirrups, by utilizing data-driven approaches. A comprehensive database of 491 shear strength tests on FRP beams was collected from the public literature for developing framework’s training and testing sets. In order to prepare the data for machine learning algorithms, exploratory data analysis (EDA) has been carried out to investigate the correlation and identify collinearity between several independent parameters. Further, different models for linear regression, decision tree regression, random forest regression, gradient boost, and XGBoost have been developed for prediction of shear strength based on twelve different independent parameters and dependent output parameters. Root mean square error (RMSE), R2 score, and mean absolute error (MAE) are used to check the performance of all the models, and the best model is chosen for forecasting the shear strength of FRP reinforced concrete beam.