Bài báo khoa học

The behavior of concrete columns reinforced with fiber reinforced polymer (FRP) bars is different from conventional reinforced concrete columns due to the mechanical properties of FRP bars. This study develops a novel machine learning (ML) model, namely gradient boosting regression tree (GBRT), for efficiently predicting the axial load-bearing capacity (ALC) of concrete columns reinforced with FRP bars. A data base containing 283 experimental results is collected to develop the ML model. Seven codebased and empirical-based equations are also included in comparison with the developed ML models. Moreover, we also propose a multiple linear regression (MLR)-based formula for calculating the ALC of the FRP-concrete column. The performance results of GBRT model are compared with those of published formulas and the proposed MLRbased formula. Statistical properties including 𝑅2, 𝑅𝑀𝑆𝐸, and 𝑎20 − 𝑖𝑛𝑑𝑒𝑥 are calculated to evaluate the accuracy of those predictive models. The comparisons demonstrate that GBRT outperforms other models with very high 𝑅2 values and small 𝑅𝑀𝑆𝐸. Moreover, the influence of input parameters on the predicted ALC isevaluated. Finally, an efficient graphical user interface tool is developed to simplify the practical design process of FRP-concrete columns.