Machine learning-driven evaluation and optimisation of compression yielded FRP-reinforced concrete beam with T section. (15th January 2023)

• Record Type: Journal Article
• Authors: Guo, Bingcheng; Lin, Xiaoshan; Wu, Yufei; Zhang, Lihai
• Is Part Of: Engineering structures. Volume 275(2023)Part A
• Publication Date: 2023-01-15
• Subjects: Compression yielding -- T section -- Moment capacity -- Ductility -- Machine learning -- Optimization; Structural engineering -- Periodicals; Structural analysis (Engineering) -- Periodicals; Construction, Technique de la -- Périodiques; Génie parasismique -- Périodiques; Pression du vent -- Périodiques; Earthquake engineering; Structural engineering; Wind-pressure; Periodicals; 624.105
• DOI: 10.1016/j.engstruct.2022.115240
• Languages: English
• ISSNs: 0141-0296

Abstract

Highlights: Predict performance of compression yielded beam with T section by machine learning. Various activation functions are tested for accurate prediction of moment capacity. Support vector machine with combined kernel functions for section effectiveness. Performance-based optimisation approach for compression yielded beam is proposed. Abstract: Fibre reinforced polymer (FRP)-reinforced concrete beams usually encounter brittle failure due to the linear behaviour of FRP. To solve this issue, compression yielding (CY) concept was proposed recently to improve the ductility of FRP-reinforced concrete beams. However, because of the complexity of the compression yielding mechanism, the calculation of flexural capacity and ductility of FRP-reinforced concrete beam with CY block (CY beam) is challenging, especially for the CY beam with T section due to the lack of closed form solution. In this study, an integrated model is proposed based on machine learning method to evaluate the moment capacity and ductility of the CY beam with T section. To improve the prediction accuracy of the artificial neural network model for moment capacity, different activation functions are tested. The section effectiveness of CY beam with T section is evaluated using the proposed support vector machine model, which combines different kernel functions. Gaussian process regression is then employed to predict the ductility of CY beam with T section. It demonstrates that the developed integrated model can produce highly accurate predictions. In addition, a genetic algorithm (GA) is developed for identifying optimal CY beam section design solutions. Finally, the robustness of the model in the optimisation design for the CY beams with rectangular section and T section is demonstrated by using two numerical examples.