Data driven strength and strain enhancement model for FRP confined concrete using Bayesian optimization. (July 2022)

• Record Type: Journal Article
• Authors: Du, Junsong; Ma, Haixing; Sun, Di; Pan, Pengfei
• Is Part Of: Structures. Volume 41(2022)
• Page Start: 1345
• Page End: 1358
• Publication Date: 2022-07
• Subjects: FRP confinement -- Bayesian -- Hyperparameter optimization -- Machine learning; Structural engineering -- Periodicals; 624.1
• DOI: 10.1016/j.istruc.2022.05.093
• Languages: English
• ISSNs: 2352-0124

Abstract

Abstract: To ensure the safe design of composite system, a precise design-based model for concrete encased in fibre-reinforced polymer (FRP) is required. An enclosure data driven model is developed in this paper to forecast the ultimate state of FRP-confined concrete using Bayesian hyperparameter optimization. For a database of 820 columns with circular cross-section, the suggested model's predictions are compared to those of six empirical models and a non-optimized machine learning regressor of XGBoost. In comparison to the previous models, statistical analysis demonstrates that the improved model (BO-XGB) more precisely forecasts the compressive strength and axial strain of concrete confined with FRP. The results further reveal that for strength and strain increase with lateral confinement, BO-XGB with weight uncertainty offers stable predictions, whereas the other models display turbulent model mistake. In comparison to previous models, the suggested model's excellent accuracy and stability are attributable to its great flexibility and resilience in capturing the impacts of lateral confinement pressure as an interaction between the concrete core and FRP shell.