Saberi, HosseinHosseinSaberiSaberi, HamidHamidSaberi2026-09-172026-09-172025-10-10Cleaner Engineering and Technology 29: 101090 (2025)https://hdl.handle.net/11420/64867Modeling the compressive behavior of fiber-reinforced polymer (FRP)-confined recycled aggregate concrete (RAC) is essential for practical engineering applications. Existing models often overlook the nonlinear effects of recycled aggregate content on concrete strength, limiting their accuracy. To address this gap and promote sustainable construction, this study proposes a novel approach for predicting the stress-strain behavior of FRP-confined RAC under compression. Tاhe method integrates clustering techniques and singular value decomposition (SVD) to extract nonlinear relationships between key system parameters and stress-strain curves. The least squares method is then used to optimize unknown system parameters. A dataset comprising 81 stress-strain curves from eight references, totaling 2452 digitized data points at a strain rate of 0.0005, was used to train the model. The proposed approach is validated against experimental results, demonstrating high accuracy in capturing the mechanical behavior of FRP-confined RAC. These findings provide a more reliable predictive tool for structural engineers and contribute to the advancement of sustainable concrete technologies.de#PLACEHOLDER_PARENT_METADATA_VALUE#2025Elsevierhttps://creativecommons.org/licenses/by/4.0/FRP-confinedMachine learningRecycled aggregate concreteStress-strain relationshipsSVD decompositionTechnology::624: Civil Engineering, Environmental Engineering::624.1: Structural Engineering::624.17: Structural Analysis and DesignJournal Article10.1016/j.clet.2025.10109010.15480/882.18434