Reliable estimation of the confined compressive strength of fiber-reinforced polymer (FRP)-wrapped concrete is essential for the safe design and assessment of strengthened structural members. This study proposes an adaptive neuro-fuzzy inference system (ANFIS) model to predict the confined compressive strength of FRP-confined circular concrete cylinders. The model is trained using the Levenberg–Marquardt backpropagation algorithm, combined with an early-stopping strategy, to enhance generalization and prevent overfitting. Four physically meaningful parameters—unconfined compressive strength, cylinder diameter, FRP thickness, and FRP elastic modulus—are employed as input variables, while the confined compressive strength is taken as the output. A comprehensive database of 812 experimental results from the literature was compiled and used for model training, validation, and testing. The predictive capability of the proposed ANFIS framework was evaluated against five widely used analytical confinement models using statistical performance indicators. The developed model demonstrated superior predictive consistency and reduced scatter relative to existing confinement equations, indicating improved reliability across a broad range of strengths. The results confirm that the proposed ANFIS approach provides a stable and practical tool for estimating the confined compressive strength of FRP-wrapped concrete, supporting preliminary structural assessment and strengthening design applications.
Tawfik et al. (Sun,) studied this question.