A machine‐learning‐based constitutive bond‐slip model for anchored CFRP strips externally bonded on concrete members

Abstract Strengthening is often required for reinforced concrete, steel, and masonry structures or structural elements when they possess insufficient performance against external loads such as earthquakes. Recently, the use of carbon fiber‐reinforced polymers (CFRP) has been considered a viable strengthening technique alternative to traditional methods. The major concern is premature debonding failure hindering the efficient use of CFRP systems. FRP anchor systems have been used to avoid this phenomenon. This paper employs a machine learning (ML)‐based algorithm (support vector regression) to propose predictive models to simulate the bond‐slip behavior of anchored CFRP strips externally bonded to the concrete surface. A comprehensive database was constructed using the previous reports on the bond‐slip behavior of FRP‐to‐concrete joints anchored with CFRP strips. Afterwards, the collected database was used to train and validate the proposed models. The input parameters cover all possible factors, that is, compressive strength of concrete, width of concrete block, anchor hole diameter, anchor hole depth, number of anchors at one row, number of anchors at one column; elastic modulus, width, bonding length, and thickness of CFRP strip. The output parameters are maximum shear capacity, residual shear capacity, displacement values at peak shear, and residual shear. Results imply that the proposed models have high prediction accuracies with low error rates. Proposed models are also presented in code format to be easily incorporated into analysis software for practical use.