Load capacity prediction of simply supported CFRP deep beams using artificial neural networks.

Anticipating the load carrying capacity of deep beams with the aim of controlling the predominant shear failure led to the consideration of the behaviour of deep beams, which is dissimilar from the behaviour of normal beams. Therefore, the usual design methods are not safe due to the fact that theoretical stresses are less than actual stresses. As a result, special design methods based on nonlinear stress distribution are used, which artificial neural network (ANN) model to predict the load capacity of simply supported deep beams reinforced with CFRP. Thorough experimental results have been compiled from several previous studies to build up a database that can be used to train and test the developed ANN model. Different input neurons were adopted including the shear span to depth ratio (a/d), the effective concrete compressive strength of concrete (fc′), the cross-sectional area of the diagonal strut (Astrut) and the sin angle of the inclined strut (sinɸ). On the other hand, the ANN output was the ultimate load of simply supported deep beams. Parametric research is also carried out to determine the extent to which certain parameters employed in the created neural network model have an impact on the load capacity. The results showed that ANN can reasonably estimate the load capacity of simply supported CFRP deep beams reinforced with CFRP. occur even in the elastic load range. This research mainly focuses on developing a multi-layer backpropagation. [ABSTRACT FROM AUTHOR]