SHEAR STRENGTH PREDICTION FOR RC BEAMS WITHOUT SHEAR REINFORCEMENT BY NEURAL NETWORK INCORPORATED WITH MECHANICAL INTERPRETATIONS

Abstract

 In this study, a shear strength prediction AI model was developed for RC beams without shear reinforcements. The AI model is a neural network model which contains explanatory variables derived from mechanical interpretation such as the neutral axis and also variables on details of tensile reinforcing bars. The AI model can predict experimental results which contains wide range of shear span to effective depth ratio with high accuracy rather than a conventional neural network model and empirical shear strength equations. The AI model can apply to new data set without dependence on the train data. Furthermore, the AI model, in which Young’s modulus of the FRP reinforcements was taken into consideration in explanatory variables, can predict shear strengths of concrete beams reinforced with FRP reinforcements correctly.