Neural network-aided prediction of post-cracking tensile strength of fibre-reinforced concrete

Structural fibres are an effective method to improve concrete post-cracking tensile strength (fctR). Currently, the characterization of this property is mainly performed experimentally. This a source uncertainties at design stages, which hinders development new fibre type and/or optimization those currently existing. paper presents multilayer perceptron neural network predict fctR fibre-reinforced (FRC) subjected Barcelona Test. The optimal architecture predictor obtained by evaluating 9216 configurations input dimension and number hidden layers neurons. generalization performance assessed using repeated random sub-sampling validation with 50 iterations. final model can high accuracy FRC for different cracking stages. A parametric analysis prove coherence between results predicted established understanding behaviour. Finally, numerical expressions provided as alternative tool traditional testing residual Test pre-design quality control purposes based on dosage, strength, specimen height geometric characteristics. These approaches found be necessary boosting technology.