Predictive modeling of the mechanical properties of concrete reinforced with steel fiber using artificial neural networks

Abstract

The aim of this paper was to estimate the mechanical properties essential to the design of concrete structures through a reliable prediction model of the compressive, tensile and flexural strengths of concrete steel fiber reinforced concrete (CFRC) using Artificial Neural Networks (ANN), and also to evaluate the influence of the fiber content on these properties. The study used a database with 57 experimental studies from the literature, implementing a neural network model with 12 input variables, 1 output and 2 hidden layers with 16 neurons. The results obtained were a mean square error (MSE) of 22.63, 0.08 and 0.80, and a mean absolute error (MAE) of 3.64, 0.24 and 0.74, respectively, for the compressive, tensile and flexural strengths. The sensitivity analysis showed that there was a considerable increase in tensile and flexural strengths with the use of fibers, which was expected. The results confirmed the model's ability to reliably reproduce the mechanical properties of the CFRC.

Keywords:
Steel fiber reinforced concrete; Mechanical properties; RNA; Sensitivity analysis; Dosage