Implementation of the use of artificial neural networks for the predictive calculation of concrete resistance, Trujillo (#488)

Authors

Cóndor Palomino, Jhordy Bryan

Huamán Sandoval, Clinton Steiner

Herrera Viloche, Alex Arquímedes

Noriega Vidal, Eduardo Manuel

Martell Ortiz, Juan Carlos

Valdiviezo Velarde, Alan Yordan

Abstract

The general objective of this research was the implementation of artificial neural networks to predict the strength of concrete. Specific objectives included evaluating the accuracy of these networks, determining their training progress, and root mean square error. The methodology adopted was of an applied type, with a descriptive cross-sectional non-experimental design. The general conclusion highlighted the successful implementation of artificial neural networks in the predictive calculation of concrete resistance. The architecture of the network, which consists of input, hidden and output layers, with 10, 19 and 1 neurons respectively, was detailed using Matlab. The specific conclusions indicated a high degree of accuracy of 99.997%, confirming the effectiveness of concrete strength prediction. In addition, a training progress of 100.00% was achieved, evidencing the ability of the neural network to adjust its internal parameters and learn to predict the resistance of the concrete with the data provided. With respect to the mean square error of the neural network, a coefficient of MSE = 1.5949 was obtained. Ultimately, it was concluded that the use of artificial neural networks is a valid approach to estimate the compressive strength of concrete. This research opens promising perspectives for its future application in concrete quality monitoring.