Journal Published Online: 26 April 2022
Volume 11, Issue 1

A Comparative Study of LASSO and ANN Regressions for the Prediction of the Direct Tensile Behavior of UHPFRC

CODEN: ACEMF9

Abstract

Direct tensile behavior is one of the most relevant properties of ultrahigh-performance fiber-reinforced concrete (UHPFRC). However, the determination of this behavior implies the realization of complex tests that must be carried out by experienced personnel because small variations could invalidate the results. This research purpose was to develop and compare two different algorithmic approaches for the direct tensile behavior of UHPFRC. Thereby, the cracking tensile strength σcc, the tensile postcracking capacity σpc and its associated strains, εcc, and εpc, were modeled by means of least absolute shrinkage and selection operator (LASSO) regression and artificial neural networks (ANN). Four ANN models, one for each response, and four LASSO models were created to produce the most accurate approaches. Results showed the reliability of the developed models through statistical indices such as the root of the mean squared error (RMSE), mean absolute error, normalized mean bias error, the ratio of the RMSE to the standard deviation of measured data, coefficient of efficiency, and coefficient of determination (R2). Besides, the analytical research also showed that the highest accuracy belongs to ANN models, with R2 values of .922, .807, .901, and .858, in forecasting the features of direct tensile behavior of UHPFRC (σcc, εcc, σpc, and εpc).

Author Information

Abellán-García, Joaquín
Department of Civil and Environmental Engineering, Barranquilla, Colombia
Ortega-Guzmán, Juan J.
Escuela Colombiana de Ingeniería Julio Garavito, Bogotá, Colombia
Chaparro-Ruiz, Diego A.
Escuela Colombiana de Ingeniería Julio Garavito, Bogotá, Colombia
García-Castaño, Eliana
Escuela Colombiana de Ingeniería Julio Garavito, Bogotá, Colombia
Pages: 28
Price: $25.00
Related
Reprints and Permissions
Reprints and copyright permissions can be requested through the
Copyright Clearance Center
Details
Stock #: ACEM20210101
ISSN: 2379-1357
DOI: 10.1520/ACEM20210101