Title: Artificial Neural Network Model for Concrete Strength Predictions based on Ultrasonic Pulse Velocity Measurement

Author(s): Fayez Moutassem and Mohamad Kharseh

Publication: Materials Journal

Volume:

Issue:

Appears on pages(s):

Keywords: artificial neural network; compressive strength model; concrete; modeling; machine learning; UPV

DOI: 10.14359/51740776

Date: 5/1/2024

Abstract:
Accurately predicting the compressive strength of concrete is crucial in ‎various fields, including construction and engineering. This research paper proposes two mathematical models based on non-linear regression and Artificial Neural Networks (ANN) to predict the compressive strength of concrete accurately based on Ultrasonic Pulse Velocity (UPV) measurements. This paper outlines the proposed models’ formulation, calibration, evaluation, and validation. An experimental program was designed to calibrate and evaluate the models, and the analysis of the results reveals the robust fit of the proposed models to the experimental data. Both models exhibit exceptional accuracy, effectively predicting compressive strength values. The ANN and non-linear regression models attained high coefficients of determination of 0.993 and 0.992, respectively, demonstrating their reliability. Additionally, the standard errors of the ANN and non-linear regression models are 2.41 MPa and 2.52 MPa, respectively. Practical applications of these models extend to concrete characterization, enabling efficient quality control and structural integrity assessment.