Modeling deflection basin using artificial neural networks with cross-validation technique in backcalculating flexible pavement layer moduli

• Published in: Advances in engineering software (1992) Vol. 39; no. 7; pp. 588 - 592
• Main Authors: Saltan, Mehmet, Terzi˙, Serdal
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.07.2008
• Subjects: Artificial neural networks; Backcalculation; Cross validation; Deflection; Flexible pavements; Non-destructive testing; Flexible pavements; Backcalculation; Cross validation; Non-destructive testing; Deflection; Artificial neural networks
• ISSN: 0965-9978
• DOI: 10.1016/j.advengsoft.2007.06.002

Through the new technological developments, for highway maintenance engineering the structural capacity of pavement is to be determined using non-destructive techniques. Up to now various methodologies have been applied based on the surface deflection bowl obtained under either a known moving wheel load or devices such as falling weight deflectometer. Backcalculating pavement layer moduli are well-accepted procedures in the evaluation of the structural capacity of pavements. The ultimate aim of the backcalculation process from non-destructive testing (NDT) results is to estimate the pavement material properties. Using backcalculation analysis, in situ material properties can be backcalculated by the measured field data for appropriate analysis techniques. To backcalculate reliable moduli, the deflection basin must be modeled more realistically. Here, in this study, the deflection basins measured on the surface of the flexible pavements are modeled using artificial neural networks (ANN) with cross-validation technique. Distances between transducers can be varied with different producer companies. The distances between transducer are used for the form deflection basin. Layer thickness and distance to loading center are used as input in the present study. Limited experimental deflection data groups from NDT are used to show the capability of the neural network technique in modeling the deflection bowl. Since enough data are not available to construct a reliable neural network, a methodology based on the cross-validation technique can be used. The results show that the proposed methodology give the deflection bowl satisfied accuracy.