Pull-off adhesion prediction of variable thick overlay to the substrate.

Non-destructive identification of the pull-off adhesion of a concrete substrate to an overlay mortar with variable thickness using artificial neural networks (ANNs) is studied in this paper. Selected ANNs with various training algorithms were tested on the basis of the parameter which describes the thickness of the overlay and also the parameters specified experimentally using non-destructive testing (NDT) methods. Real world data collected from experiments of pull-off adhesion were used for building our learner models. The tests were carried out in the same place where tests using NDT methods were performed. Three variant analyses of the possibility of such identification were conducted. The variance was calculated for these testing methods and parameters obtained with their usage, without considering the parameter that describes the thickness of the overlay in this work. [ABSTRACT FROM AUTHOR]