Neural network prediction of pullout capacity of marquee ground anchors

Abstract

Marquees and other temporary light structures are connected to the ground by tensile anchors that resist uplift forces. The existing methods for predicting the pullout capacity of these anchors are inaccurate and incomplete. As a result, failures of such structures are not rare and have resulted in deaths and tens of thousands of dollars of damage. This paper aims to increase the safety of temporary light structures, such as marquees, by developing a more accurate pullout capacity prediction method based on artificial neural networks (ANNs). Two types of ANNs are examined, namely, multi-layer perceptrons (MLPs) that are trained with the back-propagation algorithm and B-spline neurofuzzy networks that are trained with the adaptive spline modeling of observation data (ASMOD) algorithm. In order to facilitate the use of the MLP model, it is made available in a tractable equation form. Predictions of pullout capacity from the developed ANN models are obtained and compared with values predicted by traditional methods currently used in practice. The results indicate that ANNs are able to predict accurately the pullout capacity of ground anchors and outperform the existing methods.