Rapid prediction of deflections in multi-span continuous composite bridges using neural networks

Abstract

This paper proposes closed form expressions for the rapid prediction of deflections in steel-concrete composite bridges of large number of spans subjected to service load. The proposed expressions take into account shear lag effect, flexibility of shear connectors and cracking in concrete slabs. Three separate neural networks have been developed for right exterior span, left exterior span and interior spans. The closed form expressions have been obtained from the neural networks developed in the study. The training, validating and testing data sets for the neural networks are generated using finite element software ABAQUS. The proposed expressions have been validated for number of bridges and the errors are found to be small for practical purposes. Sensitivity studies have been carried out using the proposed expressions to evaluate the suitability of input parameters. The use of the proposed expressions requires a computational effort that is fraction of that required for the finite element analysis, therefore, can be used for rapid prediction of deflection for everyday design.