초록 close

The infrastructure system in the United States has been aging faster than the resourceavailable to restore them. Therefore decision for allocating the resources is based in part on the conditionof the structural system. This paper proposes to use neural network to predict the overall rating of thestructural system because of the successful applications of neural network to other fields which require aymptom-diagnostictype relationship. The goal of this paper is to illustrate the potential of using neuralnetwork in civil engineering applications and, particularly, in bridge evaluations. Data collected by theTenessee Department of Transportation were used as est bedfor the study. Multi-layer feed forwardnetworks were developed using the Levenberg-Marquardt training algorithm. All the neural networksconsisted of at least one hidden layer of neurons. Hyperbolic tangent transfer functions were used in thefirst hidden layer and log-sigmoid transfer functions were used in the subsequent hidden and outputlayers. The best performing neural network consisted of three hidden layers. This network contained threneurons in the first hidden layer, two neurons in the second hidden layer and one neuron in the thirdhidden layer. The neural network performed well based on a target eror of 10%. The results of this studyindicate that the potential for using neural networks for the evaluation of infrastructure systems is very good.