Abstract
Infrastructure databases have been widely used to obtain statistical deterioration models that assist decision makers in optimally allocating available budgets to maintain infrastructures in a functioning and at the same time safe state. Although infrastructure databases contain a plethora of information for the contained structure’s characteristics and structural condition, additional trusted data sources, should be used effectively to model environmental or other factors, which could affect deterioration rates. In most studies, the selection of deterioration factors is based on a small preselected set of factors or based on expert judgment. In this work, Artificial Neural Networks (ANN) and pattern recognition are used to capture structural deterioration information for bridges of the US National Bridge Inventory (NBI). A data-driven framework including Genetic Algorithms and ANNs, successfully utilized in other research fields, is proposed here in for simultaneously optimizing the architecture of an ANN and performing an unbiased variable selection process. The framework is applied to a large-scale bridge database and an initial selection of 52 variables. The results of the application are assessed for their accuracy and efficiency and show promising prospect for using GA-ANN pattern recognition for capturing bridge deterioration information.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call