ABSTRACT
Bridge damage detection (BDD) techniques have been widely explored over the last years driven by the advances of computational intelligence technologies. Drive-by monitoring approaches, using sensors in a vehicle, provide a promising framework for BDD. In this context, the aim of this paper is to present a novel automated methodology for damage detection based on the extraction of features from the vibration responses measured on a passing vehicle using filter-type feature selection scheme. Feature extraction is performed in the time domain and then feature selection (FS) is applied to increase the relevance of the feature set by choosing the most important features based on their ability to discriminate between different classes of data. Finally, a cascade forward neural network (CFNN) is used to classify damage severity. The method was tested using a finite element (FE) model of concrete beam-and-slab bridge by considering vehicle bridge interaction (VBI). Results show that when the Pearson correlation coefficient (PCC) is used as a feature selection algorithm and an CFNN as a classifier, 94% accuracy can be achieved for identifying the level of cracking in bridge.
