National University Transportation Consortium: a Speaker Recognition Based Damage Detection

Download Final Report

CAIT project no.: CAIT-UTC-NC47

Fiscal Year: 2015/2016

Status: Final

Rutgers-CAIT Author(s): Raimondo Betti, Ph.D., Patrick Szary, Ph.D.

External Author(s): Dyab A. Khazem

Sponsor(s): USDOT-FHWA, Parsons Transportation Group


This project has to be seen as a part of a major research project that the PI has been trying to put together over the past couple of years, with researchers at Harbin Institute of Technology (China) and at Rutgers University.  The focal idea is the use of dense arrays of small miniature sensors in assessing the damage of highway concrete/steel bridges.  Small solar-powered smart sensors that can be placed inside refractors in the road pavement have been developed at Harbin Institute of Technology. These sensors can be easily mounted on a bridge deck and record large amount of vibration data, e.g. accelerations, during the regular operation of the bridge.  Because of their small dimensions, it is possible to install tens of those sensors on a normal highway bridge deck and this will allow us to collect time histories of the bridge response at locations that are very close to each other.  Using the results of the previous UTC project, characteristic features will be extracted from these time-histories and various types of correlation will be extracted from them.  Studying the pattern of such correlation and their variation with structural damage will allow us to assess the presence of certain levels of damage in the bridge and provide indication for immediate action and repair, if needed.  Once the different methodologies have been studied and tested, the ultimate step will be to develop a prototype that can be tested in the BEAST facility at Rutgers University:  here, it is expected to install the small sensor dense array on one of the bridge deck specimens and conduct extensive tests considering different environmental conditions and damage levels.