Continuous bridge sensing and monitoring is an important component of smart infrastructure. Traditional bridge monitoring techniques require sensors to be installed on bridges, which is costly and time consuming. Also, a certain set of sensors have to be used to monitor a single bridge at a time. In order to resolve these issues, a novel bridge damage detection method focusing on monitoring a population of bridges simultaneously utilizing crowdsourcing data collected from smartphones on passing-by vehicles is developed. In this method, Mel-frequency cepstral coefficients (MFCCs) are first extracted on the acceleration data collected from smartphones in all the vehicles within a certain period. Principal component analysis (PCA) is used to transform the features so that they are linearly uncorrelated. The damage is then identified by comparing the distributions of these transformed features. The results from lab experiments show that the approach not only identifies the existence of the damage, but also provides useful information about severity.
展开▼
