Importance of Eliminating Operational Effects in Indirect Monitoring of Bridges Under Real-Life Conditions

Abstract

This paper investigates the importance of eliminating operational effects in the indirect bridge monitoring using smartphones in the passing vehicles. The vibration of a moving vehicle is affected by various operational sources originated from vehicle and road surface. This study focuses on the importance of considering operational effects under real-life conditions, and proposes an inverse-filtering-based method using Mel-frequency cepstral analysis to detect bridge abnormalities. Towards elimination of operational effects, inverse filtering employs off-bridge vibrations to design a filter capable of removing vehicle and road features affecting drive-by vibrations. To account for the change in the operational parameters during the travel, a database of vibrations is recorded and later analyzed to classify different speed and road roughness levels, which are employed in the inverse filtering process. The inverse filtered vibrations are later used in a Mel-frequency cepstral analysis, leading to the calculation of an abnormality index, representing the change in the bridge state. The performance of the proposed method in eliminating operational effects under real-life conditions is evaluated considering different vehicle types and bridge systems. It was shown that without addressing operational effects in real-life cases, extracting bridge features from drive-by vibrations may not seem to be feasible. All the required data are recorded using the built-in accelerometer and GPS sensors of the smartphone without the need for extra instruments. In addition, this approach considers each data collection device within each vehicle separately, which makes it robust against device and vehicle features. As a result of the proposed methodology, it would be possible to monitor a large number of bridges using crowdsourced data collected from the passengers’ smartphones in future smart cities.