There is a growing attention in real-time bridge condition assessment using data from drive-by vehicles as a potentially scalable approach. Most system identification methods are based on synchronized vibration data collection for this purpose. This study presents an approach for bridge modal identification that estimates high-resolution absolute value of the operational mode shapes using asynchronous mobile data. With each trip of a vehicular sensor, the spatio-temporal response of the bridge is sampled, along with various sources of noise, e.g. vehicle dynamics, environmental effects, road profile, etc. The crowdsourced modal identification using continuous wavelet (CMICW) method is proposed that gradually magnifies the bridge dynamical signatures and mitigates noise over the spatio-temporal map. The performance of the CMICW method is validated in an experimental setting. The method successfully identifies natural frequencies and absolute value of the operational mode shapes of a bridge with high resolution and accuracy. Notably, by including data collected from various bridge lanes, the method can reconstruct 3D representation of the mode shapes. The influence of the speed of the mobile sensors on the accuracy of the estimated modal properties is investigated as well. Using a hybrid simulation framework, the effect of vehicle dynamic is included in the mobile sensing data. The study shows that the CMICW method is successful in discounting the effect of vehicle dynamics, thereby strengthening the bridge modal information. Finally, a blind source separation technique is implemented to separate the effects of road irregularities, which further improves the accuracy of modal property estimates. This study contributes to the growing body of knowledge on mobile crowdsensing for physical properties of transportation infrastructure.

No abstract

This study proposes a learning-based method with domain adaptability for input estimation of vehicle suspension systems. In a crowdsensing setting for bridge health monitoring, vehicles carry sensors to collect samples of the bridge's dynamic response. The primary challenge is in preprocessing; signals are highly contaminated from road profile roughness and vehicle suspension dynamics. Additionally, signals are collected from a diverse set of vehicles vitiating model-based approaches. In our data-driven approach, two autoencoders for the cabin signal and the tire-level signal are constrained to force the separation of the tire-level input from the suspension system in the latent state representation. From the extracted features, we estimate the tire-level signal and determine the vehicle class with high accuracy (98% classification accuracy). Compared to existing solutions for the vehicle suspension deconvolution problem, we show that the proposed methodology is robust to vehicle dynamic variations and suspension system nonlinearity.

This study proposes a learning-based method with domain adaptability for input estimation of vehicle suspension systems. In a crowdsensing setting for bridge health monitoring, vehicles carry sensors to collect samples of the bridge's dynamic response. The primary challenge is in preprocessing; signals are highly contaminated from road profile roughness and vehicle suspension dynamics. Additionally, signals are collected from a diverse set of vehicles vitiating model-based approaches. In our data-driven approach, two autoencoders for the cabin signal and the tire-level signal are constrained to force the separation of the tire-level input from the suspension system in the latent state representation. From the extracted features, we estimate the tire-level signal and determine the vehicle class with high accuracy (98% classification accuracy). Compared to existing solutions for the vehicle suspension deconvolution problem, we show that the proposed methodology is robust to vehicle dynamic variations and suspension system nonlinearity.