Vehicle-based autonomous modal analysis for enhanced bridge health monitoring

Indirect health monitoring is an emerging concept in bridge engineering, aimed at identifying the modal parameters of bridges, including their natural frequencies, by utilizing acceleration data from passing vehicles. However, this approach faces significant challenges due to noise from road surface conditions and other environmental disturbances, which complicate the accurate identification of natural frequencies. The primary effect of noise is the introduction of spurious mathematical modes in the stabilization diagram, making the identification process more difficult. This paper presents a novel, fully autonomous approach to address these challenges, leveraging subspace state-space system identification within the framework of autonomous operational modal analysis. Initially, vehicle acceleration data are processed using a subspace algorithm that incorporates QR decomposition of the projected Hankel matrix. Modal parameters are identified across various model orders, and a multi-clustering algorithm is employed to filter out non-physical poles from the stabilization diagram. The key contributions of this work are threefold: (1) the development of a robust subspace framework that autonomously eliminates spurious poles from the stabilization diagram using QR decomposition, thereby improving the accuracy and interpretability of modal analysis; (2) the validation of this framework through a combination of numerical simulations and experimental data; and (3) the establishment of a foundation for future innovations in structural health monitoring for bridge infrastructure.