AI-based vibration monitoring for structural damage detection in bridges and roads

Abstract

Timely detection of structural damage in civil transportation infrastructure is critical to protect public safety, maintain serviceability, and extend the service life of structural systems. Conventional manual inspection is unsuitable for large, spatially distributed systems such as bridge superstructures and road networks; procedures are labor-intensive, require traffic control, and full coverage is rarely achievable. Structural Health Monitoring (SHM) offers a data-driven alternative, using in situ measurements to localize and quantify deterioration in such structures. In particular, vibration-based monitoring uses sensor measurements of ambient and operational responses to extract dynamic parameters that serve as damage-sensitive indicators. Artificial Intelligence (AI) models help to transform sensor measurements and their derived dynamic parameters into decision-ready outputs of damage presence, location, and severity. Despite these advances, notable gaps in practice remain: for bridge-beam components, the literature has essentially treated localization and quantification as separate problems, and end-to-end models that deliver both simultaneously are scarce. For roads, robust multi-defect classification remains limited, and integrated frameworks that carry network-level condition assessment through segment-level repair scoping with transparent, unit-rate cost estimates are not yet well established.