Spectral anomaly detection for identifying prestress loss in prestressed concrete bridges: The PSC-A16 case study

Abstract

The timely identification of damages in prestressed concrete bridges is a significant challenge for the structural health monitoring community, primarily when the malfunctions are related to the prestressing system. Despite extensive research, a shared solution for detecting tension loss in prestressed tendons is still lacking. This paper investigates the capability of Spectral Jump-Anomaly Detection (SJ-AD), a data-driven technique that directly analyzes accelerometric data in the frequency domain and emits alerts within a multi-window implicit thresholding scheme. Additionally, this work introduces the PSC-A16 Benchmark, a case study involving vibration monitoring of an Italian viaduct during strengthening interventions with external tendons, thus providing data at different prestressing levels. Evaluating SJ-AD on the PSC-A16 benchmark, this paper shows that the proposed method can successfully provide alerts related to tension losses that affect the bridge deck.