Anomaly detection with the Switching Kalman Filter for structural health monitoring

Abstract

Detecting changes in structural behaviour, i.e. anomalies over time is an important aspect in structural safety analysis. The amount of data collected from civil structures keeps expanding over years while there is a lack of data-interpretation methodology capable of reliably detecting anomalies without being adversely affected by false alarms. This paper proposes an anomaly detection method that combines the existing Bayesian Dynamic Linear Models framework with the Switching Kalman Filter theory. The potential of the new method is illustrated on the displacement data recorded on a dam in Canada. The results show that the approach succeeded in capturing the anomalies caused by refection work without triggering any false alarms. It also provided the specific information about the dam's health and conditions. This anomaly detection method offers an effective data-analysis tool for Structural Health Monitoring.