.png){kind=icon}
.png){kind=icon}
Return
Environmental effect compensation and anomaly detection in an ageing prestressed concrete girder bridge using Gaussian process regression
.png){kind=icon}
PRE
AI
H
Y
DOI：10.1080/15732479.2026.2678472.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Monitoring the dynamic characteristics of bridges is useful for assessing long-term performance and structural safety. However, they are influenced not only by structural changes but by environmental variations such as temperature and humidity, which often obscure the structural anomaly detection. Although previous studies have explored environmental effect removal using statistical and data-driven approaches, few have considered associated uncertainties in the dynamic characteristic identification in long-term monitoring. This research addresses this gap by estimating uncertainties in natural frequencies using a Gaussian Process Regression (GPR) model. A single-span prestressed concrete girder bridge was monitored over two campaigns, five years apart (2016–2017 and 2021–2022). Natural frequencies were extracted with environmental conditions obtained from the Japan Meteorological Agency (JMA) station for 2016–2017 and from both the on-site and JMA station for 2021–2022. The GPR model, trained on the 2021–2022 dataset as the baseline state, predicted natural frequencies under varying environmental conditions and was applied to the 2016–2017 data to identify deviations caused by possible structural changes. An anomaly index was developed to distinguish environmentally induced variations from possible structural degradation.
Keywords:
Age degradation
gaussian process regression
anomaly detection
prestressed concrete bridge
structural health monitoring
natural frequencies
environmental variation
Journal
IF:
2.6
Papers:
451
Citations:
5.3K
Organization
Cited Papers