Multi-dimensional dynamic deformation monitoring of long-span railway bridges using GBIR and IVM data fusion

Structural health monitoring of long-span bridges is critical to their safe operation and ensuring efficient daily traffic. Ground-based interferometric radar (GBIR) and inertial vision-based measurement (IVM) can capture linear and point deformation of long-span bridges, respectively. In this paper, we propose a framework to obtain a multi-dimensional dynamic deformation time series by fusing these two datasets with procedures of spatial-temporal alignment, interpolating, established deformation spatial-temporal correlation models, and weighting. To our knowledge, it was experimented on the Xijiang Railway Bridge, located in Guangdong, China, which is the first combination of these two data. Deformations along the vertical and lateral directions were derived when trains crossed the bridge. To validate the effectiveness of the derived results, static leveling sensors and vibrometers were employed on the bridge to obtain instantaneous measurements. The results show that the derived deformation is consistent with these in-situ measurements and the accuracy has improved by 27.4% and 27.0% compared with GBIR and IVM, respectively. The framework combining GBIR and IVM performs well in multi-dimensional dynamic deformation monitoring of long-span bridges and can play an important role in structural health monitoring of similar structures.