Non-stationary Analysis of Vehicle-Bridge Interaction for Structural Health Monitoring of a Railway Bridge

Structural Health Monitoring for bridges is increasingly used in the asset management of bridges. The dynamic properties are key elements in the structural integrity assessment and the identification of damage. Natural frequencies can provide a relatively simple high-level condition assessment, but suffer from variations in the environmental and operational conditions. Methods considering the transient vibrations and hence the dynamic interaction between the train passing and the bridge itself (Vehicle-Bridge Interaction) are based on non-physical parameters, limiting the understanding of the changes in these parameters. Here, a damage identification is proposed, based on the change of the instantaneous frequency observed during the traverse phase - the phase in which the train passes the bridge. The damage index relies in the first place on an accurate extraction of the instantaneous frequency which is extracted with the Wavelet Synchro-Squeezed Transformation (WSST). WSST was shown earlier to outperform the Continuous Wavelet Transformation in terms of resolution in the low-frequency range. Secondly, the change of the instantaneous frequency due to damage is distinguished from that due to changes in the environmental and operational conditions by calculating the correlation coefficient. The results, based on a finite element model with a mass moving over a beambased bridge, show that the changes in the instantaneous frequency can be positively linked to either damage or a change of vehicle mass, representing a change in the operational condition. Together with the earlier made observation that the extraction of the instantaneous frequency by WSST is relatively insensitive to noise, compared to CWT, provides a solid ground for more detailed investigations on the capabilities of the method proposed. This research is therefore currently being expanded to include more variations in operational and environmental conditions as well as to include more complex vehicles, such that ultimately a coupling can be made to field data and provide guidelines on how to use this method.