Time-Frequency-Based Analysis of Pedestrian Induced Vibrations Using a Two-Step Clustering Approach

Vibration of footbridges caused by walking pedestrians is still an important research topic. The excitation is very complex and recent load models have shown to be sometimes un-conservative. Furthermore, cablesupported footbridges and stress ribbon footbridges have closely spaced natural frequencies. Well-known methods for modal identification such as frequency domain decomposition, subspace identification and blind source separation require comparatively long time series to work well. They sometimes have problems with transient frequency components of pedestrian induced vibrations. This paper presents a novel method to analyze the vibration of highly flexible footbridges. This method addresses three aspects: analyzing measurement data to understand the mechanics of the excitation, comparing the quality of load models with real excitation in the time-frequency domain and identifying modal parameters such as natural frequencies and mode shapes. This method utilizes a time-frequency analysis and a two-step clustering approach of the results. The implementation and advantages of this method are discussed using vibration data from an aluminum bridge at the University of Waterloo and a stress ribbon bridge at the TU Berlin. Furthermore, some interesting findings for pedestrian induced vibrations of both bridges are presented.

This creative work is copyrighted. The copyright holder(s) do(es) not grant any usage rights other than viewing and downloading the work for personal use. Further copying or publication requires the permission of the copyright holder(s).