ABSTRACT
The present paper illustrates a wavelet-based drive-by methodology to assess bridge health condition, identifying possible damage presence and position. The case study is represented by a single-track, medium span, Warren truss bridge, whose FE model was constructed referring to a real existing structure, and dynamically validated with experimental tests. Vertical bogie accelerations are processed by Continuous Wavelet Transform and the obtained scalogram is sectioned at a specific frequency, associated to the passage of the vehicle over each module of the truss structure. A damage index is then built up starting from the wavelet coefficients referred to this frequency. Therefore, the method exploits the structural conformation of the truss bridge as a tool for detecting damages occurred to its deck. The approach is tested using a 3D train-track-bridge dynamic interaction model: different speeds, track geometrical profiles and damage scenarios are explored to assess the effectiveness of the method and its robustness.
