Diagnosis of a portal frame using advanced signal processing of laser vibrometer data

This paper introduces an original, non-destructive diagnosis tool using modal data only. Our objective is twofold: to identify damage in structural elements (beams) and to establish a classification of the connections between elements using no a priori information except the modal data. This paper introduces enhanced wavelet analysis as a tool for damage detection based on mode shape data, using an analytical mode shape fit as the undamaged state. Then we demonstrate that damage severity can be identified using a supervised neural networks approach. To supplement the local diagnosis, boundary conditions are characterised using spatial frequency. Finally, a damaged beam of a wooden portal is analysed using a high-resolution mode shape acquired with a laser Doppler vibrometer. The experimental results confirm the accuracy of the method for detecting damages and classifying boundary conditions.