Transmissibility function analysis for boundary damage identification of a two-storey framed structure using artificial neural networks

This paper presents a damage identification technique that uses output-only scalar transmissibility measurements of a structure to identify boundary conditions. A damage index is formulated based on output-only acceleration response measurements from ambient floor vibration. The damage index is analysed by a system of artificial neural networks (ANNs) to predict boundary condition changes of the structure. Using the data compression and noise filtering capabilities of principal component analysis (PCA), the size of the damage index is reduced in order to obtain suitable patterns for ANN training. To test the proposed method, it is applied to different models of a numerical two-storey framed structure with varying boundary conditions. Boundary damage is simulated by changing the condition of individual joint elements of the structure from fixed to pinned. The results of the investigation show that the proposed method is effective in identifying boundary damage in structures based on output-only response measurements.