Real-time damage detection of bridges using adaptive time-frequency analysis and ANN

Although traditional signal-based structural health monitoring algorithms have been successfully employed for small structures, their application for large and complex bridges has been challenging due to non-stationary signal characteristics with a high level of noise. In this paper, a promising damage detection algorithm is proposed by incorporation of adaptive signal processing and Artificial Neural Network (ANN). First, three adaptive signal processing techniques including Empirical Mode Decomposition (EMD), Local Mean Decomposition (LMD) and Hilbert Vibration Decomposition (HVD) are compared. The efficacy of these methods is examined for several numerically simulated signals to find a reliable signal processing tool. Then, three signal features are compared to find the most sensitive feature to damage. In the next step, an ANN ensemble is utilized as a classifier. Traditional statistical features and energy indices are used as the network input and output to make real-time detection of damage possible. The strength of this approach lies with training the network only based on healthy state of the structure. Having a trained ANN, online processing can be made to find a possible damage. Results show that the proposed algorithm has a good capacity as an online output-only damage detection method.