A new strategy for data-driven damage diagnosis of shear structures adapted to ambient vibration

Since many critical infrastructures of large civil engineering are shear structures and are exposed to ambient excitation most of the operational time, it is crucial to utilize ambient vibration data for damage diagnosis of shear structures. However, some autoregressive moving average with exogenous inputs (ARMAX) model methods will face challenges under ambient excitation due to assumptions about external loads. To address this issue, we propose a new data -driven strategy. Firstly, pseudo -free responses of shear structures are calculated from correlation signals. Subsequently, substructure ARMAX models are constructed using pseudofree responses, and model residuals are extracted as damage -sensitive features (DSFs). Furthermore, to improve the accuracy of the anomaly diagnosis, the Mel -frequency cepstral coefficients (MFCCs) are employed for dimensionality reduction of model residuals and are combined with Mahalanobis-squared distance (MSD) to form a hybrid distance method. Finally, the practicality and superiority of the strategy are validated using numerical and experimental benchmark structures.