AbstractThis article aims to present the results of an investigation on detection and quantification of damage location and severity for steel structures using wavelet packet transform for denoising the initial signals, in combination with a peak picking technique, to capture the modal parameters of the structures. In the peak picking technique, as an extended version of the basic frequency domain method, modes can be identified from the spectral density when having a white noise input and a damped structural model. Using the spectral density matrix, this method estimates the modes by a singular value decomposing. Consequently, this decomposition results in a single-degree-of-freedom identification of the system for each singular value. In other words, in terms of a proposed two-step algorithm, the modal parameters of the structures are identified by decomposing their free vibration responses. Because the decomposed signal could has the same amount of energy as the main one, it can be utilized in the peak picking technique to obtain the modal parameters. In the next stage, the obtained modal parameters are compared with those obtained from auto-regressive moving average with eXogenous input (ARMAX) method to control the accuracy of the proposed methodology. Finally, the Minokowski method is employed to determine the structural damage intensity. Moreover, the precision of the proposed algorithm was measured against the results of a famous experimental benchmark structural model using modal characteristics.