فهرست نشریات علمی دسترسی آزاد ایران

System identification, such as extracting modal frequencies of structures, is an important tool to identify a system and also to study a system before and after the earthquake to evaluate the demolition effects on the system. Using classical signal processing methods, FFT and other well-known methods, has some limitations. For example, these signal processing methods extract frequencies in only the time or only frequency domain. In this research, modern signal processing methods are performed to obtain modal frequencies of Karkheh embankment dam, and the obtained frequencies are compared with extracted frequencies with classic signal processing methods. Since the amplitude and also the frequency contents of earthquake signals vary efficiently with time, as is called non-stationary signals, the classical signal processing techniques have some limitations in extracting the exact characteristics of non-stationary signals. To overcome this limitation, TFD (Time Frequency Distribution) and Wavelet analyses are performed. In this paper, recorded earthquake signals of Karkheh dam, one of the highest rock fill dams in Iran, are used to extract dominant frequencies of the dam body. Also, the ambient signals are recorded at three different levels of this dam. Using classical and modern signal processing methods, the proposed modal frequencies of the dam body are evaluated, and the obtained results are compared. Differences between the two methods are described, and the benefits of the modern signal processing method are discussed. The obtained results show that some frequencies are extracted by modern time-frequency methods, whereas they can't be obtained based on classical methods. These results prove the benefits and advantages of time-frequency signal processing methods as a powerful method for extracting modal frequencies of different structures. This advantage is mainly obvious in processing non-stationary signals such as earthquake and blasting signals.