In this study, the classical and modern signal processing
methods are used to extract dominant frequencies of Masjed Soleiman dam,
the highest embankment dam in Iran. The signals were recorded in the gallery,
mid-height and the crest of the dam during local earthquakes. Since the
amplitude and frequency contents of earthquake acceleration time histories
vary with time, classical signal processing techniques are limited to
extracting the exact characteristics of the signal. Time-frequency distribution
and wavelet analysis were used in this study to overcome this limitation.
The proposed modal frequencies of the dam body were evaluated using both
the classical and new techniques and the results compared. Differences
between the two sets of methods are described and the benefits of the
modern signal processing methods are discussed. It is shown that, in non-stationary
signals such as earthquake records, higher frequencies are extracted by
modern methods that cannot be obtained using classical methods. Besides,
the spectral variations of the scalograms clearly indicate that lower
frequency contents become more dominant as the excitation amplitude decreases.
The lower mode shapes of dam body are excited during the weak part of
an earthquake, whereas during the stronger part, all the high and low
modes are excited.
M. Davoodi, M.A. Sakhi and M.K. Jafari, 2009. Comparing Classical and Modern Signal Processing Techniques in Evaluating Modal Frequencies of Masjed Soleiman Embankment Dam during Earthquakes. Asian Journal of Applied Sciences, 2: 36-49.