Frequency Prediction of a Von Karman Vortex Street Based on a Spectral Analysis Estimation

摘要

Spectral analysis studies the power distribution over frequency of a signal. This allows the characterization of time signals by its harmonics. This article will establish a relationship between the autocorrelation function and the spectrum. The direct implementation of the theory when analyzing a finite time signal results in a raw periodogram or first estimation of the spectrum. However, owing to the biased nature of the autocorrelation function, the periodogram obtained will not be a good estimation. Thus, several estimation techniques are needed in order to acquire a reliable spectrum. Amongst the techniques handled are the averaging Welch method, the use of window functions or tapering and the implementation of Fast Fourier Transform algorithms. To validate the accuracy and improvements made with these techniques, an algorithm is implemented in Matlab. Several synthetic signals are assessed and the classical Kármán Vortex Street is performed in a wind tunnel experiment. The results obtained are proof of the need for a careful study of the different estimation techniques when analyzing a signal.