Power-frequency resolution of Proper Orthogonal Decomposition in acoustic velocity field measurement and visualization

The paper addresses a problem of the decomposition of experimental data representing acoustic velocity field into orthogonal modes using Proper Orthogonal Decomposition (POD). The main goal was to find the conditions, which have to be fulfilled if we want to compare POD results with classical Fourier spectral analysis, which is fundamental tool used in acoustics. Some numerical experiments were done, in which artificial sound fields were generated representing one pure tone and two pure tones travelling wave. In following steps the amplitudes and signal-to-noise ratios were manipulated. The artificial data have a form of snapshots, which in real measurement conditions can be obtained using the phase-locked particle image velocimetry technique (PIV). The POD applied in postprocessing stage allows the breakdown of an observed sound field into dynamically relevant and coherent structures and thus aids in the characterization and quantification of physical phenomena in near acoustic field of real sound sources. The in-deep knowledge of POD features applied to harmonic signals is crucial. From metrological point of view the POD also helps to increase of the signal-to-noise ratio, what is especially important in noninvasive observation of low level sound fields by means of particle image velocimetry.