FREQUENCY DOMAIN DECONVOLUTION FOR ROTATING SOURCES ON AN AXIAL FAN

The capture of moving sound sources is a challenge for high-resolution beamforming algorithms. In general, time domain formulations are necessary to acoustically track a source. In the special case of rotating sources, the application of frequency-domain algorithms is possible using a circular microphone array in alignment with the rotational axis. Measured signals are interpolated between neighboring microphones in synchronization with the rotating object, simulating a rotating array. This allows the use of time-invariant steering vectors and therewith frequency domain beamforming with subsequent deconvolution. In this contribution, the virtual rotating array method is applied to analyze sources occurring on an axial fan. The technique facilitates the identification of major noise sources as well as evaluating spectra of subcomponents such as leading and trailing edges. Requirements on the array geometry and necessary adaptations of the steering vectors will be discussed.