A Bayesian approach to eliminate correlated noise using an independent reference-Application to supersonic jet noise extraction.

A Bayesian method to remove correlated noise from multi-channel measurements is introduced. It is based on Bayesian factor analysis coupled with prior but uncertain knowledge of the correlation structure of the noise. This technique is well suited to denoise cross-spectral matrices measured in the frame of aeroacoustic experiments when background noise measurements are available, because it allows separating the engine noise contribution from the turbulent boundary layer and uniform noise components that are all sensed by in-flow microphones. In-flight data measured on flush-mounted microphones on an aircraft fuselage are denoised using this method. It is shown that it has a significant benefit for studying the broadband shock-associated noise generated by the engines in realistic flight conditions.

A Bayesian approach for the separation of the acoustic and the correlated aerodynamic wall pressure fluctuations.

When performing measurements with wall-installed microphone array, the turbulent boundary layer that develops over the measuring system can induce pressure fluctuations that are much greater than those of acoustic sources. It then becomes necessary to process the data to extract each component of the measured field. For this purpose, it is proposed in this paper to decompose the measured spectral matrix into the sum of matrices associated with the acoustic and aerodynamic contributions. This decomposition exploits the statistical properties of each pressure field. On the one hand, assuming that the acoustic contribution is highly correlated over the sensors, the rank of the corresponding cross-spectral matrix is limited to a finite number. On the other hand, the correlation structure of the aerodynamic noise matrix is constrained to resemble a Corcos-like model, with physical parameters estimated within the separation procedure. This separation problem is solved by a Bayesian inference approach, which takes into account the uncertainties on each component of the model. The performance of the method is first evaluated on wind tunnel measurements and then on a particularly noisy industrial measurement setup: microphones flush-mounted on the fuselage of a large aircraft.