Source localization and power estimation in aeroacoustic noise measurements

Using microphone arrays for noise source localization and power estimation has become common practice in aeroacoustic measurements, with the ultimate goal being the development of acoustic treatments to reduce overall airframe noise. This dissertation discusses the challenges involved in aeroacoustic testing with microphone arrays and develops a number of new signal processing techniques to overcome these challenges. The proposed algorithms are validated using both simulations and experimental data acquired at the University of Florida Aeroacoustic Flow Facility (UFAFF) with a 63-element microphone array.;The standard delay-and-sum (DAS) beamformer is the most widely employed beamforming algorithm due to its simplicity and robustness, although it suffers from high sidelobe level and low resolution problems. Deconvolution can be used to eliminate the effects of the array response function from the DAS estimates. In this dissertation, the deconvolution problem is carried onto the sparse signal representation area and a sparsity constrained deconvolution approach (SC-DAMAS) as well as a sparsity preserving covariance matrix fitting approach (CMF) area presented. These algorithms are shown to offer better performance than several existing methods.;Next, a systematic experimental analysis of DAS, deconvolution approach for the mapping of acoustic sources (DAMAS), SC-DAMAS, CMF, and CLEAN based on spatial source coherence (CLEAN-SC) is presented using uncorrelated and coherent sources as well as a NACA Mod 63-215 Mod B airfoil model. The source localization and absolute signal power estimation performance of the aforementioned algorithms are analyzed.;To deal with correlated sources, the CMF-C algorithm, which is an extension to CMF, is proposed as an alternative to DAMAS-C, which is the extension of DAMAS to the correlated case. Since DAMAS-C and CMF-C are computationally impractical, an alternative algorithm, named mapping of acoustic correlated sources (MACS), is also presented. MACS is shown to work with simulated and experimental data containing correlated (or coherent) sources within a reasonable amount of time.;Furthermore, a systematic uncertainty analysis of the DAS beamformer and a widely used array calibration procedure is presented. It is shown using experimental data that the uncertainties in the DAS beamformer integrated levels can be expected to be larger than about +/-1 dB. It is also shown that the array calibration procedure is essential when the assumed steering vectors are expected to contain errors.;Most existing array processing algorithms for aeroacoustic noise measurement applications assume the presence of monopole sources. The last chapter of the dissertation addresses the problem of directive sources with unknown steering vectors. An algorithm for estimating non-diagonal measurement noise covariance matrices is also presented in this chapter as an alternative to diagonal removal.