A covariance fitting approach for correlated acoustic source mapping

A hybrid deconvolution approach to separate static and moving single-tone acoustic sources by phased microphone array measurements

A deconvolution approach for the mapping of acoustic sources (DAMAS) determined from phased microphone arrays

Inlet fan noise in turbofan engines is frequently characterized in terms of modal amplitude distribution. The modes can be measured nonintrusively using an array of wall mounted microphones. At high frequency, the number of cuton modes can be much larger than any reasonable number of microphones. This suggests a beamforming analysis technique, rather than a least squares fit, as an appropriate method to extract the mode amplitude from the data. The quality of beamforming output is limited by array resolution and sidelobes. The CLEAN deconvolution algorithm has been used in the past to attempt to improve mode maps from beamforming processing. The recently introduced DAMAS algorithm is a much more comprehensive deconvolution technique, and is well suited to processing mode maps. The benefit of DAMAS is illustrated using simulated mode data. The engine inlet is represented as a cylinder with uniform flow. Two array designs are considered. In each case, several mode distributions are postulated and used to create synthetic array data. Beamforming is applied to create mode maps, and DAMAS is used to deconvolve the array response functions from the mode maps. The results are evaluated by comparing the true mode distributions, the beamform maps, and the deconvolved maps.

Beamforming algorithms have been applied in multiple contexts in aeroacoustic applications, but difficulty arises when applying these to the partially correlated and distributed sources found in jet noise. To measure and more accurately distinguish correlated sources, cross-beamforming methods are employed to incorporate correlation information. Deconvolution methods such as DAMAS-C, an extension of the deconvolution approach for the mapping of acoustic sources (DAMAS), remove array effects from cross-beamforming applications and further resolve beamforming results. While DAMAS-C results provide insight to correlation between sources, the extent to which these results relate to source correlation remains to be analyzed. Numerical simulations of sources with varying degrees of correlation are provided to benchmark the DAMAS-C results. Finally, correlation lengths are established for DAMAS-C results from measurements for full-scale military jet noise sources. [Work supported by ONR.]

Functional delay and sum beamforming for three-dimensional acoustic source identification with solid spherical arrays

Aeroacoustic measurements in closed-section wind tunnels often employ phased microphone arrays that are mounted flush to the wall of the test section. Phased arrays can measure the directivity of multiple acoustic sources within a sound field across all frequencies, but require dozens of microphones in typical applications. Furthermore, the turbulent boundary layer developed by wind tunnels engulf the phased arrays in loud pressure fluctuations that drown out the acoustics trying to be measured. This paper describes an alternative approach for measuring the directivity of a single broadband acoustic source that significantly reduces the number of microphones needed. This approach was developed from experimental testing of an acoustic metamaterial-based pressure sensing array in Virginia Tech’s Anechoic Wall Jet Tunnel. The measurements obtained reveal that an acoustic metamaterial-based array can be designed to passively reinforce select pressure waves as a function of frequency and the acoustic source’s location.

An evaluation of low-power microphone array sound source localization for deforestation detection

Improving the efficiency of DAMAS for sound source localization via wavelet compression computational grid

Coherent multi-transducer ultrasound (CoMTUS) imaging enables the use of multiple arrays as one large effective aperture. This yields images with enlarged field-of-view, improved resolution, and higher SNR. However, creating a large but discontinuous effective aperture increases the grating and side lobe levels and generates cross-talk artifacts from the direct transmissions between arrays. These additional challenges can degrade the contrast of the images obtained through the classic Delay and Sum (DAS) beamforming algorithm. This study investigates the use of an alternative nonlinear beamforming algorithm, the Delay Multiply and Sum (DMAS), to improve contrast and reduce cross-talk artifacts in CoMTUS. We adapted the original DMAS beamforming algorithm to the specific effective aperture of CoMTUS and tested its performance in a coherent dual-array system for both 2-D and 3-D imaging. The comparison between images obtained through CoMTUS DMAS and DAS was investigated by both 2-D and 3-D simulations and experiments. DMAS was shown effective to lower the side lobes and the noise floor, thus improving the detectability of cystic lesions.

In order to solve the problem of robustness of beamforming algorithm with microphone array channel mismatch, an adaptive dynamic-weighted constrained least square algorithm-based microphone array robustness frequency invariant beamforming algorithm is proposed. In the proposed algorithm, by analyzing the microphone array model, with or without channel mismatch, the disadvantages of the constrained least square frequency invariant beamforming algorithm with channel mismatch are studied. After the probability density functions of the microphones are defined as the robustness factors and added to the constraint least square frequency invariant beamforming algorithm, the robustness is improved to a certain extent, but it is still poor. In order to further improve the robustness of the algorithm, dynamic-weighted coefficients for controlling frequency invariance in the cost function are used to regulate the sidelobe spectrum energy. The fluctuation error is defined as the ratio of the maximum to minimum value of array response formed by the same angle of arrival at different frequencies, within the frequency range of frequency invariant, to compare the proposed algorithm with the constrained least square robustness frequency invariant and minmax robustness broadband beamforming algorithm. Experimental results of the algorithms show that the fluctuation errors of the proposed algorithm are the smallest and its robustness is the best; it can effectively overcome the poor robustness of the beamforming algorithm caused by microphone array channel mismatch, and can be applied to any arbitrary array structure.

Acoustic imaging implemented by microphone arrays has been widely used in variety of areas. Aiming at overcoming the drawbacks of the exhaustive search processes for some commonly used deconvolution methods, the Partial Swarm Optimization (PSO) combined with a 3D microphone array is introduced in this work to localize the acoustic sources. Based on the analysis result of the simulated data, the proposed method presents the potential to map the acoustic sources in 3D with acceptable computation time.

In support of the QTD II (Quiet Technology Demonstrator) program, aeroacoustic measurements of a 26%-scale, Boeing 777 main landing gear model were conducted in the Virginia Tech Stability Tunnel. The objective of these measurements was to perform risk mitigation studies on noise control devices for a flight test performed at Glasgow, Montana in 2005. The noise control devices were designed to target the primary main gear noise sources as observed in several previous tests. To accomplish this task, devices to reduce noise were built using stereo lithography for landing gear components such as the brakes, the forward cable harness, the shock strut, the door/strut gap and the lower truck. The most promising device was down selected from test results. In subsequent stages, the initial design of the selected lower truck fairing was improved to account for all the implementation constraints encountered in the full-scale airplane. The redesigned truck fairing was then retested to assess the impact of the modifications on the noise reduction potential. From extensive acoustic measurements obtained using a 63-element microphone phased array, acoustic source maps and integrated spectra were generated in order to estimate the noise reduction achievable with each device.

Purpose: To develop a deconvolution algorithm to obtain the robust fluence for external beam radiation treatment under geometrical uncertainties and to demonstrate the feasibility of this deconvolution approach for clinical robust IMRT planning. Method and Materials: Usually, the geometrical uncertainty is incorporated in the doseoptimization process for inverse treatment planning to determine the fluence which is robust to the geometrical uncertainty. Most of these approaches rely on the dose‐convolution which is subject to the error caused by patient surface curvature and internal inhomogeneity. In this work, based on a 1D deconvolution algorithm developed by Ulmer and Kaissl, a fluence‐deconvolution approach was developed to obtain the robust fluence through the deconvolution of the nominal static one given by any treatment planning system. It incorporates the geometrical uncertainty outside the doseoptimization procedure. Results: Robust fluences were calculated for a 4×4 cm flat field, a prostate IMRT and a head & neck IMRT plan in a commercial treatment planning system. The corresponding doses were simulated for 30 fractions with random Gaussian distribution of the iso‐centers showing good agreement with the nominal static doses.Conclusion: The clinical applicable robust IMRT fluence can be obtained through this deconvolution approach. Because it is separated from the doseoptimization procedure, the advantages of this approach are 1) the error of dose‐convolution is avoid; 2) it is more flexible and easier to be integrated into different existing treatment planning systems to obtain robust fluence.

Microphone arrays are an efficient tool for localizing acoustic sources. Time domain beamforming is one of the techniques which can be used to build a sound source map. This technique based on the cross-correlation of the microphone signals can be seen as a broadband beamforming method. The spatial likelihood functions are averaged with the arithmetic mean. Recently, it has been shown that the harmonic and geometric means can improve the quality of the sound source map. In this study, the influence of the exponent value of the generalized mean on the sound source map is investigated. Then it is shown that the proposed technique is less sensitive to the number of microphones and therefore, it can be reduced.

Decentralized learning of randomization-based neural networks with centralized equivalence