Acoustic Source Localization Research Articles

Multiple Acoustic Source Localization in Microphone Array Networks

The problem of multiple acoustic source localization using observations from a microphone array network is investigated in this article. Multiple source signals are assumed to be window-disjoint-orthogonal (WDO) on the time-frequency (TF) domain and time delay of arrival (TDOA) measurements are extracted at each TF bin. A Bayesian network model is then proposed to jointly assign the measurements to different sources and estimate the acoustic source locations. Considering that the WDO assumption is usually violated under reverberant and noisy environments, we construct a relational network by coding the distance information between the distributed microphone arrays such that adjacent arrays have higher probabilities of observing the same acoustic source, which is able to mitigate the miss detection issues in adverse environments. A Laplace approximate variational inference method is introduced to estimate the hidden variables in the proposed Bayesian network model. Both simulations and real data experiments are performed. The results show that our proposed method is able to achieve better source localization accuracy than existing methods.

An Integrated Approach to Real-Time Acoustic Emission Damage Source Localization in Piled Raft Foundations

Acoustic emission (AE) has garnered significant interest as a promising way to detect the early-stage development of internal cracks and damage in underground and geotechnical structures, associated with natural disasters. Meanwhile, AE source localization techniques that can identify the damage location in a piled-raft foundation (PRF) are premature because of its complex geometry, although the PRF is a widely used deep foundation type for high-rise buildings. In this study, we propose an integrated approach to localize AE sources in the PRF by using the modified Akaike information criterion (AIC) method and examine its accuracy to mark with pile zones. We performed a series of experiments on a scaled PRF model at a ratio of 1:50, composed of one raft and 25 piles. The results demonstrate that the combined approach with the modified AIC method and the Simplex method can localize the AE source zones with good accuracy, greater than 95% on average. The suggested two-stage AIC picker shows accurate onset time determination, and hence, it significantly improves the accuracy, particularly effective for the signals with low signal-to-noise ratios. The approach exploiting the two-stage AIC picker can be readily used for automated real-time AE monitoring to detect crack generation and its location in buried foundations that cannot be inspected visually.

Memory-efficient approximate three-dimensional beamforming.

Localization of acoustic sources using a sensor array is typically performed by estimating direction-of-arrival (DOA) via beamforming of the signals recorded by all elements. Software-based conventional beamforming (CBF) forces a trade-off between memory usage and direction resolution, since time delays associated with a set of directions over which the beamformer is steered must be pre-computed and stored, limiting the number of look directions to available platform memory. This paper describes a DOA localization method that is memory-efficient for three-dimensional (3D) beamforming applications. Its key lies in reducing 3D look directions [described by azimuth/inclination angles (ϕ, θ) when considering the array as a whole] to a single variable (a conical angle, ζ) by treating the array as a collection of sensor pairs. This insight reduces the set of look directions from two dimensions to one, enabling computational and memory efficiency improvements and thus allowing direction resolution to be increased. This method is described and compared to CBF, with comparisons provided for accuracy, computational speedup, and memory usage. As this method involves the incoherent summation of sensor pair outputs, gain is limited, restricting its use to localization of strong sources-e.g., for real-time acoustic localization on embedded systems, where computation and/or memory are limited.

Coherent interferometric imaging in a random flow

This paper is concerned with the development of imaging methods to localize sources or reflectors in inhomogeneous moving media with acoustic waves that have travelled through them. A typical example is the localization of broadband acoustic sources in a turbulent jet flow for aeroacoustic applications. The proposed algorithms are extensions of Kirchhoff migration (KM) and coherent interferometry (CINT) which have been considered for smooth and randomly inhomogeneous quiescent media so far. They are constructed starting from the linearized Euler equations for the acoustic perturbations about a stationary ambient flow. A model problem for the propagation of acoustic waves generated by a fixed point source in an ambient flow with constant velocity is addressed. Based on this result imaging functions are proposed to modify the existing KM and CINT functions to account for the ambient flow velocity. They are subsequently tested and compared by numerical simulations in various configurations, including a synthetic turbulent jet representative of the main features encountered in actual jet flows.

A Novel Joint Localization Method for Acoustic Emission Source Based on Time Difference of Arrival and Beamforming

Time difference of arrival (TDOA) method and beamforming method are often individually utilized in the localization of acoustic emission source of a plate. Based on the performances of the two methods, a novel joint localization method for acoustic emission source is proposed in this paper. Firstly, the performances of TDOA method and beamforming method are studied based on the simulation signals. Then, aiming at the advantages and disadvantages of these two methods, a joint localization method is proposed. Finally, the performances of joint localization method are verified by simulation and experiment. Both simulation and experimental results show that the accuracy of the joint localization method is improved, and the calculation amount is greatly reduced in comparison to the TDOA method and the beamforming method.

Acoustic source imaging using densely connected convolutional networks

The localization of acoustic sources using acoustic imaging methods such as acoustic beamforming can be limited by the Rayleigh resolution limit at relatively low frequencies and the output of these methods may also produce spatial aliasing images referred to as sidelobes, particularly at high frequencies. To date, there are very few Deep Neural Network (DNN) applications for acoustic imaging to help alleviate some of these issues. In this study, several DNN models were developed with a training strategy specifically designed for an acoustic imaging task. This proposed DNN-based method is examined for various acoustic source conditions and compared with other classic acoustic imaging methods. The DNN method is able to recognize the pattern behind microphone array signals for different source positions, by using the real-component of the cross-spectral matrix of the received pressure vectors at the microphone positions. This input feature allows the DNN model to accurately locate and quantify the source strengths of complicated distributed acoustic sources, even at low frequencies that typically challenge acoustic beamforming deconvolution methods. The loss function of the DNN model is based on the difference between the estimated and true acoustic source maps, that is used to iteratively improve weighting functions within the DNN hidden layers. DNN models with both a fixed and random number of input sources are simulated for a range of specific frequencies. The DNN model with a random number of input sources is tested against conventional beamforming, CLEAN-SC and DAMAS, revealing a far improved source localization and source strength estimation. These DNN models represent a very promising proof-of-concept for the use of DNN models in the field of acoustic imaging.

Steered sample algorithm for acoustic source localization.

High-precision source localization depends on many factors, including a suitable location method. Beamforming-based methods, such as the steered response power (SRP), are a common type of acoustic localization methods. However, these methods have low spatial resolution. The SRP method with phase transform (SRP-PHAT) improves the spatial resolution of SRP and is one of the most effective and robust methods for source localization. However, the introduction of a phase transform to SRP might amplify the power of the noise and result in many local extrema in the SRP space, which has a negative impact on source localization. In this paper, a steered sample algorithm (SSA) based on the reciprocity of wave propagation for acoustic source localization is proposed. The SSA localization process is similar to the hyperbolic Radon transform, which is theoretically analyzed and is the most essential difference form the SRP/SRP-PHAT. Compared with the SRP-PHAT, the experimental results demonstrate that the SSA perform better when it comes to array signal positioning with limited array elements and narrow azimuth signal, where SSA can achieve high precision positioning with lower SNR.

Investigation of acoustic propagation and source localization in a hot jet flow

A jet flow typically occurs in open-jet wind tunnels and turbofan engines, where there are velocity and temperature gradients between the source region and the outside environment. These gradients form a thermal shear layer that determines acoustic propagation and source localization. Moreover, they induce Kelvin–Helmholtz instabilities in numerical simulations. To stabilize simulations, a gradient-term suppression and filtering method is proposed and validated using time-domain linearized Euler equations. For sound propagation, the influences of the steep gradients of the velocity and temperature together with the spreading angle and source frequency are investigated through a self-similar shear layer with superposition of compressibility and heat. Those refraction influences strongly change the beamforming-related phase rather than the directivity-related amplitude. Next, the computed database is processed by delay-and-sum beamforming to localize the source position. In the beamforming part, refraction and amplitude corrections based on Amiet’s theory with an extension for temperature differences are numerically evaluated and later applied when analyzing the localization error. This localization error is basically proportional to the jet Mach number, temperature ratio, and source frequency and is inversely proportional to the spreading rate. Finally, the localization error is fitted as a linear function of the product of the jet Mach number and the Strouhal number.

Study on acoustic time delay localization in boiler tube arrays considering effective sound velocity

Based on the study of sound propagation characteristics and acoustic radiation in periodic tube arrays, we found that the wave front differed from the no-tube arrays at the same frequency. The phenomenon indicated that there was different sound velocity in tube arrays. This research considered tube arrays as an effective medium for sound and studied the effective sound velocity at different frequencies in tube arrays. The numeral calculations and experiments shown similar conclusions that the sound velocity was slower than in no-tube arrays one. Acoustic velocity directly affects the accuracy of time delay localization method, and the time delay localization method in acoustic isotropic media is mature. The velocity of low frequency sound wave is isotropic in the arrays, we investigate the variation of sound velocity theoretically and experimentally. The sound demonstrates anisotropy at high frequency region, the effective sound velocity is related to the band gap in the tube arrays. The research has been found that the same frequency has different sound velocity values in different directions. In different directions of tube arrays, the effective sound velocity is used to conduct an one-dimensional acoustic source location experiment. This study can provide an innovative concept and method for leakage location in boiler tube arrays.

Sound source localizations using various microphone arrays

Estimating directions of arrival (DoAs) of multiple sound events still posts challenges in acoustic source localization. This paper reviews recent development on model-based direction-of-arrival analysis using three different microphone array configurations, in situations involving potentially multiple concurrent sound sources. The array configurations include a sparse arrangement consisting of two microphones, coprime linear microphone arrays, and spherical microphone arrays. Formulation and evaluation of prediction models are of central importance in coping with these challenges. These three array configurations are all used to solve a two-level inferential problem of sound source enumeration and direction of arrival estimation. The prediction models based on specific array configurations are evaluated upon experimental data derived from pertinent array processing in order to select the simplest such model that can adequately describe the experimental data, thereby enumerating first the concise number of sound sources, then their DoA information. This paper reviews the prediction models recently established for the three microphone arrays, the two levels of inference, and analysis results. This paper will highlight the importance of the prediction models for the model-based analysis of complex sound environments with potentially multiple concurrent sources.

Acoustic source localization with the angular spectrum approach in continuously stratified media.

The angular spectrum approach (ASA)-a frequency domain method to calculate the acoustic field-enables highly efficient passive source localization and modeling forward propagation in homogeneous media. If the medium is continuously stratified, a first-order analytical solution may be obtained for the field at arbitrary depth. Simulations show that the proposed stratified ASA solution enables accurate source localization as compared to the uncorrected ASA (error from 1.2 ± 0.3 to 0.49 ± 0.3 wavelengths) at scalings relevant to biomedical, underwater, and atmospheric acoustic applications, and requiring milliseconds on nonspecialized hardware. The results suggest the proposed correction enables efficient and accurate localization in stratified environments.

Acoustic source localization in imperfectly known environments using frequency-differencing techniques

Remote unknown sound-source localization is a challenging task with applications in a variety of fields, such as underwater acoustics, atmospheric acoustics, structural acoustics, bio-medical ultrasound, animal bio-acoustics, and seismology. In nearly all cases, the localization algorithm is only successful when there is good cross correlation between remotely measured and predicted acoustic fields, a possibility that exits at low frequencies and/or short ranges in imperfectly known environments. However, at sufficiently high frequency and/or long range, the requisite measured-to-predicted-field correlation might never be high enough for successful source localization. However, recent investigations of the frequency-difference autoproduct, a quadratic product of two complex field amplitudes having different frequencies, suggest that it may have the phase structure of an acoustic field at the difference frequency. Thus, using sufficiently low difference frequencies, unknown sources may be localized at ranges where conventional techniques are unsuccessful by correlating measured and predicted (ideal) autoproducts. The underlying formulation of frequency-differencing techniques is presented along with examples drawn from simulations, laboratory experiments, and ocean propagation measurements that involve frequencies from fractions of a Hertz to more than 100 kHz, and propagation distances from tens of centimeters to hundreds of kilometers. [Sponsored by ONR, NAVSEA, and NSF.]

Semi-supervised source localization in reverberant environments using deep generative modeling

We present a method for acoustic source localization in reverberant environments based on semi-supervised machine learning (ML) with deep generative models. Source localization in the presence of reverberation remains a major challenge, which recent ML techniques have shown promise in addressing. Despite often large data volumes, the number of labels available for supervised learning in reverberant environments is usually small. In semi-supervised learning, ML systems are trained using many examples with only few labels, with the goal of exploiting the natural structure of the data. We use variational autoencoders (VAEs), which are generative neural networks (NNs) that rely on explicit probabilistic representations, to model the latent distribution of reverberant acoustic data. VAEs consist of an encoder NN, which maps complex input distributions to simpler parametric distributions (e.g., Gaussian), and a decoder NN which approximates the training examples. The VAE is trained to generate the phase of relative transfer functions (RTFs) between two microphones in reverberant environments, in parallel with a DOA classifier, on both labeled and unlabeled RTF samples. The performance this VAE-based approach is compared with conventional and ML-based localization in simulated and real-world scenarios.

A classification approach to the characterization of seabed geoacoustic profiles via deep learning

Acoustic source localization algorithms in the shallow ocean require information about the ocean seabed. While past approaches have characterized seabed geoacoustic profiles via continuous physical parameters, a classification approach approximates the distribution of seabed types as a discrete set of seabed classes. To exclude acoustically indistinguishable seabeds from the discrete classes, the acoustic separation between two seabeds may be evaluated by the Pearson correlation between their characteristic transmission loss curves. This process yields a subset of discrete seabed classes. To distinguish these classes, we train a convolutional neural network classifier on the pressure time-series of SUS charges, which are simulated using the geoacoustic profiles of the seabed classes. This network successfully predicts seabed class, when tested on a holdout set where the sound speed profiles of the ocean are adequately similar to those in the training set. An acoustic separation measure between sound speed profiles, analogous to the one for seabeds, directly relates to the generalization error of the network. [Work supported by the Office of Naval Research Contract No. N00014-16-C-3065 and by the NSF REU program, Grant No. 1757998.]

Imaging With Therapeutic Acoustic Wavelets-Short Pulses Enable Acoustic Localization When Time of Arrival is Combined With Delay and Sum.

Passive acoustic mapping (PAM) is an algorithm that reconstructs the location of acoustic sources using an array of receivers. This technique can monitor therapeutic ultrasound procedures to confirm the spatial distribution and amount of microbubble activity induced. Current PAM algorithms have an excellent lateral resolution but have a poor axial resolution, making it difficult to distinguish acoustic sources within the ultrasound beams. With recent studies demonstrating that short-length and low-pressure pulses-acoustic wavelets-have the therapeutic function, we hypothesized that the axial resolution could be improved with a quasi-pulse-echo approach and that the resolution improvement would depend on the wavelet's pulse length. This article describes an algorithm that resolves acoustic sources axially using time of flight and laterally using delay-and-sum beamforming, which we named axial temporal position PAM (ATP-PAM). The algorithm accommodates a rapid short pulse (RaSP) sequence that can safely deliver drugs across the blood-brain barrier. We developed our algorithm with simulations (k-wave) and in vitro experiments for one-, two-, and five-cycle pulses, comparing our resolution against that of two current PAM algorithms. We then tested ATP-PAM in vivo and evaluated whether the reconstructed acoustic sources mapped to drug delivery within the brain. In simulations and in vitro, ATP-PAM had an improved resolution for all pulse lengths tested. In vivo, experiments in mice indicated that ATP-PAM could be used to target and monitor drug delivery into the brain. With acoustic wavelets and time of flight, ATP-PAM can locate acoustic sources with a vastly improved spatial resolution.

Iterative Modified SRP-PHAT with Adaptive Search Space for Acoustic Source Localization

The Steered Response Power with Phase Transform (SRP-PHAT) is one of the leading methods for sound source localization because of its proven robustness in a noisy and reverberant environment. However, for the higher accuracy, the maximum power location needs to be searched over a dense grid, making the computational cost a real issue for large-scale microphone arrays. Modified SRP-PHAT introduced an effective strategy that computes the SRP functional at each discrete location, also considering the volume surrounding it. Subsequently, it explores the entire search space and achieves improved robustness over a coarse spatial grid. To accomplish higher spatial resolution with minimal computations, the Modified SRP-PHAT was further extended by employing an iterative grid decomposition search strategy. However, in challenging environments, the performance of the method deteriorates considerably with the iterative approach. In this paper, we suggest furtherance to iterative modified SRP-PHAT by employing adaptive search space in each iteration. The performance of the proposed method is extensively tested and compared to the state-of-the-art techniques on a simulated dataset and SMARD database to validate its reliability. The results exhibited that the proposed method achieves higher accuracy with only a marginal increase in computation expense than that of iterative modified SRP-PHAT.

Acoustic Emission Wave Velocity Attenuation of Concrete and Its Application in Crack Localization

The accurate localization of an acoustic emission (AE) source is a vital aspect of AE nondestructive testing technology. A model of wave velocity attenuation caused by the extension of transmission distance is established to analyze the attenuation of AE wave velocities in concrete and thus improve the acoustic source localization accuracy from the perspective of modified velocity. In combination with the exhaustive and region localization methods, a region exhaustive localization method is established based on the modified wave velocity. The results indicate that the smaller the water–cement ratio, the larger the reference wave velocity, and the spatially dependent attenuation of wave velocity increase. Moreover, the larger the aggregate particle size, the larger the reference wave velocity, and the greater the attenuation of wave velocity with distance. For a propagation distance of 1000 mm, the AE wave velocity attenuation exceeds 50% compared with the AE velocity. The optimized localization method reduces the number of nodes calculated, thus improving the method’s accuracy when used for localization.

Rapid localization of acoustic source using sensor clusters in 3D homogeneous and heterogeneous structures

The acoustic source localization in 3D structures is a challenging task especially if the structure is heterogeneous. A large number of unknown parameters in a 3D heterogeneous structure require a large number of sensor arrays for accurate source localization. In this article, a localization technique is proposed using triangular pyramid shaped sensor clusters. The time difference of arrival of acoustic signals between different sensors of the clusters is analyzed to localize the acoustic source. It uses acoustic wave propagation information to rapidly localize the acoustic source. This rapid acoustic source localization technique works well for both homogeneous and heterogeneous media. It considers the refraction in heterogeneous media using Snell’s law. The proposed technique is verified experimentally and numerically. The experimental results show that the technique is effective for source localization in 3D homogeneous structures. Numerical results are generated by finite element modeling for both homogeneous and heterogeneous structures. The results show the reliability of the proposed technique. This technique helps to localize the acoustic source with only a few sensors and is indispensable for monitoring large 3D structures with monitoring equipment that can handle only a few sensors.

High-resolution acoustic localization of changes in spatially-distributed coherent sources for structural health monitoring.

Localization of acoustic sources is a common remote sensing goal. When multiple sources are present and coherent, high-resolution localization typically becomes more challenging. The spectral estimation method with additive noise (SEMWAN) is an existing technique for high-resolution localization of incoherent monopole sources in low-signal-to-noise environments. SEMWAN utilizes a reference measurement to incoherently suppress background noise, but its performance suffers in applications involving spatially-distributed coherent sources, such as like a vibrating plate. However, by subtracting a reference measurement and using subarray averaging, SEMWAN can be extended to localization of small changes in distributed coherent sources. This revised approach, the spectral estimation method with coherent background removal (SEMCBR), permits remote acoustic localization of damage in a vibrating structure. A simple multi-source experiment using an 8-by-8 planar square microphone array with 6-cm spacing in both horizontal directions was performed to validate SEMCBR at a frequency of 5.0 kHz. Additional SEMCBR experimental results are reported for the same array placed 1.0 m above a 30 cm × 30 cm vibrating plate with and without damage. Cuts, boundary failures and delamination of a composite plate were successfully located with conventional spherical-wave beamforming and SEMCBR using a 0.1 to 6.0 kHz bandwidth. However, SEMCBR provides 4 to 6 times better resolution.

Three-dimensional acoustic emission source localisation in concrete based on sparse least-squares support vector regression

In order to improve the accuracy of concrete damage localisation based on acoustic emission (AE) monitoring, a multi-output sparse least-squares support vector regression (S-LS-SVR) method is attempted for AE source localisation in concrete. The AE events are produced by pencil lead breaks and the response wave is received by piezoelectric sensors. A Newton iterative method, an improved exhaustive method and two S-LS-SVR approaches (S-LS-SVR(A) and S-LS-SVR(B)) are used to locate the AE source, then the positioning accuracies of the methods in the three coordinate directions are compared and analysed. The results show that the accuracy of AE source localisation using the S-LS-SVR(B) model is higher than that of the other methods. The accuracy of the S-LS-SVR model using the time difference of arrival (TDOA) and the sequential number of sensors that arrive successively as input parameters is higher than that of the other AE signal combination trialled as the input. This shows that the S-LS-SVR(B) model is better than the S-LS-SVR(A) model. The intelligent S-LS-SVR(B)-based localisation method provides a basis for application in actual damage detection.