CN115032591A - Broadband multi-sound-source positioning asynchronous measurement method and device and related medium - Google Patents
Broadband multi-sound-source positioning asynchronous measurement method and device and related medium Download PDFInfo
- Publication number
- CN115032591A CN115032591A CN202210625448.7A CN202210625448A CN115032591A CN 115032591 A CN115032591 A CN 115032591A CN 202210625448 A CN202210625448 A CN 202210625448A CN 115032591 A CN115032591 A CN 115032591A
- Authority
- CN
- China
- Prior art keywords
- cross
- tensor
- source
- sound
- sound source
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S5/00—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations
- G01S5/18—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations using ultrasonic, sonic, or infrasonic waves
- G01S5/22—Position of source determined by co-ordinating a plurality of position lines defined by path-difference measurements
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Measurement Of Velocity Or Position Using Acoustic Or Ultrasonic Waves (AREA)
Abstract
The invention discloses a broadband multi-sound-source positioning asynchronous measurement method, a device and a related medium, wherein the method comprises the following steps: setting a virtual microphone array, and carrying out multiple sound source measurements on the virtual microphone array; carrying out frequency analysis on the multiple sound source measurement results to obtain a cross-correlation matrix; performing completion processing on the cross-correlation matrix based on rank minimization of tensor; and solving the sound source measurement by using the supplemented cross-correlation matrix, and positioning the broadband multiple sound sources according to the solved result. The invention fully utilizes the observed tensor structure to complement target data by the principle of tensor rank minimization, and can position the broadband multi-sound source by operating the asynchronous measurement method in a reasonable frequency range, thereby reducing the selection complexity of the broadband signal aiming at effective frequency points in the positioning process and reducing the difficulty aiming at complex number operation.
Description
Technical Field
The invention relates to the technical field of sound source positioning, in particular to a broadband multi-sound-source positioning asynchronous measurement method, a broadband multi-sound-source positioning asynchronous measurement device and a related medium.
Background
In practical Sound Source Localization (SSL), the operating frequency range of a microphone array of an acoustic beamforming method is always limited by the array size and the distance between two adjacent microphones. As a new approach to break this limitation, Antoni J first used the non-synchronous measurement (NSM) method for acoustic imaging in near-field acoustic holography (NAH) in the literature "Synthetic adaptation acoustic tomography" (International Conference on Noise and simulation Engineering and International, isma2012, Leuven, belgium.2012). Asynchronous measurement allows the range of operating frequencies of the array to be extended by sequentially moving a miniature prototype array during measurement to form a virtual microphone array of larger size and higher density. Unlike Synthetic Aperture Radar (SAR) or Synthetic Aperture Sonar (SAS), microphone arrays are passive arrays. Therefore, a matrix or tensor completion method is needed to obtain missing information, which is not needed in SAR and SAS.
Ning Chu et al, in the document "A fast and robust localization method for low-frequency acquisition source: variance Bayesian inference based on non-arbitrary spectral arrays" (IEEE Transactions on instruments and measures, 2020,70:1-18.) proposes a Variational Bayes (VB) inference method based on student prior and KL divergence optimization to complete a cross-spectrum matrix (CSM) to achieve NSM for acoustic beam formation at frequencies of 800Hz and 1000 Hz. The method only tests narrow-band signals, and the positioning effectiveness of the wide-band signals is not verified.
Ning F et al, in the document "Sound source localization of non-synchronous measurements with Block Hermitian Matrix Completion (BHMC) method," Mechanical Systems and Signal Processing,2021,147:107118. The method utilizes the characteristics of a related spectrum matrix and matrix decomposition and inverse operation rules. By means of spatial continuity, the matrix can be completed without iteration, and the calculation amount is reduced to a certain extent. However, the method only tests signals of 3000Hz-5000Hz, and the experimental result shows that certain errors still exist in the positioning precision. In addition, the method does not verify the broadband voice signal, which requires precise selection of frequency points for wave number formation.
Ning Chu et al in the literature "Non-Synchronous Measurements of a microphone array at Coprime Positions" (IEEE Signal Processing Letters,2021,28: 1420-. However, this method still does not break through the limitation of NSM on the frequency point selection of wideband speech signals.
Therefore, although the NSM method can widen the operating frequency range of acoustic beamforming by virtualizing a larger and denser array, the frequency point selection problem of broadband sound source localization is still not effectively solved without the frequency domain prior information of the target sound signal.
Disclosure of Invention
The embodiment of the invention provides a broadband multi-sound-source positioning asynchronous measurement method, a broadband multi-sound-source positioning asynchronous measurement device, computer equipment and a storage medium, and aims to reduce the selection complexity of effective frequency points of broadband signals in the positioning process and reduce the difficulty of complex number operation.
In a first aspect, an embodiment of the present invention provides a broadband multi-sound-source-localization asynchronous measurement method, including:
setting a virtual microphone array, and carrying out multiple sound source measurements on the virtual microphone array;
carrying out frequency analysis on the multiple sound source measurement results to obtain a cross-correlation matrix;
performing completion processing on the cross-correlation matrix based on rank minimization of tensor;
and solving the sound source measurement by using the supplemented cross-correlation matrix, and positioning the broadband multiple sound sources according to the solved result.
In a second aspect, an embodiment of the present invention provides a broadband multi-sound-source-localization asynchronous measurement apparatus, including:
the sound source measuring unit is used for setting a virtual microphone array and carrying out multiple sound source measurements on the virtual microphone array;
the frequency analysis unit is used for carrying out frequency analysis on the multiple sound source measurement results to obtain a cross-correlation matrix;
a completion processing unit, configured to perform completion processing on the cross-correlation matrix based on rank minimization of a tensor;
and the solving and positioning unit is used for solving the sound source measurement by using the completed cross-correlation matrix and carrying out broadband multi-sound-source positioning according to the solving result.
In a third aspect, an embodiment of the present invention provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor, when executing the computer program, implements the broadband multi-sound source localization asynchronous measurement method according to the first aspect.
In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the wideband multi-localization asynchronous measurement method according to the first aspect.
The embodiment of the invention provides a broadband multi-sound-source positioning asynchronous measurement method, a broadband multi-sound-source positioning asynchronous measurement device, computer equipment and a storage medium, wherein the method comprises the following steps: setting a virtual microphone array, and carrying out multiple sound source measurements on the virtual microphone array; carrying out frequency analysis on the multiple sound source measurement results to obtain a cross-correlation matrix; performing completion processing on the cross-correlation matrix based on rank minimization of tensor; and solving the sound source measurement by using the supplemented cross-correlation matrix, and positioning the broadband multiple sound sources according to the solved result. According to the embodiment of the invention, the observed tensor structure is fully utilized to complement target data through the tensor rank minimization principle, and the asynchronous measurement method is operated in a reasonable frequency range, so that the broadband multi-sound source can be positioned, and therefore, the selection complexity of the broadband signal for effective frequency points in the positioning process is reduced, and the difficulty for complex number operation is reduced.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a schematic flowchart of a broadband multi-sound source localization asynchronous measurement method according to an embodiment of the present invention;
fig. 2 is a schematic diagram of decomposition of singular values of tensors in a broadband multi-source localization asynchronous measurement method according to an embodiment of the present invention;
fig. 3 is a schematic diagram of a simulation experiment of a broadband multi-sound-source-localization asynchronous measurement method according to an embodiment of the present invention;
FIG. 4 is a graph showing the results of the simulation experiment of FIG. 3;
FIG. 5 is a diagram illustrating the effect of the compensated simulation experiment tensor shown in FIG. 3;
fig. 6 is a schematic diagram illustrating an effect of a broadband multi-source localization asynchronous measurement method according to an embodiment of the present invention;
fig. 7 is a schematic diagram of a site experiment of a broadband multi-sound-source-localization asynchronous measurement method according to an embodiment of the present invention;
FIG. 8 is a graphical representation of the results of the field experiment of FIG. 7;
fig. 9 is a schematic block diagram of a broadband multi-sound-source-localization asynchronous measurement apparatus according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It will be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
It is also to be understood that the terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting. As used in this specification and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.
It should be further understood that the term "and/or" as used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.
Referring to fig. 1, fig. 1 is a schematic flow chart of a broadband multi-sound-source-localization asynchronous measurement method according to an embodiment of the present invention, which specifically includes: steps S101 to S104.
S101, setting a virtual microphone array, and carrying out multiple sound source measurements on the virtual microphone array;
s102, carrying out frequency analysis on the multiple sound source measurement results to obtain a cross-correlation matrix;
s103, performing completion processing on the cross-correlation matrix based on rank minimization of tensor;
and S104, solving the sound source measurement by using the supplemented cross-correlation matrix, and positioning the broadband multi-sound source according to the solving result.
In the embodiment, multiple sound source measurements are realized through the virtual microphone array, then the cross-correlation matrix is constructed according to the frequency analysis of the sound source measurements, and then the cross-correlation matrix is complemented by utilizing the rank minimization of the tensor, so that the broadband multi-sound-source positioning is realized.
According to the embodiment, a tensor rank is defined on the basis of a traditional rank minimization method, the target data is supplemented by fully utilizing an observed tensor structure through the tensor rank minimization principle, and the broadband multi-sound source can be positioned by operating the NSM method in a reasonable frequency range. Compared with the conventional broadband signal positioning method, the method provided by the embodiment reduces the selection complexity of the broadband signal, especially the voice signal, aiming at the effective frequency point in the positioning process, reduces the difficulty of complex number operation, and effectively meets the requirement on real-time performance in the actual application scene. In addition, the embodiment effectively inhibits the sidelobe energy in the beam forming process, so that the finally expressed broadband multi-information-source positioning effect graph has higher energy concentration expression, thereby reducing the selection complexity of the broadband signal for effective frequency points in the positioning process and reducing the difficulty for complex number operation.
In one embodiment, the step S101 includes:
setting M microphones to form an initial array, moving the initial array, and performing K times of asynchronous measurement on a plurality of sound sources in the initial array to form a virtual microphone array;
acquiring a sound source signal P of the kth asynchronous measurement according to the following formula k (r m )：
In the formula, r m Representing the distance between the array element and the sound source,
respectively representing the 1 st, 2 nd and mth sound source signals.
In this embodiment, for the near-field asynchronous measurement model, assuming that there is a prototype array consisting of M microphones in the sound field, the prototype array is sequentially moved, and K measurements are performed on a plurality of sound sources in the sound field, then the model can form a virtual array consisting of M × K microphones. And the virtual array has a larger array aperture and a higher array element density than the prototype array. So the kth measurement of an unsynchronized measurement can yield the following signal:
wherein r is m Representing the distance between the array element and the source.
In one embodiment, the step S102 includes:
the k-th nonsynchronized measured sound source signal P is calculated according to k (r m ) Performing frequency analysis to obtain cross-spectrum matrix
R k (f)＝E[p k (f)p k (f) H ]
In the formula, p k (f) Is to P k (r m ) F is a frequency point of data obtained after discrete Fourier transform;
k cross-spectral matrices R are obtained based on k unsynchronized measurements k (K-1, 2, … K) and constructing the cross-correlation matrix R from K cross-spectral matrices Ω ：
In this embodiment, based on the formula (1), the cross-spectrum matrix obtained by the kth measurement can be obtained
Comprises the following steps:
R k (f)＝E[p k (f)p k (f) H ] (2)
wherein p is k (f) Is to P k (r m ) And f is a frequency point of data obtained after discrete Fourier transform.
For the asynchronous measurement method of K times of measurement, K sub-cross-spectrum matrixes R can be obtained when the analysis frequency is f k (K-1, 2, … K), this results in non-synchronized measurements occurring between them, since there is no overlap in the time of each measurement by the non-synchronized measurement techniqueThe phenomenon of phase loss. Thus, the following cross-correlation matrix can be derived from the non-synchronous measurement technique:
the cross-correlation matrix has valid information only at diagonal positions, and other positions cannot be obtained through measurement due to information loss. Asynchronous measurements become a problem for cross-spectral matrix completion.
In one embodiment, the step S103 includes:
setting a cross-spectral tensor of a broadband signal
Wherein Ω is an operator of sampling diagonal block position data, and the cross-spectrum tensor is composed of S
And splicing in a third dimension according to the following formula:
wherein S is the total number of the selected frequency points,
and performing splicing operation on the representation cross-spectrum tensor in a third dimension.
In general, for positioning of a broadband source, it is difficult to correctly select an effective frequency point on the premise that no prior information exists. To address this problem, the present embodiment proposes a tensor-based rank minimization (TMN) cross-spectral matrix completion method. Definition of
Is a broadband signal cross-spectrum tensor generated by asynchronous measurement, wherein omega is an operator for sampling diagonal block position data, and the tensor is composed of S
In the third dimension, i.e.
Wherein S is the total number of selected frequency points and symbols
And the representative tensor is spliced in the third dimension.
Further, in an embodiment, the step S103 further includes:
selecting a minimized estimated cross-spectral tensor according to
Rank:
where lambda is a parameter of the regularization term,
is composed of
Item s of front side slicing | | · | luminance F Is the F norm, the tensor y is the lagrange multiplier, and μ is the penalty parameter term.
The present embodiment estimates the cross-spectral tensor by selecting the minimization
The rank is such that the optimized tensor is such that the data at the diagonal block positions is as consistent as possible with the observed data while the rank is minimized, i.e., the data at the diagonal block positions is as consistent as possible
Where lambda is a parameter of the regularization term,
is composed of
Item s of the frontal slice. I | · | purple wind F Is the F norm. The augmented Lagrangian function of equation (5) is then:
and define<A,B>＝vec{A} H vec { B }, vec { · } is a tensor vectorization operator. Tensor y is a lagrange multiplier and μ is a penalty parameter, and the optimization terms are iterated in turn using an Alternating Direction Multiplier Method (ADMM). Namely, it is
Can be updated as follows:
for three-dimensional tensors
Its t-SVD can be expressed as (as shown in FIG. 2):
wherein D is r Min { D1, D2}, is the tensor product. Similar to the matrix SVD decomposition, the tensors U and V are orthogonal, i.e. there are:
and pairIn the unit tensor I, only the first slice of the third dimension is the unit matrix, and the element values of the matrix of other slices are all 0. The tensor S is an F-diagonal tensor, and the third-dimensional slice matrixes of the tensor S are diagonal matrixes and contain all singular values of the tensor X. By FFT calculation of X
Wherein fft (X, etc]And 3) all other columns of the X are subjected to fast Fourier transform along the third dimension, namely, fast Fourier transform operation is carried out on the third dimension of the tensor. And converting the signal into a Fourier domain to carry out t-SVD decomposition so as to simplify the calculation process. And performing inverse fast Fourier transform on all the calculated results to obtain a decomposition result of the original tensor X. In contrast to the unit tensor I, the unit tensor of the Fourier domain
Each slice of its third dimension is an identity matrix. Referring to equation (8), when the tensor rank minimization method is used to recover the required tensor
In time, there are:
wherein r is d The number of non-zero ranks representing the d-th slice S (: d) of the rank tensor S in the third dimension, d ═ 1,2, …, S. the calculation procedure for t-SVD is as follows:
further, in an embodiment, the step S103 further includes:
according to the distribution optimization characteristics of the ADMM algorithm, carrying out iterative optimization on tensor M and tensor y according to the following formula:
wherein, gamma is a relaxation parameter,
representing a sampling operator;
setting a stopping condition for the iterative optimization according to the following formula:
according to the tensor discrete Fourier transform definition, the iterative formula (7) is equal to the format after FFT calculation, and then the optimization formula is converted into a singular value threshold truncation iterative calculation method. Then the optimization equation (7) can be updated as:
wherein, the formula
Is tensor SVD decomposition, subscript S is 1,2, …, S represents the selected wideband frequency bin sequence. In fact, as can be seen from equation (10), the operation of t-SVD on a tensor is to perform the FFT processing and then perform the decomposition operation on its matrix slices one by one. Different from the common matrix completion method based on the singular value threshold truncation method, the tensor is diagonalized by using FFT in the tensor cross-correlation spectrum completion, so that the t-SVD algorithm is equivalent to processing a rearranged matrix of the tensor.
According to the characteristics of step-by-step optimization of the ADMM algorithm, tensors M and y can be optimized as follows:
where γ is the set relaxation parameter. Since the iterative optimization formula (11) has a sampling operator Ω for the diagonal block elements of the tensor, it is optimized in two steps:
(symbol)
the elements representing the sampling operator, i.e. the positions of the non-diagonal blocks of the sampling tensor. When the algorithm can obtain a convergence value, setting the stopping condition as follows:
if the final result obtained by the optimization algorithm can not meet the requirement of (15), setting the stopping condition of the algorithm as N as the maximum iteration number m . To make the optimization algorithm more reasonable, the singular value threshold in equation (10)
Should be decreased as the number of iterations proceeds, the decreasing coefficient α is set so as to satisfy
Updating lambda ′(k+1) ＝αλ ′(k) And setting the minimum value thereof to be lambda m ' -0.0001. Reducing counts by discarding singular values less than a thresholdThe rank of the tensor is computed, thereby reducing the presence of noise subspaces. Furthermore, considering the requirement of spatial continuity, by introducing a spatial projection basis
Setting up
Wherein S represents a frequency bin sequence, and satisfies S1, 2, …, S.
Therefore, the flow of the cross-correlation spectrum tensor completion algorithm based on the ADMM provided by this embodiment is as follows:
in one embodiment, the step S104 includes:
using the wideband multi-signal split pair tensor as follows
From f 0 To f S Weighting the cross-correlation spectrum matrixes of all frequency points to realize broadband multi-sound source positioning:
in the formula, w (f, r) m ) For the spatial scanning vector r of the sound field at different frequency points f m Is the distance between the spatial scanning point and the array, U (f, r) m ) n Is the noise subspace eigenvector of the cross-correlation spectrum matrix at frequency f.
In this embodiment, the proposed cross-correlation spectrum tensor completion method is used to recover the calculation
Then, the tensor is subjected to Broadband multi-signal classification (Broadband-MUSIC)
From f 0 To f S Is weighted by the cross-correlation spectrum matrix of all frequency points, i.e.
In the formula, w (f, r) m ) For the spatial scanning vector r of the sound field at different frequency points f m Is the distance between the spatial scan point and the array, U (f, r) m ) n Is the noise subspace eigenvector of the cross-correlation spectrum matrix at frequency f.
The embodiment of the invention is based on an optimization method of ADMM, and an incomplete cross-correlation spectrum matrix generated by asynchronous measurement is completed on the basis of satisfying rank minimization by using tensor singular value decomposition (T-SVD). Compared with the conventional broadband signal positioning method, the method reduces the complexity of selecting the effective frequency points of the broadband signals, particularly the voice signals, in the positioning process, reduces the difficulty of complex number operation, and effectively meets the requirement on real-time performance in practical application scenes. The invention effectively inhibits the sidelobe energy in the beam forming process, so that the finally expressed broadband multi-information-source positioning effect graph has higher energy concentration expression.
The performance of the invention can be illustrated by simulation experiments, wherein the simulation platform is MATLAB, and the simulation parameters are shown in Table 1:
TABLE 1
The schematic diagram of the simulation experiment is shown in fig. 3, and the sound field is measured 5 times in the simulation test field by using an asynchronous measurement mode, and the sampling frequency is 5s every time at 16 KHz. The results of fig. 4 show that when the sound source distribution is dispersed and the array shape is too small, even with the method of wideband MUSIC, it is difficult to accurately locate all the sound sources simultaneously in a single measurement. As shown in fig. 5, after the cross-correlation spectrum tensor is complemented in the present invention, if a conventional narrowband MUSIC method is used to locate multiple sources, it is difficult for a single frequency point to simultaneously locate all the sources in the sound field in view of the difference in frequency distribution of the sources, and even at some analysis frequencies, the obtained result has strong noise interference. And for a wideband speech signal, when the analysis frequency is higher than a certain level, the distribution of signal energy will gradually decrease. In fact, it is also an option to complete the cross-correlation spectrum matrix of each frequency point by using the conventional ADMM, but the tensor completion method provided by the present invention adopts an overall rank minimization strategy, so that the incomplete cross-correlation spectrum matrix generated by the asynchronous measurement is recovered more effectively and accurately, and the result can be seen in fig. 6. The method of the invention more effectively suppresses noise and focuses the energy of the sound source more.
In order to verify the effect of the invention, experiments on actual sites were performed, and the experimental parameters are shown in table 2:
TABLE 2
The site is schematically shown in FIG. 7, and the experimental results are shown in FIG. 8 and Table 3. The result shows that compared with the conventional broadband signal positioning method, the algorithm provided by the invention reduces the selection complexity of the broadband signal, particularly the voice signal, aiming at the effective frequency point in the positioning process, reduces the difficulty aiming at complex number operation, and effectively meets the requirement on real-time property in the actual application scene. The invention has higher energy concentration performance on the finally expressed broadband multi-information-source positioning effect graph by effectively inhibiting the side lobe energy in the beam forming process.
TABLE 3
Fig. 9 is a schematic block diagram of a broadband multi-sound-source-localization asynchronous measurement apparatus 900 according to an embodiment of the present invention, where the apparatus 900 includes:
a sound source measuring unit 901, configured to set a virtual microphone array, and perform multiple sound source measurements on the virtual microphone array;
a frequency analysis unit 902, configured to perform frequency analysis on the multiple sound source measurement results to obtain a cross-correlation matrix;
a completion processing unit 903, configured to perform completion processing on the cross-correlation matrix based on rank minimization of a tensor;
and a solving and positioning unit 904, configured to solve the sound source measurement by using the completed cross-correlation matrix, and perform broadband multi-sound-source positioning according to a solving result.
In an embodiment, the sound source measuring unit 901 includes:
the array setting unit is used for setting M microphones to form an initial array, moving the initial array and carrying out K times of asynchronous measurement on a plurality of sound sources in the initial array so as to form the virtual microphone array;
a signal acquisition unit for acquiring the kth nonsynchronous measurement sound source signal P according to the following formula k (r m )：
In the formula, r m Representing the distance between the array element and the sound source,
respectively representing the 1 st, 2 nd and mth sound source signals.
In one embodiment, the frequency analysis unit 902 includes:
a first matrix building unit for the kth asynchronously measured sound source signal P according to k (r m ) Performing frequency analysis to obtain cross-spectrum matrix
R k (f)＝E[p k (f)p k (f) H ]
In the formula, p k (f) Is to P k (r m ) F is a frequency point of data obtained after discrete Fourier transform;
a second matrix establishing unit for obtaining k cross-spectrum matrices R based on k times of asynchronous measurement k (K-1, 2, … K) and constructing the cross-correlation matrix R from K cross-spectral matrices Ω ：
In one embodiment, the completion processing unit 903 comprises:
a tensor setting unit for setting a cross-spectrum tensor of the broadband signal
Wherein Ω is an operator for sampling diagonal block position data, and the cross-spectrum tensor is composed of S
And splicing in a third dimension according to the following formula:
wherein S is the total number of the selected frequency points,
and performing splicing operation on the representation cross-spectrum tensor in a third dimension.
In an embodiment, the completion processing unit 903 further includes:
a selection unit for selecting the minimized estimated cross-spectral tensor according to
Rank:
where lambda is a parameter of the regularization term,
is composed of
Item s of front section, | · | | non-woven phosphor F Is the F norm, the tensor y is the lagrange multiplier, and μ is the penalty parameter term.
In an embodiment, the completion processing unit 903 further includes:
the iterative optimization unit is used for performing iterative optimization on the tensor M and the tensor y according to the distribution optimization characteristics of the ADMM algorithm and the following formula:
wherein, gamma is a relaxation parameter,
representing a sampling operator;
a condition setting unit configured to set a stop condition for the iterative optimization according to:
in an embodiment, the solution positioning unit 904 comprises:
a weighting unit for utilizing the broadband multi-signal splitting pair tensor according to
From f 0 To f S Weighting the cross-correlation spectrum matrixes of all frequency points to realize broadband multi-sound-source positioning:
in the formula, w (f, r) m ) For the spatial scanning vector r of the sound field at different frequency points f m Is the distance between the spatial scanning point and the array, U (f, r) m ) n Is the noise subspace eigenvector of the cross-correlation spectrum matrix at frequency f.
Since the embodiment of the apparatus portion and the embodiment of the method portion correspond to each other, please refer to the description of the embodiment of the method portion for the embodiment of the apparatus portion, and details are not repeated here.
Embodiments of the present invention further provide a computer-readable storage medium, on which a computer program is stored, where the computer program can implement the steps provided in the foregoing embodiments when executed. The storage medium may include: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
The embodiment of the present invention further provides a computer device, which may include a memory and a processor, where the memory stores a computer program, and the processor may implement the steps provided in the above embodiments when calling the computer program in the memory. Of course, the computer device may also include various network interfaces, power supplies, and the like.
The embodiments are described in a progressive manner in the specification, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description. It should be noted that, for those skilled in the art, it is possible to make several improvements and modifications to the present application without departing from the principle of the present application, and such improvements and modifications also fall within the scope of the claims of the present application.
It is further noted that, in the present specification, relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
Claims (10)
1. A broadband multi-sound-source positioning asynchronous measurement method is characterized by comprising the following steps:
setting a virtual microphone array, and carrying out multiple sound source measurements on the virtual microphone array;
carrying out frequency analysis on the multiple sound source measurement results to obtain a cross-correlation matrix;
performing completion processing on the cross-correlation matrix based on rank minimization of tensor;
and solving the sound source measurement by using the supplemented cross-correlation matrix, and positioning the broadband multiple sound sources according to the solved result.
2. The broadband multi-sound-source-localization asynchronous measurement method according to claim 1, wherein the setting of the virtual microphone array and the multiple sound source measurements of the virtual microphone array comprise:
setting M microphones to form an initial array, moving the initial array, and performing K times of asynchronous measurement on a plurality of sound sources in the initial array to form the virtual microphone array;
acquiring a sound source signal P of the kth asynchronous measurement according to the following formula k (r m )：
In the formula, r m Representing the distance between the array element and the sound source,
respectively representing the 1 st, 2 nd and mth sound source signals.
3. The method of claim 2, wherein the frequency analyzing the multiple acoustic source measurements to obtain a cross-correlation matrix comprises:
the source signal P of the kth unsynchronized measurement is measured according to k (r m ) Performing frequency analysis to obtain cross-spectrum matrix
R k (f)＝E[p k (f)p k (f) H ]
In the formula, p k (f) Is to P k (r m ) F is a frequency point of data obtained after discrete Fourier transform;
k cross-spectral matrices R are obtained based on k unsynchronized measurements k (K-1, 2, … K) and constructing the cross-correlation matrix R from K cross-spectral matrices Ω ：
4. The wideband multi-source localization asynchronous measurement method according to claim 3, wherein the tensor-based rank minimization completes the cross-correlation matrix and comprises:
setting a cross-spectral tensor of a broadband signal
Wherein Ω is an operator for sampling diagonal block position data, and the cross-spectrum tensor is composed of S
And splicing in a third dimension according to the following formula:
wherein S is the total number of the selected frequency points,
and performing splicing operation on the representation cross-spectrum tensor in a third dimension.
5. The wideband multi-source localization asynchronous measurement method according to claim 4, wherein the tensor-based rank minimization completes the cross-correlation matrix, and further comprising:
selecting a minimized estimated cross-spectral tensor according to
Rank:
where lambda is a parameter of the regularization term,
is composed of
Item s of front section, | · | | non-woven phosphor F Is the F norm, the tensor y is the lagrange multiplier, and μ is the penalty parameter term.
6. The wideband multi-source localization asynchronous measurement method according to claim 5, wherein the tensor-based rank minimization completes the cross-correlation matrix, and further comprising:
according to the distribution optimization characteristics of the ADMM algorithm, carrying out iterative optimization on tensor M and tensor y according to the following formula:
wherein, gamma is a relaxation parameter,
representing a sampling operator;
setting a stopping condition for the iterative optimization according to the following formula:
7. the broadband multi-sound-source-localization asynchronous measurement method according to claim 6, wherein the solving of the sound source measurement by using the complemented cross-correlation matrix and the broadband multi-sound-source localization according to the solving result comprises:
using the wideband multi-signal split pair tensor as follows
From f 0 To f S Weighting the cross-correlation spectrum matrixes of all frequency points to realize broadband multi-sound-source positioning:
in the formula, w (f, r) m ) For the spatial scanning vector r of the sound field at different frequency points f m Is the distance between the spatial scanning point and the array, U (f, r) m ) n Is the noise subspace eigenvector of the cross-correlation spectrum matrix at frequency f.
8. A broadband multi-source positioning asynchronous measurement device, comprising:
the sound source measuring unit is used for setting a virtual microphone array and carrying out multiple sound source measurements on the virtual microphone array;
the frequency analysis unit is used for carrying out frequency analysis on the multiple sound source measurement results to obtain a cross-correlation matrix;
a completion processing unit, configured to perform completion processing on the cross-correlation matrix based on rank minimization of a tensor;
and the solving and positioning unit is used for solving the sound source measurement by using the completed cross-correlation matrix and carrying out broadband multi-sound-source positioning according to the solving result.
9. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the wideband multi-source localization non-synchronous measurement method according to any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, wherein the computer-readable storage medium has a computer program stored thereon, and when the computer program is executed by a processor, the method for broadband multi-sound-source-localization non-synchronous measurement according to any one of claims 1 to 7 is implemented.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210625448.7A CN115032591A (en) | 2022-06-02 | 2022-06-02 | Broadband multi-sound-source positioning asynchronous measurement method and device and related medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210625448.7A CN115032591A (en) | 2022-06-02 | 2022-06-02 | Broadband multi-sound-source positioning asynchronous measurement method and device and related medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115032591A true CN115032591A (en) | 2022-09-09 |
Family
ID=83122412
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210625448.7A Pending CN115032591A (en) | 2022-06-02 | 2022-06-02 | Broadband multi-sound-source positioning asynchronous measurement method and device and related medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115032591A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115508780A (en) * | 2022-11-23 | 2022-12-23 | 杭州兆华电子股份有限公司 | Synthetic aperture acoustic imaging method |
| CN117825898A (en) * | 2024-03-04 | 2024-04-05 | 国网浙江省电力有限公司电力科学研究院 | GIS distributed vibration and sound combined monitoring method, device and medium |
| CN117825898B (en) * | 2024-03-04 | 2024-06-11 | 国网浙江省电力有限公司电力科学研究院 | GIS distributed vibration and sound combined monitoring method, device and medium |
-
2022
- 2022-06-02 CN CN202210625448.7A patent/CN115032591A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115508780A (en) * | 2022-11-23 | 2022-12-23 | 杭州兆华电子股份有限公司 | Synthetic aperture acoustic imaging method |
| CN117825898A (en) * | 2024-03-04 | 2024-04-05 | 国网浙江省电力有限公司电力科学研究院 | GIS distributed vibration and sound combined monitoring method, device and medium |
| CN117825898B (en) * | 2024-03-04 | 2024-06-11 | 国网浙江省电力有限公司电力科学研究院 | GIS distributed vibration and sound combined monitoring method, device and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8874439B2 (en) | Systems and methods for blind source signal separation | |
| CN107247251B (en) | Three-dimensional sound source positioning method based on compressed sensing | |
| CN109407045B (en) | Non-uniform sensor array broadband signal direction-of-arrival estimation method | |
| KR100959050B1 (en) | System and method for generating a separated signal | |
| US10237676B2 (en) | Sparse decomposition of head related impulse responses with applications to spatial audio rendering | |
| CN109616138B (en) | Voice signal blind separation method based on segmented frequency point selection and binaural hearing aid system | |
| Wang et al. | A region-growing permutation alignment approach in frequency-domain blind source separation of speech mixtures | |
| CN109343003B (en) | Method for identifying sound source formed by fast iterative shrinking wave beams | |
| US10818302B2 (en) | Audio source separation | |
| CN101667425A (en) | Method for carrying out blind source separation on convolutionary aliasing voice signals | |
| CN105556260B (en) | broadband acoustical holography | |
| CN110109058A (en) | A kind of planar array deconvolution identification of sound source method | |
| CN104539340A (en) | Steady direction of arrival estimation method based on sparse representation and covariance fitting | |
| CN115032591A (en) | Broadband multi-sound-source positioning asynchronous measurement method and device and related medium | |
| CN116068493A (en) | Passive sound source positioning method for deep sea large-depth vertical distributed hydrophone | |
| CN113805139A (en) | Broadband signal sparse representation direction-of-arrival estimation method based on focusing transformation | |
| Chen et al. | Broadband sound source localisation via non-synchronous measurements for service robots: A tensor completion approach | |
| Maazaoui et al. | Adaptive blind source separation with HRTFs beamforming preprocessing | |
| JP5387442B2 (en) | Signal processing device | |
| Mazur et al. | Robust room equalization using sparse sound-field reconstruction | |
| Fontaine et al. | Scalable source localization with multichannel α-stable distributions | |
| Al-Sadoon et al. | Construction of projection matrices based on non-uniform sampling distribution for AoA estimation | |
| CN113406560A (en) | Angle and frequency parameter estimation method of incoherent distributed broadband source | |
| Ning et al. | Acoustic imaging with compressed sensing and microphone arrays | |
| Mallis et al. | Convolutive audio source separation using robust ICA and an intelligent evolving permutation ambiguity solution |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |