An improved function beamforming method and related device

Abstract

The present invention provides an improved function beamforming method and related devices, which are related to sound source identification based on a microphone array. By establishing an array signal receiving model; calculating the FB output according to the array signal receiving model; calculating the PSF matrix of power-raising space transfer invariance according to the array signal receiving model; determining the spatial convolution information according to the FB output and the PSF matrix; transforming the spatial convolution information into a wave number domain product through FFT operation under periodic boundary conditions; constructing a difference function according to the wave number domain product and the FB output; calculating the difference function through a fast iterative shrinkage threshold method, solving the wave number leakage problem caused by the zero-filled boundary, improving the calculation accuracy, and further improving the imaging performance; sharpening the main lobe of the point spread function through exponential operation, and expanding the applicability of the assumption of spatial transfer invariance of the point spread function. Improve the imaging spatial resolution and dynamic range, and expand the effective imaging area. A technical effect of avoiding performance loss in the process of sound source identification is achieved.

Description

Technical Field

[0001] The present application relates to the technical field of sound source identification based on microphone arrays, and in particular to an improved function beamforming method and related devices. Background Art

[0002] The sound source identification technology based on microphone array has the advantages of fast measurement speed, high calculation efficiency, and suitability for medium and long distance measurement. It has become a common method for noise source identification in industrial fields such as gas leakage, mechanical fault detection, and high-speed trains, wind turbines, and automobiles. Among them, conventional beamforming (CBF) is the basic algorithm of beamforming. It has the advantages of fast calculation speed, long measurement distance, and easy deployment. However, due to the limited array aperture and signal frequency, its spatial positioning resolution is low and the side lobes are prominent, resulting in many false sound sources in the imaging, which limits its application. In order to improve the imaging performance, experts and scholars at home and abroad have proposed many improved algorithms, mainly including functional beamforming (FB) algorithm, deconvolution algorithm and fast deconvolution algorithm.

[0003] The fast deconvolution beamforming algorithm is based on the assumption of PSF spatial transfer invariance and Fast Fourier Transform (FFT). The main algorithms include DAMAS2, DAMAS3, FFT-NNLS, and FFT-FISTA. FFT is used to convert the spatial convolution of the sound source distribution and PSF into the product of the wave number domain to improve the computational efficiency.

[0004] Compared with deconvolution algorithms, fast deconvolution algorithms ignore the changes in PSF spatial transfer and have warping errors caused by FFT operations, resulting in a wider imaging main lobe and limited dynamic performance. The warping error caused by FFT needs to be suppressed by the boundary zero padding method. However, zero padding during the FFT operation may cause spectrum leakage, which is extended to the wavenumber domain, that is, wavenumber leakage. When the sound source point is far away from the center of the focal plane, there will be errors in the positioning result. At present, an effective way to expand the imaging range of the sound source is to introduce an irregular clustering point grid generation method to improve the invariance of the actual PSF spatial transfer, but compared with deconvolution algorithms, its imaging performance is inferior.

[0005] Therefore, how to avoid performance loss in the sound source identification process has become a technical problem that needs to be solved urgently. Summary of the invention

[0006] In order to avoid performance loss during sound source identification, the present application provides an improved function beamforming method and related devices.

[0007] In the first aspect, the present application provides an improved method for function beamforming using the following technical solution:

[0008] A function beamforming improvement method, comprising:

[0009] Establish array signal receiving model;

[0010] Calculate FB output according to the array signal receiving model;

[0011] Calculate a PSF matrix with invariant power spatial transfer according to the array signal receiving model;

[0012] Determine spatial domain convolution information according to the FB output and the PSF matrix;

[0013] According to the spatial domain convolution information, the information is transformed into a wave number domain product through FFT operation under periodic boundary conditions;

[0014] constructing a difference function according to the wavenumber domain product and the FB output;

[0015] The difference function is calculated by a fast iterative shrinkage threshold method to determine the true sound source distribution.

[0016] Optionally, the step of calculating the FB output according to the array signal receiving model includes:

[0017] Calculate based on the array signal Dimensional CSM;

[0018] Regarding the dimensional CSM to perform eigenvalue decomposition to determine the corresponding cross-spectral index function;

[0019] The FB output is calculated according to the cross-spectral index function.

[0020] Optionally, the step of calculating a PSF matrix with invariant power-raising spatial transfer according to the array signal reception model comprises:

[0021] The array signal receiving model is combined with the sound source incoherence assumption to determine the The target cross-spectral matrix corresponding to the dimensional CSM;

[0022] According to the relationship between the FB output and the power point spread function, the PSF matrix that is invariant to the power space transfer is calculated based on the Loener-Henry inequality.

[0023] Optionally, the step of transforming the spatial domain convolution information into a wave number domain product through FFT operation under periodic boundary conditions includes:

[0024] Introducing periodic boundary conditions;

[0025] The error is uniformly convolved in the spatial domain convolution information by combining the periodic boundary conditions through FFT operation to transform into a wavenumber domain product.

[0026] Optionally, the step of constructing a difference function according to the wavenumber domain product and the FB output comprises:

[0027] Determine a periodic power-raising space transfer invariant matrix according to the wave number domain product;

[0028] Acquire the sound source distribution, and construct a difference function according to the FB output and the periodic power-raising space transfer invariant matrix.

[0029] Optionally, the expression of the difference function is:

[0030]

[0031] in, represents the Frobenius norm, is the periodic power-raising space transfer invariant matrix, is the sound source distribution, Output result for FB.

[0032] Optionally, the step of calculating the difference function by a fast iterative shrinkage threshold method to determine the real sound source distribution includes:

[0033] By the difference function about Iterate the search in the negative gradient direction to obtain ;

[0034] Set the initial matrix ,make , , perform iterative calculation:

[0035]

[0036]

[0037]

[0038] in, represents the Euclidean projection in the non-negative quadrant, is the Lipschitz constant, Equal to the matrix The maximum eigenvalue, gradient The expression is

[0039]

[0040] Obtain iterative calculation results and Remove boundaries to determine the true sound source distribution.

[0041] In a second aspect, the present application provides a function beamforming improvement device, comprising:

[0042] A model building module, used to build an array signal receiving model;

[0043] An FB output module, used for calculating FB output according to the array signal receiving model;

[0044] A PSF matrix module, used for calculating a PSF matrix with invariant power space transfer according to the array signal receiving model;

[0045] A spatial convolution information module, used to determine spatial convolution information according to the FB output and the PSF matrix;

[0046] A wave number domain product module, used for transforming the spatial domain convolution information into a wave number domain product through FFT operation under periodic boundary conditions;

[0047] A difference function module, used for constructing a difference function according to the wave number domain product and the FB output;

[0048] The sound source distribution module is used to calculate the difference function by a fast iterative shrinkage threshold method to determine the real sound source distribution.

[0049] In a third aspect, the present application provides a computer device, comprising: a memory and a processor, wherein the processor executes the method described above when running computer instructions stored in the memory.

[0050] In a fourth aspect, the present application provides a computer-readable storage medium, comprising instructions, which, when executed on a computer, enable the computer to execute the method described above.

[0051] In summary, the present application includes the following beneficial technical effects:

[0052] This application establishes an array signal receiving model; calculates the FB output according to the array signal receiving model; calculates the PSF matrix of power-raising spatial transfer invariance according to the array signal receiving model; determines the spatial convolution information according to the FB output and the PSF matrix; transforms the spatial convolution information into a wavenumber domain product through FFT operation under periodic boundary conditions; constructs a difference function according to the wavenumber domain product and the FB output; calculates the difference function through a fast iterative shrinkage threshold method, solves the wavenumber leakage problem caused by the zero-filled boundary, improves the calculation accuracy, and further improves the imaging performance; sharpens the main lobe of the point spread function through exponential operation, and expands the applicability of the assumption of spatial transfer invariance of the point spread function. Improve the imaging spatial resolution and dynamic range, and expand the effective imaging area. Achieve the technical effect of avoiding performance loss in the process of sound source identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a schematic diagram of the computer device structure of the hardware operating environment involved in the embodiment of the present application;

[0054] Figure 2 It is a flowchart of the first embodiment of the improved function beamforming method of the present application;

[0055] Figure 3 The microphone array receiving signal model of the first embodiment of the improved function beamforming method of the present application;

[0056] Figure 4 This is an algorithm principle diagram of the first embodiment of the improved function beamforming method of the present application;

[0057] Figure 5 It is a structural block diagram of the first embodiment of the function beamforming improvement device of the present application. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below through the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0059] Reference Figure 1 , Figure 1 A schematic diagram of the computer device structure of the hardware operating environment involved in the embodiment of the present application.

[0060] like Figure 1As shown, the computer device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (Wireless-Fidelity, Wi-Fi) interface). The memory 1005 may be a high-speed random access memory (Random Access Memory, RAM), or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk storage. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0061] Those skilled in the art will understand that Figure 1 The structure shown in the figure does not constitute a limitation on the computer device, and may include more or less components than shown in the figure, or combine some components, or arrange the components differently.

[0062] like Figure 1 As shown, the memory 1005 as a storage medium may include an operating system, a network communication module, a user interface module, and a function beamforming improvement program.

[0063] exist Figure 1 In the computer device shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in the present application can be set in the computer device, and the computer device calls the function beamforming improvement program stored in the memory 1005 through the processor 1001, and executes the function beamforming improvement method provided in the embodiment of the present application.

[0064] The present application embodiment provides an improved method for function beamforming, referring to Figure 2 , Figure 2 This is a flow chart of the first embodiment of the improved function beamforming method of the present application.

[0065] In this embodiment, the function beamforming improvement method includes the following steps:

[0066] Step S10: Establishing an array signal receiving model.

[0067] In a specific implementation, this embodiment is based on a microphone array and a structure such as Figure 3 The signal model shown. The focal plane is divided into N grid points. The planar microphone array is arranged with M array elements and the aperture is D. The frequency domain model of the array signal

[0068]

[0069] in, is the signal beat number, is the sound source frequency, for dimensional transfer matrix,

[0070]

[0071] for dimensional sound source signal vector, representing The sound source at the location is mapped to the sound source amplitude at each focusing grid point, for dimensional array vector, representing The sound source mapping from the sound source to each microphone, for Gaussian noise signal, represents the transposition operation, which will be omitted in the following text for the sake of convenience. The elements in the array vector are

[0072]

[0073] in, Focusing grid points The distance to the mth array element, is the speed of sound.

[0074] It should be noted that for One of the 1~N vectors is a matrix The specific description of the vector in the focal plane. There are N grid points in total. for dimensional sound source signal vector, representing The sound source at the location is mapped to the sound source amplitude at each focusing grid point, ~ Indicates the amplitude of the sound source mapped on 1~N grid points.

[0075] It is understandable that the FB algorithm reduces the main lobe width and suppresses the side lobes by performing exponential operations on the cross spectral matrix (CSM) of the CBF. The larger the exponent, the higher the dynamic range and spatial resolution performance of the imaging result. In practical applications, factors such as array errors and phase mismatches cause the actual array model to be inconsistent with the theoretical model. Excessively high exponents will cause exponential amplification of algorithm errors, limiting the FB algorithm's ability to improve spatial resolution performance.

[0076] Deconvolution beamforming algorithms mainly include the deconvolution approach for the mapping of acoustic sources (DAMAS) algorithm, the non-negative least squares (NNLS) algorithm, the fast iterative shrinkage-thresholding algorithm (FISTA) algorithm, and the CLEAN algorithm. The output of the CBF algorithm is the spatial convolution of the point spread function (PSF) and the sound pressure contribution of the real sound source. Based on this, deconvolution algorithms obtain the real sound source distribution through various iterative operations and remove the interference of non-ideal PSF on imaging.

[0077] Step S20: Calculate the FB output according to the array signal receiving model.

[0078] It should be noted that the step of calculating the FB output according to the array signal receiving model includes: calculating according to the array signal CSM The dimensional CSM performs eigenvalue decomposition to determine the corresponding cross-spectral index function; and the FB output is calculated based on the cross-spectral index function.

[0079] In a specific implementation, the array signal is further calculated CSM

[0080]

[0081] in, is the noise power, for dimensional identity matrix, and perform eigenvalue decomposition on CSM. is the definition The cross-spectral index function on , then the cross-spectral index function of CSM is

[0082]

[0083] Where v is the exponent, is a unitary matrix composed of the eigenvectors of CSM; is a diagonal matrix composed of the eigenvalues ​​of CSM.

[0084] Calculate the grid points using the cross-spectral index function FB output at

[0085]

[0086] in, for dimensional weight vector, Expressed as

[0087]

[0088] in, Focusing grid points The distance to the mth array element.

[0089] The FB algorithm attenuates the non-sound source points by performing CSM exponential operations. value to achieve the purpose of sharpening the main lobe and suppressing the side lobes.

[0090] Step S30: Calculate the PSF matrix that is invariant to power-raising space transfer according to the array signal reception model.

[0091] It can be understood that the step of calculating the PSF matrix of the power-raising space transfer invariant according to the array signal receiving model includes: determining according to the array signal receiving model in combination with the sound source incoherence assumption The target cross-spectral matrix corresponding to the dimensional CSM; according to the relationship between the FB output and the power point spread function, the PSF matrix that is invariant to the power space transfer is calculated based on the Loener-Heinz inequality.

[0092] It should be noted that in the ideal case without noise interference, the array signal is

[0093]

[0094] Based on the assumption of incoherent sound sources, the cross-spectral matrix corresponding to the FB algorithm is transformed into

[0095]

[0096] According to the relationship between FB output and the power point spread function, based on the Löwner-Heinz inequality, it is proved that

[0097]

[0098]

[0099] in, Represents a grid point The sound source intensity at

[0100]

[0101] for Dimensional power , denoted as , represents the array's contribution to the beamforming of a single-point unit amplitude sound source in space at the grid point of the power v (v>1). Its physical meaning is the response of the array beam, reflecting the performance of the array's resolution. In special cases, when v=1, FB becomes CBF. Conventional is a Hermitian matrix, and all diagonal elements are 1, and their diagonal values ​​remain unchanged after exponential operation; the modulus of the off-diagonal elements is less than 1, and their values ​​decrease after exponential operation. It can be seen that Using v times the exponential operation can sharpen the regular The above formula can be expressed as

[0102]

[0103] in, for dimensional unknown sound source distribution.

[0104] Step S40: Determine spatial convolution information according to the FB output and the PSF matrix.

[0105] This embodiment assumes that the PSF is spatially invariant, that is, it depends only on the relative position between the observation point and the sound source point, and has nothing to do with the specific position. The PSF at the center of the focusing grid plane is generally used as the spatially invariant PSF. The improved algorithm in this embodiment performs an exponential operation on the spatially invariant PSF to form a "power-raising spatially invariant PSF", denoted as . Assume that the number of grid points in each row of the focal plane is , the number of grid points in each column is ,but for dimensional matrix. Applying power-raising space transfer is invariant , transform the above formula into

[0106]

[0107] in, represents the focal plane center coordinate vector, Indicated by The elements in dimensional matrix, for dimensional unknown sound source distribution.

[0108] Through FFT operation, the sound source distribution and (Abbreviated as ) is transformed into a wave number domain product, namely

[0109]

[0110] in," ” represents the Hadamard product, and Represents fast Fourier transform and inverse transform respectively. The original linear convolution becomes circular convolution.

[0111] Step S50: transform the spatial domain convolution information into a wave number domain product through FFT operation under periodic boundary conditions.

[0112] It should be noted that the step of transforming the spatial domain convolution information into the wave number domain product through FFT operation under periodic boundary conditions includes: introducing periodic boundary conditions; uniformly wrapping the error in the spatial domain convolution information through FFT operation combined with the periodic boundary conditions to transform it into the wave number domain product.

[0113] In specific implementations, the zero boundary condition FFT needs to suppress the wrapping error generated by the FFT by using the boundary zero padding method. Dimensional Matrix For example, to fill the border with zeros, Embedded into the boundary matrix filled with 0, we get , for Dimensional matrix.

[0114]

[0115] However, the zero-padding operation of FFT may cause spectrum leakage. To solve this problem, for single-point frequency signals, the sampling frequency can be adjusted to form a full-cycle sampling. Extended to the wave number domain, for non-periodic functions, periodic boundary conditions can be introduced to make them become periodic. Then use Fourier transform theory. At this time, the period is very large, and the difference between the wave numbers in the wave number domain is very small, so for non-periodic functions, their wave number domain can be regarded as continuous.

[0116] right Complement the periodic boundary to obtain , for matrix

[0117]

[0118] Based on the assumption that array PSF space transfer remains unchanged, at this time dimension The matrix can be viewed as a circulant matrix and can be diagonalized by Fourier matrix similarity, that is,

[0119]

[0120] in, for The eigenvalue matrix of . is a two-dimensional unitary discrete Fourier matrix, whose elements are

[0121]

[0122] matrix satisfy , for According to the Fourier matrix property, the discrete Fourier transform of a discrete signal vector can be expressed as This operation can be implemented using FFT. and matrix There are the following relationships.

[0123]

[0124] remember The characteristic value of . You can use the matrix The discrete Fourier transform of the first column element is

[0125]

[0126] in, Indicates taking the first Column, we can know the matrix The first column element .

[0127]

[0128]

[0129] in," " indicates an equivalence relationship, for The eigenvalues ​​of dimensional vector, For the matrix A vector consisting of the elements in column 1, for The matrix consists of the elements in column 1. Circulant matrix properties, can be The elements in the matrix are moved upwards in a circular manner , move left in a circular motion get. and are the row and column numbers of the center of the focus plane grid point.

[0130]

[0131] Based on the above analysis.

[0132]

[0133] Step S60: construct a difference function based on the wavenumber domain product and the FB output.

[0134] It should be noted that the step of constructing a difference function based on the wave number domain product and the FB output includes: determining a periodic power-raising spatial transfer invariant matrix based on the wave number domain product; obtaining the sound source distribution, and constructing a difference function based on the FB output and the periodic power-raising spatial transfer invariant matrix.

[0135] It should be noted that the expression of the difference function is:

[0136]

[0137] in, represents the Frobenius norm, is the periodic power-raising space transfer invariant matrix, is the sound source distribution, Output results for FB.

[0138] Step S70: Calculate the difference function by using a fast iterative shrinkage threshold method to determine the actual sound source distribution.

[0139] In a specific implementation, the step of calculating the difference function by a fast iterative shrinkage threshold method to determine the real sound source distribution includes: about Iterate the search in the negative gradient direction to obtain ; Set the initial matrix ,make , , perform iterative calculation:

[0140]

[0141]

[0142]

[0143] in, represents the Euclidean projection in the non-negative quadrant, is the Lipschitz constant, Equal to the matrix The maximum eigenvalue, gradient The expression is

[0144]

[0145] Get the iterative calculation results and Remove boundaries to determine the true sound source distribution.

[0146] It should be noted that the improved algorithm in this embodiment uses FB output to replace CBF output, and improves the performance of the algorithm by performing exponential operation on PSF and based on fast Fourier transform under periodic boundary conditions. Figure 4 As shown. Exponential operation can reduce the side lobes to improve the imaging dynamic range; at the same time, Sharpen the main lobe of the beam to improve the spatial resolution of imaging; based on the properties of fast Fourier transform, Matrix and sound source distribution The high-dimensional spatial convolution operation is converted into the wave number domain product, which effectively improves the calculation efficiency; based on FFT under periodic boundary conditions, the wave number leakage problem caused by zero padding is solved, which can improve the calculation accuracy and further improve the imaging performance; exponential operation makes It is close to the ideal DF and can suppress the position solution error of the sound source at the edge of the focusing plane. The improved algorithm expands the effective imaging range of the FFT-FISTA algorithm.

[0147] It is understandable that the fast iterative shrinkage threshold algorithm (FFT-FISTA) based on fast Fourier transform has high computational efficiency, but it ignores the spatial variation and warping error of the point spread function, resulting in the loss of sound source identification performance. Therefore, an improved FFT-FISTA algorithm based on function beamforming is proposed. The improved algorithm uses the function beamforming output as the iterative input of the FFT-FISTA algorithm, establishes a linear equation system of function beamforming, sound source distribution and power-up spatial transfer invariant point spread function, and iteratively solves it based on the fast Fourier transform under periodic boundary conditions, so that the non-periodic function being calculated becomes a periodic function, solves the wave number leakage problem caused by the zero-filling boundary, improves the calculation accuracy, and further improves the imaging performance; the main lobe of the point spread function is sharpened by exponential operation, and the applicability of the assumption of spatial transfer invariance of the point spread function is expanded. The simulation and experimental results show that compared with the conventional FFT-FISTA algorithm, the improved algorithm can improve the imaging spatial resolution and dynamic range, expand the effective imaging area of ​​the FFT-FISTA algorithm, and the compressed gas leakage experiment results verify the effectiveness of the improved algorithm.

[0148] This embodiment establishes an array signal receiving model; calculates the FB output according to the array signal receiving model; calculates the PSF matrix of power-raising spatial transfer invariance according to the array signal receiving model; determines the spatial convolution information according to the FB output and the PSF matrix; transforms the spatial convolution information into a wave number domain product through FFT operation under periodic boundary conditions; constructs a difference function according to the wave number domain product and the FB output; calculates the difference function through a fast iterative shrinkage threshold method, solves the wave number leakage problem caused by the zero-filled boundary, improves the calculation accuracy, and further improves the imaging performance; sharpens the main lobe of the point spread function through exponential operation, and expands the applicability of the assumption of spatial transfer invariance of the point spread function. Improve the imaging spatial resolution and dynamic range, and expand the effective imaging area. A technical effect of avoiding performance loss in the process of sound source identification is achieved.

[0149] In addition, an embodiment of the present application further proposes a computer-readable storage medium, on which a program for improving function beamforming is stored. When the program for improving function beamforming is executed by a processor, the steps of the method for improving function beamforming as described above are implemented.

[0150] Reference Figure 5 , Figure 5 This is a structural block diagram of the first embodiment of the function beamforming improvement device of the present application.

[0151] like Figure 5 As shown, the function beamforming improvement device proposed in the embodiment of the present application includes:

[0152] A model building module 10 is used to build an array signal receiving model;

[0153] The FB output module 20 is used to calculate the FB output according to the array signal receiving model;

[0154] A PSF matrix module 30, for calculating a PSF matrix with invariant power spatial transfer according to an array signal receiving model;

[0155] A spatial convolution information module 40, for determining spatial convolution information according to the FB output and the PSF matrix;

[0156] The wave number domain product module 50 is used to transform the spatial domain convolution information into the wave number domain product through FFT operation under periodic boundary conditions;

[0157] A difference function module 60 is used to construct a difference function according to the wave number domain product and the FB output;

[0158] The sound source distribution module 70 is used to calculate the difference function by using a fast iterative shrinkage threshold method to determine the actual sound source distribution.

[0159] It should be understood that the above is only an example and does not constitute any limitation on the technical solution of the present application. In specific applications, technicians in this field can make settings as needed, and the present application does not impose any limitation on this.

[0160] This embodiment establishes an array signal receiving model; calculates the FB output according to the array signal receiving model; calculates the PSF matrix of power-raising spatial transfer invariance according to the array signal receiving model; determines the spatial convolution information according to the FB output and the PSF matrix; transforms the spatial convolution information into a wave number domain product through FFT operation under periodic boundary conditions; constructs a difference function according to the wave number domain product and the FB output; calculates the difference function through a fast iterative shrinkage threshold method, solves the wave number leakage problem caused by the zero-filled boundary, improves the calculation accuracy, and further improves the imaging performance; sharpens the main lobe of the point spread function through exponential operation, and expands the applicability of the assumption of spatial transfer invariance of the point spread function. Improve the imaging spatial resolution and dynamic range, and expand the effective imaging area. A technical effect of avoiding performance loss in the process of sound source identification is achieved.

[0161] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of the present application. In practical applications, technicians in this field can select part or all of it according to actual needs to achieve the purpose of the present embodiment, and no limitation is made here.

[0162] In addition, for technical details not described in detail in this embodiment, reference can be made to the method for improving function beamforming provided in any embodiment of the present application, which will not be repeated here.

[0163] In addition, it should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0164] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0165] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory (ROM) / RAM, a disk, or an optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods of each embodiment of the present application. The above is only a preferred embodiment of the present application, and does not limit the patent scope of the present application. All equivalent structures or equivalent process changes made using the contents of the present application specification and drawings, or directly or indirectly used in other related technical fields, are similarly included in the patent protection scope of the present application.

Claims

1. A method for improving function beamforming, characterized in that: include: Establish array signal receiving model; Calculate FB output according to the array signal receiving model; Calculate a PSF matrix with invariant power spatial transfer according to the array signal receiving model; Determine spatial domain convolution information according to the FB output and the PSF matrix; According to the spatial domain convolution information, the information is transformed into a wave number domain product through FFT operation under periodic boundary conditions; constructing a difference function according to the wavenumber domain product and the FB output; The difference function is calculated by a fast iterative shrinkage threshold method to determine the real sound source distribution; The step of constructing a difference function according to the wavenumber domain product and the FB output comprises: Determine a periodic power-raising space transfer invariant matrix according to the wave number domain product; Acquire the sound source distribution, and construct a difference function according to the FB output and the periodic power-raising space transfer invariant matrix; Wherein, the expression of the difference function is: in, represents the Frobenius norm, is the periodic power-raising space transfer invariant matrix, is the sound source distribution, Output results for FB; The step of calculating the difference function by a fast iterative shrinkage threshold method to determine the real sound source distribution includes: By the difference function about Iterate the search in the negative gradient direction to obtain ; Set the initial matrix ,make , , perform iterative calculation: in, represents the Euclidean projection in the non-negative quadrant, is the Lipschitz constant, Equal to the matrix The maximum eigenvalue, gradient The expression is Obtain iterative calculation results and Remove boundaries to determine the true sound source distribution.

2. The improved function beamforming method according to claim 1, characterized in that: The step of calculating the FB output according to the array signal receiving model comprises: Calculate based on the array signal Dimensional CSM; Regarding the dimensional CSM to perform eigenvalue decomposition to determine the corresponding cross-spectral index function; The FB output is calculated according to the cross-spectral index function.

3. The improved function beamforming method according to claim 2, characterized in that: The step of calculating the PSF matrix with invariant power-raising spatial transfer according to the array signal receiving model comprises: The array signal receiving model is combined with the sound source incoherence assumption to determine the The target cross-spectral matrix corresponding to the dimensional CSM; According to the relationship between the FB output and the power point spread function, the PSF matrix that is invariant to the power space transfer is calculated based on the Loener-Henry inequality.

4. The improved function beamforming method according to claim 1, characterized in that: The step of transforming the spatial domain convolution information into a wave number domain product through FFT operation under periodic boundary conditions comprises: Introducing periodic boundary conditions; The error is uniformly convolved in the spatial domain convolution information by combining the periodic boundary conditions through FFT operation to transform into a wavenumber domain product.

5. A function beamforming improvement device, characterized in that: Executing the method according to claim 1, comprising: A model building module, used for building an array signal receiving model; An FB output module, used for calculating FB output according to the array signal receiving model; A PSF matrix module, used for calculating a PSF matrix with invariant power spatial transfer according to the array signal receiving model; A spatial convolution information module, used to determine spatial convolution information according to the FB output and the PSF matrix; A wave number domain product module, used for transforming the spatial domain convolution information into a wave number domain product through FFT operation under periodic boundary conditions; A difference function module, used for constructing a difference function according to the wave number domain product and the FB output; The sound source distribution module is used to calculate the difference function by a fast iterative shrinkage threshold method to determine the real sound source distribution.

6. A computer device, characterized in that: The device comprises: a memory and a processor, wherein the processor executes the method according to any one of claims 1 to 4 when running computer instructions stored in the memory.

7. A computer-readable storage medium, characterized in that: The method comprises instructions, which, when executed on a computer, cause the computer to execute the method according to any one of claims 1 to 4.

Images