Statistical optimal near-field acoustical holography method, system, medium and equipment

Abstract

The invention discloses a statistical optimal near-field acoustical holography method, system, medium and equipment based on random total least squares, and the method comprises the steps: collecting the near-field data of a housing structure in a wavelength region away from a sound source; aiming at the appearance characteristics of the sound source, reasonably selecting and counting an optimal near-field acoustical holographic unit wave function, constructing a holographic surface, and reconstructing a surface wave function model; the random matrix is used for reducing the holographic surface and reconstruction surface wave function model matrix dimension, and the calculation efficiency is effectively improved; noise is cut off through a total least square method, and high-precision and high-robustness reconstruction of a sound field is achieved. And the anti-noise performance of the statistical optimal near-field acoustical holography method is obviously improved.

Description

Technical Field

[0001] The present invention relates to the field of acoustic signal technology, and in particular to a statistically optimal near-field acoustic holography method, system, medium and equipment based on random total least squares. Background Art

[0002] Accurately identifying noise sources is a prerequisite for noise control, low-noise design, and sound quality design of electromechanical products. Statistical optimal near-field acoustic holography is a very effective technology for noise source identification, positioning, and sound field visualization. This method regards the sound pressure at any point in space as a linear superposition of the sound pressure collected by the holographic surface, and can predict many acoustic quantities such as the sound pressure, particle velocity, sound intensity, and sound source radiation power at any point in the entire three-dimensional sound field. However, today's statistical optimal near-field acoustic holography only considers the modeling error of the holographic surface wave function, and ignores the uncertainties introduced in the process of reconstructing the surface wave function modeling, resulting in low robustness and unstable reconstruction process.

[0003] The above information disclosed in this Background section is only for enhancement of understanding of the background of the invention and therefore it may contain information that does not form the prior art that is already known to a person skilled in the art. Summary of the invention

[0004] The present invention provides a statistically optimal near-field acoustic holography method, system, medium and equipment based on random total least squares, which utilizes a small random matrix to simplify the wave function model, and then adopts a truncated total least squares method to truncate high-order noise, while taking into account the modeling errors of the holographic surface and the reconstructed surface wave functions, thereby significantly improving the accuracy and noise resistance of the original statistically optimal near-field acoustic holography method and improving stability.

[0005] A statistically optimal near-field acoustic holography method based on random population least squares includes:

[0006] Step a, arranging M microphones in a wavelength region near the sound source, collecting shell radiation sound pressure data, and extracting the holographic surface sound pressure pH using Fourier transform and Kalman filter;

[0007] Step b, selecting the statistically optimal near-field acoustic holographic unit wave function based on the sound source shape characteristics to construct a holographic surface and a reconstruction surface wave function model;

[0008] Step c, using a random matrix to reduce the dimensions of the holographic surface and the reconstructed surface wave function model matrix;

[0009] Step d: QR decomposition and singular value decomposition (SVD) are used to truncate the noise, and the sound field transfer coefficient is solved by the total least squares method to reconstruct the target sound field.

[0010] In the statistically optimal near-field acoustic holography method based on random total least squares, an array consisting of M microphones covers the entire sound source.

[0011] In the statistically optimal near-field acoustic holography method based on random population least squares, the array shapes include rectangular plane, cylindrical surface, and ring shape.

[0012] In the statistically optimal near-field acoustic holography method based on random total least squares, in step b, constructing the holographic surface and reconstructing the surface wave function model includes:

[0013] Get the three-dimensional coordinates of each measuring point on the holographic surface and the reconstructed surface, where the coordinates of the holographic surface measuring point r H , reconstruct the surface measurement point coordinates r R ,

[0014] In the statistically optimal near-field acoustic holography unit wave function, the unit plane wave function at the spatial position r in the rectangular coordinate system is as follows:

[0015]

[0016] In the formula, the wave number vector k in the rectangular coordinate system is (k x , k y , k z ), the spatial coordinates in the rectangular coordinate system r = (x, y, z), i is the imaginary unit, z s is the source surface position in the rectangular coordinate system,

[0017] The wave function of the unit cylinder at the spatial position r in the cylindrical coordinate system is as follows:

[0018]

[0019] In the formula, the wave number vector k in the cylindrical coordinate system is (n, k z , k r ), the spatial coordinates in the cylindrical coordinate system r = (r, φ, z), i is the imaginary unit, H n (1) represents the nth-order first-kind Hankle function, r S is the source surface radius,

[0020] The unit spherical wave function at the spatial position r in the spherical coordinate system is as follows:

[0021]

[0022] Where, the wave number vector k in the spherical coordinate system is (n, m), the spatial coordinate r in the spherical coordinate system is (r, θ, φ), i is the imaginary unit, H n (1) represents the n-th order first kind Hankle function, represents the associated Legendre function, the wave number vector k and the spatial coordinate r represent the same sampling position in the wave number domain and the spatial domain, the difference is that they are projected in different spatial coordinate systems,

[0023] The unit wave function is used to construct the holographic surface and reconstruction surface acoustic field wave function matrix A and a, which are in the following form:

[0024]

[0025]

[0026] Where k represents one of the wave number vectors in the rectangular coordinate system, cylindrical coordinate system or spherical coordinate system, r represents the spatial coordinate in the rectangular coordinate system, cylindrical coordinate system or spherical coordinate system, the following table "R" represents the reconstruction surface, Φ(k, r) represents several or several combinations of the wave number vector k and the plane, cylindrical or spherical unit wave function under the spatial coordinate r, r H，M represents the coordinates of the Mth microphone in three-dimensional space, r R，Q Represents the coordinates of the Qth reconstruction point in space.

[0027] In the statistically optimal near-field acoustic holography method based on random total least squares, in step c, the random matrix Ω is a (M+1)×L-order complex Gaussian random matrix, and L is less than the total number of microphones M.

[0028] In the statistically optimal near-field acoustic holography method based on random population least squares, in step d,

[0029] The wave function matrix of the holographic surface and the reconstructed surface based on statistical optimal near-field acoustic holography is as follows:

[0030] Ac=a,

[0031] Among them, A is the holographic surface wave function matrix, a is the reconstructed surface wave function matrix, c is the sound field transfer coefficient matrix to be solved, and the random total least squares method is used to solve the sound field transfer coefficient matrix c. The solution process of the i-th column is as follows:

[0032] Establish an augmented matrix C = [A, b], where A is the holographic surface wave function matrix, b is the i-th column in the reconstructed surface wave function matrix a, then the dimension of the matrix C is N × (M + 1),

[0033] The reduced-dimensional data matrix Y=CΩ is obtained, whose dimension is N×L.

[0034] Perform QR decomposition on the reduced-dimensional data matrix Y, that is, Y = QR, where the dimension of the matrix Q is N × L.

[0035] Construct the intermediate matrix Z = Q T C, whose dimension is L×(M+1),

[0036] Calculate the singular value decomposition of the intermediate matrix Z and get Z = UΣV T , where the dimension of the unitary matrix V is (M+1)×L,

[0037] Extract the solution matrix V 11 =V(1:M, 1:k) and solution vector v 21 =V(end,1:k), where k represents the cutoff order, and its value range is 0<k<L,

[0038] The solution vector x of the i-th column of the sound field transfer coefficient matrix c is:

[0039]

[0040] The superscript "T" indicates the matrix transpose, and the superscript "+" indicates the generalized inverse of the matrix / vector. Repeat the above steps to obtain the final sound field transfer coefficient matrix c.

[0041] Calculate the reconstruction surface sound pressure p R , the calculation formula is as follows:

[0042] p R =c T p H .

[0043] In the statistically optimal near-field acoustic holography method based on random population least squares, the overall relative error of the reconstructed surface sound pressure is obtained, which is:

[0044]

[0045] In the statistically optimal near-field acoustic holography method based on random population least squares, the overall relative error is less than 50%.

[0046] In the statistically optimal near-field acoustic holography method based on random population least squares, the noise source includes electromechanical product noise.

[0047] In the statistically optimal near-field acoustic holography method based on random population least squares, the sound source is a point sound source.

[0048] A statistically optimal near-field acoustic holography system based on random total least squares includes:

[0049] A collection unit, which includes M microphones arranged in a wavelength region near the sound source to collect shell radiation sound pressure data, a Fourier transform unit for extracting the holographic surface sound pressure pH, and a Kalman filter;

[0050] A modeling unit, which is used to: select the statistically optimal near-field acoustic holographic unit wave function based on the sound source shape characteristics to construct a holographic surface and a reconstruction surface wave function model;

[0051] A dimension reduction unit, which is used to: reduce the dimension of the matrix of the holographic surface and the reconstructed surface wave function model by using a random matrix;

[0052] The reconstruction unit is used to: truncate the noise by using QR decomposition and singular value decomposition (SVD), solve the sound field transfer coefficient by total least square method, and then reconstruct the target sound field.

[0053] A computer storage medium includes computer instructions, which, when executed on a computer, cause the computer to execute the method described.

[0054] An electronic device, comprising:

[0055] A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein:

[0056] When the processor executes the program, the method described is implemented.

[0057] Compared with the prior art, the present invention has the following advantages: based on the principle of statistical optimal near-field acoustic holography, the present invention constructs a small random matrix to achieve effective dimension reduction of the holographic surface and the reconstructed surface wave function matrix, significantly improving the efficiency of the algorithm; the truncated total least squares method is used to solve the sound field transfer coefficient, while considering the modeling errors of the holographic surface and the reconstructed surface wave function, avoiding the risk of failure of sound field reconstruction due to ignoring the modeling errors of the reconstructed surface wave function. Compared with the traditional statistical optimal near-field acoustic holography, the method provided by the present invention has the advantages of high reconstruction accuracy, high efficiency, strong anti-noise performance, etc., and can scientifically guide the noise control and low-noise design of electromechanical products. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] By reading the detailed description of the preferred specific embodiments below, various other advantages and benefits of the present invention will become clear to those of ordinary skill in the art. The drawings in the specification are only for the purpose of illustrating the preferred embodiments and are not considered to be limitations of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative work. Moreover, the same reference numerals are used to represent the same components throughout the drawings.

[0059] In the attached picture:

[0060] Figure 1 Flow chart for implementing the present invention;

[0061] Figure 2The sound source arrangement diagram in the simulation example used in the present invention;

[0062] Figure 3 Schematic diagram of sound pressure of holographic surface test when the sound source frequency f=1000Hz;

[0063] Figure 4 It is a schematic diagram of the sound pressure of the holographic surface test when the sound source frequency f=1000Hz and SNR=15dB;

[0064] Figure 5 It is a schematic diagram of the sound pressure of the holographic surface test when the sound source frequency f=1000Hz and SNR=5dB;

[0065] Figure 6 Schematic diagram of theoretical sound pressure on the reconstruction surface when the sound source frequency f = 1000Hz;

[0066] Figure 7 Schematic diagram of sound pressure reconstruction using the traditional method when the sound source frequency is f = 1000 Hz and SNR = 15 dB;

[0067] Figure 8 Schematic diagram of sound pressure reconstruction using the proposed method when the sound source frequency is f = 1000 Hz and SNR = 15 dB;

[0068] Fig. 9 Schematic diagram of sound pressure reconstruction using the traditional method when the sound source frequency is f = 1000 Hz and SNR = 5 dB;

[0069] Fig.10 Schematic diagram of sound pressure reconstructed by the proposed method when the sound source frequency is f = 1000 Hz and SNR = 5 dB.

[0070] The present invention is further explained below in conjunction with the accompanying drawings and embodiments. DETAILED DESCRIPTION

[0071] The specific embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the specific embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0072] It should be noted that certain words are used in the specification and claims to refer to specific components. Those skilled in the art should understand that technicians may use different nouns to refer to the same component. This specification and claims do not use the difference in nouns as a way to distinguish components, but use the difference in the functions of the components as the criterion for distinction. As mentioned throughout the specification and claims, "including" or "comprising" is an open term, so it should be interpreted as "including but not limited to". The subsequent description of the specification is a preferred embodiment of the present invention, but the description is based on the general principles of the specification and is not intended to limit the scope of the present invention. The scope of protection of the present invention shall be determined by the attached claims.

[0073] To facilitate understanding of the embodiments of the present invention, further explanation will be given below by taking specific embodiments as examples in conjunction with the accompanying drawings, and each of the accompanying drawings does not constitute a limitation on the embodiments of the present invention.

[0074] like Figures 1 to 10 As shown, the statistically optimal near-field acoustic holography method based on random population least squares includes the following steps:

[0075] Step a, arranging M microphones in a wavelength region near the sound source, collecting shell radiation sound pressure data, and extracting the holographic surface sound pressure pH using Fourier transform and Kalman filter;

[0076] Step b, selecting the statistically optimal near-field acoustic holographic unit wave function based on the sound source shape characteristics to construct a holographic surface and a reconstruction surface wave function model;

[0077] Step c, using a random matrix to reduce the dimensions of the holographic surface and the reconstructed surface wave function model matrix;

[0078] Step d: QR decomposition and singular value decomposition (SVD) are used to truncate the noise, and the sound field transfer coefficient is solved by the total least squares method to reconstruct the target sound field.

[0079] In a preferred implementation of the statistically optimal near-field acoustic holography method based on random total least squares, an array consisting of M microphones covers the entire sound source.

[0080] In a preferred implementation of the statistically optimal near-field acoustic holography method based on random population least squares, the array shapes include rectangular plane, cylinder, and ring.

[0081] In a preferred implementation of the statistically optimal near-field acoustic holography method based on random total least squares, in step b, constructing the holographic surface and reconstructing the surface wave function model includes:

[0082] Get the three-dimensional coordinates of each measuring point on the holographic surface and the reconstructed surface, where the coordinates of the holographic surface measuring point r H, reconstruct the surface measurement point coordinates r R ,

[0083] In the statistically optimal near-field acoustic holography unit wave function, the unit plane wave function at the spatial position r in the rectangular coordinate system is as follows:

[0084]

[0085] In the formula, the wave number vector k in the rectangular coordinate system is (k x , k y , k z ), the spatial coordinates in the rectangular coordinate system r = (x, y, z), i is the imaginary unit, z S is the source surface position in the rectangular coordinate system,

[0086] The wave function of the unit cylinder at the spatial position r in the cylindrical coordinate system is as follows:

[0087]

[0088] In the formula, the wave number vector k in the cylindrical coordinate system is (n, k z , k r ), the spatial coordinates in the cylindrical coordinate system r = (r, φ, z), i is the imaginary unit, H n (1) represents the nth-order first-kind Hankle function, r S is the source surface radius,

[0089] The unit spherical wave function at the spatial position r in the spherical coordinate system is as follows:

[0090]

[0091] Where, the wave number vector k in the spherical coordinate system is (n, m), the spatial coordinate r in the spherical coordinate system is (r, θ, φ), i is the imaginary unit, H n (1) represents the n-th order first kind Hankle function, represents the associated Legendre function, the wave number vector k and the spatial coordinate r represent the same sampling position in the wave number domain and the spatial domain, the difference is that they are projected in different spatial coordinate systems,

[0092] The unit wave function is used to construct the holographic surface and reconstruction surface acoustic field wave function matrix A and a, which are in the following form:

[0093]

[0094] Where k represents one of the wave number vectors in the rectangular coordinate system, cylindrical coordinate system or spherical coordinate system, r represents the spatial coordinate in the rectangular coordinate system, cylindrical coordinate system or spherical coordinate system, the following table "R" represents the reconstruction surface, Φ(k, r) represents several or several combinations of the wave number vector k and the plane, cylindrical or spherical unit wave function under the spatial coordinate r, r H，M represents the coordinates of the Mth microphone in three-dimensional space, r R，Q Represents the coordinates of the Qth reconstruction point in space.

[0095] In a preferred implementation of the statistically optimal near-field acoustic holography method based on random total least squares, in step c, the random matrix Ω is a (M+1)×L order complex Gaussian random matrix, and L is less than the total number of microphones M.

[0096] In a preferred implementation of the statistically optimal near-field acoustic holography method based on random population least squares, in step d,

[0097] The wave function matrix of the holographic surface and the reconstructed surface based on statistical optimal near-field acoustic holography is as follows:

[0098] Ac=a,

[0099] Among them, A is the holographic surface wave function matrix, a is the reconstructed surface wave function matrix, c is the sound field transfer coefficient matrix to be solved, and the random total least squares method is used to solve the sound field transfer coefficient matrix c. The solution process of the i-th column is as follows:

[0100] Establish an augmented matrix C = [A, b], where A is the holographic surface wave function matrix, b is the i-th column in the reconstructed surface wave function matrix a, then the dimension of the matrix C is N × (M + 1),

[0101] The reduced-dimensional data matrix Y=CΩ is obtained, whose dimension is N×L.

[0102] Perform QR decomposition on the reduced-dimensional data matrix Y, that is, Y = QR, where the dimension of the matrix Q is N × L.

[0103] Construct the intermediate matrix Z = Q T C, whose dimension is L×(M+1),

[0104] Calculate the singular value decomposition of the intermediate matrix Z and get Z = UΣV T , where the dimension of the unitary matrix V is (M+1)×L,

[0105] Extract the solution matrix V 11 =V(1:M, 1:k) and solution vector v 21 =V(end,1:k), where k represents the cutoff order, and its value range is 0 <k<L,

[0106] The solution vector x of the i-th column of the sound field transfer coefficient matrix c is:

[0107]

[0108] The superscript "T" indicates the matrix transpose, and the superscript "+" indicates the generalized inverse of the matrix / vector. Repeat the above steps to obtain the final sound field transfer coefficient matrix c.

[0109] Calculate the reconstruction surface sound pressure p R , the calculation formula is as follows:

[0110] p R =c T p H .

[0111] In a preferred implementation of the statistically optimal near-field acoustic holography method based on random total least squares, the overall relative error of the reconstructed surface sound pressure is obtained, which is:

[0112]

[0113] In a preferred implementation of the statistically optimal near-field acoustic holography method based on random population least squares, the overall relative error is less than 50%.

[0114] In a preferred implementation of the statistically optimal near-field acoustic holography method based on random total least squares, the noise source includes electromechanical product noise.

[0115] In a preferred implementation of the statistically optimal near-field acoustic holography method based on random total least squares, the sound source is a point sound source.

[0116] In one embodiment, Figure 1 As shown, the following steps are included:

[0117] Step 1: Arrange M microphones in the near field (within one wavelength region) of the sound source, collect the shell radiation sound pressure data, and use Fourier transform and Kalman filter to extract the complex sound pressure of a specific frequency, that is, the holographic surface sound pressure pH. The array composed of M microphones should cover the entire sound source, and the array shape includes but is not limited to rectangular plane, cylinder, ring, etc.

[0118] As attached Figure 2As shown in the figure, this simulation constructs a three-point sound source structure, and the three point sound sources are located at (0, 0, 0), (-0.6, -0.6, 0) and (-0.6, 0.6, 0). Both the holographic surface and the reconstruction surface are 25×25 uniformly distributed grids, and the length in both the x and y directions is 1.2m. The holographic surface and the reconstruction surface are parallel to the xOy plane, the holographic surface is located at z=0.4m, and the reconstruction surface is located at z=0.25m. The holographic surface test sound pressure is calculated according to the sound source arrangement of the simulation example as shown in the attached figure. Figure 3 As shown, to verify the anti-noise ability of this method, 15dB and 5dB white noise are applied respectively as shown in Figure 4 With attached Figure 5 The theoretical sound pressure of the calculated reconstruction surface is shown in the attached Figure 6 shown.

[0119] Step 2: According to the shape characteristics of the sound source, reasonably select the statistically optimal near-field acoustic holographic unit wave function to construct the holographic surface and reconstruction surface wave function model. Obtain the three-dimensional coordinates of each measuring point on the holographic surface and the reconstruction surface, where the coordinates of the holographic surface measuring point r H , reconstruct the surface measurement point coordinates r R .

[0120] The statistical optimal near-field acoustic holography method provides three basic unit wave functions. The unit plane wave function at the spatial position r in the rectangular coordinate system is as follows:

[0121]

[0122] In the formula, the wave number vector k in the rectangular coordinate system is (k x , k y , k z ), the spatial coordinates in the rectangular coordinate system r = (x, y, z), i is the imaginary unit, z S is the source surface position in the rectangular coordinate system.

[0123] The wave function of the unit cylinder at the spatial position r in the cylindrical coordinate system is as follows:

[0124]

[0125] In the formula, the wave number vector k in the cylindrical coordinate system is (n, k z , k r ), the spatial coordinates in the cylindrical coordinate system r = (r, φ, z), i is the imaginary unit, H n (1) represents the nth-order first-kind Hankle function, r S is the source surface radius.

[0126] The unit spherical wave function at the spatial position r in the spherical coordinate system is as follows:

[0127]

[0128] Where, the wave number vector k in the spherical coordinate system is (n, m), the spatial coordinate r in the spherical coordinate system is (r, θ, φ), i is the imaginary unit, H n (1) represents the n-th order first kind Hankle function, represents the associated Legendre function, the wave number vector k and the spatial coordinate r represent the same sampling position in the wave number domain and the spatial domain, the difference is that they are projected in different spatial coordinate systems. The above unit wave function is used to construct the holographic surface and reconstruction surface acoustic field wave function matrix A and a, and its form is shown below.

[0129]

[0130] Where k represents one of the wave number vectors in the rectangular coordinate system, cylindrical coordinate system or spherical coordinate system, r represents the spatial coordinate in the rectangular coordinate system, cylindrical coordinate system or spherical coordinate system, the following table "R" represents the reconstruction surface, Φ(k, r) represents several or several combinations of the wave number vector k and the plane, cylindrical or spherical unit wave function under the spatial coordinate r, r H，M represents the coordinates of the Mth microphone in three-dimensional space, r R，Q Indicates the coordinates of the Qth reconstruction point in space. Taking this simulation as an example, this simulation can select three sets of unit spherical wave functions or one set of unit plane wave functions. Considering the amount of calculation, unit plane waves are used to construct the unit wave function matrix of the holographic surface and the reconstruction surface.

[0131] Step 3: Use random matrices to reduce the dimensions of the holographic surface and reconstructed surface wave function model matrices, effectively improving the computational efficiency. Use random matrices to reduce the dimensions of the holographic surface and reconstructed surface wave function model matrices, where the small random matrix Ω is a (M+1)×L-order complex Gaussian random matrix, where L is much smaller than the total number of microphones M, and L=50 is taken in this simulation.

[0132] Step 4: Use QR decomposition and singular value decomposition (SVD) to truncate the noise, solve the sound field transfer coefficient by total least squares method, and then reconstruct the target sound field. Use QR decomposition and singular value decomposition (SVD) to truncate the noise, solve the sound field transfer coefficient by total least squares method, and then reconstruct the target sound field.

[0133] Based on the principle of statistical optimal near-field acoustic holography, the wave function matrix of the combined holographic surface and the reconstructed surface is as follows:

[0134] Ac=a,

[0135] Among them, A is the holographic surface wave function matrix, a is the reconstructed surface wave function matrix, and c is the sound field transfer coefficient matrix to be solved. The random total least squares method is used to solve the sound field transfer coefficient matrix c, and the solution process of the i-th column is as follows:

[0136] Step a, establish an augmented matrix C = [A, b], where A is the holographic surface wave function matrix, b is the i-th column in the reconstructed surface wave function matrix a, and the dimension of the matrix C is N×(M+1).

[0137] Step b, obtaining the reduced-dimensional data matrix Y=CΩ, whose dimension is N×L.

[0138] Step c, performing QR decomposition on the reduced-dimensional data matrix Y, that is, Y=QR, where the dimension of the matrix Q is N×L.

[0139] Step d, construct the intermediate matrix Z = Q T C, whose dimension is L×(M+1).

[0140] Step e, calculate the singular value decomposition of the intermediate matrix Z, and obtain Z = U∑V T , where the dimension of the unitary matrix V is (M+1)×L.

[0141] Step f, extract the solution matrix V 11 =V(1:M, 1:k) and solution vector v 21 =V(end,1:k), where k represents the cutoff order, and its value range is 0<k<L. In this simulation, k=5.

[0142] Step g, the solution vector x of the i-th column of the sound field transfer coefficient matrix c is:

[0143]

[0144] The superscript "T" indicates the matrix transpose, and the superscript "+" indicates the generalized inverse of the matrix / vector. Repeat the above steps to obtain the final sound field transfer coefficient matrix c.

[0145] Finally, calculate the reconstructed surface sound pressure p R , the calculation formula is as follows:

[0146] p R =c T p H .

[0147] In order to verify the effectiveness of the statistically optimal near-field acoustic holography method based on random population least squares proposed in the present invention, the overall relative error is used as the reconstruction error to quantify the accuracy of the reconstruction result. In order to prevent the experimental results from being accidental due to Gaussian noise, the experiment is repeated 10 times to obtain the average error. Among them, the overall relative error is defined as:

[0148]

[0149] The traditional method is used to reconstruct the sound pressure under 15dB and 5dB signal-to-noise ratio interference as shown in the attached figure. Figure 7 With attached Fig. 9 As shown in the figure, the overall relative errors are 58.68% and 89.78% respectively; the sound pressure reconstructed by the method proposed in the present invention under 15dB and 5dB signal-to-noise ratio interference is shown in the figure. Figure 8 With attached Fig.10 As shown in the figure, the overall relative errors are 38.45% and 44.18%, respectively, which are improved by 20.23% and 45.60%. Figure 8 With attached Fig.10 It can be seen that compared with the traditional method, although the method of the present invention still cannot achieve ultra-high precision reconstruction of the sound field under strong noise interference, the reconstruction result of the present invention accurately reflects the spatial position of the three point sound sources, which can strongly support the noise source positioning and low-noise design of the equipment.

[0150] In one embodiment, the method includes:

[0151] The near-field data of the shell structure is collected within a wavelength region away from the sound source, and the complex sound pressure of a specific frequency, i.e., the holographic surface sound pressure, is extracted using Fourier transform and Kalman filter; according to the shape characteristics of the sound source, the statistically optimal near-field acoustic holography unit wave function is reasonably selected to construct the holographic surface and reconstructed surface wave function model; the random matrix is ​​used to reduce the matrix dimension of the holographic surface and reconstructed surface wave function model, effectively improving the calculation efficiency; the noise is truncated by the total least squares method to achieve high-precision and high-robustness reconstruction of the sound field. Compared with the traditional statistically optimal near-field acoustic holography method, the present invention combines the random algorithm with the total least squares method, uses a small random matrix to effectively improve the calculation efficiency, and uses the truncated total least squares method to simultaneously limit the modeling errors of the holographic surface and reconstructed surface wave functions, which significantly improves the noise resistance of the statistically optimal near-field acoustic holography method, and provides strong support for the engineering promotion of the statistically optimal near-field acoustic holography method.

[0152] A statistically optimal near-field acoustic holography system based on random total least squares includes:

[0153] A collection unit, which includes M microphones arranged in a wavelength region near the sound source to collect shell radiation sound pressure data, a Fourier transform unit for extracting the holographic surface sound pressure pH, and a Kalman filter;

[0154] A modeling unit, which selects the statistically optimal near-field acoustic holographic unit wave function based on the sound source shape characteristics to construct a holographic surface and a reconstruction surface wave function model;

[0155] A dimension reduction unit, which uses a random matrix to reduce the dimension of the holographic surface and the reconstructed surface wave function model matrix;

[0156] The reconstruction unit uses QR decomposition and singular value decomposition (SVD) to truncate the noise, solves the sound field transfer coefficient through the total least squares method, and then reconstructs the target sound field.

[0157] A computer storage medium includes computer instructions, which, when executed on a computer, cause the computer to execute the method described.

[0158] An electronic device, comprising:

[0159] A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein:

[0160] When the processor executes the program, the method described is implemented.

[0161] Although the embodiments of the present invention are described above in conjunction with the accompanying drawings, the present invention is not limited to the above specific embodiments and application fields, and the above specific embodiments are only illustrative and instructive, rather than restrictive. A person of ordinary skill in the art can also make many forms under the guidance of this specification and without departing from the scope of protection of the claims of the present invention, all of which belong to the protection of the present invention.

Claims

1. A statistically optimal near-field acoustic holography method based on random population least squares, characterized in that: The steps include: Step a, arranging M microphones in a wavelength region near the sound source, collecting shell radiation sound pressure data, and extracting the holographic surface sound pressure pH using Fourier transform and Kalman filter; Step b, selecting the statistically optimal near-field acoustic holographic unit wave function based on the sound source shape characteristics to construct a holographic surface and a reconstruction surface wave function model; Step c, using a random matrix to reduce the dimensions of the holographic surface and the reconstructed surface wave function model matrix; Step d: QR decomposition and singular value decomposition (SVD) are used to truncate the noise, and the sound field transfer coefficient is solved by the total least squares method to reconstruct the target sound field.

2. The statistically optimal near-field acoustic holography method based on random total least squares according to claim 1, characterized in that: Preferably, the array consisting of M microphones covers the entire sound source.

3. The statistically optimal near-field acoustic holography method based on random total least squares according to claim 2 is characterized in that: The array shapes include rectangular plane, cylindrical surface and ring shape.

4. The statistically optimal near-field acoustic holography method based on random total least squares according to claim 1, characterized in that: In step b, constructing the holographic surface and reconstructing the surface wave function model includes: Get the three-dimensional coordinates of each measuring point on the holographic surface and the reconstructed surface, where the coordinates of the holographic surface measuring point r H , reconstruct the coordinates of the surface measurement points r R , In the statistically optimal near-field acoustic holography unit wave function, the unit plane wave function at the spatial position r in the rectangular coordinate system is as follows: In the formula, the wave number vector k in the rectangular coordinate system is (k x , k y , k z ), the spatial coordinates in the rectangular coordinate system r = (x, y, z), i is the imaginary unit, z S is the source surface position in the rectangular coordinate system, The wave function of the unit cylinder at the spatial position r in the cylindrical coordinate system is as follows: In the formula, the wave number vector k in the cylindrical coordinate system is (n, k z , k r ), the spatial coordinates in the cylindrical coordinate system r = (r, φ, z), i is the imaginary unit, H n (1) represents the n-order first-kind Hankle function, r S is the source surface radius, The unit spherical wave function at the spatial position r in the spherical coordinate system is as follows: Where, the wave number vector k in the spherical coordinate system is (n, m), the spatial coordinate r in the spherical coordinate system is (r, θ, φ), i is the imaginary unit, H n (1) represents the n-th order first kind Hankle function, represents the associated Legendre function, the wave number vector k and the spatial coordinate r represent the same sampling position in the wave number domain and the spatial domain, the difference is that they are projected in different spatial coordinate systems, The unit wave function is used to construct the holographic surface and reconstruction surface acoustic field wave function matrix A and a, which are in the following form: Where k represents one of the wave number vectors in rectangular coordinate system, cylindrical coordinate system or spherical coordinate system, r represents the spatial coordinate in rectangular coordinate system, cylindrical coordinate system or spherical coordinate system, "R" in the following table represents the reconstruction surface, Φ(k, r) represents several or several combinations of the wave number vector k and the plane, cylindrical or spherical unit wave function in the spatial coordinate r, r H，M represents the coordinates of the Mth microphone in three-dimensional space, r R，Q Represents the coordinates of the Qth reconstruction point in space.

5. The statistically optimal near-field acoustic holography method based on random total least squares according to claim 1, characterized in that: In step c, the random matrix Ω is a (M+1)×L order complex Gaussian random matrix, where L is less than the total number M of microphones.

6. The statistically optimal near-field acoustic holography method based on random total least squares according to claim 5, characterized in that: In step d, The wave function matrix of the holographic surface and the reconstructed surface based on statistical optimal near-field acoustic holography is as follows: Ac=a, Among them, A is the holographic surface wave function matrix, a is the reconstructed surface wave function matrix, c is the sound field transfer coefficient matrix to be solved, and the random total least squares method is used to solve the sound field transfer coefficient matrix c. The solution process of the i-th column is as follows: Establish an augmented matrix C = [A, b], where A is the holographic surface wave function matrix, b is the i-th column in the reconstructed surface wave function matrix a, then the dimension of the matrix C is N × (M + 1), The reduced-dimensional data matrix Y=CΩ is obtained, whose dimension is N×L. Perform QR decomposition on the reduced-dimensional data matrix Y, that is, Y = QR, where the dimension of the matrix Q is N × L. Construct the intermediate matrix Z = Q T C, whose dimension is L×(M+1), Calculate the singular value decomposition of the intermediate matrix Z and get Z = U∑V T , where the dimension of the unitary matrix V is (M+1)×L, Extract the solution matrix V 11 =V(1:M, 1:k) and solution vector v 21 =V(end,1:k), where k represents the cutoff order, and its value range is 0<k<L, The solution vector x of the i-th column of the sound field transfer coefficient matrix c is: Among them, the superscript "T" represents the matrix transpose, and the superscript "+" represents the generalized inverse of the matrix / vector. Repeat the above steps to obtain the final sound field transfer coefficient matrix c. Calculate the reconstruction surface sound pressure p R , the calculation formula is as follows: p R =c T p H 。 7. The statistically optimal near-field acoustic holography method based on random total least squares according to claim 1, characterized in that: The overall relative error of the reconstructed surface sound pressure is obtained as:

8. A statistically optimal near-field acoustic holography system based on random total least squares, characterized in that: It includes: The collection unit includes M microphones arranged in a wavelength region near the sound source to collect the shell radiation sound pressure data, extract the holographic surface sound pressure p H Fourier transform unit and Kalman filter; A modeling unit, which is used to: select the statistically optimal near-field acoustic holographic unit wave function based on the sound source shape characteristics to construct a holographic surface and a reconstruction surface wave function model; A dimension reduction unit, which is used to: reduce the dimension of the matrix of the holographic surface and the reconstructed surface wave function model by using a random matrix; The reconstruction unit is used to: truncate the noise by using QR decomposition and singular value decomposition (SVD), solve the sound field transfer coefficient by total least square method, and then reconstruct the target sound field.

9. A computer storage medium, characterized in that: The storage medium includes computer instructions, which, when executed on a computer, enable the computer to execute the method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: The electronic device comprises: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

Images