Multi-source statistical optimal near-field acoustical holography method, system, medium and equipment

Abstract

The invention discloses a cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, system, medium and equipment, and the method comprises the steps: arranging M microphones in a wavelength region away from a sound source, and obtaining the near-field holographic surface sound pressure pH of a housing structure; selecting planar, cylindrical and spherical unit wave functions based on the sound source appearance features to construct a multi-source wave function model; calculating the cosine similarity between the multi-source wave function model of each order and the holographic surface sound pressure pH, and updating based on the cosine similarity to obtain a weighted multi-source wave function model; and solving a transfer coefficient of the weighted multi-source wave function model of the holographic surface and the reconstruction surface through a Tikhonov regularization method, thereby realizing high-precision reconstruction of the multi-source sound field.

Description

Technical Field

[0001] The present invention relates to the field of acoustic signal technology, and in particular to a cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, system, medium and equipment. Background Art

[0002] Statistical optimal near-field acoustic holography is a common local sound field reconstruction method. This method obtains the spatial sound field transfer coefficient matrix by constructing the wave function model of the holographic surface and the reconstruction surface measurement point, thereby realizing the spatial transformation of the sound field. Among them, the multi-source statistical optimal near-field acoustic holography method uses a combination wave function model of radiation from a variety of sound sources of different shapes to simulate the combined sound field, breaking through the limitation of the traditional statistical optimal near-field acoustic holography method that can only match a single plane, cylinder, sphere and other regular shapes, and has been widely used in engineering. However, the traditional multi-source statistical optimal near-field acoustic holography method assigns the same weight to all wave function models, and cannot make adaptive updates for slight changes in the shape of the sound source, resulting in low accuracy of sound field reconstruction without prior knowledge.

[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 cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, system, medium and equipment, which utilizes cosine similarity to measure the matching degree between a wave function model and a sound field shape, and effectively reduces the wave function modeling error by assigning different weights to multi-source wave function models, thereby achieving accurate reconstruction of the sound source of the combined sound field.

[0005] A chord similarity weighted multi-source statistical optimal near-field acoustic holography method includes:

[0006] Step a, arranging M microphones within a wavelength region away from the sound source to obtain the near-field holographic surface sound pressure pH of the shell structure;

[0007] Step b, selecting plane, cylindrical and spherical unit wave functions based on the shape characteristics of the sound source to construct a multi-source wave function model;

[0008] Step c, calculating the cosine similarity between each order of multi-source wave function model and the holographic surface sound pressure pH, and updating the weighted multi-source wave function model based on the cosine similarity;

[0009] Step d, solving the transfer coefficient of the weighted multi-source wave function model between the holographic surface and the reconstruction surface by the Tikhonov regularization method, thereby achieving high-precision reconstruction of the multi-source sound field.

[0010] In the cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, an array consisting of M microphones covers the entire sound source.

[0011] In the cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, in step b, constructing a multi-source wave function model includes:

[0012] 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 ,

[0013] The unit plane wave function at the spatial position r in the rectangular coordinate system is as follows:

[0014]

[0015] 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,

[0016] The wave function of the unit cylinder at the spatial position r in the cylindrical coordinate system is as follows:

[0017]

[0018] 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,

[0019] The unit spherical wave function at the spatial position r in the spherical coordinate system is as follows:

[0020]

[0021] 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,

[0022] The i-th holographic surface wave function matrix A is obtained by combining three different unit wave functions based on multi-source statistical optimal near-field acoustic holography. i And reconstruct the surface wave function matrix a i The dimensions are N×M and N×R respectively. For S different sound source shapes, the holographic surface multi-source wave function matrix A and the reconstruction surface multi-source wave function matrix a are constructed respectively:

[0023]

[0024] The obtained holographic surface multi-source wave function matrix A∈R (S×N)×M , reconstruct the multi-source wave function matrix a∈R (S×N)×R .

[0025] In the multi-source statistical optimal near-field acoustic holography method weighted by cosine similarity, in step c, the i-th cosine similarity calculation formula is as follows:

[0026]

[0027] Among them, A i，. represents the i-th row of the holographic multi-source wave function matrix A, p H represents the holographic surface sound pressure, ||·|| represents the second norm, and based on the cosine similarity, the i-th weight calculation formula is as follows:

[0028]

[0029] Among them, k is the weight control hyperparameter,

[0030] All weights are combined to obtain the weight matrix W = diag (w1, w2, ..., w i , …), where i = 1, 2, …, S × N, diag (·) represents the diagonal elements of the matrix, then the holographic surface weighted multi-source wave function model A weight = WA, reconstruction surface weighted multi-source wave function model a weight =Wa.

[0031] In the cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, the weight controls the hyperparameter k==i.

[0032] In the multi-source statistical optimal near-field acoustic holography method weighted by cosine similarity, in step d,

[0033] Combining the weighted multi-source wave function model, we get the following formula:

[0034] A weight c=a weight ,

[0035] In the formula, A weightis the weighted multi-source wave function matrix of the holographic surface, a weight is the weighted multi-source wave function matrix of the reconstruction surface, c is the sound field transfer coefficient, and the sound field transfer coefficient c is solved according to the following formula:

[0036] c=(A weight H A weight +λ 2 I) -1 A weight H a weight ,

[0037] In the formula, λ is the regularization parameter, and the sound field reconstruction is calculated as follows:

[0038] p R =c T p H ,

[0039] In the formula, p R To reconstruct the surface sound pressure, p H is the holographic surface sound pressure.

[0040] In the multi-source statistical optimal near-field acoustic holography method weighted by cosine similarity, the regularization parameter λ is calculated according to the Hald criterion.

[0041]

[0042] In the formula, is the spatial wave number, f is the sound source frequency, c0 is the sound wave velocity, d is the distance between the holographic surface and the reconstruction surface, and SNR is the signal-to-noise ratio.

[0043] A cosine similarity weighted multi-source statistical optimal near-field acoustic holography system includes:

[0044] A collection unit, comprising M microphones arranged within a wavelength region from the sound source, for obtaining the near-field holographic surface sound pressure pH of the shell structure;

[0045] A modeling unit, which is used to select plane, cylindrical and spherical unit wave functions based on the shape characteristics of the sound source to construct a multi-source wave function model;

[0046] A cosine similarity calculation unit is used to calculate the cosine similarity between each order of multi-source wave function model and the holographic surface sound pressure pH, and update the weighted multi-source wave function model based on the cosine similarity;

[0047] The reconstruction unit is used to solve the transfer coefficient of the weighted multi-source wave function model of the holographic surface and the reconstruction surface through the Tikhonov regularization method, so as to achieve high-precision reconstruction of the multi-source sound field.

[0048] A computer storage medium includes computer instructions, which, when executed on a computer, cause the computer to execute the method described.

[0049] An electronic device, comprising:

[0050] A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein:

[0051] When the processor executes the program, the method described is implemented.

[0052] Compared with the prior art, the present invention has the following advantages: the present invention uses cosine similarity to assign different weights to multi-source wave function models, thereby solving the inherent problem that the traditional multi-source statistical optimal near-field acoustic holography method cannot adaptively match the sound field shape, and provides theoretical support for the engineering application of the multi-source statistical optimal near-field acoustic holography method. It can reduce the difficulty of wave function selection, adaptively reduce the modeling error between the multi-source wave function model and the sound source shape, and reconstruct the multi-source combined sound field with high precision in the absence of effective prior information. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] 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.

[0054] In the attached picture:

[0055] Figure 1 Flow chart for implementing the present invention;

[0056] Figure 2 The sound source arrangement diagram in the simulation example used in the present invention;

[0057] Figure 3 It is a schematic diagram of the sound pressure of the holographic surface test when the sound source frequency f=1000Hz and SNR=15dB;

[0058] Figure 4 Schematic diagram of theoretical sound pressure on the reconstruction surface when the sound source frequency f = 1000Hz;

[0059] Figure 5 Schematic diagram of reconstructed sound pressure on the reconstruction surface when the sound source frequency f = 1000Hz and SNR = 15dB;

[0060] Figure 6Schematic diagram of weight factors after normalization for implementation cases;

[0061] Figure 7 This is a line graph showing the reconstruction error changing with the sound source frequency when SNR=15dB;

[0062] Figure 8 This is a line graph showing the reconstruction error changing with the signal-to-noise ratio when the sound source frequency f=1000Hz.

[0063] The present invention is further explained below in conjunction with the accompanying drawings and embodiments. DETAILED DESCRIPTION

[0064] 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.

[0065] 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.

[0066] 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.

[0067] like Figures 1 to 8 As shown, the multi-source statistical optimal near-field acoustic holography method weighted by cosine similarity includes the following steps:

[0068] Step a, arranging M microphones within a wavelength region away from the sound source to obtain the near-field holographic surface sound pressure pH of the shell structure;

[0069] Step b, selecting plane, cylindrical and spherical unit wave functions based on the shape characteristics of the sound source to construct a multi-source wave function model;

[0070] Step c, calculating the cosine similarity between each order of multi-source wave function model and the holographic surface sound pressure pH, and updating the weighted multi-source wave function model based on the cosine similarity;

[0071] Step d, solving the transfer coefficient of the weighted multi-source wave function model between the holographic surface and the reconstruction surface by the Tikhonov regularization method, thereby achieving high-precision reconstruction of the multi-source sound field.

[0072] In a preferred implementation of the cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, an array consisting of M microphones covers the entire sound source.

[0073] In a preferred implementation of the cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, in step b, constructing a multi-source wave function model includes:

[0074] 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 ,

[0075] The unit plane wave function at the spatial position r in the rectangular coordinate system is as follows:

[0076]

[0077] 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,

[0078] The wave function of the unit cylinder at the spatial position r in the cylindrical coordinate system is as follows:

[0079]

[0080] 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,

[0081] The unit spherical wave function at the spatial position r in the spherical coordinate system is as follows:

[0082]

[0083] 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,

[0084] The i-th holographic surface wave function matrix A is obtained by combining three different unit wave functions based on multi-source statistical optimal near-field acoustic holography. i And reconstruct the surface wave function matrix a i The dimensions are N×M and N×R respectively. For S different sound source shapes, the holographic surface multi-source wave function matrix A and the reconstruction surface multi-source wave function matrix a are constructed respectively:

[0085]

[0086] The obtained holographic surface multi-source wave function matrix A∈R (S×N)×M , reconstruct the multi-source wave function matrix a∈R (S×N)×R .

[0087] In a preferred implementation of the cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, in step c, the i-th cosine similarity calculation formula is as follows:

[0088]

[0089] Among them, A i，· represents the i-th row of the holographic multi-source wave function matrix A, p H represents the holographic surface sound pressure, ||·|| represents the second norm, and based on the cosine similarity, the i-th weight calculation formula is as follows:

[0090]

[0091] Among them, k is the weight control hyperparameter,

[0092] All weights are combined to obtain the weight matrix W = diag (w1, w2, ..., w i , …), where i = 1, 2, …, S × N, diag (·) represents the diagonal elements of the matrix, then the holographic surface weighted multi-source wave function model A weight = WA, reconstruction surface weighted multi-source wave function model a weight =Wa.

[0093] In a preferred implementation of the cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, the weight control hyperparameter k=1.

[0094] In a preferred implementation of the cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, in step d,

[0095] Combining the weighted multi-source wave function model, we get the following formula:

[0096] A weight c=a weight ,

[0097] In the formula, A weight is the weighted multi-source wave function matrix of the holographic surface, a weight is the weighted multi-source wave function matrix of the reconstruction surface, c is the sound field transfer coefficient, and the sound field transfer coefficient c is solved according to the following formula:

[0098] c=(A weight H A weight +λ 2 I) -1 A weight H a weight ,

[0099] In the formula, λ is the regularization parameter, and the sound field reconstruction is calculated as follows:

[0100] p R =c T p H ,

[0101] In the formula, p R To reconstruct the surface sound pressure, p H is the holographic surface sound pressure.

[0102] In a preferred implementation of the cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, the regularization parameter λ is calculated according to the Hald criterion.

[0103]

[0104] In the formula, is the spatial wave number, f is the sound source frequency, c0 is the sound wave velocity, d is the distance between the holographic surface and the reconstruction surface, and SNR is the signal-to-noise ratio.

[0105] In a preferred implementation of the cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, when the signal-to-noise ratio is unknown, the SNR is 15.

[0106] In a preferred implementation of the cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, a holographic plane H is arranged within half the distance of the wavelength corresponding to the sound source frequency, M microphones are arranged at the nodes of the holographic plane H, and the time domain sound pressure is measured by the microphones. The complex sound pressure corresponding to the frequency is Fourier transformed and extracted to obtain the holographic surface test sound pressure pH.

[0107] In a preferred implementation of the cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, the sound source is a point sound source.

[0108] In one embodiment, Figure 1 As shown, the method comprises the following steps:

[0109] Step 1: Arrange M microphones within a wavelength region away from the sound source to obtain the near-field holographic surface sound pressure pH of the shell structure. Arrange a holographic plane H within half the distance of the wavelength corresponding to the research frequency of the sound source, arrange M microphones at the nodes of the holographic surface, use the microphones to measure the time domain sound pressure, Fourier transform and extract the complex sound pressure corresponding to the research frequency to obtain the holographic surface test sound pressure pH. Theoretically, the holographic surface can be arranged in any shape. For the convenience of calculation, take the plane array as the holographic surface. As shown in the attached figure Figure 2 As shown in the figure, this simulation constructs a dual-line sound source structure. Each line sound source is composed of multiple point sound sources, and the distance between point sound sources is less than half of the wavelength of the sound source. The two line sound sources are parallel to the y-axis and located at x=±0.3m. The holographic surface and the reconstruction surface are both 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.

[0110] According to the simulation example, the sound source arrangement is used to calculate the 1000Hz sound source radiation holographic surface test sound pressure and the reconstruction surface theoretical sound pressure as shown in the attached figure. Figure 3 and attached Figure 4 As shown, a test error of 15dB signal-to-noise ratio is applied to the holographic surface test sound pressure.

[0111] Step 2: According to the shape characteristics of the sound source, the plane, cylinder and spherical unit wave functions are reasonably selected to construct a multi-source wave function model. The three-dimensional coordinates of each measuring point on the holographic surface and the reconstruction surface are obtained, where the coordinates of the measuring point on the holographic surface r H , reconstruct the surface measurement point coordinates r R According to the shape characteristics of the sound source, the plane, cylinder and spherical unit wave functions are reasonably selected, among which the unit plane wave function at the spatial position r in the rectangular coordinate system is as follows:

[0112]

[0113] 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.

[0114] The wave function of the unit cylinder at the spatial position r in the cylindrical coordinate system is as follows:

[0115]

[0116] 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.

[0117] The unit spherical wave function at the spatial position r in the spherical coordinate system is as follows:

[0118]

[0119] 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.

[0120] Based on multi-source statistical optimal near-field acoustic holography, the i-th holographic surface wave function matrix A is obtained by combining the above three different unit wave functions. i And reconstruct the surface wave function matrix a i The dimensions are N×M and N×R respectively. For S different sound source shapes, the holographic surface multi-source wave function matrix A and the reconstruction surface multi-source wave function matrix a are constructed respectively:

[0121]

[0122] The obtained holographic surface multi-source wave function matrix A∈R (S×N)×M , reconstruct the multi-source wave function matrix a∈R (S×N)×R .like Figure 3 As shown, the simulation object of this example is a two-line sound source. When the sound source composition is known, the holographic surface multi-source wave function matrix A is obtained based on the appearance characteristics of the two cylindrical sound sources. 先验 ∈R (2×2401)×625 and the reconstruction surface multi-source wave function matrix a 先验 ∈R (2 ×2401)×625 However, due to the holographic surface sound pressure Figure 3 It can be seen that at x = 0,

[0123] The same cylindrical radiation sound field shape characteristics are presented at ±0.3m. In the absence of prior information, the holographic surface multi-source wave function matrix A is obtained. 无先验 ∈R (3×2401)×625 and the reconstruction surface multi-source wave function matrix a 无先验 ∈R (3×2401)×625 The multi-source wave function models constructed above are respectively called a priori model and no priori model.

[0124] Step 3, calculate the cosine similarity between each order of multi-source wave function model and the holographic surface sound pressure, and update the weighted multi-source wave function model based on the cosine similarity. The calculation formula for the i-th cosine similarity is as follows:

[0125]

[0126] Among them, A i，· Represents the multi-source wave function matrix A of the holographic surface without prior information 无先验 The i-th row, p H represents the holographic surface sound pressure, and ||·|| represents the second norm. Based on the cosine similarity, the calculation formula for the i-th weight is as follows:

[0127]

[0128] Among them, k is a weight control hyperparameter, generally, k=1.

[0129] All weights are combined to obtain the weight matrix W = diag (w1, w2, ..., w i ,…,w 7203 ), where i = 1, 2, ..., 7203, and diag(·) represents the diagonal elements of the matrix. Then the holographic surface weighted multi-source wave function model A weight =WA 无先验 , reconstruction surface weighted multi-source wave function model a weight =Wa 无先验 .

[0130] Step 4: Solve the transfer coefficient of the weighted multi-source wave function model of the holographic surface and the reconstruction surface by using the Tikhonov regularization method, so as to achieve high-precision reconstruction of the multi-source sound field. The three multi-source wave function models obtained in the above steps are combined to obtain the following formula:

[0131] Ac=a,

[0132] The sound field transfer coefficient c under three different multi-source wave function models is solved according to the following formula:

[0133] c=(A H A+λ 2 I) -1 A H a,

[0134] Where λ is the regularization parameter. The regularization parameter is calculated according to the Hald criterion, that is:

[0135]

[0136] In the formula, is the spatial wave number, f is the sound source frequency, c0 is the sound wave velocity, d is the distance between the holographic surface and the reconstruction surface, and SNR is the signal-to-noise ratio. When the signal-to-noise ratio is unknown, SNR=15 is usually taken. The sound field reconstruction calculation is as follows:

[0137] p R =c T p H ,

[0138] In the formula, p R To reconstruct the surface sound pressure, p H is the holographic surface sound pressure. The weighted multi-source wave function model is used to solve the sound field transfer coefficient and calculate the reconstructed surface sound pressure. The results after visualization are as follows: Figure 5 As shown. In order to verify the effectiveness of the multi-source statistical optimal near-field acoustic holography method weighted by cosine similarity 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. The overall relative error is defined as:

[0139]

[0140] According to the above steps, calculate Figure 4 and Figure 5 The average overall relative error between them is 7.32%, which strongly verifies the effectiveness and robustness of this method in reconstructing multi-source combined sound fields in the absence of accurate prior information about the sound sources. Figure 6 The weight factors obtained after normalization of cosine similarity of multi-source wave function models of different orders are given, among which the 2401 weights on the left are all holographic surfaces. Figure 3 The virtual sound source at x=0m corresponds to the wave function model, and the 2401 weights in the middle and on the far right are the wave function models corresponding to sound source 1 and sound source 2. Figure 6 It can be seen that the proposed method effectively reduces the influence of the virtual source matching wave function model on the results, and only part of the model is still involved in the calculation of the final sound field transfer coefficient. Figure 6 The minimum weight factor is shown to be greater than 0.5, which indicates that this method can only limit the modeling error rather than completely remove it. The minimum weight factor can be adjusted by changing the weight control hyperparameter.

[0141] Figure 7 A line graph showing the reconstruction error changing with the source frequency when SNR = 15dB is given. Figure 8A line graph showing the reconstruction error versus signal-to-noise ratio when the sound source frequency is f = 1000 Hz is given. Figure 7 It can be seen that in the entire research frequency band, by reasonably adjusting the weights, the proposed method can achieve a similar effect to that of establishing a multi-source wave function model under prior information, and the worst result is still due to the lack of prior information. Figure 8 It can be seen that when the signal-to-noise ratio is >5dB, the effect of this method is equivalent to the result of the prior model; when the signal-to-noise ratio is less than 5dB, the effect of this method is equivalent to that without a priori model. This is because, when the signal-to-noise ratio is low, there is a large error in the sound pressure of the holographic surface obtained by the test, and the cosine similarity cannot effectively reflect the modeling error, which leads to a decrease in the effect of the method proposed in the present invention. However, the method proposed in the present invention can ensure that the sound field reconstruction result is better than that without prior information, and provides theoretical support for the engineering application of the multi-source statistical optimal near-field acoustic holography method, which has technical reference value.

[0142] In one embodiment, the method includes,

[0143] M microphones are arranged in a wavelength region away from the sound source to obtain the pH of the near-field holographic surface sound pressure of the shell structure; according to the shape characteristics of the sound source, the plane, cylindrical and spherical unit wave functions are reasonably selected to construct a multi-source wave function model; the cosine similarity between each order of the multi-source wave function model and the holographic surface sound pressure is calculated, and the weighted multi-source wave function model is obtained based on the cosine similarity update; the transfer coefficient of the weighted multi-source wave function model between the holographic surface and the reconstruction surface is solved by the Tikhonov regularization method, so as to achieve high-precision reconstruction of the multi-source sound field. Compared with the traditional multi-source statistical optimal near-field acoustic holography method, the present invention uses cosine similarity to assign different weights to the multi-source wave function model, effectively reducing the modeling error caused by the mismatch between the wave function model and the actual sound source, providing theoretical support for the engineering application of the multi-source statistical optimal near-field acoustic holography method, and has technical reference value.

[0144] A cosine similarity weighted multi-source statistical optimal near-field acoustic holography system includes:

[0145] A collection unit, which includes M microphones arranged in a wavelength region away from the sound source, to obtain the near-field holographic surface sound pressure pH of the shell structure;

[0146] A modeling unit, which selects plane, cylindrical and spherical unit wave functions based on the shape characteristics of the sound source to construct a multi-source wave function model;

[0147] A cosine similarity calculation unit is used to calculate the cosine similarity between each order of multi-source wave function model and the holographic surface sound pressure pH, and to obtain a weighted multi-source wave function model based on the cosine similarity update;

[0148] The reconstruction unit solves the transfer coefficient of the weighted multi-source wave function model between the holographic surface and the reconstruction surface through the Tikhonov regularization method, thereby achieving high-precision reconstruction of the multi-source sound field.

[0149] A computer storage medium includes computer instructions, which, when executed on a computer, cause the computer to execute the method described.

[0150] An electronic device, comprising:

[0151] A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein:

[0152] When the processor executes the program, the method described is implemented.

[0153] 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 cosine similarity weighted multi-source statistical optimal near-field acoustic holography method, characterized in that: The steps include: Step a, arranging M microphones within a wavelength region away from the sound source to obtain the near-field holographic surface sound pressure pH of the shell structure; Step b, selecting plane, cylindrical and spherical unit wave functions based on the shape characteristics of the sound source to construct a multi-source wave function model; Step c, calculating the cosine similarity between each order of multi-source wave function model and the holographic surface sound pressure pH, and updating the weighted multi-source wave function model based on the cosine similarity; Step d, solving the transfer coefficient of the weighted multi-source wave function model between the holographic surface and the reconstruction surface by the Tikhonov regularization method, thereby achieving high-precision reconstruction of the multi-source sound field.

2. The multi-source statistical optimal near-field acoustic holography method weighted by cosine similarity according to claim 1 is characterized in that: Preferably, the array consisting of M microphones covers the entire sound source.

3. The multi-source statistical optimal near-field acoustic holography method weighted by cosine similarity according to claim 1 is characterized in that: In step b, constructing a multi-source 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 , 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 i-th holographic surface wave function matrix A is obtained by combining three different unit wave functions based on multi-source statistical optimal near-field acoustic holography. i And reconstruct the surface wave function matrix a i The dimensions are N×M and N×R respectively. For S different sound source shapes, the holographic surface multi-source wave function matrix A and the reconstruction surface multi-source wave function matrix a are constructed respectively: The obtained holographic surface multi-source wave function matrix A∈R (S×N)×N , reconstruct the multi-source wave function matrix a∈R (S×N)×R .

4. The multi-source statistical optimal near-field acoustic holography method weighted by cosine similarity according to claim 1 is characterized in that: In step c, the calculation formula for the i-th cosine similarity is as follows: Among them, A i，· represents the i-th row of the holographic multi-source wave function matrix A, p H represents the holographic surface sound pressure, ||·|| represents the second norm, and based on the cosine similarity, the i-th weight calculation formula is as follows: Among them, k is the weight control hyperparameter, All weights are combined to obtain the weight matrix W = diag (w1, w2, ..., w i , …), where i = 1, 2, …, S × N, diag (·) represents the diagonal elements of the matrix, then the holographic surface weighted multi-source wave function model A weight = WA, reconstruction surface weighted multi-source wave function model a weight =Wa.

5. The cosine similarity weighted multi-source statistical optimal near-field acoustic holography method according to claim 4, characterized in that: The weight control hyperparameter k==1.

6. The cosine similarity weighted multi-source statistical optimal near-field acoustic holography method according to claim 5, characterized in that: In step d, Combining the weighted multi-source wave function model, we get the following formula: A weight c=a weight , In the formula, A weight is the weighted multi-source wave function matrix of the holographic surface, a weight is the weighted multi-source wave function matrix of the reconstruction surface, c is the sound field transfer coefficient, and the sound field transfer coefficient c is solved according to the following formula: c=(A weight H A weight +λ 2 I) -1 A weight H a weight , In the formula, λ is the regularization parameter, and the sound field reconstruction is calculated as follows: p R =c T p H , In the formula, p R To reconstruct the surface sound pressure, p H is the holographic surface sound pressure.

7. The cosine similarity weighted multi-source statistical optimal near-field acoustic holography method according to claim 6, characterized in that: The regularization parameter λ is calculated according to the Hald criterion, In the formula, is the spatial wave number, f is the sound source frequency, c0 is the sound wave velocity, d is the distance between the holographic surface and the reconstruction surface, and SNR is the signal-to-noise ratio.

8. A cosine similarity weighted multi-source statistical optimal near-field acoustic holography system, characterized in that: include, The acquisition unit includes M microphones arranged in a wavelength region away from the sound source, which are used to obtain the near-field holographic surface sound pressure p of the shell structure. H ; A modeling unit, which is used to select plane, cylindrical and spherical unit wave functions based on the shape characteristics of the sound source to construct a multi-source wave function model; A cosine similarity calculation unit is used to calculate the cosine similarity between each order of multi-source wave function model and the holographic surface sound pressure pH, and update the weighted multi-source wave function model based on the cosine similarity; The reconstruction unit is used to solve the transfer coefficient of the weighted multi-source wave function model of the holographic surface and the reconstruction surface through the Tikhonov regularization method, so as to achieve high-precision reconstruction of the multi-source 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