Space projection regularization method for sound source localization and sound field reconstruction and related device

Abstract

The invention discloses a spatial projection regularization method for sound source localization and sound field reconstruction and a related device, and belongs to the technical field of sound source localization. Specifically, according to the method, an equivalent source method model is utilized, and an external radiation sound field of a sound source to be measured is simulated through a series of virtual sources. Then, according to the first sound pressure transfer matrix from the virtual source arrangement surface to the holographic measuring point plane and the sound pressure of the holographic measuring point, a space projection regularization algorithm is used for solving the inverse problem, and therefore the equivalent source intensity is obtained; and then, reconstructing an external radiation sound field of the sound source to be detected by using the second sound pressure transfer matrix and the equivalent source intensity. And finally, by taking the maximum sound pressure amplitude as a standard, screening field points in an external radiation sound field so as to determine the position of the sound source to be detected. According to the space projection regularization algorithm, truncation processing of the vector space is carried out with the projection size of the measurement information in the vector space as the criterion, real information is reserved, and noise interference is restrained to the maximum extent.

Description

Technical Field

[0001] The present application relates to the technical field of sound source localization, and in particular to a spatial projection regularization method and related devices for sound source localization and sound field reconstruction. Background Art

[0002] With the continuous progress of society and the significant improvement in people's living standards, electromechanical products have been widely integrated into our daily lives, and the resulting noise problems have gradually spread from the industrial and transportation fields to all aspects of life. At present, the noise problem has become one of the key indicators for evaluating environmental quality and product quality. Therefore, the global manufacturing industry has invested huge amounts of technology and funds to control noise pollution in order to enhance the market competitiveness of its products. Generally speaking, the most fundamental and effective way to control noise is to control the sound source. In sound source control technology, the most critical links are the identification and positioning of the noise source and the accuracy of sound field reconstruction. At present, the algorithm most likely to meet the future demand for high-precision and high-efficiency sound field reconstruction is the Equivalent Source Method (ESM).

[0003] However, the primary challenge facing the ESM algorithm is the ill-posed inverse problem that is prevalent in sound field reconstruction algorithms. The ill-posed inverse problem manifests itself in two main aspects. First, due to objective conditions, the input data of the inverse problem is often insufficient or excessive, which leads to the non-uniqueness or non-existence of the solution to the problem. Second, the solution to the inverse problem is discontinuously dependent on the input data. When there are errors in the measured input information, it may have a significant impact on the solution and may even cause the solution to lose its meaning. Therefore, solving the ill-posed inverse problem of the equivalent source method is crucial for achieving high-precision sound field reconstruction and sound source localization using the equivalent source method. Summary of the Invention

[0004] The purpose of this application is to provide a spatial projection regularization method and related devices for sound source localization and sound field reconstruction, which can maximize the suppression of noise interference while retaining real information and greatly improve the computational efficiency of solving the acoustic inverse problem of the equivalent source method.

[0005] To achieve the above objectives, this application provides the following solutions:

[0006] In a first aspect, the present application provides a spatial projection regularization method for sound source localization and sound field reconstruction, comprising:

[0007] At the sound source to be measured, the equivalent source method is used to construct an equivalent source method model; in the equivalent source method model, a series of virtual sources inside the sound source are used to equivalently simulate the external radiation sound field of the sound source to be measured, and a microphone array is arranged in the near field of the sound source to construct a holographic measurement point plane.

[0008] According to the position and number of virtual sources, the first sound pressure transfer matrix from the virtual source arrangement surface to the holographic measurement point plane is constructed; the first sound pressure transfer matrix is ​​used to characterize the sound pressure transfer relationship between any virtual source and any holographic measurement point on the holographic measurement point plane.

[0009] According to the first sound pressure transfer matrix and the sound pressure of the holographic measurement point, the spatial projection regularization algorithm is used to solve the inverse problem to obtain the equivalent source intensity; the spatial projection regularization algorithm truncates the vector space based on the projection size of the holographic measurement point sound pressure data in the vector space, and in the spatial projection regularization algorithm, the double generalized cross-validation algorithm is used to automatically select the appropriate regularization parameter.

[0010] The sound pressure at any point in the external radiation sound field of the sound source to be measured is solved according to the second sound pressure transfer matrix and the equivalent source intensity, and the external radiation sound field of the sound source to be measured is reconstructed; the second sound pressure transfer matrix is ​​used to characterize the sound pressure transfer relationship between any virtual source and any point in the external radiation sound field.

[0011] In the reconstructed external radiation sound field, the field points in the external radiation sound field are screened based on the maximum sound pressure amplitude as a standard to determine the position of the sound source to be measured.

[0012] In a second aspect, the present application provides a spatial projection regularization device for sound source localization and sound field reconstruction, comprising the following functional modules:

[0013] The equivalent source method model modeling module is used to construct an equivalent source method model at the sound source to be measured using the equivalent source method; in the equivalent source method model, a series of virtual sources inside the sound source are used to equivalently simulate the external radiation sound field of the sound source to be measured, and a microphone array is arranged in the near field of the sound source to construct a holographic measurement point plane.

[0014] The sound pressure transfer matrix construction module is used to construct a first sound pressure transfer matrix from the virtual source arrangement surface to the holographic measurement point plane according to the position and number of virtual sources; the first sound pressure transfer matrix is ​​used to characterize the sound pressure transfer relationship between any virtual source and any holographic measurement point on the holographic measurement point plane.

[0015] The spatial projection regularization solution module is used to solve the inverse problem using the spatial projection regularization algorithm based on the first sound pressure transfer matrix and the sound pressure of the holographic measurement point to obtain the equivalent source intensity; the spatial projection regularization algorithm truncates the vector space based on the projection size of the holographic measurement point sound pressure data in the vector space, and in the spatial projection regularization algorithm, a double generalized cross-validation algorithm is used to automatically select the appropriate regularization parameter.

[0016] The sound source external sound field reconstruction module is used to solve the sound pressure of any point in the external radiation sound field of the sound source to be measured based on the second sound pressure transfer matrix and the equivalent source intensity, and reconstruct the external radiation sound field of the sound source to be measured; the second sound pressure transfer matrix is ​​used to characterize the sound pressure transfer relationship between any virtual source and any point in the external radiation sound field.

[0017] The sound source positioning module to be measured is used to screen the field points in the external radiation sound field based on the maximum sound pressure amplitude in the reconstructed external radiation sound field to determine the position of the sound source to be measured.

[0018] In a third aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the spatial projection regularization method for sound source localization and sound field reconstruction described above.

[0019] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the spatial projection regularization method for sound source localization and sound field reconstruction described above.

[0020] In a fifth aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of the spatial projection regularization method for sound source localization and sound field reconstruction described above.

[0021] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0022] The present application provides a spatial projection regularization method and related device for sound source localization and sound field reconstruction. In this method, an equivalent source method model is first used to simulate the external radiation sound field of the sound source to be measured through a series of virtual sources inside the sound source, and a microphone array is arranged in the near field of the sound source to construct a holographic measurement point plane. Then, based on the first sound pressure transfer matrix from the virtual source arrangement surface to the holographic measurement point plane and the holographic measurement point sound pressure, a spatial projection regularization algorithm is used to solve the inverse problem to obtain the equivalent source intensity. A double generalized cross-validation algorithm is also used to automatically select a suitable regularization parameter to maximize the suppression of noise interference. Subsequently, the sound pressure of any field point in the external radiation sound field of the sound source to be measured is solved based on the second sound pressure transfer matrix and the equivalent source intensity, and the external radiation sound field of the sound source to be measured is reconstructed. Finally, the field points in the external radiation sound field are screened based on the maximum sound pressure amplitude as a standard to determine the position of the sound source to be measured. The spatial projection regularization algorithm adopted in the above-mentioned solution of this application truncates the vector space based on the projection size of the measurement information in the vector space. It can retain the real information while maximally suppressing the interference of noise. At the same time, in order to meet the needs of automated calculation, the double generalized cross-validation algorithm that automatically selects the appropriate regularization parameter based on the characteristics of the spatial projection regularization algorithm can greatly improve the computational efficiency of solving the acoustic inverse problem of the equivalent source method. Compared with the traditional regularization method, the average relative error of the spatial projection regularization algorithm at all nodes is significantly better than the truncated singular value decomposition algorithm TSVD and the Tikhonov regularization algorithm, and the relative error is smaller at locations with smaller sound pressure amplitudes, better restoring the detailed information of the sound field. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 This is a flowchart of a spatial projection regularization method for sound source localization and sound field reconstruction provided in one embodiment of the present application.

[0025] Figure 2 A schematic diagram of a technical route of a spatial projection regularization method for sound source localization and sound field reconstruction provided in one embodiment of the present application.

[0026] Figure 3 A schematic diagram of the model distribution when a spatial projection regularization method for sound source localization and sound field reconstruction provided in this application is applied in a simulation example.

[0027] Figure 4This is a schematic diagram comparing the regularization parameters selected by the spatial projection regularization algorithm SPR provided in this application and the traditional regularization algorithms TSVD and Tikhonov under four types of sound fields.

[0028] Figure 5 Error cloud diagrams of the reconstructed sound fields and the corresponding analytical solutions of the spatial projection regularization algorithm SPR provided in this application and the traditional regularization algorithms TSVD and Tikhonov under the sound fields of four sources.

[0029] Figure 6 The following are line graphs of the reconstruction node errors of the spatial projection regularization algorithm SPR provided in this application and the traditional regularization algorithms TSVD and Tikhonov under four sound fields of 100 to 2000 Hz.

[0030] Figure 7 This is a schematic diagram of experimental model parameter settings for a spatial projection regularization method for sound source localization and sound field reconstruction provided in this application when applied to a dual sound source localization experiment.

[0031] Figure 8 The sound source localization cloud images of the spatial projection regularization algorithm SPR and the traditional regularization algorithms TSVD and Tikhonov provided in this application at 400Hz, 800Hz, 1200Hz and 1600Hz respectively.

[0032] Figure 9 A schematic diagram of the functional modules of a spatial projection regularization system for sound source localization and sound field reconstruction provided in one embodiment of the present application.

[0033] Figure 10 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0035] For the equivalent source method (ESM), the first problem to overcome is the ill-posedness in the sound field reconstruction. The ill-posedness is characterized in two aspects. On one hand, due to the limitation of objective conditions, the input data of the inverse problem is often underdetermined or overdetermined, leading to non-uniqueness or nonexistence of the solution. On the other hand, the solution of the inverse problem does not have a continuous dependence on the input data. When the input information measured has errors, the solution will be significantly affected, and even the solution can be meaningless. Therefore, under the actual background noise interference, a regularization method needs to be used to stabilize the solution process. The traditional regularization methods such as Tikhonov regularization and truncated singular value decomposition (TSVD) all filter out the space corresponding to the small singular values to suppress the influence of measurement errors on the parameters to be solved. In fact, due to the random distribution characteristics of noise, the noise is distributed in the space corresponding to each singular value, not only in the space corresponding to the small singular values. When a large amount of noise is distributed in a large space, the noise in the space corresponding to the large singular values will significantly interfere with the solution. Therefore, a regularization method with full-space filtering must be constructed to solve the above problems.

[0036] The present application aims to analyze the projection size of real information and noise in the space vector based on the basic equation of the inverse problem, and proposes a spatial projection regularization (SPR) algorithm with full-space filtering. At the same time, in order to meet the needs of automatic calculation, a double generalized cross-validation method for automatically selecting the regularization parameter is further constructed according to the characteristics of the spatial projection regularization algorithm. The above spatial projection regularization based on double generalized cross-validation has stronger spatial filtering capability and can adapt to sound field reconstruction and sound source positioning under different noise interference.

[0037] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below in combination with the drawings and specific embodiments.

[0038] The spatial projection regularization method for sound source positioning and sound field reconstruction provided by the embodiments of the present application includes the following steps in an exemplary embodiment, as shown in Figure 1

[0039] ​S1. At the sound source to be measured, an equivalent source method model is constructed using the equivalent source method. In this equivalent source method model, a series of virtual sources within the sound source are used to simulate the external sound field radiated by the sound source to be measured. A microphone array is arranged in the near field of the sound source to construct a holographic measurement point plane. Generally speaking, the series of virtual sources in the equivalent source method model are monopole sources, dipole sources, or a combination of multipole sources. The virtual sources are located within the sound source or in the near field to avoid overlap between the virtual source positions and the external sound field points.

[0040] In this embodiment, the equivalent source method can be expressed by the following formula:

[0041] p(r)=∫∫μ(r s' )G(r,r s′ )ds′.

[0042] Wherein, p(r) is the external radiation sound field of the sound source to be measured, μ(r s′ ) is the virtual source strength, G(r,r s' ) is the free space Green's function s′ is the virtual source layout surface. The virtual source layout surface s′ grid is discretized into D units and approximated by constant units. The above formula can be transformed into the following form:

[0043]

[0044] Among them, q(r s' ) is the equivalent source intensity of the virtual source, the subscript i represents the grid unit of the ith virtual source layout surface of the grid discretization, and Δs' is the area of ​​the discrete unit.

[0045] Converting the above formula into matrix form is as follows:

[0046] P H =G H Q.

[0047] Among them, P H is the column vector form of the sound pressure at the holographic measurement point, P H =[p(r H1 ),p(r H2 ),…,p(r HM )] T , M is the number of holographic measurement points, G H is the first sound pressure transfer matrix from the virtual source layout plane to the holographic measurement point plane, Q is the column vector form of the equivalent source intensity, Q=[q(r E1 ),q(r E2 ),…,q(r EN )], N is the number of virtual sources.

[0048] S2, constructing a first sound pressure transfer matrix from the virtual source arrangement surface to the holographic measurement point plane according to the positions and the number of the virtual sources; the first sound pressure transfer matrix is used to represent the sound pressure transfer relationship between any virtual source and any holographic measurement point on the holographic measurement point plane. Wherein the element G ij represents the sound pressure contribution coefficient of the jth virtual source to the ith holographic measurement point.

[0049] S3, solving the inverse problem by using a spatial projection regularization algorithm according to the first sound pressure transfer matrix and the holographic measurement point sound pressure, to obtain the equivalent source strength; the spatial projection regularization algorithm truncates the vector space according to the projection size of the holographic measurement point sound pressure data in the vector space, and in the spatial projection regularization algorithm, a double generalized cross-validation algorithm is used to automatically select a suitable regularization parameter. The process of solving the source strength Q is a typical inverse problem, and a regularization method is needed to stabilize the solving process.

[0050] Specifically in this embodiment, the projection size of the holographic measurement point sound pressure data in the vector space is |u i T (b+ε)|, because |u i T (b+ε)| satisfies:

[0051] |u i T b|+|u i T b|+|u i T ε|.

[0052] When ui satisfies |u i T b|≈|u i T ε| and |u i T b|<|u i T ε|, since the noise ε is small, the projection |u i ε| of the vector space u i T ε| is also small. Then the projection |u i (b+ε)| of the measurement information in the vector space u i T (b+ε)| must also be small. At this time, the projection of the real information b in the vector space u i (b+ε)| is easily covered by noise. Therefore, truncating the vector space according to the size of |u i T (b+ε)| can maximize the suppression of noise interference. Truncating the vector space according to the projection size of the holographic measurement point sound pressure data in the vector space is shown in the following formula:

[0053] |u i T (b+ε)|≤t.

[0054] Among them, u i is the vector space, b is the sound pressure data of the holographic measurement point in vector form, ε is the noise vector, and t is the first truncation threshold.

[0055] In order to realize the automatic selection of regularization parameters, the double generalized cross-validation algorithm is used in the spatial projection regularization algorithm to automatically select the appropriate regularization parameters. In the double generalized cross-validation algorithm, the formula for the first generalized cross-validation is:

[0056]

[0057] Among them, G1(i) is the formula of generalized cross validation, A is the first sound pressure transfer matrix, A i is a regularization matrix constructed by retaining the first i vector spaces in order according to the size of the singular values, I is the unit matrix, is the mean of the holographic measurement point sound pressure data in vector form, Tr() is the trace of the matrix; according to t=(i,min(G1(i))), the i when G1(i) is the smallest is used as the first truncation threshold, and the first truncation threshold is used to screen the vector space whose singular value is greater than the threshold.

[0058] The formula for the second generalized cross validation is:

[0059]

[0060] Among them, G1(i) is the new generalized cross-validation formula, ε is the noise vector, according to t sort =(i,min(G2(i))) takes i when G2(i) is the smallest as the second truncation threshold, which is used to filter out vector spaces whose projections are greater than the threshold.

[0061] The equivalent source strength is obtained according to the following formula:

[0062]

[0063] Among them, U c and V is the unitary matrix obtained by singular value decomposition, ∑ is the singular value matrix, X SPR is the equivalent source strength.

[0064] In the above-mentioned spatial projection regularization algorithm combined with double generalized cross-validation, the input b is equivalent to the measurement information P of the equivalent source method. H , the input A represents the transfer matrix G H , calculated X SPRThe corresponding output is the equivalent source strength Q.

[0065] S4. Solve the sound pressure at any point in the external radiation sound field of the sound source to be measured based on the second sound pressure transfer matrix and the equivalent source intensity, and reconstruct the external radiation sound field of the sound source to be measured; the second sound pressure transfer matrix is ​​used to characterize the sound pressure transfer relationship between any virtual source and any point in the external radiation sound field. The external radiation sound field of the sound source to be measured is obtained according to the following formula:

[0066] P r =G r Q.

[0067] Among them, P r is the external radiation sound field of the sound source to be measured, G r is the second sound pressure transfer matrix, Q is the equivalent source intensity X obtained in the above formula SPR .

[0068] S5. In the reconstructed external radiated sound field, the field points in the external radiated sound field are screened based on the maximum sound pressure amplitude to determine the location of the sound source to be measured. In this embodiment, the location of the actual sound source can be determined by retaining the sound field contour map within the -6dB dynamic range of the sound pressure difference based on the maximum sound pressure amplitude.

[0069] The above embodiments of the present application provide a spatial projection regularization method for sound source localization and sound field reconstruction, such as Figure 2 The schematic diagram of the technical route shown in the figure first establishes an equivalent source method model, and uses a series of virtual sources inside the sound source to equivalently simulate the external radiation sound field of the sound source to be measured, and arranges a microphone array in the near field of the sound source to construct a holographic measurement point plane. Then, based on the first sound pressure transfer matrix from the virtual source arrangement surface to the holographic measurement point plane and the holographic measurement point sound pressure, the spatial projection regularization algorithm is used to solve the inverse problem to obtain the equivalent source strength. The double generalized cross-validation algorithm is also used to automatically select the appropriate regularization parameter to maximize the suppression of noise interference. Subsequently, the sound pressure of any field point in the external radiation sound field of the sound source to be measured is solved according to the second sound pressure transfer matrix and the equivalent source strength, and the external radiation sound field of the sound source to be measured is reconstructed. Finally, the field points in the external radiation sound field are screened based on the maximum sound pressure amplitude as the standard to determine the position of the sound source to be measured.

[0070] The following simulation example verifies the spatial projection regularization method for sound source localization and sound field reconstruction provided by the above embodiment of the present application, and uses the equivalent source method to analyze simple monopole, dipole, quadrupole and asymmetric sound sources. Figure 3As shown, the reconstructed surface is a regular hexahedron with edge length of 0.8 m, the internal equivalent source sphere radius a = 0.3 m, and there are 225 equivalent source points. The holographic sphere radius is 1 m, and there are 400 scanning measurement points. Gaussian white noise with a signal-to-noise ratio of 10 dB is added to the holographic measurement points, and the calculation frequency is 1000 Hz.

[0071] The generalized cross-validation algorithm GCV is used to select the truncation point and regularization parameter in the two traditional regularization algorithms TSVD and Tikhonov, respectively. In the spatial projection regularization algorithm SPR proposed in this embodiment, the double generalized cross-validation algorithm D-GCV is used to select the regularization parameter. As shown in Figure 4 As shown, under the sound field of the four sources, the truncation points selected by TSVD using GCV are 45, 20, 56 and 56, respectively. The regularization parameter a selected by Tikhonov using GCV is not selected according to the singular value, in order to facilitate comparison, the singular value number closest to the selected regularization parameter is given, which is 65, 51, 80 and 66, respectively. The SPR algorithm retains the front 4, 3, 4 and 16 projection vector spaces using double generalized cross-validation, respectively. The position is basically the same as the best regularization parameter, which verifies the effectiveness and stability of the double generalized cross-validation automatic parameter selection. It needs to be specially pointed out that the vector space retained by D-GCV using two GCV cross-validation methods is not necessarily the minimum point of the G(i) function.

[0072] The source intensity solution Q calculated according to the three algorithms TSVD , Q Tikhonov , Q SPR The sound field on the surface of the reconstructed regular hexahedron is reconstructed, and the sound field cloud map solution and error cloud map of the three regularization methods are as shown in Figure 5 As can be seen from the figure, in the monopole sound source, the average relative errors of all nodes of TSVD and Tikhonov are 12.19% and 13.90%, respectively, while the error of SPR is only 3.82%, and the calculation accuracy is significantly improved. From the error cloud map, SP algorithm is significantly better than TSVD and Tikhonov, and the error of each node is less than 10%, which shows stronger anti-interference ability. In the dipole sound source, the average relative errors of all nodes of TSVD and Tikhonov are 8.11% and 15.31%, respectively, and the error of SPR is 3.86%. Secondly, when close to the xoy plane, the sound pressure amplitude is small and is easily disturbed by noise, and the relative error of TSVD and Tikhonov in the xoy plane is large, while the SPR algorithm still ensures good calculation accuracy. In the four-pole and asymmetric sound source, the average relative error of all nodes of the SPR algorithm is significantly better than that of the TSVD and Tikhonov algorithms, and the relative error is smaller in the position with small sound pressure amplitude, which better restores the detail information of the sound field.

[0073] In order to further verify the stability and applicability of the SPR algorithm, the relative average error comparison of the sound field reconstruction nodes of monopole, dipole, quadrupole and asymmetric sound source at 100~2000Hz is given. Figure 6 In the 100-2000Hz range, the SPR algorithm demonstrates significant advantages across all four acoustic environments, achieving significantly higher accuracy than TSVD and Tikhonov. The errors at 1716Hz for a monopole source, 1961Hz for a dipole source, 1667Hz for a quadrupole source, and 1912Hz for asymmetric sources are significantly greater than those at other frequencies. This is due to the non-uniqueness of the solutions at these eigenfrequencies, which can be addressed using methods such as the CHIEF method and the Burton-Miller method. Since this non-uniqueness issue is less relevant to this paper, it will not be discussed in detail here; please refer to the references above for details. Based on the above analysis, the double generalized cross-validation method constructed in this paper selects an appropriate regularization parameter, enabling the SPR algorithm to achieve high accuracy across the frequency range, significantly outperforming traditional regularization methods such as TSVD and Tikhonov, which filter based on singular value size. These examples fully validate the correctness of the SPR theory.

[0074] In another exemplary embodiment of the present application, a dual sound source localization test is used to verify the effectiveness of the above algorithm. The dual sound source localization test first simulates the sound field radiated from the outside of the actual sound source through two speakers at known positions, and then uses a microphone array to scan the sound pressure on the holographic surface outside the sound source. Finally, the measured sound pressure data is used in the sound field reconstruction algorithm to identify and locate the sound source. The equipment used in the test is H&K's fan-shaped wheel-shaped microphone array and LAN-Ⅺ data collector. The center position of the microphone array is the origin of the coordinate system. The speaker and the microphone array are 0.2m apart. The center coordinates of the two speakers are (0.15, 0, 0.2)m and (-0.15, 0, 0.2)m respectively. In order to ensure the accuracy of the sound field calculation of the equivalent source method, the sound pressure information of 90 holographic measurement points is obtained by translating the microphone array four times up, down, left and right. The specific test arrangement and corresponding coordinates are as follows. Figure 7 The calculated frequencies are 400Hz, 800Hz, 1200Hz, and 1600Hz.

[0075] The fixed microphone is used as the reference signal, and the reference microphone cross-spectrum method is used to obtain the amplitude and phase of the sound pressure at all holographic points.

[0076]

[0077] S r,h (ω) is the cross spectrum of the sound pressure of the reference microphone and the sound pressure of the microphone array, Sr,r (ω) is the autospectrum of the sound pressure of the reference microphone, S h,h (ω) is the autospectrum of the sound pressure from the microphone array. The complex sound pressure p(r) on all holographic surfaces is obtained by processing the collected time-domain signals through cross- and autospectral processing. The equivalent source method is used to reconstruct the sound field on the source surface based on the calculated sound pressure, and a contour map with a sound pressure difference of -6dB is used for sound source localization.

[0078] In order to verify the calculation effect of the three algorithms, the sound field is reconstructed and the sound field cloud map within the dynamic range of -6dB sound pressure difference is retained with the maximum sound pressure amplitude as the standard. The calculation results are as follows Figure 8 As shown in the figure. At a frequency of 400Hz, the TSVD and Tikhonov algorithms have similar cloud patterns due to their closely preserved vector spaces. However, due to the interference of low-frequency reflection noise, the reconstructed cloud patterns of these two algorithms exhibit asymmetric shapes, making it impossible to confirm the actual sound source location. The SPR algorithm outperforms the TSVD and Tikhonov algorithms, with better filtering performance. Its cloud pattern is symmetrical, but the center is offset. This may be due to the upward shift of the maximum sound pressure amplitude due to ground reflection. At 800Hz, the maximum sound pressure amplitude of the TSVD and Tikhonov algorithms is far away from the left sound source location, while the right sound source location can be basically determined. The SPR algorithm can basically determine the location of both left and right sound sources, and the calculation results are also better than those of the TSVD and Tikhonov algorithms. However, all three algorithms exhibit significant sidelobe interference in both the left and right directions. At a frequency of 1200Hz, the localization performance of all three algorithms is significantly improved, but the TSVD and Tikhonov algorithms still have some sidelobe interference. At a frequency of 1600 Hz, all three algorithms can accurately locate the sound source. However, the maximum sound pressure amplitude of the TSVD and Tikhonov algorithms deviates significantly from the actual sound source location, while the SPR algorithm achieves a smaller deviation. Based on the above analysis, the SPR algorithm, by retaining less vector space, better suppresses noise interference and achieves significantly better computational accuracy than the TSVD and Tikhonov algorithms. This further demonstrates the theoretical validity of the spatial projection algorithm and the effectiveness of the D-GCV parameter selection.

[0079] Based on the same inventive concept, the embodiment of the present application also provides a system for implementing the above-mentioned spatial projection regularization method for sound source localization and sound field reconstruction. The solution provided by the system is similar to the solution described in the above-mentioned method. In an exemplary embodiment, Figure 9 As shown, a spatial projection regularization system for sound source localization and sound field reconstruction is provided, comprising:

[0080] The equivalent source method model modeling module is used to construct an equivalent source method model at the sound source to be measured using the equivalent source method; in the equivalent source method model, a series of virtual sources inside the sound source are used to equivalently simulate the external radiation sound field of the sound source to be measured, and a microphone array is arranged in the near field of the sound source to construct a holographic measurement point plane.

[0081] The sound pressure transfer matrix construction module is used to construct a first sound pressure transfer matrix from the virtual source arrangement surface to the holographic measurement point plane according to the position and number of virtual sources; the first sound pressure transfer matrix is ​​used to characterize the sound pressure transfer relationship between any virtual source and any holographic measurement point on the holographic measurement point plane.

[0082] The spatial projection regularization solution module is used to solve the inverse problem using the spatial projection regularization algorithm based on the first sound pressure transfer matrix and the sound pressure of the holographic measurement point to obtain the equivalent source intensity; the spatial projection regularization algorithm truncates the vector space based on the projection size of the holographic measurement point sound pressure data in the vector space, and in the spatial projection regularization algorithm, a double generalized cross-validation algorithm is used to automatically select the appropriate regularization parameter.

[0083] The sound source external sound field reconstruction module is used to solve the sound pressure of any point in the external radiation sound field of the sound source to be measured based on the second sound pressure transfer matrix and the equivalent source intensity, and reconstruct the external radiation sound field of the sound source to be measured; the second sound pressure transfer matrix is ​​used to characterize the sound pressure transfer relationship between any virtual source and any point in the external radiation sound field.

[0084] The sound source positioning module to be measured is used to screen the field points in the external radiation sound field based on the maximum sound pressure amplitude in the reconstructed external radiation sound field to determine the position of the sound source to be measured.

[0085] certainly, Figure 9 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different functions. Figure 9 One or at least two components of the system shown.

[0086] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 10As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a spatial projection regularization method for sound source localization and sound field reconstruction provided in the above embodiment can be implemented.

[0087] Those skilled in the art will understand that Figure 10 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0088] In an exemplary embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0089] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0090] In an exemplary embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0091] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0092] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0093] The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may include, but are not limited to, general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units based on quantum computing, and the like.

[0094] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0095] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A spatial projection regularization method for sound source localization and sound field reconstruction, characterized in that: include: At the sound source to be measured, an equivalent source method model is constructed using the equivalent source method; in the equivalent source method model, a series of virtual sources inside the sound source are used to equivalently simulate the external radiation sound field of the sound source to be measured, and a microphone array is arranged in the near field of the sound source to construct a holographic measurement point plane; According to the positions and number of the virtual sources, a first sound pressure transfer matrix from the virtual source arrangement plane to the holographic measurement point plane is constructed; the first sound pressure transfer matrix is ​​used to characterize the sound pressure transfer relationship between any virtual source and any holographic measurement point on the holographic measurement point plane; A spatial projection regularization algorithm is used to solve an inverse problem based on the first sound pressure transfer matrix and the sound pressure at the holographic measurement point to obtain an equivalent source intensity; the spatial projection regularization algorithm truncates the vector space based on the projection size of the holographic measurement point sound pressure data in the vector space, and a double generalized cross-validation algorithm is used in the spatial projection regularization algorithm to automatically select an appropriate regularization parameter; The sound pressure of any point in the external radiation sound field of the sound source to be measured is solved according to the second sound pressure transfer matrix and the equivalent source intensity, and the external radiation sound field of the sound source to be measured is reconstructed; the second sound pressure transfer matrix is ​​used to characterize the sound pressure transfer relationship between any virtual source and any point in the external radiation sound field; In the reconstructed external radiation sound field, the field points in the external radiation sound field are screened based on the maximum sound pressure amplitude as a standard to determine the position of the sound source to be measured.

2. The spatial projection regularization method for sound source localization and sound field reconstruction according to claim 1, characterized in that: The series of virtual sources in the equivalent source method model are monopole sources, dipole sources or multipole source combinations; the virtual sources are located inside the sound source or in the near field to avoid the virtual source position coinciding with the external radiation sound field point.

3. The spatial projection regularization method for sound source localization and sound field reconstruction according to claim 1, characterized in that: The equivalent source method can be expressed as follows: p(r)=∫∫μ(r s' )G(r,r s' )ds'; Wherein, p(r) is the external radiation sound field of the sound source to be measured, μ(r s' ) is the virtual source strength, G(r,r s' ) is the free space Green's function, s' is the virtual source layout surface, the virtual source layout surface s' grid is discretized into D units, and the constant unit approximation is used. The above formula can be transformed into the following form: Among them, q(r s' ) is the equivalent source intensity of the virtual source, the subscript i represents the grid unit of the ith virtual source layout surface of the grid discretization, and Δs' is the area of ​​the discrete unit; Converting the above formula into matrix form is as follows: P H =G H Q; Among them, P H is the column vector form of the sound pressure at the holographic measurement point, P H =[p(r H1 ),p(r H2 ),…,p(r HM )] T , M is the number of holographic measurement points, G H is the first sound pressure transfer matrix from the virtual source layout plane to the holographic measurement point plane, Q is the column vector form of the equivalent source intensity, Q=[q(r E1 ),q(r E2 ),…,q(r EN )], N is the number of virtual sources.

4. The spatial projection regularization method for sound source localization and sound field reconstruction according to claim 1, characterized in that: The projection size of the holographic measurement point sound pressure data in the vector space is |u i T (b+ε)|, the vector space is truncated based on the projection size of the holographic measurement point sound pressure data in the vector space as shown in the following formula: |u i T (b+e)|≤t; Among them, u i is the vector space, b is the sound pressure data of the holographic measurement point in vector form, ε is the noise vector, and t is the first truncation threshold.

5. The spatial projection regularization method for sound source localization and sound field reconstruction according to claim 1, characterized in that: In the spatial projection regularization algorithm, a double generalized cross-validation algorithm is used to automatically select an appropriate regularization parameter. In the double generalized cross-validation algorithm, the formula for the first generalized cross-validation is: Among them, G1(i) is the formula of generalized cross validation, A is the first sound pressure transfer matrix, A i is a regularization matrix constructed by retaining the first i vector spaces in order according to the size of the singular values, I is the unit matrix, is the mean of the sound pressure data of the holographic measurement points in vector form, Tr() is the trace of the matrix; according to t = (i, min (G1 (i))), the i where G1 (i) is the smallest is used as the first truncation threshold, which is used to filter the vector space with singular values ​​greater than the threshold; The formula for the second generalized cross validation is: Among them, G1(i) is the new generalized cross-validation formula, ε is the noise vector, according to t sort =(i,min(G2(i))) takes i when G2(i) is the smallest as the second truncation threshold, which is used to filter out vector spaces whose projections are greater than the threshold.

6. The spatial projection regularization method for sound source localization and sound field reconstruction according to claim 5, characterized in that: The equivalent source strength is obtained according to the following formula: Among them, U c and V is the unitary matrix obtained by singular value decomposition, ∑ is the singular value matrix, X SPR is the equivalent source strength; The external radiation sound field of the sound source to be measured is obtained according to the following formula: P r =G r Q; Among them, P r is the external radiation sound field of the sound source to be measured, G r is the second sound pressure transfer matrix, Q is the equivalent source intensity X obtained in the above formula SPR .

7. A spatial projection regularization device for sound source localization and sound field reconstruction, characterized in that: include: An equivalent source method modeling module is used to construct an equivalent source method model at the sound source to be measured using the equivalent source method; in the equivalent source method model, a series of virtual sources inside the sound source are used to equivalently simulate the external radiation sound field of the sound source to be measured, and a microphone array is arranged in the near field of the sound source to construct a holographic measurement point plane; A sound pressure transfer matrix construction module is used to construct a first sound pressure transfer matrix from the virtual source arrangement plane to the holographic measurement point plane based on the positions and number of the virtual sources; the first sound pressure transfer matrix is ​​used to characterize the sound pressure transfer relationship between any virtual source and any holographic measurement point on the holographic measurement point plane; a spatial projection regularization solution module, configured to solve an inverse problem using a spatial projection regularization algorithm based on the first sound pressure transfer matrix and the holographic measurement point sound pressure to obtain an equivalent source intensity; the spatial projection regularization algorithm truncates the vector space based on the projection size of the holographic measurement point sound pressure data in the vector space, and automatically selects an appropriate regularization parameter using a double generalized cross-validation algorithm in the spatial projection regularization algorithm; a sound source external sound field reconstruction module, configured to solve the sound pressure at any point in the external radiation sound field of the sound source to be measured based on a second sound pressure transfer matrix and the equivalent source intensity, and reconstruct the external radiation sound field of the sound source to be measured; the second sound pressure transfer matrix is ​​used to characterize the sound pressure transfer relationship between any virtual source and any point in the external radiation sound field; The sound source positioning module to be measured is used to screen the field points in the reconstructed external radiation sound field based on the maximum sound pressure amplitude as a standard to determine the position of the sound source to be measured.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the spatial projection regularization method for sound source localization and sound field reconstruction according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the spatial projection regularization method for sound source localization and sound field reconstruction according to any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the spatial projection regularization method for sound source localization and sound field reconstruction according to any one of claims 1 to 6 is implemented.

Images