阅读说明：本技术 基于机器视觉和全息方法的声场测试分析方法及系统 (Sound field test analysis method and system based on machine vision and holographic method ) 是由 吴海军 李豆 蒋伟康 于 2020-07-06 设计创作，主要内容包括：一种基于机器视觉和全息方法的声场测试分析方法及系统,通过预先采集待测区域的几何形状得到区域网格信息,并结合传声器阵列位置建立阵列与待测区域组成的边界元模型；再以传声器阵列进行声压测量,通过自由场格林函数计算边界元模型中的全息面与重建面之间的阻抗矩阵并采用正则化方法反向求解重建面法向振速,得到声源的精确位置。本发明能够更方便地在噪声环境下重建不规则声源的局部表面振速。(A sound field test analysis method and system based on machine vision and holographic method, through gathering the geometric shape of the area to be measured in advance and getting the regional grid information, and combine the microphone array position to set up the boundary element model that array and area to be measured make up; and then, carrying out sound pressure measurement by using a microphone array, calculating an impedance matrix between a holographic surface and a reconstruction surface in the boundary element model by using a free field Green function, and reversely solving the normal vibration velocity of the reconstruction surface by using a regularization method to obtain the accurate position of the sound source. The invention can more conveniently reconstruct the local surface vibration velocity of the irregular sound source in a noise environment.)

1. A sound field test analysis method based on machine vision and holographic method is characterized in that area grid information is obtained by collecting the geometric shape of a region to be tested in advance, and a boundary element model composed of an array and the region to be tested is established by combining the position of a microphone array; then, sound pressure measurement is carried out by a microphone array, an impedance matrix between a holographic surface and a reconstruction surface in the boundary element model is calculated through a free field Green function, and the normal vibration velocity of the reconstruction surface is reversely solved by adopting a regularization method, so that the accurate position of a sound source is obtained;

the boundary meta-model is specifically as follows: the device comprises a rigid acoustic shielding cover, an acoustic surface reconstruction region and a closed cavity formed by gaps among the rigid acoustic shielding cover and the acoustic surface reconstruction region, wherein the grid of a measuring surface is divided based on the position of a microphone, and the grid of the gap is divided based on the grid of a sound source reconstruction region and the grid of the measuring surface;

the surface of the cavity meets a Helmholtz integral equation, which specifically comprises the following steps:

the sound pressure and the speed on the boundary of the cavity meet the following requirements:

2. The sound field test analysis method according to claim 1, wherein the reconstruction adopts a BundleFusion real-time global three-dimensional reconstruction method, the input is RGB-D data collected by a camera, each 11 frames constitute a data block, when the acquired frames are accumulated into a data block, feature matching and pose optimization are performed in the data block, the first frame in the data block represents the data block, the features of all the frames are combined together to represent the features in the data block, the data block is matched with all the previous data blocks, and the final reconstruction result is represented by a grid and includes node and unit information.

3. The sound field test analysis method of claim 1, wherein the free field Green's functionWherein: i is the imaginary unit, k is the wavenumber, and k ═ ω/c, ω ═ 2 π f is the circle frequency, f is the analysis frequency, r is the distance between the Q and Q' points.

4. The sound field testing and analyzing method of claim 1, wherein the impedance matrix Z ═ ik ρ cH^-1G, wherein:ρ is the air density, c is the speed of sound,_QQ′is a dirac function, G_QQ′And H_QQ′Is the transfer coefficient between the points Q and Q', G and H are the matrix of coefficients between all points of the cavity surface, H^-1Is the generalized inverse of the coefficient matrix H.

5. A sound field test analysis system for implementing the method of any preceding claim, comprising: consecutive camera, sound source reconstruction module, data acquisition module and microphone array, wherein: the method comprises the steps that a camera is connected with a sound source reconstruction module and transmits grid information of an area to be reconstructed and microphone array position information, a microphone array is arranged in a rigid acoustic shielding cover and is connected with a data acquisition module to transmit acquired sound pressure information, the acoustic shielding cover vertically projects towards a sound source surface along the normal direction of the acoustic shielding cover, the projection area is a sound source reconstruction area, the data acquisition module transmits acquired sound pressure to the sound source reconstruction module, the sound source reconstruction module integrates the grid information of the area to be reconstructed and the microphone array position, then a boundary element model is constructed and an impedance matrix is calculated, Fourier change is carried out on the acquired sound pressure to obtain frequency domain sound pressure, and finally the normal vibration velocity of the reconstruction surface is reversely solved to obtain the accurate position of.

6. The sound field testing and analyzing system of claim 5, wherein the rigid acoustic shielding case is a planar structure and is discretely provided with a plurality of openings, the number of the openings is consistent with the number of the microphones, and each opening corresponds to each microphone one by one;

the microphone array comprises a plurality of microphones, each microphone is specifically embedded in a corresponding opening of the rigid acoustic shielding case, and the microphones and the shielding case are arranged flush with the surface of one side close to the sound source.

7. The sound field test analysis system of claim 5, wherein the rigid acoustic shield and the microphone array are secured and supported by a base that includes a lift mechanism for adjusting the vertical height of the rigid acoustic shield and the microphone array.

8. The sound field test analysis system of claim 5, wherein the sound source reconstruction module comprises: boundary element modeling unit, impedance matrix calculation unit, acoustic pressure data processing unit and sound source rebuild unit, wherein: the boundary element modeling unit is connected with the impedance matrix calculation and transmits boundary element model information, the impedance matrix calculation unit is connected with the sound source reconstruction unit and transmits impedance matrix information, and the sound pressure data processing unit is connected with the sound source reconstruction unit and transmits frequency domain sound pressure information.

Technical Field

The invention relates to a technology in the field of noise processing, in particular to a sound field test analysis method and a sound field test analysis system based on machine vision and a holographic method.

Background

The acoustic holography method under the environment of the non-free sound field can identify the noise source at the working site of the mechanical equipment, which is necessary for the mechanical equipment which is large or cannot move into the anechoic chamber. In the prior art, a sound source is reconstructed by an inverse block transfer function method based on a rigid acoustic array, and the surface vibration velocity of the sound source can be reconstructed by measuring a sound pressure value and calculating an impedance matrix according to a modal superposition method without measuring a particle velocity on a holographic surface or using a double-layer sound pressure measuring surface. However, when the technology is used for reconstructing the local vibration velocity of the irregular sound source surface, the application of the method in engineering practice is limited because the local shape of the sound source surface is difficult to obtain.

Disclosure of Invention

Aiming at the defects in the prior art, the invention provides a sound field test analysis method and system based on machine vision and a holographic method, which can more conveniently reconstruct the local surface vibration velocity of an irregular sound source in a noise environment.

The invention is realized by the following technical scheme:

the invention relates to a sound field test analysis method based on machine vision and a holographic method, which is characterized in that area grid information is obtained by acquiring the geometric shape of a region to be tested in advance, and a boundary element model consisting of an array and the region to be tested is established by combining the position of a microphone array; and then, carrying out sound pressure measurement by using a microphone array, calculating an impedance matrix between a holographic surface and a reconstruction surface in the boundary element model by using a free field Green function, and reversely solving the normal vibration velocity of the reconstruction surface by using a regularization method to obtain the accurate position of the sound source.

The area grid information is obtained by reconstruction through scanning geometric shapes of the camera at the pre-judged position of the sound source surface to be reconstructed.

The reconstruction adopts but is not limited to a Bundlefusion real-time global three-dimensional reconstruction method, and RGB-D data collected by a camera is input. And (2) forming a data block by every 11 frames, when the acquired frames are accumulated into a data block, firstly performing feature matching and pose optimization in the data block, representing the data block by using a first frame in the data block, combining the features of all the frames to represent the features in the data block, and then matching the data block with all the previous data blocks. The final reconstruction result is represented as a grid, including node and cell information.

The boundary meta-model is specifically as follows: the device comprises a rigid acoustic shielding cover, an acoustic surface reconstruction region and a closed cavity formed by gaps among the rigid acoustic shielding cover and the acoustic surface reconstruction region, wherein the grid of a measuring surface is divided based on the position of a microphone, and the grid of the gap is divided based on the grid of a sound source reconstruction region and the grid of the measuring surface.

The surface of the cavity meets a Helmholtz integral equation, which specifically comprises the following steps: wherein: s_mFor measuring the surface, S_vThe acoustic source surface and the gap between the measurement surface and the acoustic source surface are included, Q is any point of the cavity surface, p (Q) is the sound pressure at the point Q, Q ' is a point on the cavity surface, G (Q, Q ') is a free field Green function, n is the normal direction of the cavity surface at the point Q ' and points to the outer side of the cavity, and p (Q ') is the sound pressure at the point Q '.

The sound pressure and the speed on the boundary of the cavity meet the following requirements:wherein: impedance matrix

p_mFor measuring sound pressure, p, in the plane_vIs the sound pressure, z, at the sound source plane and at the gap_llTo measure the self-impedance of the surface, z_klIs the mutual impedance between the sound pressure at the sound source plane and the gap and the sound pressure at the measurement plane, z_lkTo measure the mutual impedance between the velocity of the plane and the sound pressure at the sound source plane and gap, z_kkIs the self-impedance of the sound source plane and gap, v_mTo measure the velocity over the surface, i.e. zero, v_vThe normal speed of the sound source surface and the gap is as follows: v. of_v＝z_lk ^-1p_m。

The free field Green function

Wherein: i is an imaginary unit, k is a wave number, and k ═ ω/c, ω ═ 2 π f is the circular frequencyThe ratio, f, the analysis frequency, and r, the distance between the points Q and Q'.

The impedance matrix Z is ik rho cH^-1G, wherein:

ρ is the air density, c is the speed of sound,_QQ′is a dirac function, G_QQ′And H_QQ′Is the transfer coefficient between the points Q and Q', G and H are the matrix of coefficients between all points of the cavity surface, H^-1Is the generalized inverse of the coefficient matrix H.

The invention relates to a sound field test analysis system for realizing the method, which comprises the following steps: consecutive camera, sound source reconstruction module, data acquisition module and microphone array, wherein: the method comprises the steps that a camera is connected with a sound source reconstruction module and transmits grid information of an area to be reconstructed and microphone array position information, a microphone array is arranged in a rigid acoustic shielding cover and is connected with a data acquisition module to transmit acquired sound pressure information, the acoustic shielding cover vertically projects towards a sound source surface along the normal direction of the acoustic shielding cover, the projection area is a sound source reconstruction area, the data acquisition module transmits acquired sound pressure to the sound source reconstruction module, the sound source reconstruction module integrates the grid information of the area to be reconstructed and the microphone array position, then a boundary element model is constructed and an impedance matrix is calculated, Fourier change is carried out on the acquired sound pressure to obtain frequency domain sound pressure, and finally the normal vibration velocity of the reconstruction surface is reversely solved to obtain the accurate position of.

The rigid acoustic shielding cover is of a plane structure, a plurality of openings are discretely formed in the plane, the number of the openings is consistent with that of the microphones, and the openings correspond to the microphones one by one.

The microphone array comprises a plurality of microphones, each microphone is specifically embedded in a corresponding opening of the rigid acoustic shielding case, and the microphones and the shielding case are arranged flush with the surface of one side close to the sound source.

The holes are arranged in a row a, and each row is b.

The rigid acoustic shield and microphone array are preferably secured to and supported by a base that includes a lift mechanism for adjusting the vertical height of the rigid acoustic shield and microphone array.

The sound source reconstruction module comprises: boundary element modeling unit, impedance matrix calculation unit, acoustic pressure data processing unit and sound source rebuild unit, wherein: the boundary element modeling unit is connected with the impedance matrix calculation and transmits boundary element model information, the impedance matrix calculation unit is connected with the sound source reconstruction unit and transmits impedance matrix information, and the sound pressure data processing unit is connected with the sound source reconstruction unit and transmits frequency domain sound pressure information.

Technical effects

The invention integrally solves the problem that the prior art can not realize the local vibration velocity reconstruction of the irregular-shaped sound source; the method is based on a machine vision method to reconstruct the geometric shape of the sound source surface and obtain the position of a rigid acoustic array, so that a boundary element model of a closed cavity formed by a rigid acoustic shielding cover, an acoustic surface reconstruction region and gaps among the rigid acoustic shielding cover and the acoustic surface reconstruction region is established and used for calculating an impedance matrix. Further by optimizing the construction of the rigid acoustic shield. By adopting the structure of the plane rigid acoustic shielding cover, the distance between the sound source plane and the microphone array is shortened, on one hand, evanescent waves with higher orders can be collected, the reconstruction resolution is improved, on the other hand, the signal-to-noise ratio can be improved to a certain extent, and therefore the reconstruction error is reduced.

Drawings

FIG. 1 is a schematic diagram of a testing system according to the present invention

FIG. 2 is an isometric view of a rigid acoustic shield;

FIG. 3 is a schematic diagram illustrating the positions of a sound source plane and a testing device in an embodiment;

FIG. 4 shows the reconstruction result in the embodiment; wherein: 1. the device comprises a camera, 2, a microphone array, 3, a rigid acoustic shielding case, 4, a base, 5, a data acquisition module, 6 and a sound source reconstruction module.

Detailed Description

As shown in fig. 1 and fig. 2, the present embodiment relates to a sound field test analysis system based on a machine vision method and a holographic method in a noisy environment, which includes: the device comprises a camera 1, a microphone array 2, a rigid acoustic shielding cover 3, a base 4, a data acquisition module 5 and a sound source reconstruction module 6 directly connected with the camera 1.

The microphone array 2 comprises a plurality of microphones, the rigid acoustic shielding case 3 is a plane, a plurality of openings are discretely formed in the plane, the number of the openings is consistent with that of the microphones, the openings correspond to the microphones one by one, the microphones are embedded in the corresponding openings of the rigid acoustic shielding case 3, the microphones are flush with the inner surface of the rigid acoustic shielding case 3, and the normal vibration speed of a sound pressure collecting point is guaranteed to be zero.

The holes are preferably a rows and b holes in each row.

The microphone array 2, the data acquisition module 5 and the sound source reconstruction module 6 are sequentially connected, and the rigid acoustic shielding case 3 and the microphone array 2 are fixed and supported by the base 4.

The base comprises a lifting mechanism for adjusting the height of the rigid acoustic shielding case and the microphone array in the vertical direction.

And vertically projecting towards the sound source surface along the normal direction of the rigid acoustic shielding cover 3, wherein the projection area is a sound source reconstruction area, and a certain gap is formed between the acoustic shielding cover and the sound source surface.

The embodiment relates to a detection method of the system, which comprises the following steps:

step S1: as shown in fig. 3, the predicted position of the sound source plane to be reconstructed is determined, the Kinect V2 camera is used to scan near the reconstruction region, the geometry of the scanning region is reconstructed, and the mesh information of the scanning region is obtained.

The three-dimensional reconstruction method adopts a Bundlefusion algorithm, and the algorithm is a real-time global three-dimensional reconstruction method. The basic idea of the BundleFusion algorithm is a local to global pose optimization strategy. The input to the algorithm is the RGB-D data acquired by the camera. And (2) forming a data block by every 11 frames, when the acquired frames are accumulated into a data block, firstly performing feature matching and pose optimization in the data block, representing the data block by using a first frame in the data block, combining the features of all the frames to represent the features in the data block, and then matching the data block with all the previous data blocks. The final reconstruction result is represented by a triangular mesh, which includes node and cell information.

Step S2: and adjusting the position of the base to ensure that the microphone array and the rigid acoustic shielding cover are just opposite to the scanning area for sound pressure measurement. And position information of four vertices of the microphone array is obtained using camera scanning.

The rigid acoustic shielding case in this embodiment is an aluminum plate with a thickness of 10mm, a length and a width of 300mm and 400mm, and holes are uniformly formed in the aluminum plate for 8 rows, 6 columns are formed in each row, and the diameter of each hole is the same as that of the microphone. Using camera scanning, position information of four vertices of the microphone array is obtained.

Step S3: and projecting to a sound source plane based on the four vertexes, wherein the projection area is a sound source reconstruction area.

Step S4: and carrying out grid optimization on the sound source reconstruction area to improve the grid quality.

Step S5: and establishing a boundary meta-model of a closed cavity consisting of the rigid acoustic shielding cover, the sound source surface reconstruction region and a gap between the rigid acoustic shielding cover and the sound source surface reconstruction region, specifically, dividing the mesh of the measuring surface based on the position of the microphone when establishing the boundary meta-model of the cavity, and obtaining the mesh of the sound source surface based on a three-dimensional reconstruction result.

The Helmholtz integral equation satisfied by the cavity surface is as follows:

Q,Q′∈S_m∪S_vwherein: s_mFor measuring the surface, S_vThe acoustic source surface and the gap between the measurement surface and the acoustic source surface are included, Q is any point of the cavity surface, p (Q) is the sound pressure at the point Q, Q ' is a point on the cavity surface, G (Q, Q ') is a free field Green function, n is the normal direction of the cavity surface at the point Q ' and points to the outer side of the cavity, and p (Q ') is the sound pressure at the point Q '.

The green function is specifically as follows:

wherein: i is the imaginary unit, k is the wavenumber, and k ═ ω/c, ω ═ 2 π f is the circle frequency, f is the analysis frequency, r is the distance between the Q and Q' points.

The impedance matrix Z is ik rho cH-G

Wherein: ρ is the air density, c is the speed of sound,_QQ′is a dirac function, G_QQ′And H_QQ′Is the transfer coefficient between the points Q and Q', G and H are the matrix of coefficients between all points of the cavity surface, H^-1Is the generalized inverse of the coefficient matrix H;

the relationship between the sound pressure and the velocity at the boundary of the cavity is:wherein:

p_mfor measuring sound pressure, p, in the plane_vIs the sound pressure, z, at the sound source plane and at the gap_llTo measure the self-impedance of the surface, z_klIs the mutual impedance between the sound pressure at the sound source plane and the gap and the sound pressure at the measurement plane, z_lkTo measure the mutual impedance between the velocity of the plane and the sound pressure at the sound source plane and gap, z_kkIs the self-impedance of the sound source plane and gap, v_mTo measure the velocity over the surface, i.e. zero, v_vThe normal speed of the sound source surface and the gap is as follows: v. of_v＝z_lk ^-1p_m

Step S5: and calculating an impedance matrix between the holographic surface and the reconstruction surface based on the free field Green function.

Step S6: and reversely solving the normal vibration speed of the reconstruction surface by adopting a regularization method.

The reconstruction result of the embodiment at 322Hz is shown in FIG. 4, the signal-to-noise ratio is-11 dB, the error is 11% when the reconstruction result is compared with the result of the acceleration sensor, and the sound source distribution can be reconstructed more accurately.

Signal to noise ratio

Wherein p is_siDenotes the measured sound pressure without interference source, p_niThe sound pressure generated at the measurement location for a coherent noise source.

Error of the measurementWherein v is_vFor the reconstructed surface normal vibration velocity, v, of the sound source_eThe normal vibration speed of the surface of the sound source is obtained for testing.

Through specific practical experiments, the experimental result of the surface vibration velocity reconstruction of the fan shell shows that the surface vibration velocity reconstruction error is 11% when the signal-to-noise ratio is-11 dB under 322 Hz.

Compared with the prior art, the invention adopts the structure of the plane rigid acoustic shielding cover, shortens the distance between the sound source plane and the microphone array, can collect evanescent waves with higher orders on one hand and improve the reconstruction resolution ratio on the other hand, and can improve the signal-to-noise ratio to a certain extent on the other hand, thereby reducing the reconstruction error.

The foregoing embodiments may be modified in many different ways by those skilled in the art without departing from the spirit and scope of the invention, which is defined by the appended claims and all changes that come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.