Acoustic field reconstruction method based on equivalent source activation theory under distributed load conditions

Abstract

In order to solve the technical problem that the traditional sound field reconstruction technology based on inverse wave modeling theory is unable to accurately restore the distributed load due to the restoration of simple point sound sources as the main link, thus having poor accuracy, the present invention provides a sound field reconstruction method based on equivalent source activation theory under distributed load conditions. On the basis of the sound field reconstruction technology of inverse wave modeling in closed space, a series of equivalent sources are introduced into the closed space. Through limited sampling information, calculations are performed under the current distributed load conditions to activate some equivalent sources, and equivalent loads are assigned to these activated equivalent sources, thereby realizing the reconstruction of the sound field in the closed space under the most economical calculation conditions.

Description

Technical Field

[0001] The invention relates to a sound field reconstruction method based on equivalent source activation theory under distributed load conditions. Background Art

[0002] Enclosed space sound field reconstruction involves the process of predicting and reconstructing the sound field outside the sampling point using limited sampling and a sound field transformation algorithm. This technique is often used to analyze the sound field and provide a basis for achieving optimal noise control and acoustic design results in spatial noise control and acoustic design. Due to the inherent diversity of sound fields, the target of sound field reconstruction varies. However, the overall development trend is from simple single-source sound field reconstruction to complex sound field reconstruction, and from steady-state sound field reconstruction to unsteady sound field reconstruction.

[0003] Acoustic holography is a typical approach for existing sound field reconstruction algorithms. After years of development, near-field acoustic holography has become relatively mature and widely used. However, because this method is developed within a free-field framework, it is no longer applicable to enclosed spaces with acoustic reverberation.

[0004] Several methods have been proposed in recent years to address the problem of reconstructing the sound field in enclosed spaces. For example, the plane wave expansion method represents the sound pressure at a point in space as the sum of the products of a series of plane waves and their coefficients. The coefficients are then recovered using limited sampling, allowing the reconstruction of the sound field at non-sampled points. However, because this method only utilizes plane wave information near the sampling point, it can only accurately reconstruct the sound field near the sampling point. Large errors occur when the reconstruction point is far from the sampling point. A recently developed method for reconstructing the sound field in enclosed spaces is the sound field reconstruction method based on inverse wave modeling theory. This method uses wave modeling to establish a transfer function model between any two points in the space. Using the sampling information, it can reconstruct the global sound field. Because this method exploits the wave properties of sound, it can still maintain high reconstruction accuracy when the reconstruction point is far from the sampling point. However, this method is only applicable to situations where point sound sources are present. In practical engineering applications, enclosed spaces often experience distributed loads such as wall panel vibration and large-scale acoustic disturbances, resulting in poor accuracy under such conditions. Summary of the Invention

[0005] In actual scenarios, the noise sources in closed spaces often have more complex forms. For example, an aircraft cabin will be affected by boundary layer turbulence and wall panel vibration to form an internal sound field. These loads are usually distributed. The traditional sound field reconstruction technology based on inverse wave modeling theory is unable to accurately restore the distributed loads because it takes the recovery of simple point sound sources as the main link, resulting in poor accuracy. In response to this problem, the present invention provides a sound field reconstruction method based on equivalent source activation theory under distributed load conditions. On the basis of the closed space inverse wave modeling sound field reconstruction technology, a series of equivalent sources are introduced into the closed space. Through limited sampling information, calculations are performed under the current distributed load conditions to activate some equivalent sources, and equivalent loads are assigned to these activated equivalent sources, thereby realizing the reconstruction of the sound field in the closed space under the most economical calculation conditions.

[0006] The technical solution of the present invention is:

[0007] The sound field reconstruction method based on equivalent source activation theory under distributed load conditions is special in that it includes the following steps:

[0008] Step 1: Arrange n nodes, q acoustic signal acquisition devices, and m equivalent sources in a closed space of the sound field to be reconstructed, and discretize the closed space;

[0009] Step 2: Constructing shape functions corresponding to the n nodes and the q acoustic signal acquisition devices respectively;

[0010] Step 3: Based on the shape function obtained in step 2, establish the inherent properties of the closed space, including the stiffness matrix K, the mass matrix M, and the damping matrix C;

[0011] Step 4: Based on the inherent properties of the enclosed space obtained in step 3, establish the system integral equation of the meshless acoustic field calculation method and construct the control matrix D of the enclosed space wave model;

[0012] Step 5: Based on the closed space wave model control matrix D constructed in step 4, construct the equivalent source reconstruction equation of the distributed load acoustic field:

[0013] Step 6: Solve the equivalent source reconstruction equation constructed in step 5 to solve the acoustic load vector F;

[0014] Step 7: Substitute the acoustic load vector F into the system integral equation established in step 4, derive the sound pressure values ​​distributed on the nodes in the entire enclosed space, and then multiply it by the corresponding node shape function to obtain the sound field distribution information in the entire enclosed space to achieve sound field reconstruction.

[0015] Furthermore, in step 1:

[0016] The layout principle of nodes is: evenly distributed, and they need to be arranged at the boundaries of closed spaces; when there are objects inside the closed space, they need to be arranged on the boundaries of the objects, and the nodes on the boundaries of the objects can outline the shape of the objects.

[0017] Furthermore, the m equivalent sources are a series of point sound sources with fixed positions, or a series of point sound sources with variable positions; when the m equivalent sources are a series of point sound sources with fixed positions, their layout principle is: uniform setting and one-to-one overlap with the n nodes.

[0018] Furthermore, in step 2, a spatially global and sparse shape function is constructed based on the moving least squares method.

[0019] Furthermore, in step 3:

[0020] Stiffness matrix

[0021] Mass Matrix

[0022] Damping Matrix

[0023] Where: N is the shape function; is the derivative matrix of the shape function; v is the medium particle velocity; c0 represents the speed of sound; ζ is called the specific acoustic impedance; s is the area of ​​the rigid wall; Γ is the boundary of the closed space and the objects inside it, and Ω is the fluid area surrounded by the boundary Γ.

[0024] Furthermore, in step 4:

[0025] The system integral equation of the meshless acoustic field calculation method is:

[0026] (K+jωC-ω 2 M)p=F

[0027] Where: p represents the sound pressure value vector at each node, and its relationship with the sound signal collected at the collection point is p s =p·N s ;p s N is the observed sound signal vector collected by a series of sound signal collection devices set up in the closed space; s is the shape function corresponding to the acquisition point; ω is the resonant frequency; F is the acoustic load vector;

[0028] The control matrix D of the closed space wave model is:

[0029] D=-jρ0ωN s (K+jkC-k 2 M) -1

[0030] Where: j is an imaginary number, ρ0 represents the static density of the fluid in the closed space, ω represents the frequency of the sound wave in the closed space, and k represents the wave number.

[0031] Furthermore, in step 5:

[0032] The equivalent source reconstruction equation of the distributed load acoustic field is:

[0033]

[0034] Where: Q1, Q2, ..., Q m are the equivalent source strengths respectively; N1, N2...N m are the shape functions corresponding to each equivalent source.

[0035] Furthermore, when the m equivalent sources are a series of point sound sources with fixed positions, a sparse recovery algorithm is used in step 6 to solve the equivalent source reconstruction equation; when the m equivalent sources are a series of point sound sources with variable positions, a sparse recovery algorithm and a dictionary learning algorithm are used in step 6 to solve the equivalent source reconstruction equation.

[0036] The present invention also provides an electronic device, comprising a processor and a storage medium; the storage medium stores a computer program; the electronic device is special in that the computer program executes the above method when executed by the processor.

[0037] The present invention also provides a non-volatile storage medium on which a computer program is stored; the special feature of the non-volatile storage medium is that the computer program executes the above method when executed by a processor.

[0038] The beneficial effects of the present invention are:

[0039] 1. The sound field reconstruction method proposed in the present invention utilizes sound field wave modeling technology and takes inverse wave modeling as the basic algorithm. It fully utilizes the inherent properties of the closed space sound field environment to construct a closed space wave model control matrix, and takes into account its impact on the sound field, and imposes corresponding constraints on the reconstruction process. Therefore, only partial sampling is required in the space to achieve accurate reconstruction of the global sound field of the closed space. Combined with the idea of ​​equivalent sources, a series of equivalent sources are set in the closed space to simulate equivalent complex sound field loads. Through local acoustic sampling, it is calculated which of the preset equivalent sources can be activated, and the load recovery is realized on these activated equivalent sources, thereby realizing stable sound field reconstruction of the closed space under distributed load conditions, effectively avoiding the problems of low accuracy and consumption of computing resources in existing sound field reconstruction methods.

[0040] 2. The sound field reconstruction method proposed in the present invention, due to the introduction of the idea of ​​equivalent source, can be applied to more common distributed complex sound source situations. After method research and simulation testing, it has been verified that it has high accuracy and good optimization effect. It has important theoretical significance in the reconstruction of the sound field of closed spaces with distributed loads and makes its application range wider, and has important practical engineering application value.

[0041] 3. The present invention can achieve global, stable and accurate reconstruction of the complex acoustic environment of various enclosed spaces such as cabins and rooms, and can provide decision-making information on the spatial and temporal distribution of the sound field for spatial noise control, especially low-frequency active noise control design.

[0042] 4. The present invention can economically and efficiently realize real-time monitoring of the spatial acoustic environment, providing technical support for advanced applications such as abnormal acoustic event prediction and real-time sound quality assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 This is the method flow chart of the present invention Figure 1 (When the point sound source position is fixed).

[0044] Figure 2 It is a schematic diagram of the principle that the distributed load sound field is composed of equivalent superposition of some equivalent sources (when the position of the point sound source is fixed).

[0045] Figure 3 It is a schematic diagram of the geometric simulation model, sound source and microphone.

[0046] Figure 4 It is a schematic diagram of the nodes of the geometric simulation model.

[0047] Figure 5 and Figure 6 They are the sound field reconstruction effects at two exemplary frequencies when the position of the point sound source is fixed. The left column is the reconstructed sound field, and the right column is the real sound field.

[0048] Figure 7 This is the method flow chart of the present invention Figure 2 (When the position of the point sound source is not fixed).

[0049] Figure 8 It is a schematic diagram of the principle that the distributed load sound field is composed of equivalent superposition of some equivalent sources (when the position of the point sound source is not fixed).

[0050] Figure 9 and Figure 10 They are the sound field reconstruction effects at two exemplary frequencies when the position of the point sound source is not fixed. The left column is the reconstructed sound field, and the right column is the real sound field. DETAILED DESCRIPTION

[0051] The present invention will be further described below with reference to the accompanying drawings.

[0052] The sound field reconstruction method of the present invention is mainly divided into two stages:

[0053] The first stage is the modeling stage, which mainly involves establishing a numerical model for a certain closed space and obtaining a system control matrix that contains the closed space's own properties such as impedance and mass; setting a series of nodes and equivalent sources evenly distributed in the space (which can be a series of point sound sources with fixed positions or a series of point sound sources with variable positions) and solving their corresponding shape functions to form a shape function matrix; finally, splicing and combining the above matrices into the corresponding sound field reconstruction equation.

[0054] The second stage is the reconstruction stage, which is mainly based on the sound field reconstruction equation formed in the first stage. It is solved to obtain the equivalent point sound source under distributed load conditions and achieve global sound field reconstruction.

[0055] The specific implementation steps of the present invention are described in detail below:

[0056] Step 1: Discretize the closed space

[0057] According to specific requirements, n nodes, q sound signal acquisition devices (microphones or sensors, etc., used to obtain the sound intensity value at a certain point) and m equivalent sources are set in the closed space of the sound field to be reconstructed;

[0058] ①Node layout principles:

[0059] The boundary of the closed space and the objects inside it is represented by Γ, and the fluid area enclosed by the boundary Γ is represented by Ω. The n nodes are approximately evenly distributed in the entire fluid area Ω, and nodes must be arranged at the boundary of the closed space. When there is an object inside the closed space, no nodes are arranged inside the object, but nodes should be arranged on the boundary of the object. The nodes on the boundary of the object should be able to outline the shape of the object.

[0060] ②Layout principles of acoustic signal acquisition equipment:

[0061] There are no special requirements for the layout of the q sound signal collection devices, as long as they meet the requirements of multi-point limited sampling, forming a total of q sound signal receiving points.

[0062] ③Layout principles for m fixed-position equivalent sources:

[0063] The m equivalent sources are evenly distributed in the entire fluid region Ω. In order to facilitate subsequent calculations and save computing resources, the s equivalent sources are coincident with the positions of the n nodes.

[0064] As attached Figure 3In the simulation example shown, the enclosed space is a box with a length, width, and height of 1m, 0.9m, and 0.6m respectively. The sound source is an iron plate excited by a distributed load, which is attached to one surface of the enclosed space. The sound signal acquisition device is a microphone, and the microphone array is a 3×6×5 cubic array with a total of 90 receiving points in the space. A node model containing the geometric information of the enclosed space is established, and the nodes are evenly distributed, with a total of 315 nodes set, as shown in the following figure. Figure 4 As shown; the schematic diagram of the enclosed space, sound source and sampling points is attached Figure 3 .

[0065] Step 2: Construct the shape function corresponding to each node and each acoustic signal acquisition device in step 1

[0066] The shape function is the only parameter for constructing system matrices such as the stiffness matrix, mass matrix, and damping matrix. It is also the main parameter for constructing the acoustic load. Therefore, its performance will affect the overall reconstruction performance.

[0067] The present invention constructs a spatially global, sparse shape function based on the moving least squares method. The moving least squares method can apply lower-order basis functions to obtain shape functions with higher continuity and compatibility. In this method, the approximate value of a field function u(x) at a point can be expressed as:

[0068]

[0069] in:

[0070] is the coordinates of each node in the neighborhood of the calculation point x;

[0071] is the basis function vector;

[0072] m is the number of basis functions;

[0073] a(x)=[a1(x),a2(x),...a m (x)] is the unknown coefficient vector.

[0074] Monomial basis functions can usually be used for calculations. The commonly used linear and quadratic monomial basis functions in three-dimensional space are:

[0075]

[0076] After discretizing the solution domain with nodes, a weight function w is defined at each node. I (xx I), the function is non-zero only within a finite region (the support domain) and is zero outside the region. In three dimensions, the support domain of the weight function is usually spherical. Commonly used weight functions include Gaussian functions and spline functions. After selecting the weight function, the weighted sum of squares of the error at the node of the approximate function can be calculated:

[0077]

[0078] Where n is the number of nodes, x I is a node, u I For node x I The field value at p i (x I ) is the node x I The basis function, w I (xx I ) is the node x I The weight function at x I The support domain is greater than zero, and it is zero outside the support domain. Let J take the minimum value, that is:

[0079]

[0080] After sorting, we can get the following formula:

[0081] A(x)a(x)=B(x)u (5)

[0082] Where A(x) and B(x) mean:

[0083]

[0084] B(x)=[w1(xx I )p(x1),w2(xx I )p(x2),…,w N (xx I )p(x n )]

[0085] From formula (5), we can get a(x), which can be substituted into formula (1) to obtain:

[0086] u h (x) = p T (x)A -1 (x)B(x)u=N(x) T u (7)

[0087] Where, N(x)=[N1(x),N2(x),...,N n (x)] T , which is the shape function vector; u is the node x I The field value u atI The vectorized result represents the contribution of a node to the field value of all other nodes. Since it is determined only by the geometry of the enclosed space, the node positions, and the sound source positions, it is often called a shape function.

[0088] Since the weight function w is constructed in the above process I When the weight function is defined within a support domain, the weight function has a valid value within the support domain, but has no valid value outside the support domain. By relying on this method, the value of the shape function can be mainly limited to the support domain space. In this way, among all the nodes, only a small number of nodes have the main contribution to the target field point, thus forming a shape function with sparse properties.

[0089] That is, this step uses the moving least squares method to establish a sparse shape function that can be applied to the reconstruction equation, making the reconstruction equation data sparse, which is convenient for subsequent sparse solution.

[0090] Step 3: Establish the inherent properties of the enclosed space

[0091] Based on the shape function N(x) constructed in step 2, the inherent properties of the closed space are established, including the stiffness matrix K, the mass matrix M, and the damping matrix C;

[0092] Assuming that the position of the sound source in the closed space is r, it provides a medium mass of ρ0q(r,t) to the space within the unit volume per unit time. According to the law of conservation of mass, the continuity equation of the sound wave in the medium can be written as:

[0093]

[0094] Where ρ' is the density increment of the medium, ρ0 represents the static density of the medium, q is the abbreviation of the sound source intensity q(r,t), v is the velocity of the medium particle, t represents time, and div is the divergence operator. In the three-dimensional Cartesian coordinate system,

[0095] In addition to the continuity equation, there are two other basic equations used to describe sound waves in a medium. They are not affected by the sound source and are the equations of motion:

[0096]

[0097] and the equation of state:

[0098] p=c0 2 ρ' (10)

[0099] In the above two equations, p represents the sound pressure, c0 represents the sound speed, and grad is the gradient operator. In the three-dimensional Cartesian coordinate system,

[0100] Using a method similar to that used to derive the passive wave equation, the three basic equations for sound waves in a medium can be used to derive the wave equation for the sound pressure p in a closed space in the active case:

[0101]

[0102] Where, is the Laplace operator, in the three-dimensional Cartesian coordinate system,

[0103] In the frequency domain, the sound source intensity q(r,t) can be expressed as:

[0104] q(r,t)=q ω (r)e jωt (12)

[0105] Where ω is the resonant frequency, q ω (r) is the sound source intensity in the frequency domain at position r.

[0106] Since the sound field in a closed space is usually considered as a linear system, the frequency of the sound pressure at each point in the space is the same as that of the sound source, and the sound pressure can be expressed as:

[0107] p(r,t)=p ω (r)e jωt (13)

[0108] Where p ω (r) is the sound pressure in the frequency domain at position r.

[0109] Substituting equations (12) and (13) into equation (11), the acoustic wave equation under the excitation of a simple harmonic sound source can be obtained as follows:

[0110]

[0111] Let k = ω / c0, which is called the wave number, and eliminate e jωt , we can get the part of the equation that only depends on the spatial coordinates, that is, the indoor active Helmholtz equation:

[0112]

[0113] In this way, the time domain problem of sound pressure is converted into a frequency domain problem, and Equation (15) is the governing equation of the sound field in a closed space.

[0114] In a closed space, the boundary usually has a certain sound absorption capacity, and its sound pressure gradient can be expressed as:

[0115]

[0116] Where n is the normal direction of the wall surface of the closed space, ζ is called the specific acoustic impedance, which satisfies the following formula:

[0117]

[0118] Where Z is the interface acoustic impedance, which can be obtained by looking up the impedance table of various materials.

[0119] According to the Galerkin-type weighted residual method, in order to solve Equation (15), we first set a test function as Substituting the active Helmholtz equation and its boundary conditions, since the trial function is usually not an exact solution, the residual R and

[0120]

[0121]

[0122] According to the Galerkin method, we can determine the weight function:

[0123]

[0124] From Green's first formula:

[0125]

[0126] Formula (20) can be simplified to

[0127]

[0128] The sound pressure at any point in the sound field can be expressed by the sound pressure at each node, that is,

[0129]

[0130] Where N is an n*1 vector, which is an interpolation function and represents the influence of all nodes in the space on the field value of a certain spatial position; N i is the shape function at node i, p i is the sound pressure at node i.

[0131] Substituting formula (23) into formula (22), we can get

[0132]

[0133] Where ▽N is the derivative matrix of the shape function, and its expression is:

[0134]

[0135] Arranging formula (24), we can get

[0136]

[0137] make

[0138]

[0139]

[0140]

[0141] In formulas (27), (28), and (29), K is called the stiffness matrix, M is called the mass matrix, C is called the damping matrix, s is the rigid wall area, v is the medium particle velocity, c0 represents the speed of sound, ζ is called the specific acoustic impedance, and N is the shape function N(x) obtained in step 2. is the derivative matrix of the shape function, Γ is the boundary of the closed space and the objects inside it, and Ω is the fluid area surrounded by the boundary Γ. From these formulas, it can be seen that the shape function is the most important parameter in the process of constructing the parameter matrices related to the system equations such as the acoustic field stiffness matrix, mass matrix, and damping matrix, and has a great influence on the overall sound field reconstruction.

[0142] Step 4: Using the inherent properties of the closed space obtained in step 3 (stiffness matrix K, mass matrix M and damping matrix C), establish the system integral equation of the meshless acoustic field calculation method and construct the closed space wave model control matrix D

[0143] In formula (26), let

[0144] ∫ Ω jρ0ωN T q ω dv=F (30)

[0145] F is called the acoustic load vector. When the sound source is a simple harmonic point sound source located at a specific position r0 (x0, y0, z0), the sound source intensity in the frequency domain can be expressed as:

[0146] q ω (r) = q ω δ(r-r0) (31)

[0147] in

[0148]

[0149] Substituting formula (31) into formula (30), we can get

[0150]

[0151] Let Q0 = q ω (r0)dv, then we can get

[0152]

[0153] Finally, substituting equations (27), (28), (29), and (30) into equation (26) and sorting them out, we can obtain

[0154] (K+jωC-ω 2 M)p=F (35) This formula is the system integral equation of the meshless sound field calculation method. In the formula, p represents the sound pressure value vector at each node, and its relationship with the sound signal collected at the collection point is p s =p·N s , N s is the shape function corresponding to the collection point, which can be obtained in advance.

[0155] Finally, in order to obtain the reconstruction equation p s =DF form, combining all known quantities together to obtain the closed space wave model control matrix D:

[0156] D=-jρ0ωN s (K+jkC-k 2 M) -1 (36)

[0157] Where j is an imaginary number, ρ0 represents the static density of the fluid in the closed space, N s is the matrix composed of the shape functions corresponding to each microphone, ω represents the sound wave frequency in the closed space, k represents the wave number; K, M, and C represent the stiffness matrix, mass matrix, and damping matrix of the closed space, respectively.

[0158] Step 5: Based on the closed space wave model control matrix D, construct the equivalent source reconstruction equation of the distributed load acoustic field

[0159] Arrange the system integral equation in step 4 to obtain the sound field reconstruction equation of the point sound source: s =DF

[0160] Then order:

[0161]

[0162] That is, the part after removing the known -jρ0ω part in the acoustic load vector F (34). Since the acoustic load vector F is an acoustic load matrix composed of several equivalent sources, F' can be expressed as:

[0163]

[0164] Where m is the number of equivalent sources and n is the number of nodes, which can be further sorted out to obtain:

[0165] F'=[N1,N2...N m ]Q (39)

[0166] Where Q1, Q2, ..., Q m are the equivalent source strengths respectively; N1, N2...N m are the shape functions corresponding to each equivalent source, which are also known and computable. Then the equivalent source reconstruction equation is derived:

[0167]

[0168] Where:

[0169] p s The observed sound signal vectors collected by a series of sound signal collection devices set up in the closed space;

[0170] D is the control matrix of the closed space wave model, obtained through step 4.

[0171] Step 6: Sound field reconstruction

[0172] Case ①: If the equivalent sources are a series of fixed point sound sources, the sound field reconstruction method is as follows (refer to Figure 1 ):

[0173] Step 6.1 Solve the equivalent source reconstruction equation constructed in step 5 to obtain the vector Q;

[0174] Use the array of acoustic signal acquisition devices to locally sample the sound field to obtain sampling information p s , the sparse recovery algorithm (SAMP) is used to solve Equation (40) and the source intensity data of the m equivalent sources set in step 1 are obtained, that is, the source intensity vector Q, that is, [Q1 Q2 ... Q m ] T .

[0175] According to the equivalent source method, the radiated sound field generated by any structure (sound radiator of any shape and any complex structure) can be replaced by the superposition of the sound fields generated by several simple sources of different source strengths placed inside the structure. Figure 2 As shown in the circle.

[0176] The vector Q obtained by the sparse recovery algorithm (SAMP) in the present invention contains only a finite number of non-zero values, that is, only a finite number of equivalent sources provide equivalent loads among the m equivalent sources, that is, the non-zero elements in the vector Q represent that the corresponding equivalent sources are activated (the activated equivalent sources are as follows Figure 2 In this way, the global sound field of the closed space can be reconstructed by superimposing the sound fields generated by fewer equivalent sources.

[0177] Step 6.2: Based on the vector Q obtained in step 6.1, solve for the acoustic load vector F;

[0178] Utilization:

[0179] F=-jρ0ωF′=-jρ0ω[N1,N2...N m ]Q (41)

[0180] Get vector F.

[0181] Step 6.3: Calculate the sound field distribution in the entire space based on the sound load vector F obtained in step 6.2

[0182] According to the system integral equation obtained in step 4, we can get:

[0183] p=(K+jkC-k 2 M) -1 F (42)

[0184] Formula (42) can be used to deduce the sound pressure values ​​distributed on the nodes in the entire enclosed space. Then, the sound pressure values ​​on each node are multiplied by the corresponding node shape function to obtain the sound field distribution information in the entire enclosed space, thus realizing sound field reconstruction.

[0185] Simulation verification of case ①:

[0186] Figure 5 and Figure 6 The figures show the simulation results of the sound field reconstruction of a closed space using the present invention at 10 Hz and 60 Hz respectively. The left column shows the reconstructed sound field, and the right column shows the real sound field. As can be seen from the figure, the reconstructed sound field obtained using the present invention is very close to the real sound field, which proves the effectiveness of the present invention.

[0187] Case ②: If the equivalent sources are a series of point sound sources with variable positions, the sound field reconstruction method is as follows (refer to Figure 7 ):

[0188] Step 6.1: Use the sparse recovery algorithm (SAMP) to solve equation (41) and obtain the vector F';

[0189] Step 6.2: Based on the vector F' obtained in step 6.1 and equations (34) and (37), solve the acoustic load vector F;

[0190] Step 6.3: Calculate the sound field distribution in the entire space based on the sound load vector F obtained in step 6.2

[0191] According to the system integral equation obtained in step 4, we can get:

[0192] p=(K+jkC-k 2 M) -1 F (43)

[0193] Formula (42) can be used to deduce the sound pressure values ​​distributed on the nodes in the entire enclosed space. Then, the sound pressure values ​​on each node are multiplied by the corresponding node shape function to obtain the sound field distribution information in the entire enclosed space, thus realizing sound field reconstruction.

[0194] Simulation verification of case ②:

[0195] Figure 9 and Figure 10 The figures show the simulation results of the sound field reconstruction of a closed space using the present invention at 10 Hz and 60 Hz respectively. The left column shows the reconstructed sound field, and the right column shows the real sound field. As can be seen from the figure, the reconstructed sound field obtained using the present invention is very close to the real sound field, which proves the effectiveness of the present invention.

[0196] In addition, since the vector F' can be expressed as shown in formula (38), the matrix [N1, N2...N m ] has a great influence on the reconstruction accuracy, that is, the position of the virtual source affects the reconstruction effect, so the dictionary learning algorithm can be used to calculate the matrix [N1, N2...N m ] is updated alternately with the vector Q. Since the matrix [N1, N2...N m ] is updated by the algorithm, so that the equivalent source position is also adaptively updated. After the final update is completed, the position and source strength of the equivalent source that meet the sound field reconstruction accuracy can be obtained. Based on the equivalent source position and source strength finally obtained from the update, the corresponding sound field can be reconstructed, providing the acoustic environment required for some simulation implementations.

[0197] Moreover, the vector Q obtained by using the dictionary learning algorithm contains only a finite number of non-zero values, that is, only a finite number of equivalent sources provide equivalent loads among the m equivalent sources, that is, the non-zero elements in the vector Q represent that the corresponding equivalent sources are activated (the activated equivalent sources are as follows Figure 8 In this way, the global sound field of the closed space can be reconstructed by superimposing the sound fields generated by fewer equivalent sources.

Claims

1. The acoustic field reconstruction method based on equivalent source activation theory under distributed load conditions is characterized by: The following steps are involved: Step 1: Arrange n nodes, q acoustic signal acquisition devices, and m equivalent sources in a closed space of the sound field to be reconstructed, and discretize the closed space; the m equivalent sources are a series of point sound sources with variable positions; Step 2: Constructing shape functions corresponding to the n nodes and the q acoustic signal acquisition devices respectively; Step 3: Based on the shape function obtained in step 2, establish the inherent properties of the closed space, including the stiffness matrix K, the mass matrix M, and the damping matrix C; Step 4: Based on the inherent properties of the enclosed space obtained in step 3, establish the system integral equation of the meshless acoustic field calculation method and construct the control matrix D of the enclosed space wave model; The system integral equation of the meshless acoustic field calculation method is: (K+jωC-ω 2 M)p=F Where: p represents the sound pressure value vector at each node, and its relationship with the sound signal collected at the collection point is p s =p·N s ;p s N is the observed sound signal vector collected by a series of sound signal collection devices set up in the closed space; s is the shape function corresponding to the acquisition point; ω is the resonant frequency; F is the acoustic load vector; The control matrix D of the closed space wave model is: D=-jρ0ωN s (K+jkC-k 2 M) -1 Where: j is an imaginary number, ρ0 represents the static density of the fluid in the closed space, ω represents the frequency of the sound wave in the closed space, and k represents the wave number; Step 5: Based on the closed space wave model control matrix D constructed in step 4, construct the equivalent source reconstruction equation of the distributed load acoustic field: Where: Q1, Q2, ..., Q m are the equivalent source strengths respectively; N1, N2...N m are the shape functions corresponding to each equivalent source; Step 6: Use the sparse recovery algorithm and dictionary learning algorithm to solve the equivalent source reconstruction equation constructed in step 5 and solve the acoustic load vector F; the dictionary learning algorithm can be used to recover the matrix [N1, N2...N m ] is updated alternately with the vector Q, thereby adaptively updating the equivalent source position; Step 7: Substitute the acoustic load vector F into the system integral equation established in step 4, derive the sound pressure values ​​distributed on the nodes in the entire enclosed space, and then multiply it by the corresponding node shape function to obtain the sound field distribution information in the entire enclosed space to achieve sound field reconstruction.

2. The sound field reconstruction method based on equivalent source activation theory under distributed load conditions according to claim 1 is characterized in that: In step 1: The layout principle of nodes is: evenly distributed, and they need to be arranged at the boundaries of closed spaces; when there are objects inside the closed space, they need to be arranged on the boundaries of the objects, and the nodes on the boundaries of the objects can outline the shape of the objects.

3. The sound field reconstruction method based on equivalent source activation theory under distributed load conditions according to claim 1 or 2 is characterized in that: In step 2, a spatially global and sparse shape function is constructed based on the moving least squares method.

4. The sound field reconstruction method based on equivalent source activation theory under distributed load conditions according to claim 3 is characterized in that: In step 3: Stiffness matrix Mass Matrix Damping Matrix Where: N is the shape function; ▽N is the derivative matrix of the shape function; v is the velocity of the medium particle; c0 represents the speed of sound; ζ is called the specific acoustic impedance; s is the area of ​​the rigid wall; Γ is the boundary of the closed space and the objects inside it, and Ω is the fluid area surrounded by the boundary Γ.

5. An electronic device comprising a processor and a storage medium; the storage medium storing a computer program; characterized in that: When the computer program is executed by the processor, the method according to any one of claims 1 to 4 is executed.

6. A non-volatile storage medium having a computer program stored thereon; characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 4 is executed.

Images