Method for separating acoustics and related pneumatic wall pressure fluctuation in closed wind tunnel

Abstract

The invention relates to the technical field of wind tunnel flow sound separation, in particular to a method for separating acoustics and related pneumatic wall pressure fluctuation in a closed wind tunnel, which comprises the following steps: acquiring sound pressure of a measurement signal in the closed wind tunnel, and storing the sound pressure in a cross-spectrum matrix; constructing a related pneumatic wall pressure fluctuation interference model of the closed wind tunnel according to physical factors such as convection velocity in the closed wind tunnel, a related transverse attenuation degree and a related longitudinal attenuation degree; constructing a related pneumatic wall pressure fluctuation interference model and a related pneumatic wall pressure fluctuation interference-acoustic separation model of probability factor analysis according to the low-rank property and the sparsity; according to the sparse prior information, constructing a prior model to constrain a related pneumatic wall pressure fluctuation interference-acoustic separation model; and obtaining maximum likelihood estimation of the parameters through an expectation maximization algorithm, and determining contribution degrees of acoustics and related pneumatic wall pressure fluctuation pressure of the to-be-tested component in the closed wind tunnel. The problem that an existing method does not consider interference physical information and is difficult to be directly applied to closed wind tunnel flow-sound separation is solved.

Description

Technical Field

[0001] The invention relates to the technical field of wind tunnel flow-acoustic separation, and in particular to a method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel. Background Art

[0002] Wind tunnel testing is one of the most critical testing methods in the research of low-noise control and design of aviation aircraft equipment. Although a lot of research has been done in the field of closed wind tunnel acoustic testing at home and abroad, in general, researchers still tend to use methods similar to those for open wind tunnel acoustic data processing to process closed wind tunnel acoustic test data, which makes it difficult to obtain high-precision aerodynamic acoustic information under closed wind tunnel testing; the separation of flow-induced interference such as turbulent boundary layer in the wind tunnel and aerodynamic noise includes physical means and signal processing.

[0003] The physical method used by Bahe.CJ et al. is to perform sound absorption treatment on the wall of closed wind tunnel; Huang Xun et al. conducted relevant research on the problem of acoustic cloak scattering in low-speed non-steady fluids; but both physical methods must meet many key criteria: have good mechanical resistance to flow, do not cause additional noise and do not filter out acoustic contributions, so they have certain limitations.

[0004] The signal processing algorithm does not require optimized design of the microphone array or the wind tunnel itself, and can directly process the acoustic test data, thereby achieving effective separation of flow-induced interference and noise signals; Huang Xun et al. used the iterative recursive method of the state observer in classical control theory to derive a new beamformer to suppress coherent background noise interference; however, this method fails to fully consider the physical information of the interference and is still difficult to directly apply to the study of flow-acoustic separation in closed wind tunnels. Summary of the invention

[0005] In view of the shortcomings of the existing methods, the present invention solves the problem that the existing methods do not consider interfering physical information and are difficult to be directly applied to closed wind tunnel flow-sound separation.

[0006] The technical solution adopted by the present invention is: a method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel comprises the following steps:

[0007] Step 1: Obtain the sound pressure of the measurement signal in the closed wind tunnel and pre-process the sound pressure;

[0008] As a preferred embodiment of the present invention, the preprocessing includes: truncating the time domain signal using a Hanning window function;

[0009] As a preferred implementation of the present invention, the time domain signal is converted into a frequency domain signal through FFT.

[0010] Step 2: storing the preprocessed frequency domain signal in a cross-spectrum matrix;

[0011] As a preferred embodiment of the present invention, the formula of the cross-spectrum matrix of the relevant aerodynamic wall pressure fluctuations is:

[0012] S yy =S aa +S nn +S εε (3)

[0013] Among them, S aa is the cross-spectral matrix of the acoustic contribution, S nn is the cross-spectral matrix of the aerodynamic contribution, S εε is the cross-spectral matrix of the additive source noise terms including all noise and errors.

[0014] Step 3: Construct the cross-spectrum matrix of the relevant aerodynamic wall pressure fluctuations based on the physical factors such as the convection velocity in the closed wind tunnel, the lateral attenuation of the correlation, and the longitudinal attenuation of the correlation;

[0015] As a preferred embodiment of the present invention, the formula of the cross-spectrum matrix of the relevant aerodynamic wall pressure fluctuations is:

[0016]

[0017] Where p is the complex scalar amplitude; U c is the eddy velocity in the boundary layer in the case of turbulent flow; α x is the longitudinal decay rate of coherence; α y is the transverse decay rate of coherence; x k is the lateral coordinate of the kth microphone; x l is the lateral coordinate of the lth microphone; y k is the longitudinal coordinate of the kth microphone; y l is the longitudinal coordinate of the lth microphone.

[0018] Step 4: Constructing the related aerodynamic wall pressure fluctuation interference model and the related aerodynamic wall pressure fluctuation interference-acoustic separation model of probability factor analysis based on low rank and sparsity;

[0019] As a preferred embodiment of the present invention, step 4 specifically includes:

[0020] Step 41, write the acoustic contribution as a linear combination of k' latent variables, modeled as:

[0021] a j =Lc j (5)

[0022] In the formula, a j is the acoustic contribution of the jth snapshot, c jis the vector of k' Fourier coefficients at the jth snapshot; L is the mixing matrix;

[0023] The cross-spectral matrix of the acoustic contribution is constructed as:

[0024] S aa =LS cc L H (6)

[0025] In the formula, S cc is the sound source c j The cross-spectral matrix of

[0026] Step 42: Construct the microphone array pressure field, modeled as:

[0027] n j =pv j (7)

[0028] Where p is the complex scalar amplitude; v is the j is a complex random vector;

[0029] The cross-spectral matrix of the relevant aerodynamic wall pressure fluctuations is constructed as:

[0030] S nn =|p| 2 S vv (8)

[0031] In the formula, S vv is the normalized aerodynamic wall pressure fluctuation pressure v j The cross-spectral matrix of

[0032] Step 43: construct a cross-spectral matrix of the additional source noise terms including all noise and errors as follows:

[0033]

[0034] In the formula, is the diagonal cross-spectral matrix of the microphone self-noise;

[0035] Step 44, the relevant aerodynamic wall pressure fluctuation interference-acoustic separation model is obtained as follows:

[0036] y j =Lc j +pv j +ε j (10)

[0037] In the formula, y j is the measurement data of the jth snapshot, ε j is the noise of the jth snapshot containing all additional sources of noise and error.

[0038] Step 5: construct a priori model constraint-related aerodynamic wall pressure fluctuation interference-acoustic separation model based on sparse prior information;

[0039] As a preferred implementation of the present invention, step five specifically includes:

[0040] Step 51: Construct the prior of the sound source c, the formula is:

[0041]

[0042] In the formula, is the variance of the sound source c;

[0043] Construct the prior of the mixing matrix L, the formula is:

[0044]

[0045] In the formula, I MK is the identity matrix of dimension MK×MK;

[0046] Step 52: construct a priori of the complex scalar amplitude p, the formula is:

[0047]

[0048] Among them, μ p is the mean of p, is the variance of p;

[0049] Construct the prior of aerodynamic wall pressure fluctuation pressure, the formula is:

[0050]

[0051] in,

[0052] Step 53: construct the noise prior of the additional sources of noise and error, the formula is:

[0053]

[0054] in, is the inverse gamma distribution of the prior variance, a ε is the shape parameter, b ε is the inverse scale parameter.

[0055] Step 6: Obtain the maximum likelihood estimation of the parameters through the expectation maximization algorithm to determine the contribution of the acoustic and related aerodynamic wall pressure fluctuation pressure of the tested component in the closed wind tunnel;

[0056] As a preferred implementation of the present invention, step six specifically includes:

[0057] Step 61, setting the flatness prior of acoustic and microphone self-noise;

[0058] Step 62: Initialize parameters;

[0059] Step 63, build about [c j |y j The expectation E{P} of the complete data log likelihood P of ] is given by And simplify E{P};

[0060] Step 64, find the parameter that maximizes the expected value of the simplified E{P};

[0061] Step 65, after taking the derivative of E{p} and the derivative is empty, we get:

[0062]

[0063]

[0064]

[0065]

[0066]

[0067]

[0068]

[0069] Step 66: Using the convergence criterion Perform iterative control;

[0070] Step 67: Output parameter estimates

[0071] As a preferred embodiment of the present invention, a system for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel includes: a memory for storing instructions executable by a processor; and a processor for executing the instructions to implement a method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel.

[0072] As a preferred embodiment of the present invention, a computer readable medium stores computer program codes, and when the computer program codes are executed by a processor, a method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel is implemented.

[0073] Beneficial effects of the present invention:

[0074] 1. Based on the statistical characteristics of the acoustic and related aerodynamic wall pressure fluctuation pressure field, the measured spectrum matrix is ​​decomposed into the sum of matrices related to the acoustic and related aerodynamic wall pressure fluctuation pressure contributions, taking into account the correlation of the aerodynamic wall pressure fluctuation on the sensor, and requiring little prior knowledge about the characteristics of the aerodynamic wall pressure fluctuation. In addition, the present invention obtains the maximum likelihood estimation of the parameters through the expectation maximization algorithm, determines the contribution of the acoustic and related aerodynamic wall pressure fluctuation pressure of the tested component in the closed wind tunnel, and realizes the precise separation of the acoustic and related aerodynamic wall pressure fluctuation pressure interference. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 is a flow chart of the method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel of the present invention;

[0076] Figure 2 This is a flowchart of the expectation maximization algorithm;

[0077] Figure 3 (a)-(d) are the beamforming diagrams of the original acoustic contribution, PFA-EM algorithm, and RPCA algorithm at a signal-to-noise ratio of -8dB at 1000Hz, respectively;

[0078] Figure 4 (a)-(d) are the beamforming diagrams of the original acoustic contribution, PFA-EM algorithm, and RPCA algorithm at 1000Hz with a signal-to-noise ratio of 0dB;

[0079] Figure 5 (a)-(d) are the beamforming diagrams of the measured signal, original acoustic contribution, PFA-EM algorithm, and RPCA algorithm at a signal-to-noise ratio of -8dB at 3000Hz;

[0080] Figure 6 (a)-(d) are the beamforming diagrams of the measured signal, original acoustic contribution, PFA-EM algorithm, and RPCA algorithm at a signal-to-noise ratio of 0dB at 3000Hz, respectively. DETAILED DESCRIPTION

[0081] The present invention is further described below in conjunction with the accompanying drawings and embodiments. This figure is a simplified schematic diagram, which only illustrates the basic structure of the present invention in a schematic manner, and therefore it only shows the components related to the present invention.

[0082] like Figure 1 As shown, a method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel includes the following steps:

[0083] Step 1: Obtain the sound pressure of the measurement signal in the closed wind tunnel and pre-process the sound pressure;

[0084] A micro-electromechanical system (MEMS) microphone array is arranged in a closed wind tunnel. The microphone array can measure the sound pressure of the signal. The signal collected by the microphone array is a time domain signal, which needs to be converted to the frequency domain for calculation in the separation step. The classic Hanning window function is selected to truncate the time domain signal, and then the truncated time domain signal is converted into a frequency domain signal through the fast Fourier transform (FFT). The Fourier transform cutoff frequency f is set, and each segment is a snapshot. In this way, the microphone array frequency domain signal data y can be obtained.

[0085] Step 2: storing the obtained frequency domain signal in a cross-spectral matrix (CSM);

[0086] When dealing with temporally statistically stationary pressure fields, the measurements are usually stored in the form of a cross-spectral matrix (CSM); s The cross-spectral matrix of the measured signal is expressed as follows:

[0087]

[0088] Among them, S yy is the cross-spectral matrix of the measured signal, N s is the number of snapshots, y j is the measurement data of the jth snapshot.

[0089] At a frequency, the microphone array frequency domain signal data can be arranged in a vector y of M complex values. The sound pressure vector can be written as the sum of the acoustic contribution a, the aerodynamic contribution n, and the noise term containing all additional sources of noise and error:

[0090] y=a+n+ε (2)

[0091] Where y is the measured data at a frequency, a is the acoustic contribution, n is the aerodynamic contribution, and ε is the noise term that includes all additional sources of noise and errors.

[0092] Substituting y into the cross-spectrum matrix, the theoretical cross-spectrum matrix expression of the measured data is:

[0093] S yy =S aa +S nn +S εε (3)

[0094] Assuming an infinite number of snapshots, the cross term tends to zero; where S yy is the cross-spectral matrix of the measured signal, S aa is the cross-spectral matrix of the acoustic contribution, S nn is the cross-spectral matrix of the aerodynamic contribution, S εε is the cross-spectral matrix of the additive source noise terms including all noise and errors.

[0095] Step 3: Based on the physical factors such as the convection velocity in the closed wind tunnel, the lateral attenuation of the correlation, and the longitudinal attenuation of the correlation, a model of the aerodynamic wall pressure fluctuation interference in the closed wind tunnel is constructed;

[0096] In the current separation model, the pressure field of the correlated aerodynamic wall pressure fluctuations is spatially stable and uniform, and can be statistically modeled by cross-correlation of the pressure field. The model of the correlated aerodynamic wall pressure fluctuation interference depends on the physical parameters, and the expression for constructing the cross-spectral matrix of the correlated aerodynamic wall pressure fluctuations is:

[0097]

[0098] In the formula, S nn is the cross-spectral matrix of the associated aerodynamic wall pressure fluctuations; f is the Fourier transform cutoff frequency; p is the complex scalar amplitude, which is a scalar that depends on the frequency f; U c is the eddy velocity in the boundary layer in the case of turbulent flow; α x is the longitudinal decay rate of coherence, typically ranging from 0.1 to 0.12; α y is the transverse decay rate of coherence, typically in the range of 0.7 to 1.2; x k is the lateral coordinate of the kth microphone, x l is the lateral coordinate of the lth microphone, y k is the longitudinal coordinate of the kth microphone, y l is the longitudinal coordinate of the lth microphone.

[0099] The problem of the correlation of the pressure field of the aerodynamic wall pressure fluctuation on the microphone is solved through the model of related aerodynamic wall pressure fluctuation interference, providing a model for constructing a more accurate aerodynamic wall pressure fluctuation pressure field.

[0100] Step 4: According to the low rank and sparsity of each component of the model, a related aerodynamic wall pressure fluctuation interference model and a related aerodynamic wall pressure fluctuation interference-acoustic separation model of probability factor analysis are constructed;

[0101] Step 4 specifically includes:

[0102] Step 41: Construct an acoustic signal model based on the low rank and sparsity of the acoustic signal. The sound source is usually unknown and the propagation may be uncertain. The acoustic contribution is written as a linear combination of k' potential variables and is modeled as:

[0103] a j =Lc j (5)

[0104] In the formula, a j is the acoustic contribution of the jth snapshot, c jis the vector of k' Fourier coefficients at the jth snapshot; L is the mixing matrix, which is an M×k complex matrix.

[0105] The cross-spectral matrix of the acoustic signal is:

[0106] S aa =LS cc L H (6)

[0107] In the formula, S aa is the cross-spectral matrix of the acoustic contribution, S cc is the sound source c j The cross-spectral matrix of , L is the mixing matrix.

[0108] Step 42, the aerodynamic wall pressure fluctuation disturbance should develop along the non-moving rigid wall and be caused by the incompressible flow parallel to the wall, which should produce a fixed and uniform pressure field on the microphone array, modeled as:

[0109] n j =pv j (7)

[0110] Where p is the complex scalar amplitude; v is the j is a complex random vector, representing the normalized aerodynamic wall pressure fluctuation pressure at the jth snapshot.

[0111] The cross-spectral matrix of the relevant aerodynamic wall pressure fluctuations is:

[0112] S nn =|p| 2 S vv (8)

[0113] In the formula, S nn is the cross-spectral matrix of the aerodynamic contribution, p is the complex scalar amplitude, S vv is the normalized aerodynamic wall pressure fluctuation pressure v j The cross-spectral matrix of .

[0114] Step 43: Microphone self-noise may be caused by measurement settings, environment, etc., and is modeled as being completely uncorrelated with the microphone and heteroscedastic, and is statistically described by a diagonal cross-spectrum matrix, modeled as:

[0115]

[0116] In the formula, S εε is the cross-spectral matrix of the additive source noise terms including all noise and errors, is the diagonal cross-spectral matrix of the microphone self-noise.

[0117] Step 44, at a frequency, the expression of the relevant aerodynamic wall pressure fluctuation interference-acoustic separation model, i.e., the measurement data of the microphone is:

[0118] y j =Lc j +pv j +ε j (10)

[0119] In the formula, y j is the measurement data of the jth snapshot, L is the mixing matrix, c j is the vector of k' Fourier coefficients at the jth snapshot, p is the complex scalar amplitude, v j is the normalized aerodynamic wall pressure fluctuation pressure at the jth snapshot, ε j is the noise of the jth snapshot containing all additional sources of noise and error.

[0120] The present invention models each component in microphone measurement data at a frequency, forms a related aerodynamic wall pressure fluctuation interference-acoustic separation optimization problem, and provides an accurate model for the related aerodynamic wall pressure fluctuation interference-acoustic separation.

[0121] Step 5: construct a priori model constraint-related aerodynamic wall pressure fluctuation interference-acoustic separation model based on sparse prior information;

[0122] Specifically include:

[0123] Step 51, the latent variables are assigned a Gaussian distribution and are considered to be a priori independent heteroskedastic, and the model is:

[0124]

[0125] Where c is the sound source, is the variance of c.

[0126] Due to the potential factor The variance of is also unknown, and a prior distribution needs to be assigned, which is an inverse gamma distribution, and the formula is:

[0127]

[0128] The prior for the mixing matrix L is assigned as a central complex multivariate Gaussian distribution with normalized variance, modeled as:

[0129]

[0130] In the formula, I MK is the identity matrix of dimension MK×MK.

[0131] Step 52: The priori value of the complex scalar amplitude p in the aerodynamic wall pressure fluctuation interference model should be Gaussian distribution, and the model is:

[0132]

[0133] The vector v in the aerodynamic wall pressure fluctuation model is modeled as a central Gaussian prior, and the covariance is selected based on the physically relevant aerodynamic wall pressure fluctuation covariance model:

[0134]

[0135] in, Following the empirically related aerodynamic wall pressure fluctuation interference model, the model is:

[0136]

[0137] Step 53, the microphone self-noise is still considered as Gaussian noise according to the central limit theorem applied to Fourier coefficients, and the prior model is:

[0138]

[0139] and choose the inverse gamma distribution for the prior variance:

[0140]

[0141] Among them, a ε , b ε is the parameter of the inverse gamma distribution, a ε is the shape parameter, b ε is the inverse scale parameter.

[0142] The fifth step is to construct a priori model for the model in step four based on the sparse prior information, which is used to constrain the related aerodynamic wall pressure fluctuation interference-acoustic separation optimization problem. The inverse gamma model is further selected for the priori model because the positive support of the inverse gamma distribution and its conjugation with Gaussian provide convenience for algebra.

[0143] Step 6: Obtain the maximum likelihood estimate of the parameters through the expectation maximization algorithm to determine the contribution of the acoustic and related aerodynamic wall pressure fluctuations of the tested component in the closed wind tunnel.

[0144] like Figure 2 As shown, specifically including:

[0145] Step 61: Set a flat prior for the acoustic part and the microphone self-noise; wherein the prior parameters of the factor variance and the uncorrelated variance are: a c =b c =a ε =b ε =10 4 ; Select information priors for the relevant aerodynamic wall pressure fluctuation disturbances,

[0146] Step 62, initialize the parameters; set k'=5; the initial value of parameter p is p0=a p;parameter The initial value is parameter The initial value is Among them, eig k' (A) refers to the k' highest eigenvalue of A; S vv The initial value is The initial value of L is L0=N c (0,I Mk' / k');

[0147] Step 63, build about [c j |y j The expectation of the complete data log-likelihood P is:

[0148]

[0149] in,

[0150]

[0151] After simplification, we get:

[0152]

[0153] In the formula, E{c j} is represented by c j expectations; Expressed as The expectation of , tr{A} represents the trace of A,

[0154] {A} H represents the conjugate transpose of A.

[0155] Step 64: Find the parameter that maximizes the expected value of step 63. The formula is:

[0156]

[0157] In the formula, It is expressed as the estimated value of L after maximizing the expected value; Represented as c j The estimated value after maximizing the expected value; Expressed as The estimated value after maximizing the expected value; Expressed as The estimated value after maximizing the expected value; It is expressed as the estimated value of p after maximizing the expected value; Represented as v j The estimated value after maximizing the expected value It is expressed as the estimated value of L after maximizing the expected value.

[0158] Step 65: When the derivative of E{p} is empty, the updated parameters are obtained. The derivative formula of E{p} with respect to each parameter is:

[0159]

[0160]

[0161]

[0162]

[0163]

[0164] The updated parameter formula obtained by leaving the parameter derivative formula of step 65 empty is:

[0165]

[0166]

[0167]

[0168]

[0169]

[0170]

[0171]

[0172] In the formula, Expressed as E{c j}The estimated value after maximizing the expected value.

[0173] Step 66, repeating steps 64 to 66 until convergence;

[0174] The convergence criterion is:

[0175] when When it is less than the set threshold z, 0<z<1, the iteration stops;

[0176] Step 67, the parameter estimation output after convergence is based on the cross-spectrum formula of the acoustic signal: Ω cj It is c j The variance of the acoustic signal is obtained, and the cross-spectrum of the acoustic signal is obtained to achieve the separation of the acoustic and related aerodynamic wall pressure fluctuation pressure in the closed wind tunnel.

[0177] Step 6: Based on the joint estimation of the models of each part to be tested in the closed wind tunnel and the prior, the contribution of the sound source and the related aerodynamic wall pressure fluctuation pressure in the closed wind tunnel is gradually determined in the iterative process, so as to achieve the accurate separation of the sound source and the related aerodynamic wall pressure fluctuation in the closed wind tunnel.

[0178] The separated sound source contribution wavenumber formation of the method of the present invention (PFA-EM) and the RPCA algorithm at different frequencies and different signal-to-noise ratios is compared; Figure 3-Figure 6 As shown, it can be seen that the two algorithms separate the sound source and the related aerodynamic wall pressure fluctuation in the closed wind tunnel, and the number of separated acoustic contribution sources is very accurate; from the results of the acoustic contribution beamforming effect, the separation effect of the present invention is better, while the RPCA algorithm does not take into account the correlation of the related aerodynamic wall pressure fluctuation on the microphone, resulting in a lower separation effect than the present invention; Figure 3 and Figure 4 The comparison, Figure 5 and Figure 6 In the comparison, due to the low analysis signal-to-noise ratio, the separation effect of the RPCA algorithm is relatively poor; as the signal-to-noise ratio increases, the separation effect of the RPCA algorithm improves; the method of the present invention always maintains a high separation effect; Figure 3 and Figure 5 The comparison, Figure 4 and Figure 6 In the comparison, due to the low analysis frequency, the separation effect of the RPCA algorithm is relatively poor; with the increase of frequency, the separation effect of the RPCA algorithm improves; the method of the present invention always maintains a high separation effect. As far as the separation effect is concerned, the method of the present invention always maintains the best separation level because it takes into account the correlation of the relevant aerodynamic wall pressure fluctuations on the microphone.

[0179] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel, characterized in that: The following steps are involved: Step 1: Obtain the sound pressure of the measurement signal in the closed wind tunnel and pre-process the sound pressure; Step 2: storing the preprocessed frequency domain signal in a cross-spectrum matrix; Step 3: construct the cross-spectrum matrix of the relevant aerodynamic wall pressure fluctuations according to the physical factors of the convection velocity, the lateral attenuation of the correlation, and the longitudinal attenuation of the correlation in the closed wind tunnel; Step 4: Constructing the related aerodynamic wall pressure fluctuation interference model and the related aerodynamic wall pressure fluctuation interference-acoustic separation model of probability factor analysis based on low rank and sparsity; Step 5: construct a priori model constraint-related aerodynamic wall pressure fluctuation interference-acoustic separation model based on sparse prior information; Step 6: Obtain the maximum likelihood estimate of the parameters through the expectation maximization algorithm to determine the contribution of the acoustic and related aerodynamic wall pressure fluctuations of the tested component in the closed wind tunnel.

2. The method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel according to claim 1, characterized in that: The formula for the cross-spectral matrix of the relevant aerodynamic wall pressure fluctuations is: S yy =S aa +S nn +S εε (3) Among them, S aa is the cross-spectral matrix of the acoustic contribution, S nn is the cross-spectral matrix of the associated aerodynamic wall pressure fluctuations, S εε is the cross-spectral matrix of the additive source noise terms including all noise and errors.

3. The method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel according to claim 2, characterized in that: The formula for the cross-spectral matrix of the relevant aerodynamic wall pressure fluctuations is: Where p is the complex scalar amplitude; U c is the eddy velocity in the boundary layer in the case of turbulent flow; α x is the longitudinal decay rate of coherence; α y is the transverse decay rate of coherence; x k is the lateral coordinate of the kth microphone; x l is the lateral coordinate of the lth microphone; y k is the longitudinal coordinate of the kth microphone; y l is the longitudinal coordinate of the lth microphone.

4. The method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel according to claim 1, characterized in that: Step 4 specifically includes: Step 41, write the acoustic contribution as a linear combination of latent variables, the formula is: a j =Lc j (5) In the formula, a j is the acoustic contribution of the jth snapshot, c j is the vector of k' Fourier coefficients at the jth snapshot; L is the mixing matrix; The cross-spectral matrix of the acoustic contribution is constructed as: S aa =LS cc L H (6) In the formula, S cc is the sound source c j The cross-spectral matrix of Step 42: Construct the pressure field of the microphone array as: n j =pv j (7) Where p is the complex scalar amplitude; v is the j is a complex random vector; The cross-spectral matrix of the relevant aerodynamic wall pressure fluctuations is constructed as: S nn =|p| 2 S vv (8) In the formula, S vv is the normalized aerodynamic wall pressure fluctuation pressure v j The cross-spectral matrix of Step 43: construct a cross-spectral matrix of the additional source noise terms including all noise and errors as follows: In the formula, is the diagonal cross-spectral matrix of the microphone self-noise; Step 44, the relevant aerodynamic wall pressure fluctuation interference-acoustic separation model is obtained as follows: y j =Lc j +pv j +ε j (10) In the formula, ε j is the noise of the jth snapshot containing all additional sources of noise and error.

5. The method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel according to claim 1, characterized in that: Step 5 specifically includes: Step 51: Construct the prior of the sound source c, the formula is: In the formula, is the variance of the sound source c; Construct the prior of the mixing matrix L, the formula is: In the formula, I MK is the identity matrix of dimension MK×MK; Step 52: construct a priori of the complex scalar amplitude p, the formula is: Among them, μ p is the mean of p, is the variance of p; Construct the prior of aerodynamic wall pressure fluctuation pressure, the formula is: in, Step 53: construct the noise prior of the additional sources of noise and error, the formula is: in, is the inverse gamma distribution of the prior variance, a ε is the shape parameter, b ε is the inverse scale parameter.

6. The method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel according to claim 1, characterized in that: Step 6 specifically includes: Step 61, setting the flatness prior of acoustic and microphone self-noise; Step 62: Initialize parameters; Step 63, build [c j |y j The expected value E{P} of the log-likelihood P of ] is given by And simplify E{P}; Step 64, find the parameter that maximizes the expected value of the simplified E{P}; Step 65, derive E{p}, and when the derivative is empty, we get: Step 66: Using the convergence criterion Perform iterative control; Step 67: Output parameter estimates 7. The method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel according to claim 1, characterized in that: The time domain signal is truncated using the Hanning window function.

8. The method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel according to claim 7, characterized in that: The time domain signal is converted into the frequency domain signal through FFT.

9. Separation acoustics and related aerodynamic wall pressure fluctuation system in a closed wind tunnel, characterized in that: include: a memory for storing instructions executable by a processor; A processor, configured to execute instructions to implement the method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel as described in any one of claims 1 to 8.

10. A computer readable medium storing computer program code, characterized in that: The computer program code, when executed by a processor, implements the method for separating acoustics and related aerodynamic wall pressure fluctuations in a closed wind tunnel as claimed in any one of claims 1 to 8.

Images