Acoustic vector circular array orientation estimation method based on cross covariance matrix construction, program, equipment and storage medium

Abstract

The invention belongs to the technical field of underwater acoustic array signal processing, and particularly relates to an acoustic vector circular array orientation estimation method based on cross covariance matrix construction, a program, equipment and a storage medium. The method comprises the following steps: introducing a preprocessing matrix to transform the output of an acoustic vector circular array, and constructing a cross covariance matrix of sound pressure and different vibration velocity components according to the transformed vector; combining and arranging the cross covariance matrixes of the sound pressure and the different vibration velocity components to obtain a combined arrangement matrix; performing singular value decomposition on the combined permutation matrix, and reconstructing a cross covariance matrix by using a unitary matrix formed by left singular vectors of the combined permutation matrix and a diagonal matrix formed by singular values; and implementing a spatial spectrum estimation method by using the cross covariance matrix, and taking the scanning angle corresponding to the maximum value in the spatial spectrum as an orientation estimation result of the detection target. The method is suitable for a column barrier condition, and can effectively improve the dual-target resolution and orientation estimation performance of the acoustic vector circular array under a low signal-to-noise ratio condition.

Description

Technical Field

[0001] The present invention belongs to the technical field of underwater acoustic array signal processing, and particularly relates to a method, program, device and storage medium for azimuth estimation of an acoustic vector circular array based on the construction of a cross-covariance matrix. Background Technique

[0002] In a sonar system, as a common and typical array form, a circular array can provide 360° all-round and unambiguous azimuth angle information, and has wide applications in sonar systems. In recent years, the azimuth estimation method of an acoustic vector circular array has become a research hotspot and rich research results have been achieved. However, there are still problems such as low azimuth estimation accuracy and poor target resolution of the azimuth estimation method. Therefore, it is of great significance to carry out research on the array signal processing technology of an acoustic vector circular array.

[0003] In recent years, since an acoustic vector sensor can obtain more sound field information compared with a sound pressure sensor, the acoustic vector sensor array has high azimuth estimation performance, and the acoustic vector array signal processing technology has developed rapidly. Hui Junying et al. (Hui Junying, Liu Hong, Yu Huabing. Preliminary study on the joint information processing of sound pressure and particle velocity and its physical basis. Acta Acustica, 2000, 25(4): 303-307.) studied the correlation between sound pressure and particle velocity and revealed the anti-noise principle of joint sound pressure and particle velocity processing. In an isotropic noise field, the sound pressure and particle velocity components of co-point noise are uncorrelated. Bai Xingyu et al. (Bai Xingyu, Jiang Yu, Zhao Chunhui. Source number detection and azimuth estimation of an acoustic vector array based on joint sound pressure and particle velocity processing [J]. Acta Acustica, 2008(01): 56-61.) realized the azimuth estimation of a remote target by constructing a cross-covariance matrix of sound pressure and particle velocity and taking advantage of the good noise suppression ability of the joint sound pressure and particle velocity processing method. Yang Desen et al. (Yang Desen, Zhu Zhongrui, Shi Shengguo et al. Azimuth estimation of a target in the phase mode domain of an acoustic vector circular array. Acta Acustica, 2014, 39(1): 19-26.) extended the joint sound pressure and particle velocity processing theory to an acoustic vector circular array by using the phase mode transformation principle. The main content is to construct a cross-covariance matrix of the components of sound pressure and radial and tangential particle velocities respectively based on the sound intensity principle, add the cross-covariance matrices of the components, obtain a new cross-covariance matrix, and then perform target azimuth estimation. However, in the methods proposed in the above-mentioned literature, only a simple addition process is performed on the cross-covariance matrices of sound pressure and different particle velocity components, and no systematic and in-depth theoretical research has been carried out on different construction methods of the cross-covariance matrix of sound pressure and particle velocity components and their azimuth estimation performance. Summary of the Invention

[0004] The object of the present invention is to provide a method, program, device and storage medium for azimuth estimation of a vector hydrophone circular array based on the construction of the cross-covariance matrix. Aiming at the problem of constructing the cross-covariance matrix in the joint processing of sound pressure and particle velocity, on the basis of the traditional joint processing method of sound pressure and particle velocity, the present invention constructs different extended cross-covariance matrices by performing longitudinal, transverse and diagonal arrangement processing on the cross-covariance matrix of the sound pressure and particle velocity components, and uses the spatial spectrum estimation method to realize the azimuth estimation of the target by the vector hydrophone circular array. The present invention is applicable to the columnar obstacle condition and can effectively improve the dual-target resolution and azimuth estimation performance of the vector hydrophone circular array under low signal-to-noise ratio conditions.

[0005] A method for azimuth estimation of a vector hydrophone circular array based on the construction of the cross-covariance matrix, comprising the following steps:

[0006] Step 1: Arrange a vector hydrophone circular array in the fluid domain where the detection target is located. The vector hydrophone circular array outputs a sound pressure vector, a radial particle velocity vector and a tangential particle velocity vector according to the received signal;

[0007] Step 2: Introduce a preprocessing matrix to transform the output of the vector hydrophone circular array, convert the uniform circular array into a virtual uniform linear array, and obtain the transformed sound pressure vector, radial particle velocity vector and tangential particle velocity vector;

[0008] Step 3: According to the transformed sound pressure vector and radial particle velocity vector, construct the cross-covariance matrix of the received data of the sound pressure and radial particle velocity; according to the transformed sound pressure vector and tangential particle velocity vector, construct the cross-covariance matrix of the received data of the sound pressure and tangential particle velocity;

[0009] Step 4: Combine and arrange the cross-covariance matrix of the received data of the sound pressure and radial particle velocity and the cross-covariance matrix of the received data of the sound pressure and tangential particle velocity to obtain a combined arrangement matrix;

[0010] Step 5: Perform singular value decomposition on the combined arrangement matrix, and reconstruct the cross-covariance matrix by using the unitary matrix composed of the left singular vectors of the combined arrangement matrix and the diagonal matrix composed of the singular values;

[0011] Step 6: Use the cross-covariance matrix to implement the spatial spectrum estimation method, and take the scanning angle corresponding to the maximum value in the spatial spectrum as the azimuth estimation result of the detection target.

[0012] Further, in the step 1, the vector hydrophone circular array has M array elements uniformly arranged in a circle with a radius of r. The vector hydrophone circular array outputs a sound pressure vector p, a radial particle velocity vector v r and a tangential particle velocity vector

[0013] In the step 3, according to the transformed sound pressure vector p c and the radial particle velocity vector v rc, construct the cross-covariance matrix of the received sound pressure and radial vibration velocity According to the transformed sound pressure vector p c and the tangential vibration velocity vector construct the cross-covariance matrix of the received sound pressure and tangential vibration velocity E{·} represents the expectation operation.

[0014] Furthermore, the preprocessing matrix T p , T vr and are introduced in step 2 to transform the output of the acoustic vector circular array, convert the uniform circular array into a virtual uniform linear array, and obtain the transformed sound pressure vector p c , the radial vibration velocity vector v rc and the tangential vibration velocity vector

[0015]

[0016] p c = T p p, v rc = T vr v r ,

[0017] where F = [ω -K , ω -K+1 , …, ω K , j is the imaginary symbol, j 2 = -1; q = -K, …, K; B p = diag[b -K , …, b K , B vr = diag[b′ -K / jkρc, …, b′ K / jkρc], b q is the modal intensity; b′ q represents the derivative of b q with respect to the radius r; ρ is the density of the fluid medium in the fluid domain, k is the wave number of the fluid medium in the fluid domain, k = ω / c, ω is the angular velocity of the fluid medium in the fluid domain, c is the sound speed; K is the maximum modal number.

[0018] Furthermore, if the acoustic vector circular array is directly arranged in the fluid domain or installed on the near surface of the acoustically transparent cylindrical barrier housing, then b q = j q J q (kr), J q (·) represents the q-th order Bessel function of the first kind;

[0019] If the acoustic vector circular array is directly arranged in the fluid domain, then K = [kr]; if the acoustic vector circular array is installed on the near surface of the cylindrical barrier shell, then K = [ka], where a is the radius of the cylindrical barrier shell.

[0020] The spatial spectrum estimation methods in step 6 include CBF, MVDR, and MUSIC methods.

[0021] Further, in step 4, the cross-covariance matrix R of the received data of sound pressure and radial vibration velocity pvr and the cross-covariance matrix of the received data of sound pressure and tangential vibration velocity are arranged longitudinally in combination to obtain a combined arrangement matrix

[0022] In step 5, the combined arrangement matrix R V is subjected to singular value decomposition, and the unitary matrix U composed of the left singular vectors of R V and the diagonal matrix Λ composed of singular values V,1 are used to reconstruct the cross-covariance matrix V,1

[0023]

[0024] Among them, the combined arrangement matrix R V has N singular values, 2N left singular vectors, and N right singular vectors; U V,1 is the unitary matrix composed of the first N left singular vectors of R V ; Λ V,1 is the diagonal matrix composed of the N singular values of R V ; D p is the unitary matrix composed of the N right singular vectors of R V .

[0025] Let Reconstruct the cross-covariance matrix

[0026] In step 6, the cross-covariance matrix is used to implement the spatial spectrum estimation method.

[0027] Further, in step 4, the cross-covariance matrix R of the received data of sound pressure and radial vibration velocity pvr and the cross-covariance matrix of the received data of sound pressure and tangential vibration velocity are arranged transversely in combination to obtain a combined arrangement matrix

[0028] In step 5, the combined arrangement matrix R H is subjected to singular value decomposition, and the unitary matrix U composed of the left singular vectors of R H ​​H The diagonal matrix Λ composed of singular values H,1 Reconstructed cross-covariance matrix

[0029]

[0030] Among them, the combined permutation matrix R H has N singular values, N left singular vectors, and 2N right singular vectors; U H is the unitary matrix composed of the N left singular vectors of R H ; Λ H,1 is the diagonal matrix composed of the N singular values of R H ; D pH,1 is the unitary matrix composed of the first N right singular vectors of R H .

[0031] Let Reconstructed cross-covariance matrix

[0032] In step 6, the cross-covariance matrix is used to implement the spatial spectrum estimation method.

[0033] Furthermore, in step 4, the cross-covariance matrix R of the sound pressure and radial vibration velocity received data pvr , the cross-covariance matrix of the sound pressure and tangential vibration velocity received data are diagonally combined and arranged to obtain the combined permutation matrix

[0034] In step 5, the combined permutation matrix R D in R pvr and are respectively subjected to singular value decomposition, and the unitary matrix U pvr composed of the left singular vectors of R pvr and the diagonal matrix Λ composed of singular values pvr are used to reconstruct the cross-covariance matrix Using the unitary matrix composed of the left singular vectors of and the diagonal matrix composed of singular values to reconstruct the cross-covariance matrix

[0035]

[0036] The reconstructed cross-covariance matrix and are diagonally combined and arranged to reconstruct the cross-covariance matrix

[0037]

[0038] In step 6, the cross-covariance matrix is used to implement the spatial spectrum estimation method.

[0039] A computer device / system, including a memory, a processor, and a computer program stored on the memory, wherein the processor executes the computer program to implement the steps of the above-mentioned sound vector circular array azimuth estimation method based on the construction of the cross-covariance matrix.

[0040] A computer-readable storage medium, on which a computer program / instructions are stored, and when the computer program / instructions are executed by a processor, the steps of the above-mentioned sound vector circular array azimuth estimation method based on the construction of the cross-covariance matrix are implemented.

[0041] A computer program product, including a computer program / instructions, and when the computer program / instructions are executed by a processor, the steps of the above-mentioned sound vector circular array azimuth estimation method based on the construction of the cross-covariance matrix are implemented.

[0042] The beneficial effects of the present invention are as follows:

[0043] In the present invention, by performing longitudinal, transverse, and diagonal arrangement processing on the cross-covariance matrix of the sound pressure and vibration velocity components, different extended cross-covariance matrices are constructed, and the spatial spectrum estimation method is used to realize the azimuth estimation of the sound vector circular array for the target. Through systematic theoretical research on the target azimuth estimation performance of different construction methods, the present invention is applicable to the columnar barrier condition and can effectively improve the double-target resolution and azimuth estimation performance of the sound vector circular array under low signal-to-noise ratio conditions. Brief Description of the Drawings

[0044] Figure 1 is the overall flowchart of the embodiment of the present invention.

[0045] Figure 2 is the model diagram of the cylindrical baffle array.

[0046] Figure 3 is the spatial spectrum comparison diagram at different signal-to-noise ratios, where (a) SNR = 0dB; (b) SNR = -8dB.

[0047] Figure 4 is the curve graph of the double-target resolution probability changing with the angular interval.

[0048] Figure 5 is the curve graph of the root mean square error changing with the signal-to-noise ratio. Detailed Embodiment

[0049] The following further describes the present invention with reference to the drawings.

[0050] Taking the sound vector circular array under the columnar barrier condition as an example, an array output signal model is established, such asFigure 2 As shown, the uniform acoustic vector circular array is installed on the surface of the cylindrical barrier housing and is located in the z = 0 plane; the positive directions of the x, y, and z axes of each vector hydrophone coincide with the radial, tangential, and axial directions at that point, respectively; element 0 is located on the x-axis, the center of the circle coincides with the coordinate origin, and the right-handed coordinate system is adopted. Considering an isotropic noise field, there are H far-field uncorrelated signal sources incident on the above cylindrical barrier array model, and the output data of the acoustic vector circular array is written in vector form:

[0051]

[0052] where:

[0053] F = [ω -K , ω -K+1 , …, ω K ,

[0054] B p = diag[b -K , …, b K , B vr = diag[b′ -K / jkρc, …, b′ K / jkρc],

[0055]

[0056] p(t), v r (t) and respectively represent the acoustic pressure and the radial and tangential vibration velocity output vectors of the acoustic vector sensor; s(t) represents the incident source vector; n p (t), n vr (t) and respectively represent the noise vectors of the acoustic pressure and the radial and tangential vibration velocities; j is the imaginary symbol, j 2 = -1; q = -K, …, K; b q is the modal strength; b′ q represents the derivative of b q with respect to the radius r; ρ is the density of the fluid medium in the fluid domain, k is the wave number of the fluid medium in the fluid domain, k = ω / c, ω is the angular velocity of the fluid medium in the fluid domain, c is the sound speed; K is the maximum modal number; the superscript H represents the conjugate transpose; the superscript T represents the transpose operation;

[0057] If the acoustic vector circular array is directly arranged in the fluid domain or installed on the near surface of the acoustically transparent cylindrical barrier housing, then b q = j q J q (kr), J q (·) represents the Bessel function of the first kind of order q;​

[0058] If the acoustic vector circular array is directly arranged in the fluid domain, then K = [kr]; if the acoustic vector circular array is installed on the near surface of the cylindrical barrier shell, then K = [ka], where a is the radius of the cylindrical barrier shell.

[0059] Introduce the preprocessing matrix:

[0060]

[0061] Perform the following transformation on the sound pressure data of the acoustic vector circular array:

[0062]

[0063] Among them, Since the steering vector matrix A(φ) has a Vandermonde form and A(φ) is similar to the steering vector matrix of a uniform linear array, the complex circular array manifold can be transformed into a relatively simple new matrix A(φ). The preprocessing matrix T P Can transform the sound pressure circular array in the element domain into a uniform linear array in the phase mode domain, completing the transformation from a uniform circular array to a virtual uniform linear array.

[0064] According to Obtain the added cross-covariance matrix:

[0065]

[0066] Among them, R s = E{s(t)s H (t)} is the signal covariance matrix; Represents the cross-covariance matrix of the sound pressure and the received data of the radial and tangential vibration velocities respectively; Is the noise covariance matrix. In an isotropic noise field, the noise is uncorrelated, then there is R npvr = 0 N×N , N = 2K + 1, so the simplified cross-covariance matrix of the acoustic vector circular array sound pressure and vibration velocity joint processing method is:

[0067] R Sum = 2A(φ)R s A H (φ)

[0068] Arrange R pvr and Vertically, horizontally, and diagonally respectively, to obtain:

[0069]

[0070] Respectively for R Sum, R V , R H , R D Perform singular value decomposition and reconstruction:

[0071] (1) Perform singular value decomposition on R Sum :

[0072]

[0073] In the formula, Λ pvs is a diagonal matrix composed of the singular values of R Sum , U pvs is a unitary matrix composed of the left singular vectors of R Sum , D pvs is a unitary matrix composed of the right singular vectors of R Sum . Use U pvs and Λ pvs to reconstruct R Sum :

[0074]

[0075] |(·)| represents taking the modulus value of each element in (·), is the cross-covariance matrix after singular value reconstruction of R Sum .

[0076] (2) Perform singular value decomposition on R V :

[0077]

[0078] In the formula, A V (φ) = [A(φ), A(φ)] T , μ V，1 > μ V,2 > … > μ V,H > μ V,H+1 = … = μ V,N = 0, μ V,i is the i-th singular value of matrix R V , i = 1, …, N; u v,ii is the ii-th left singular vector of matrix R V , ii = 1, …, 2N; d p,i is the i-th right singular vector of matrix R V ; Λ V = [Λ V,1 , 0 N×N T , where Λ V，1 = diag(μ V,1 , …, μ V,N ); U V = [u​v,1 ,…,u v,2N and D p = [d p,1 ,…,d p,N represent 2N×2N and N×N dimensional unitary matrices respectively; U V,1 = [u v，1 ,…,u v,N = U V [I N ,0 N×N T .

[0079] Multiply both sides of the above equation by the matrix D p , we get:

[0080]

[0081] Furthermore, we can obtain:

[0082]

[0083] In the formula,

[0084] Since R s is a Hermitian matrix, that is, Therefore, the matrix is equal to:

[0085]

[0086] In the formula, R z = ZZ H , indicating that the non-zero singular values in Λ V,1 are proportional to the signal power.

[0087] Construct the matrix Furthermore, use the matrix r V to construct the cross-covariance matrix:

[0088]

[0089] In the formula,

[0090] (3) Perform singular value decomposition on the horizontally arranged cross-covariance matrix R H :

[0091]

[0092] In the formula, μ H,1 > μ H,2 > … > μ H,H > μ H,H+1 = … = μ​H,N = 0, μ H,i is the i-th singular value of matrix R H , i = 1, …, N; u H,i is the i-th left singular vector of matrix R H ; d pH,i is the ii-th right singular vector of matrix R H , ii = 1, …, 2N; Λ H = [Λ H,1 0 N×N , where Λ H,1 = diag(μ H,1 , …, μ H,N ); U H = [u H,1 , …, u H,N and D pH = [d pH,1 , …, d pH,2N represent N×N and 2N×2N dimensional unitary matrices respectively. D pH,1 = [d pH,1 , …, d pH,N = D pH [I N , 0 N×N T .

[0093] Similar to the method of the vertically arranged cross-covariance matrix, construct a matrix r H using U H,1 and Λ H = U H Λ H,1 1 / 2 , and then construct a Hermitian matrix R H through r Hp :

[0094]

[0095] (4) Respectively perform singular value decomposition and reconstruction on the cross-covariance matrix R D in R pvr and :

[0096]

[0097] Combine the reconstructed cross-covariance matrices and to form a block cross-covariance matrix

[0098]

[0099] In summary, R​Sum , R V , R H and R D The new matrix constructed using the left singular vectors and the diagonal matrix of non - zero singular values is R Vp , R Hp and

[0100] Using the covariance matrix R Vp , R Hp and Implement the spatial spectrum estimation method, which includes but is not limited to methods such as CBF, MVDR, and MUSIC methods. Here, taking the spatial spectrum of the MVDR method as an example, its expression is:

[0101]

[0102] where, a S (ψ), a V (ψ), a H (ψ), a D (ψ) are respectively R Vp , R Hp , 's steering vectors, and ψ is the scanning angle. Take the scanning angle ψ corresponding to the maximum value in the spatial spectrum as the azimuth estimation result of the detection target.

[0103] According to the above - mentioned theoretical derivation, a method for azimuth estimation of an acoustic vector circular array based on the construction of the cross - covariance matrix provided by the present invention in practical applications includes the following steps:

[0104] Step 1: Deploy an acoustic vector circular array in the fluid domain where the detection target is located. The acoustic vector circular array has M array elements evenly arranged in a circle with a radius of r. The acoustic vector circular array outputs an acoustic pressure vector p, a radial vibration velocity vector v r and a tangential vibration velocity vector

[0105] Step 2: Introduce the pre - processing matrices T p , T vr and to transform the output of the acoustic vector circular array, converting the uniform circular array into a virtual uniform linear array, and obtaining the transformed acoustic pressure vector p c , radial vibration velocity vector v rc and tangential vibration velocity vector

[0106]

[0107] p c = Tp p, v rc = T vr v r ,

[0108] Step 3: According to the transformed sound pressure vector p c and the radial vibration velocity vector v rc , construct the cross-covariance matrix of the received data of sound pressure and radial vibration velocity According to the transformed sound pressure vector p c and the tangential vibration velocity vector construct the cross-covariance matrix of the received data of sound pressure and tangential vibration velocity

[0109] Step 4: Combine and arrange the cross-covariance matrix of the received data of sound pressure and radial vibration velocity and the cross-covariance matrix of the received data of sound pressure and tangential vibration velocity to obtain a combined arrangement matrix;

[0110] The combined arrangement methods designed in the present invention include vertical combined arrangement, horizontal combined arrangement, and diagonal combined arrangement;

[0111] I. If R pvr and are vertically combined and arranged to obtain

[0112] Perform singular value decomposition on R V , and use the unitary matrix U V constituted by the left singular vectors of R V,1 and the diagonal matrix Λ V,1 constituted by the singular values to reconstruct the cross-covariance matrix

[0113] Use the cross-covariance matrix to implement the spatial spectrum estimation method.

[0114] II. If R pvr and are horizontally combined and arranged to obtain a combined arrangement matrix

[0115] Perform singular value decomposition on R H , and use the unitary matrix U H constituted by the left singular vectors of R H and the diagonal matrix Λ H,1 constituted by the singular values to reconstruct the cross-covariance matrix

[0116] Use the cross-covariance matrix to implement the spatial spectrum estimation method.

[0117] III. If Rpvr Combine diagonally with to obtain

[0118] For R pvr Combine respectively with Perform singular value decomposition on them respectively. Utilize the unitary matrix U pvr constituted by the left singular vectors of R pvr and the diagonal matrix Λ pvr to reconstruct the cross-covariance matrix Utilize the unitary matrix constituted by the left singular vectors of and the diagonal matrix to reconstruct the cross-covariance matrix Combine the reconstructed cross-covariance matrix diagonally with

[0119]

[0120] Utilize the cross-covariance matrix to implement the spatial spectrum estimation method.

[0121] Example 1:

[0122] Take the acoustic vector circular array installed on the elastic column barrier as the model. Assume the number of array elements M = 8, the radius of the circular array r = 0.35 m, the radius of the column barrier shell a = 0.25 m, the length of the shell is 2 m, the thickness of the shell is 0.002 m, the outside of the shell is water with the sound speed c = 1500 m / s, the inside is air with the sound speed c0 = 346 m / s, the density ρ0 = 1.29 kg / m 3 , the material of the shell is steel with the density ρ s = 7850 kg / m 3 , the Poisson's ratio σ = 0.3, the Young's modulus E = 2.1×10 11 Pa. The signal frequency is 2 kHz, the incident angle of the single target signal is 100°, the incident angles of the two incoherent signals are 90° and 150° respectively, and the number of Monte Carlo tests is 100. Assume the noise field is isotropic zero-mean Gaussian white noise, that is, the sound pressure and vibration velocity components of the noise received by the acoustic vector sensor are uncorrelated; the noise received by any two acoustic vector sensors is also uncorrelated. Denote the MVDR method of the conventional additive cross-covariance matrix as "S-MVDR", and denote the MVDR azimuth estimation algorithms of the longitudinal, transverse, and diagonal arranged cross-covariance matrices in the present invention as "V-MVDR", "H-MVDR", and "D-MVDR" respectively.

[0123] Figure 3Figures (a) and (b) are the spatial spectrum comparison diagrams of each method under the conditions of signal-to-noise ratio SNR = 0 dB and SNR = -8 dB, respectively. It can be found that under different signal-to-noise ratio conditions, the V-MVDR in the present invention has a lower spatial spectrum background level and a narrower main lobe width that are basically the same as those of the S-MVDR.

[0124] Figure 4 It is the curve of the double-target resolution probability of each method changing with the angular interval. It can be found that as the angular interval between two uncorrelated signals increases, the target resolution probabilities of the four methods gradually increase. The four methods of S-MVDR, V-MVDR, H-MVDR, and D-MVDR respectively satisfy the signal angular interval and when they can completely resolve two uncorrelated signals. The V-MVDR method in the present invention can achieve a target resolution success probability of 1 with a smaller double-target interval and has a higher target resolution ability.

[0125] Figure 5 It is the curve of the root mean square error (RMSE) of each method changing with the signal-to-noise ratio. Through analysis, it can be found that the signal-to-noise ratio has a relatively obvious impact on the four algorithms. Generally speaking, as the signal-to-noise ratio increases, the RMSE of the four methods all shows a significant downward trend. Among them, the V-MVDR method proposed in the present invention always has the best performance under different signal-to-noise ratio conditions. In particular, when the signal-to-noise ratio is low (SNR is lower than -14 dB), the performance of the V-MVDR method is more excellent and can provide higher estimation accuracy. When the signal-to-noise ratio is -20 dB, the RMSE of the V-MVDR method in the present invention can basically be guaranteed to be about 3°, which can be regarded as being able to effectively achieve azimuth estimation, while the RMSE of the other methods has exceeded 7° and cannot guarantee high accuracy. In summary, it can be considered that the V-MVDR method in the present invention can better be applied to the more common low signal-to-noise ratio situations in engineering and has high practical value.

[0126] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for azimuth estimation of a sound vector circular array based on the construction of a cross-covariance matrix, characterized in that Including the following steps: Step 1: Deploy an acoustic vector circular array in the fluid domain where the detection target is located. The acoustic vector circular array outputs an acoustic pressure vector, a radial vibration velocity vector, and a tangential vibration velocity vector according to the received signals. Step 2: Introduce a preprocessing matrix to transform the output of the acoustic vector circular array, converting the uniform circular array into a virtual uniform linear array, and obtaining the transformed acoustic pressure vector, radial vibration velocity vector, and tangential vibration velocity vector. Step 3: According to the transformed acoustic pressure vector and radial vibration velocity vector, construct the cross-covariance matrix of the acoustic pressure and radial vibration velocity received data; according to the transformed acoustic pressure vector and tangential vibration velocity vector, construct the cross-covariance matrix of the acoustic pressure and tangential vibration velocity received data. Step 4: Combine and arrange the cross-covariance matrix of the acoustic pressure and radial vibration velocity received data and the cross-covariance matrix of the acoustic pressure and tangential vibration velocity received data to obtain a combined arrangement matrix. Step 5: Perform singular value decomposition on the combined arrangement matrix, and reconstruct the cross-covariance matrix using the unitary matrix composed of the left singular vectors of the combined arrangement matrix and the diagonal matrix composed of the singular values. Step 6: Implement a spatial spectrum estimation method using the cross-covariance matrix, and take the scanning angle corresponding to the maximum value in the spatial spectrum as the azimuth estimation result of the detection target.

2. The azimuth estimation method of an acoustic vector circular array based on the construction of a cross-covariance matrix according to claim 1, characterized in that: In step 1, the acoustic vector circular array consists of M elements evenly arranged in a circle with a radius of r. The acoustic vector circular array outputs an acoustic pressure vector p, a radial vibration velocity vector v, r and a tangential vibration velocity vector In step 3, according to the transformed sound pressure vector p c and the radial vibration velocity vector v rc , construct the cross-covariance matrix of the received data of the sound pressure and the radial vibration velocity According to the transformed sound pressure vector p c and the tangential vibration velocity vector Construct the cross-covariance matrix of the received data of the sound pressure and the tangential vibration velocity E{·} represents the expectation operation.

3. A method for azimuth estimation of a sound vector circular array based on the construction of a cross-covariance matrix according to claim 2, characterized in that: Introduce the preprocessing matrix T in step 2 p , T vr and Transform the output of the acoustic vector circular array to convert the uniform circular array into a virtual uniform linear array, and obtain the transformed acoustic pressure vector p c , radial vibration velocity vector v rc and tangential vibration velocity vector p c = T p p, v rc = T vr v r , Where F = [ω -K ,ω -K+1 ,…,ω K ], j is the imaginary number symbol, j 2 = -1; q = -K, ..., K; B p =diag[b -K ,…,b K ], B vr =diag[b - ' K / jkρc,…,b′ K / jkρc], b q is the modal strength; b q ′ represents b q The derivative with respect to radius r; ρ is the density of the fluid medium in the fluid domain, k is the wave number of the fluid medium in the fluid domain, k = ω / c, ω is the angular velocity of the fluid medium in the fluid domain, c is the speed of sound; K is the maximum mode number.

4. A method for azimuth estimation of a sound vector circular array constructed based on a cross-covariance matrix according to claim 3, characterized in that: If the acoustic vector circular array is directly arranged in the fluid domain or installed near the surface of the acoustically transparent cylindrical barrier shell, then b q = j q J q (kr), where J q (·) represents the Bessel function of the first kind of order q; If the acoustic vector circular array is directly arranged in the fluid domain, then K = [kr]; if the acoustic vector circular array is installed on the near surface of the cylindrical barrier shell, then K = [ka], where a is the radius of the cylindrical barrier shell. The spatial spectrum estimation method in Step 6 includes CBF, MVDR, and MUSIC methods.

5. A method for azimuth estimation of a sound vector circular array based on the construction of a cross-covariance matrix according to claim 2, characterized in that: In step 4, the cross-covariance matrix R of the received sound pressure and radial vibration velocity pvr , and the cross-covariance matrix of the received sound pressure and tangential vibration velocity are arranged longitudinally in combination to obtain a combined arrangement matrix In step 5, for the combined permutation matrix R V perform singular value decomposition, and use the unitary matrix U V formed by the left singular vectors of R V,1 and the diagonal matrix Λ V,1 formed by the singular values to reconstruct the cross-covariance matrix Among them, the combined permutation matrix R V has N singular values, 2N left singular vectors, and N right singular vectors; U V,1 is the unitary matrix formed by the first N left singular vectors of R V ; Λ V,1 is the diagonal matrix formed by the N singular values of R V ; D p is the unitary matrix formed by the N right singular vectors of R V . Let Reconstruct the cross-covariance matrix In step 6, the cross-covariance matrix is used to implement the spatial spectrum estimation method.

6. A method for azimuth estimation of a sound vector circular array based on the construction of a cross-covariance matrix according to claim 2, characterized in that: In step 4, the cross-covariance matrix R of the received sound pressure and radial vibration velocity pvr , and the cross-covariance matrix of the received sound pressure and tangential vibration velocity are arranged horizontally in combination to obtain a combined arrangement matrix In step 5, the combined permutation matrix R H is subjected to singular value decomposition, and the unitary matrix U H formed by the left singular vectors of R H and the diagonal matrix Λ H,1 formed by the singular values are used to reconstruct the cross-covariance matrix Among them, the combined permutation matrix R H has N singular values, N left singular vectors, and 2N right singular vectors; U H is the unitary matrix composed of the N left singular vectors of R H ; Λ H,1 is the diagonal matrix composed of the N singular values of R H ; D pH,1 is the unitary matrix composed of the first N right singular vectors of R H . Let Reconstruct the cross-covariance matrix In step 6, the cross-covariance matrix is used to implement the spatial spectrum estimation method.

7. A method for azimuth estimation of a sound vector circular array constructed based on a cross-covariance matrix according to claim 2, characterized in that: In step 4, the cross-covariance matrix R of the received sound pressure and radial vibration velocity pvr , the cross-covariance matrix of the received sound pressure and tangential vibration velocity are arranged in a diagonal combination to obtain a combined permutation matrix In the said step 5, for the combined permutation matrix R D where R pvr and are respectively subjected to singular value decomposition, and the unitary matrix U pvr constituted by the left singular vectors of R pvr and the diagonal matrix Λ pvr constituted by the singular values are used to reconstruct the cross-covariance matrix Using the unitary matrix constituted by the left singular vectors of and the diagonal matrix The reconstructed cross-covariance matrix and are arranged in a diagonal combination to reconstruct the cross-covariance matrix In step 6, the cross-covariance matrix is used to implement the spatial spectrum estimation method.

8. A computer device / equipment / system, comprising a memory, a processor, and a computer program stored on the memory, characterized in that: The processor executes the computer program to implement the steps of the method described in any one of claims 1 to 7.

9. A computer-readable storage medium having computer programs / instructions stored thereon, characterized in that: When the computer program / instructions are executed by the processor, the steps of the method described in any one of claims 1 to 7 are implemented.

10. A computer program product, comprising a computer program / instructions, characterized in that: When the computer program / instructions are executed by the processor, the steps of the method described in any one of claims 1 to 7 are implemented.

Images