A target detection method, device, medium and equipment for under-ice environment

Abstract

The present invention discloses a target detection method, device, medium and equipment for an under-ice environment, which relates to the field of target detection, in order to solve the problem of passive detection of weak targets in complex under-ice sound field environments such as polar regions. The target detection method comprises: performing sound pressure and velocity joint processing on the received signal of a vector circular array receiver to obtain the sound pressure velocity cross-covariance matrix at each frequency point; transforming the sound pressure velocity cross-covariance matrix into the sound pressure velocity cross-covariance matrix of a virtual linear array receiver in the modal domain; performing Toeplitz reconstruction and eigendecomposition according to the sound pressure velocity cross-covariance matrix of the virtual linear array receiver to obtain noise subspace signals corresponding to different scanning orientations; obtaining the MUSIC spatial spectrum corresponding to the received signal using the MUSIC algorithm according to the noise subspace signal, and determining the target orientation based on the spectrum peak of the MUSIC spatial spectrum. The target detection method provided by the present invention realizes passive detection of weak targets under ice.

Description

Technical Field

[0001] The present invention relates to the field of target detection technology, and in particular to a target detection method, device, medium and equipment for an under-ice environment. Background Art

[0002] In the sub-ice environment, especially in the sub-ice environment of the polar waters, the sound velocity distribution shows the characteristics of a dual-channel sound velocity distribution. The dual-channel waveguide effect limits the propagation of ice-source noise to the seabed, making the noise near the seabed Gaussian distributed, which has obvious advantages in receiving signals on the seabed. First of all, we only need to consider the case where the noise is Gaussian noise, which makes signal processing easier. Secondly, the array near the seabed can achieve blind spot detection within a certain distance. However, this seabed array method also has limitations. When the target is far away, due to the large loss of sound wave propagation, the signal-to-noise ratio of the seabed received signal is low. When the traditional azimuth estimation method is used, the weak target signal will be submerged in the background noise, and the weak target cannot be directionally measured. Therefore, passive detection of long-distance weak targets is a problem that needs to be solved urgently.

[0003] Traditional conventional beamforming (CBF) achieves the purpose of canceling out the noise of the signals added in the same direction by compensating the phase difference of the signals received by different array elements. This is equivalent to spatial filtering, which improves the signal-to-noise ratio at the output. This beamforming method can also be applied to target azimuth estimation. By designing different beam directions and traversing all spatial azimuths, the azimuth-power spectrum, i.e., the spatial spectrum, can be calculated. First, the autocorrelation matrix of the received array element domain signal x(t) is calculated to obtain the received signal cross-covariance matrix R x , and then calculate the weighted vector w of the mth array element at different orientations m (θ), the expression is as follows m (θ) = e j2πRcos(2πm / M-θ) / λ , where R represents the radius of the circular array, M represents the number of array elements, θ represents the scanning direction that needs to be traversed in all directions during calculation, and λ represents the wavelength. The conventional beamforming spatial spectrum is obtained by traversing all directions θ. The direction of the spectrum peak maximum is the direction of the target. Conventional beamforming methods are relatively robust, but when the number of array elements is small, they have defects such as wide main lobe, low spatial resolution, and high side lobe, which are not ideal when performing direction finding for weak targets.

[0004] The disadvantages of conventional beamforming methods are wide main lobe, low spatial resolution, high side lobes and background, which can easily drown out weak targets. The reason is that the spatial resolution of conventional beamforming methods depends on the array aperture, which is limited by the number of array elements and the wavelength corresponding to the highest frequency of the received signal. According to the Nyquist sampling theorem, it can be deduced that spatial sampling needs to satisfy the requirement that the spacing between adjacent array elements does not exceed half the wavelength corresponding to the highest frequency of the signal, otherwise aliasing will occur in the spatial spectrum. In addition, the number of array elements cannot be infinite in actual engineering, so the array aperture is always limited, which leads to low resolution of conventional beamforming, and the limited number of array elements will lead to high side lobes and background, and weak targets are easily drowned.

[0005] The MUSIC (Multiple Signal Classification) algorithm performs eigendecomposition on the cross-covariance matrix of the received data to obtain the noise subspace and uses the orthogonality of the noise subspace and the signal subspace to achieve direction finding. The cross-covariance matrix of the received signal can be eigendecomposed to obtain K (K is the number of signal sources) large eigenvalues ​​and MK (M is the number of array elements) small eigenvalues. The eigenvectors corresponding to the K large eigenvalues ​​constitute the signal subspace, and the eigenvectors corresponding to the MK small eigenvalues ​​constitute the noise subspace. The eigenvectors corresponding to the K large eigenvalues ​​are in the same subspace as the signal incident steering vector a(θ), so the signal steering vector and the noise subspace E n It is also orthogonal, using The MUSIC spatial spectrum can be calculated, and the maximum value of the spectrum peak is the direction of the target. The MUSIC algorithm breaks through the limitation of array aperture and can achieve high-resolution direction finding, but the algorithm performance is seriously degraded under low signal-to-noise ratio and small snapshot conditions, and is not suitable for direction finding of weak targets.

[0006] The disadvantage of the MUSIC algorithm is that the performance of the algorithm decreases under low signal-to-noise ratio and small block conditions, and the error in direction finding for weak targets is large. The reasons are as follows: the performance of the MUSIC algorithm decreases under small snapshots and low signal-to-noise ratio. When the number of snapshots is small, there will be a large error between the calculated signal covariance and the ideal signal autocorrelation matrix, which will cause the algorithm performance to decrease. When the signal-to-noise ratio is low, the division of the signal subspace and the noise subspace will be blurred during the eigendecomposition of the signal cross-covariance matrix, which will cause the obtained noise subspace and signal subspace to not meet strict orthogonality. When calculating the MUSIC spatial spectrum, the peak value is low due to the weak orthogonality between the signal steering vector and the noise subspace, and the contrast with the background is not obvious. At this time, the MUSIC algorithm cannot measure the direction of the weak target for direction estimation.

[0007] The widespread ice-source noise interference in the polar sub-ice sound field brings difficulties to passive target detection. The traditional target azimuth estimation method is no longer applicable under ice-source noise. In order to achieve passive detection of sub-ice targets, the passive sonar array is placed near the seabed away from the ice layer to avoid the adverse effects of ice-source noise on passive detection. However, this array arrangement faces the problem of low signal-to-noise ratio caused by the long propagation distance of the target source radiation sound wave. In summary, this application is committed to solving the problem of low signal-to-noise ratio detection in sub-ice environments, especially the problem of weak target detection. Summary of the invention

[0008] The purpose of the present invention is to address the problem that passive target detection in complex acoustic field environments under ice often cannot be detected, and propose a target detection technical solution for an under-ice environment, in particular, a method for passive detection of weak targets under ice. By giving full play to the advantages of the array, the vector array and the circular array are combined into a vector circular array, and then the weak target azimuth estimation method of the adapted vector circular array is combined to achieve passive detection of weak targets under ice. In an under-ice environment, the use of a vector circular array for passive target detection solves the problem of difficulty in detecting weak targets with low signal-to-noise ratio.

[0009] In order to achieve the above object, the present invention provides the following technical solutions:

[0010] A target detection method for an under-ice environment, comprising:

[0011] Perform sound pressure and velocity joint processing on the received signal of the vector circular array receiver to obtain the cross-covariance matrix of sound pressure and velocity at each frequency point;

[0012] The cross-covariance matrix of the sound pressure and velocity at each frequency point at different scanning positions is transformed into the cross-covariance matrix of the sound pressure and velocity of the virtual linear array receiver in the modal domain;

[0013] Toeplitz reconstruction and eigendecomposition are performed based on the sound pressure velocity cross-covariance matrix of the virtual linear array receiver to obtain the noise subspace signals corresponding to different scanning directions.

[0014] Obtaining a MUSIC spatial spectrum corresponding to the received signal by using a MUSIC algorithm according to the noise subspace signal;

[0015] The target position is determined according to the spectrum peak of the MUSIC spatial spectrum.

[0016] Compared with the prior art, the target detection method for under-ice environment provided by the present invention provides a weak target azimuth estimation algorithm for polar under-ice environment, which organically combines Toeplitz reconstruction technology, circular array modal domain transformation technology, vector array sound pressure and velocity joint processing method and MUSIC method on the basis of vector circular array, and can realize accurate azimuth estimation of long-distance weak targets under ice.

[0017] The Toeplitz matrix reconstruction method can perform Toeplitz reconstruction on the cross-covariance matrix of the received signal, and then combined with the MUSIC algorithm, it can realize target direction estimation under low signal-to-noise ratio. However, the Toeplitz reconstruction method is only applicable to linear arrays whose array manifold satisfies the Vandermonde matrix form. For circular arrays, it is necessary to first perform a circular array modal domain transformation, transform the circular array into a "virtual linear array", and then perform Toeplitz processing on the cross-covariance matrix.

[0018] The sound pressure and vibration velocity joint processing method of the vector array can better suppress noise and improve the signal-to-noise ratio. The x and y channel vibration velocities are linearly combined to construct the combined vibration velocity Vc and the sound pressure channel is combined to calculate the sound pressure and vibration velocity cross-covariance matrix. The irrelevance of the noise between the vibration velocity channel and the sound pressure channel is used to suppress the isotropic noise. The artificial selection of the observation direction of the vibration velocity Vc will lead to azimuth estimation errors. The present invention combines the observation direction of the vibration velocity Vc with the array scanning direction to avoid the azimuth estimation error caused by the artificial selection of the Vc observation direction.

[0019] Through the joint processing of sound pressure and vibration velocity and Toeplitz reconstruction based on the modal domain, the direction estimation of weak targets with low signal-to-noise ratio under ice can be achieved.

[0020] The present invention also provides a target detection device for an under-ice environment, comprising:

[0021] A joint processing unit is configured to perform sound pressure and vibration velocity joint processing on the received signal of the vector circular array receiver to obtain a sound pressure and vibration velocity cross-covariance matrix at each frequency point;

[0022] A transformation unit is configured to transform the sound pressure velocity cross-covariance matrix of each frequency point at different scanning directions into the sound pressure velocity cross-covariance matrix of the virtual linear array receiver in the modal domain;

[0023] A reconstruction unit is configured to perform Toeplitz reconstruction and eigendecomposition according to the sound pressure velocity cross-covariance matrix of the virtual linear array receiver to obtain noise subspace signals corresponding to different scanning orientations;

[0024] A MUSIC processing unit is configured to obtain a MUSIC spatial spectrum corresponding to the received signal using a MUSIC algorithm according to the noise subspace signal;

[0025] The direction identification unit is configured to determine the target direction according to the spectrum peak of the MUSIC spatial spectrum.

[0026] Compared with the prior art, the beneficial effects of the target detection device for an under-ice environment provided by the present invention are the same as the beneficial effects of the target detection method for an under-ice environment described in the above technical solution, and will not be elaborated here.

[0027] The present invention also provides a target detection device for an under-ice environment, comprising: a vector circular array receiver for receiving signals and the target detection device as described above.

[0028] The present invention also provides a computer storage medium, in which instructions are stored. When the instructions are executed, the target detection method for an under-ice environment as described above is implemented.

[0029] The present invention also provides a computing device, including a processor and a communication interface coupled to the processor; the processor is used to run a computer program or instruction to implement the target detection method for an under-ice environment as described above. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0031] Figure 1 Schematic diagram of a receiving signal model of a vector circular array receiver in an embodiment of the present invention.

[0032] Figure 2 Schematic diagram of the flow of a target detection method for an under-ice environment in an embodiment of the present invention.

[0033] Figure 3 Schematic diagram of a module of a target detection device for an under-ice environment according to an embodiment of the present invention.

[0034] Figure 4 It is a schematic diagram of the process from receiving a signal from a vector circular array to outputting a MUSIC spatial spectrum in an embodiment of the present invention.

[0035] Figure 5 Schematic diagram of the output spatial spectrum comparison between the target detection scheme in the embodiment of the present invention and various existing target detection algorithms.

[0036] Figure 6 Schematic diagram of the comparison of the root mean square error of the azimuth estimation of each algorithm in the embodiment of the present invention.

[0037] Figure 7 Schematic diagram of average background level comparison of azimuth estimation using various algorithms in an embodiment of the present invention.

[0038] Figure 8 This is a schematic diagram of the spatial spectrum comparison results obtained when the observation direction of the combined vibration velocity Vc is fixed and when the observation direction of the combined vibration velocity Vc matches the scanning direction in an embodiment of the present invention. DETAILED DESCRIPTION

[0039] In order to clearly describe the technical solutions of the embodiments of the present invention, in the embodiments of the present invention, words such as "first" and "second" are used to distinguish the same items or similar items with basically the same functions and effects. For example, the first threshold and the second threshold are only used to distinguish different thresholds, and their order is not limited. Those skilled in the art can understand that words such as "first" and "second" do not limit the quantity and execution order, and words such as "first" and "second" do not necessarily limit them to be different.

[0040] It should be noted that, in the present invention, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0041] In the present invention, "at least one" means one or more, and "plurality" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, the combination of a and b, the combination of a and c, the combination of b and c, or the combination of a, b and c, where a, b, c can be single or multiple.

[0042] In the field of under-ice target detection technology, vector hydrophones have sound pressure and multiple vibration velocity channels compared to pressure hydrophones, and can obtain more target information. By utilizing the difference in spatial correlation between sound pressure and particle vibration velocity in the target radiation noise field and the ambient noise field, higher spatial gain and stronger noise suppression can be obtained. Circular arrays have similar azimuth estimation accuracy and spatial resolution in all directions compared to linear arrays, and are simpler to deploy in practice. The spatial resolution of uniform circular arrays is very similar within the scanning range of 0° to 360°, but the spatial resolution of uniform linear arrays is only high within the scanning range of 0° to 180°, and the closer to 90°, the higher the spatial resolution, and the closer to 0° or 180°, the lower the spatial resolution.

[0043] The vector circular array receiver includes M array elements, wherein the M array elements are vector array elements, and together they form a circular vector circular array receiver. The vector circular array receiver is a vector circular array passive sonar array. The appearance of grating lobes in the spatial spectrum will have an adverse effect on azimuth estimation. In order to avoid the appearance of grating lobes, when deploying the passive sonar array, the spacing between two adjacent array elements is required to be less than half of the wavelength corresponding to the highest frequency of the received signal. Based on this, the relationship between the number of array elements M of the circular array, the radius R of the circular array, and the wavelength λ corresponding to the highest frequency of the received signal can be deduced, and it is required to satisfy

[0044] For uniform vector circular array, vector is relative to scalar, scalar array is sound pressure array; circular array is the array formation, there are many array formations, including linear array, arc array, cross array, etc. Uniform means that the array elements are equally spaced, and the opposite are non-uniform array, sparse array, etc.

[0045] like Figure 1 As shown, the receiving signal model of the vector circular array receiver is given. The vector circular array with M array elements is located in the xOy plane, with a radius of R and the positive direction of the x-axis as the 0° direction. The vector circular array is an array composed of many array elements, and a single array element is called a vector hydrophone or an acoustic vector sensor. Vector hydrophones include acoustic pressure hydrophones and / or vibration velocity hydrophones, for example, including one acoustic pressure hydrophone, two or three vibration velocity hydrophones. The element for obtaining the sound pressure is the acoustic pressure hydrophone, and the elements for obtaining the x and y vibration velocities are the x vibration velocity hydrophone and the y vibration velocity hydrophone, respectively. Vector sensors include acoustic pressure sensors (i.e., acoustic pressure hydrophones) and vibration velocity sensors. Vibration velocity sensors include displacement sensors, velocity sensors, acceleration sensors, etc. Although the principles are different and the functions are similar, they can all be called vector hydrophones.

[0046] The channel directions of the velocity components x and y of the vector circular array are consistent with the x-axis and y-axis directions respectively. S represents the incident signal source. In the passive sonar detection mode, the signal source is mainly the noise signal radiated by the target ship, which is a broadband signal with the main frequency components concentrated below 1kHz; the three symbols x, y, z, and O constitute a spatial rectangular coordinate system, which is used here to describe a three-dimensional space; θ represents the angle between the straight line where the projection of the incident direction of the signal source S on the xOy plane is located and the positive direction of the x-axis. The angle is determined by rotating counterclockwise from the positive direction of the x-axis to the projection direction, ranging from 0° to 360°; Indicates the angle between the incident direction of the signal source S and the positive direction of the z-axis, ranging from 0° to 180°.

[0047] The scanning direction refers to the scanning azimuth or scanning direction angle of the vector circular array, while the observation direction refers to the direction of the combined vibration velocity Vc. The observation direction is essentially the electronic rotation direction of the vector hydrophone. The x-channel vibration velocity and the y-channel vibration velocity of the vector hydrophone can be linearly combined through the sine and cosine values ​​of the electronic rotation angle to obtain the combined vibration velocity Vc. In this way, the combined vibration velocity Vc has a direction. The direction angle is essentially the electronic rotation angle of the vector hydrophone, which is uniformly defined as the "observation direction" and represented by the symbol φ. In this application, the observation direction is creatively kept consistent with the scanning direction, and changes with the change of the scanning direction.

[0048] like Figure 2 As shown, a schematic diagram of the target detection method for an under-ice environment is given, including the following steps:

[0049] Step S1, performing sound pressure and vibration velocity joint processing on the received signal of the vector circular array receiver to obtain the sound pressure and vibration velocity cross-covariance matrix at each frequency point;

[0050] Step S2, transforming the sound pressure velocity cross-covariance matrix of each frequency point at different scanning directions into the sound pressure velocity cross-covariance matrix of the virtual linear array receiver in the modal domain;

[0051] Step S3, performing Toeplitz reconstruction and eigendecomposition according to the sound pressure velocity cross-covariance matrix of the virtual linear array receiver to obtain noise subspace signals corresponding to different scanning orientations;

[0052] Step S4, obtaining a MUSIC spatial spectrum corresponding to the received signal by using a MUSIC algorithm according to the noise subspace signal;

[0053] Step S5: determining the target orientation according to the spectrum peak of the MUSIC spatial spectrum.

[0054] The step of performing sound pressure and vibration velocity joint processing on the received signal of the vector circular array receiver comprises:

[0055] Construct a broadband receiving signal model based on vector circular array under ice to obtain the sound pressure receiving signal and x, y vibration velocity receiving signals;

[0056] The vibration velocity signal in the x direction and the vibration velocity signal in the y direction are combined to obtain a combined vibration velocity signal;

[0057] The combined vibration velocity signal and the sound pressure signal at each frequency point are cross-correlated to obtain the cross-covariance matrix of the sound pressure and vibration velocity at each frequency point;

[0058] The combined vibration velocity observation direction φ is traversed in all directions in the range of 0 to 360° to obtain the cross-covariance matrix of the sound pressure and combined vibration velocity in each scanning direction at different frequencies.

[0059] The combined vibration velocity observation direction φ is consistent with the scanning direction of the vector circular array.

[0060] The method of transforming the sound pressure velocity cross-covariance matrix of each frequency point at different scanning positions into the sound pressure velocity cross-covariance matrix of the virtual linear array receiver in the modal domain includes:

[0061] Perform Fourier transform processing on the received signal of the vector circular array;

[0062] According to the signal after the spatial discrete Fourier transform, the modal transformation matrix is ​​determined;

[0063] Based on the modal transformation matrix, the sound pressure velocity cross-covariance matrix at different frequency points and scanning directions is transformed in the modal domain, and the sound pressure velocity cross-covariance matrix of the virtual linear array receiver obtained by the modal domain transformation is output respectively.

[0064] The Toeplitz reconstruction and eigendecomposition are performed according to the sound pressure velocity cross-covariance matrix of the virtual linear array receiver, including:

[0065] The cross-covariance matrix of the sound pressure and velocity of the virtual linear array receiver is averaged over time to obtain the cross-covariance matrix of the sound pressure and velocity in each scanning direction at different frequencies.

[0066] The cross-covariance matrices of different frequency points at each scanning direction are frequency-averaged in turn to obtain a frequency-averaged cross-covariance matrix;

[0067] Perform Toeplitz reconstruction on the frequency-averaged cross-covariance matrix;

[0068] The cross-covariance matrix obtained by Toeplitz reconstruction is subjected to eigendecomposition to obtain the noise subspace corresponding to different scanning orientations.

[0069] like Figure 3 As shown, a schematic diagram of a target detection device 100 for an under-ice environment is provided, including:

[0070] The joint processing unit 101 is configured to perform sound pressure and velocity joint processing on the received signal of the vector circular array receiver to obtain the cross-covariance matrix of sound pressure and velocity at each frequency point;

[0071] The transformation unit 102 is configured to transform the sound pressure velocity cross-covariance matrix of each frequency point at different scanning directions into the sound pressure velocity cross-covariance matrix of the virtual linear array receiver in the modal domain;

[0072] The reconstruction unit 103 is configured to perform Toeplitz reconstruction and eigendecomposition according to the sound pressure velocity cross-covariance matrix of the virtual linear array receiver to obtain noise subspace signals corresponding to different scanning orientations;

[0073] A MUSIC processing unit 104 is configured to obtain a MUSIC spatial spectrum corresponding to the received signal using a MUSIC algorithm according to the noise subspace signal;

[0074] The direction identification unit 105 is configured to determine the target direction according to the spectrum peak of the MUSIC spatial spectrum.

[0075] The wood strip detection technical solution of the present application is described in detail below with reference to the accompanying drawings.

[0076] In step S1, the received signal of the vector circular array receiver is subjected to joint processing of sound pressure and velocity to obtain the cross-covariance matrix of sound pressure and velocity at each frequency point. Step S1 specifically includes:

[0077] S11. Construct a broadband receiving signal model based on vector circular array under ice to obtain the sound pressure receiving signal P(f j ) and x, y vibration speed receiving signal V x (f j ), V y (f j );

[0078] In the passive sonar detection mode, the signal source is the broadband signal of the target ship radiation noise. To estimate the direction of the broadband signal, the signal must be segmented in the time domain to increase the number of snapshots in the frequency domain (the number of time domain segments corresponds to the number of frequency domain snapshots). Too few snapshots will lead to poor performance of MUSIC methods.

[0079] Assume that H uncorrelated far-field broadband signals are incident on the vector circular array, and the received broadband signal is divided into L segments. Each broadband signal is subjected to a discrete Fourier transform at J points (J is the number of time domain snapshots of each signal segment), where the frequency f j The received signal at is expressed as:

[0080]

[0081] In formula 1, j = 1, 2, ..., J, P (f j ) represents the frequency f j The sound pressure signal received at x (f j ) and V y (f j ) represent the frequency f j The received vibration velocity signals of the x and y channels, the sound pressure signal and the vibration velocity signal together constitute the received signal of the vector circular array. A(f j , θ) represents the frequency f j Steering vector of the lower sound pressure channel, N p (f j ) represents the sound pressure signal of the noise, S(f j) represents the frequency f j The target source radiated noise signal under x (f j ) and N y (f j ) represent the noise velocity signals of x and y channels respectively, and θ represents the direction of the incident signal.

[0082] S12, the vibration speed signal V in the x direction x (f j ) and the vibration velocity signal V in the y direction y (f j ) to obtain the combined vibration speed signal V c (f j )=V x (f j )cosφ+V y (f j )sinφ, where φ is the combined velocity observation direction. The combined velocity observation direction φ is consistent with the scanning direction of the vector circular array to avoid detection errors caused by the preset φ direction. If the combined velocity observation direction φ adopts a preset value, the preset value deviates from the true target direction, which will cause direction finding errors.

[0083] S13, the combined vibration speed signal V at each frequency point c (f j ) and the sound pressure signal P(f j ) to perform cross-correlation processing to obtain the cross-covariance matrix of sound pressure and vibration velocity at each frequency point

[0084] The sound pressure signal and vibration velocity signal of the far-field target radiated noise are coherent. Signal coherence means that the two signals meet the following conditions in the encounter area: ① the same vibration direction; ② the same vibration frequency; ③ the same phase or the phase difference remains constant.

[0085] For an isotropic noise field, the sound pressure signal and vibration velocity signal of the noise measured at the same point are uncorrelated. The same point synchronous measurement means that the two signals meet at the measurement point, and the two signals arrive at the measurement point at the same time and at the same position. (The measurement point can be arbitrary). The sound pressure and vibration velocity of the noise signal measured at different array elements are also uncorrelated.

[0086] The vector hydrophone is a single element of the vector circular array. One set of vector hydrophones includes one pressure hydrophone and two to three velocity hydrophones. By utilizing the uncorrelation between the sound pressure and velocity of the noise signal of the vector hydrophone in the isotropic noise field, the joint information processing of pressure and velocity is performed to suppress the isotropic noise and give full play to the detection performance of the vector array.

[0087] S14, traverse the combined vibration velocity observation direction φ in the range of 0 to 360° in all directions to obtain the cross-covariance matrix of the sound pressure and combined vibration velocity in each scanning direction at different frequencies. It is expressed as:

[0088]

[0089] Where E(·) means taking the mean of the matrix, (P+Vc) means the sum of the data matrix P of the sound pressure channels of all array elements and the calculated combined velocity data matrix Vc of all array elements, (·) H Represents the conjugate transposed matrix of the computation matrix. It is M rows and M columns, where M is the number of array elements; P represents the matrix composed of the sound pressure channel data of all array elements, assuming it is M rows and N columns (N represents the number of snapshots), Vc represents the matrix composed of the vibration velocity data of all array elements, which is also M rows and N columns. Multiplying each column in (P+Vc) and each row in (P+Vc)H can obtain an M×M matrix, and a total of N M×M matrices are obtained. The average of these N matrices is calculated as

[0090] Each segment of the received signal divided into L segments is processed according to steps S12-S13 to obtain the cross-covariance matrix of sound pressure and vibration velocity at each frequency point

[0091] In step S2, the sound pressure velocity cross-covariance matrix of each frequency point at different scanning positions is transformed into a modal domain to obtain the sound pressure velocity cross-covariance matrix of the virtual linear array receiver.

[0092] To calculate the cross-covariance matrix of sound pressure and velocity, we first need to obtain the combined velocity Vc. The combined velocity Vc has an observation direction. Artificially selecting the observation direction will lead to azimuth estimation errors. Keeping the observation direction consistent with the scanning azimuth of the array can avoid this error, but this process comes at the cost of increasing the amount of calculation. Therefore, it should be pointed out that here we calculate the cross-covariance matrix of sound pressure and velocity "under different scanning azimuths". For example, the scanning azimuth changes from 0° to 360°, with an interval of 1°, so here we need to calculate the cross-covariance matrix of sound pressure and velocity corresponding to every 1° in the range of 0° to 360°.

[0093] When there are H far-field incident sources incident on the vector circular array from different directions, the receiving signal of the mth element of the vector circular array with M receiving elements is expressed as:

[0094]

[0095] In formula 2, j represents the imaginary unit, x m represents the received signal of the mth array element, θh represents the incident angle of the hth signal, s h Represents the source signal of the hth incident signal.

[0096] Step S2 specifically includes:

[0097] S21, perform spatial discrete Fourier transform processing on the received signal of the mth array element of the vector circular array to obtain a signal v after Fourier transform processing i , expressed as:

[0098]

[0099] In the above formula, j represents an imaginary number, h = 1, 2, ..., H, v i is x m The signal after spatial discrete Fourier transform, J -i (·) represents the -ith order Bessel function, and K is a constant.

[0100] S22. Determine a modal transformation matrix according to the signal after spatial discrete Fourier transform.

[0101] When the number of array elements M meets the conditions: hour, ([·] indicates rounding).

[0102] At this time, Formula 3 is simplified to:

[0103]

[0104] Let u i =v i , rewrite the above formula into the following matrix form:

[0105]

[0106] Formula 5 can be simplified as It is the array manifold of the “virtual linear array” obtained after the modal domain transformation.

[0107] According to the spatial discrete Fourier transform formula 3, we can know that:

[0108]

[0109] make

[0110] Then formula 6 can be simplified to:

[0111]

[0112] Formula 8 can be simplified as u=F H x, combined with the above available:

[0113]

[0114] In formula 9, x is the received signal of the vector circular array with M array elements. Then the modal transformation matrix T is expressed as:

[0115]

[0116] Where M is the number of elements of the vector circular array, J and F both represent matrices, and the calculation is as follows:

[0117]

[0118]

[0119] In the above formula, j represents an imaginary number, J -K (·) represents the K-th order Bessel function, and R represents the radius of the vector circle array.

[0120] Then the output signal of the vector circular array can be transformed into a "virtual linear array" output signal:

[0121]

[0122] in

[0123]

[0124] Formula 12 represents the array manifold of the “virtual linear array” obtained after the modal domain transformation.

[0125] S23. Based on the modal transformation matrix, the cross-covariance matrix of sound pressure and vibration velocity at different frequency points and scanning directions is transformed into the modal domain, and the cross-covariance matrix after the modal domain transformation is obtained as follows:

[0126]

[0127] Among them, f j represents the corresponding frequency, and θ represents the scanning direction.

[0128] The cross-covariance matrix after modal domain transformation is:

[0129]

[0130] That is the output signal of the “virtual linear array”. The cross-covariance matrix of the sound pressure and vibration velocity after the modal transformation becomes R e (f j ,θ).

[0131] In step S3, Toeplitz reconstruction and eigendecomposition are performed based on the sound pressure velocity cross-covariance matrix of the virtual linear array receiver to obtain noise subspace signals corresponding to different scanning directions. Step S3 specifically includes:

[0132] S31, perform time averaging on the sound pressure velocity cross-covariance matrix of the virtual linear array receiver to obtain the processed cross-covariance matrix R eL (f j ,θ); Specifically: perform time averaging on the modal domain transformation of the sound pressure and vibration velocity cross-covariance matrix at different frequencies and scanning directions to obtain the processed cross-covariance matrix R eL (f j ,θ);

[0133] S32, respectively calculate the cross-covariance matrix of the L-segment signal after modal domain transformation and perform time averaging to obtain the cross-covariance matrix R of each scanning direction at different frequencies eL (f j ,θ);

[0134] S33, the cross-covariance matrix R of different frequency points at each scanning direction eL (f j ,θ) perform frequency averaging in sequence;

[0135] The cross-covariance matrix R of different frequencies at each scanning position after the above modal transformation and time averaging processing is eL (f j ,θ) performs the following frequency averaging

[0136]

[0137] In the above formula, Indicates the upper frequency limit of the processing frequency band f h The corresponding frequency points, [·] indicates rounding. s represents the sampling rate, J represents the number of discrete Fourier points, Indicates the lower limit frequency f of the processing frequency band l The corresponding frequency points, R eLf (θ) represents the cross-covariance matrix at different scanning positions after averaging different frequencies.

[0138] S34, performing Toeplitz reconstruction on the frequency-averaged cross-covariance matrix;

[0139] Under ideal conditions, the cross-covariance matrix of the uniform linear array received data is a Toeplitz matrix. The ideal cross-covariance matrix of the virtual linear array obtained after the circular array is transformed in the modal domain should also conform to the Toeplitz matrix form. The matrix R eLf(θ) is obviously not in Toeplitz matrix form in the case of low signal-to-noise ratio, and Toeplitz reconstruction is required. The ideal Toeplitz matrix R0 satisfies the following form

[0140]

[0141] The cross-covariance matrix R eLf The main diagonal of (θ) and the elements parallel to the main diagonal are Toeplitz processed. The processed matrix is The process of Toeplitz reconstruction is expressed as follows:

[0142]

[0143] Where M is the number of array elements, k = mn. mn and Represents R and The m-th row and n-th column element of .

[0144] S35. Cross-covariance matrix obtained by Toeplitz reconstruction Perform feature decomposition to obtain the noise subspace E corresponding to different scanning orientations n (θ).

[0145] The cross-covariance matrix after Toeplitz reconstruction Perform eigendecomposition and select the eigenvector corresponding to the noise eigenvalue to obtain the noise subspace E corresponding to different scanning directions n (θ).

[0146] The eigenvalues ​​obtained after eigendecomposition are sorted by size, and the corresponding number of source eigenvalues ​​are selected for the source in order from large to small, and the remaining eigenvalues ​​are used as noise eigenvalues. The total number of eigenvalues ​​is equal to the number of array elements, and the number of source eigenvalues ​​is consistent with the number of sources. For example, if the number of array elements is 8 and the number of sources (targets) is 2, the eigenvalues ​​are sorted from large to small, and the first two are selected in order as source eigenvalues, and the remaining 6 eigenvalues ​​are used as noise eigenvalues. The noise subspace corresponding to different scanning directions can be obtained according to the eigenvectors corresponding to the selected noise eigenvalues.

[0147] In step S4, the noise subspace E obtained in S3 at different scanning directions is n (θ) is substituted into the calculation expression of the output power of the MUSIC algorithm:

[0148]

[0149] Among them, a p (θ) represents the scanning vector of the sound pressure array, E n(θ) represents the noise subspace under different scanning orientations.

[0150] Normalize the MUSIC output power of all scanning directions and take the logarithm to obtain the spatial spectrum P MUSIC-out (θ) is:

[0151]

[0152] In the above formula, P MUSIC-out (θ) represents the output spatial spectrum, and max(·) represents the maximum value.

[0153] In step S5, after obtaining the spatial spectrum output by MUSIC, the target orientation is determined based on the spectrum peak. For example, Figure 5 The two spectrum peaks appearing at -30° and 60° are the directions of the two targets.

[0154] like Figure 4 As shown, a flow chart from receiving signals from a vector circular array to outputting a MUSIC spatial spectrum is given. The signals received from the vector circular array are sequentially segmented in the time domain; the received signals in each time domain segment are subjected to a fast Fourier transform (FFT); the sound pressure velocity cross-covariance matrix is ​​calculated at the frequency division point and scanning azimuth; the sound pressure velocity cross-covariance matrix is ​​modally transformed; the cross-covariance matrix of each time period of the frequency division point and scanning azimuth is time averaged, and the cross-covariance matrix of different frequency points in the scanning azimuth is frequency averaged; the frequency-averaged cross-covariance matrix is ​​subjected to Toeplitz reconstruction and eigendecomposition to obtain a noise subspace; the MUSIC spatial spectrum is obtained based on the noise subspace using the MUSIC algorithm; the target azimuth is determined based on the spatial spectrum (the target azimuth represents the horizontal angle of incidence).

[0155] like Figure 5 As shown, a schematic diagram of the output spatial spectrum comparison between the target detection scheme of the present application and various existing target detection algorithms is given. In the simulation process, a vector circular array with a radius of 0.75 meters including 16 (M=16) array elements is set, and independent dual-target broadband signals are incident from -30° and 60° directions, with in-band signal-to-noise ratios of -10dB and -15dB respectively, and a signal duration of 1s. The time domain signal is divided into 100 (L=100) segments for processing, with a sampling frequency of 32kHz, a FFT point number of 320, a scanning azimuth range of 0 to 360°, and an angle interval of 1°.

[0156] The target detection method of this application is marked as Toeplitz reconstruction modal domain (P+V c )*(P+V c ) H MUSIC, the algorithms used for comparison include array element domain sound pressure MUSIC, array element domain vector independent channel MUSIC, array element domain (P+V c)*(P+V c ) H MUSIC, modal domain (P+V c )*(P+V c ) H MUSIC.

[0157] pass Figure 5 It can be seen that the array element domain sound pressure MUSIC, the array element domain vector independent channel MUSIC, the array element domain (P+V c )*(P+V c ) H The background level of the MUSIC algorithm is very high and it is unable to measure the direction of a weak target with a -15dB in-band signal-to-noise ratio in the 60° direction. c )*(P+V c ) H The MUSIC algorithm can distinguish weak targets, but the spatial spectrum background level is high and the spectrum peak is wide. c )*(P+V c ) H The simulation results of MUSIC show that it has the lowest background level and can accurately measure the direction of weak targets, while the spectrum peak is the narrowest.

[0158] like Figure 6 As shown in the figure, the root mean square error (RMSE) of the orientation estimation of each algorithm is given. The root mean square error is defined as the orientation estimation value The square root of the ratio of the sum of squares of deviations from the true azimuth value θ to the number of estimates χ is:

[0159]

[0160] exist Figure 6 In the simulation, a vector circular array with 16 elements and a radius of 0.75m was set up. The single target broadband signal was incident from a 60° direction. The element domain sound pressure MUSIC, element domain vector independent channel MUSIC, element domain (P+V c )*(P+V c ) H MUSIC, modal domain (P+V c )*(P+V c ) H MUSIC and the Toeplitz reconstruction modal domain of this application (P+V c )*(P+V c ) HThe root mean square error of the MUSIC algorithm at different signal-to-noise ratios shows that when the signal-to-noise ratio is high, the Toeplitz reconstruction modal domain (P+V c )*(P+V c ) H The estimated root mean square errors of the MUSIC algorithm and other algorithms are very close. When the signal-to-noise ratio is low, the Toeplitz reconstruction modal domain (P+V c )*(P+V c ) H Compared with the array element domain vector independent channel MUSIC and array element domain (P+V c )*(P+V c ) H MUSIC, modal domain (P+V c )*(P+V c ) H The MUSIC algorithm estimates that the root mean square error is slightly higher, but it is still within 3°.

[0161] Figure 6 It shows that the Toeplitz reconstruction modal domain (P+V c )*(P+V c ) H Compared with other algorithms, the MUSIC algorithm has the advantages of high resolution and low background noise, while still having high estimation accuracy.

[0162] like Figure 7 As shown in the figure, the average background level of the azimuth estimation of each algorithm is shown. The average background level (Average Background Level, ABL) is defined as the average value of the normalized spatial spectrum (expressed in logarithm) except the angle range corresponding to the first side lobe on both sides of the target azimuth, which is expressed by the following formula:

[0163]

[0164] In the above formula, θ min is the minimum scanning angle, θ max is the maximum scanning angle; θ v1 is the angle corresponding to the first side lobe on the left side of the target azimuth, θ v2 is the angle corresponding to the first side lobe on the right side of the target azimuth; num θ It is the number of scanning angle points after removing the range surrounded by the first side lobe. The average background level can represent the characteristics of the spatial spectrum side lobe, reflecting the algorithm's ability to suppress noise interference and improve the signal-to-noise ratio.

[0165] exist Figure 7In the simulation, a 16-element vector circular array with a radius of 0.75 m was set, and a single target broadband signal was incident from a 60° direction. The array element domain sound pressure MUSIC, array element domain vector independent channel MUSIC, and array element domain (P+V c )*(P+V c ) H MUSIC, modal domain (P+V c )*(P+V c ) H MUSIC and the Toeplitz reconstruction modal domain (P+V c )*(P+V c ) H The average background level of the MUSIC algorithm under different signal-to-noise ratios. It can be seen that the average background level of each algorithm increases with the decrease of the signal-to-noise ratio. Overall, the Toeplitz reconstruction modal domain (P+V c )*(P+V c ) H The average background level of the MUSIC algorithm is much lower than that of other algorithms, and this gap increases with the increase of the signal-to-noise ratio, which shows that the algorithm proposed in this application can reduce the spatial spectrum background to a very low level, reduce the possibility of low signal-to-noise ratio weak targets being submerged in the background noise, and has better weak target detection capability.

[0166] Figure 8 Shows the combined vibration speed V c Observation direction fixed and combined vibration velocity V c The spatial spectrum comparison results of the algorithm of this application are as follows: the observation direction matches the scanning direction. In the simulation, a 16-element vector circular array with a radius of 0.75m is set, and independent dual-target broadband signals are incident from -30° and 60° directions. The in-band signal-to-noise ratio is -10dB and -15dB respectively, the signal duration is 1s, the time domain is divided into 100 segments, the sampling rate is 32kHz, the FFT points are 320, the scanning azimuth range is 0~360°, and the angle interval is 1°. The algorithm of this application is compared in the combined vibration velocity V c Observation direction fixed and combined vibration velocity V c The spatial spectrum of the two cases where the observation direction matches the scanning direction is compared. It is found that the combined vibration velocity V c The direction finding result of the weak target at 60° is 66° when the observation direction is fixed, and there is a 6° direction finding error. c The direction finding result of the weak target at 60° when the observation direction matches the scanning direction is 62°, with a 2° direction finding error, indicating that the combined vibration velocity V c The observation direction matches the scanning direction, which results in smaller direction finding errors than the fixed direction.

[0167] The target detection technology solution of the present application combines the Toeplitz reconstruction technology used to process the mutual covariance matrix of the linear array received signal with the modal domain transformation technology of the circular array. By transforming the circular array into a "virtual linear array", the Toeplitz reconstruction technology is indirectly applied to process the mutual covariance matrix of the circular array received signal, and then combined with the vector circular array sound pressure and velocity joint processing technology, the algorithm's noise suppression ability is further enhanced, thereby improving the weak target detection capability.

[0168] The weak target background ratio is defined as the difference between the weak target output power and the background average power on the MUSIC spatial spectrum, which is used to describe the algorithm's detection capability for weak targets with low signal-to-noise ratios. The higher the weak target background ratio, the stronger the algorithm's detection capability for weak targets, and the easier it is to detect the position of weak targets with low signal-to-noise ratios. Figure 5 It can be seen that when the in-band signal-to-noise ratio of the dual-target signal input is -10dB and -15dB respectively, the weak target background ratio of the algorithm proposed in this application is improved by about 30dB compared with the array element domain sound pressure MUSIC algorithm and the array element domain vector independent channel MUSIC algorithm, and is 100dB higher than that of the array element domain (P+V c )*(P+V c ) H Compared with the MUSIC algorithm, the performance is improved by about 24dB and compared with the modal domain (P+V c )*(P+V c ) H Compared with the MUSIC algorithm, it is improved by about 21dB, indicating that the algorithm proposed in this application has stronger weak target detection capability.

[0169] The target detection technical solution of this application combines the vibration speed V c The observation direction φ is combined with the array scanning direction to avoid the problem of target orientation estimation deviation caused by the artificial selection of observation direction φ that does not match the actual target orientation (can the invention steps, drawings, tables, and performance data be combined to elaborate on the significant improvement in target orientation estimation deviation performance compared with existing methods?).

[0170] exist Figure 8 It can be seen from the simulation results that when the combined vibration speed V c When the observation direction is fixed, the direction finding result of the weak target at 60° is 66°, and the direction finding error is 6°. c When the observation direction matches the array scanning direction, the direction finding result for the 60° weak target is 62°, and the direction finding error is 2°. The direction finding error is reduced by 4° when the observation direction matches the scanning direction compared with the fixed observation direction. c The 2° direction finding error when the observation direction matches the scanning direction is caused by the algorithm itself. This result shows that by combining the combined vibration velocity V cMatching the observation direction with the scanning direction can avoid the direction-finding error introduced by the fixed observation direction.

[0171] The technical solution of the present application combines the Toeplitz matrix reconstruction technology with the vector circular array sound pressure and velocity joint processing technology, making full use of both algorithm denoising and array denoising to more fully suppress noise. In addition, as the Toeplitz reconstruction technology cannot be directly applied to circular arrays, the modal domain transformation technology is applied, so that various technical methods can be organically combined, thereby improving the robustness of the algorithm.

[0172] In the existing vector array sound pressure and vibration velocity joint processing method, when calculating the combined vibration velocity V c When calculating the target direction, the observation direction φ is set manually. If the selected observation direction φ deviates greatly from the actual target direction, a certain direction estimation error will be generated when calculating the sound pressure velocity cross-covariance matrix for target direction estimation. The technical solution of the present application makes the observation direction φ consistent with the array scanning vector direction to avoid this error. Because when the array scanning direction is the same as the actual target direction, the combined velocity observation direction φ is also the same as the actual target direction. At this time, the problem of direction estimation error caused by the inconsistency between the observation direction φ and the actual target direction will not occur. The remaining scanning directions themselves are not the directions where the actual targets are located, and therefore will not affect the target direction finding.

[0173] Compared with the prior art, the target detection method in the technical solution of the present application has the advantages of strong noise suppression capability and good weak target detection performance. Specifically, under low signal-to-noise ratio conditions, the spatial spectrum background level is lower and the spatial resolution is higher. The low spatial background level is due to the organic combination of Toeplitz matrix reconstruction technology and modal domain transformation technology, which can be fully applied to vector circular arrays whose array manifolds do not satisfy the Vandermonde matrix form. Toeplitz matrix reconstruction removes part of the noise and improves the signal-to-noise ratio.

[0174] In addition, the characteristics of the joint processing of sound pressure and velocity of the vector circular array are fully utilized. The characteristics that the signal sound pressure velocity is coherent while the noise sound pressure velocity is uncorrelated are used to calculate the cross-covariance matrix, which fully suppresses the isotropic noise and improves the signal-to-noise ratio. The noise suppression effect is superimposed by Toeplitz reconstruction and joint processing of sound pressure and velocity, so that most of the noise is suppressed. Therefore, the background level of the output spatial spectrum is very low, making it difficult for weak targets to be submerged in the background noise, thereby improving the detection capability of weak targets.

[0175] The conventional MUSIC algorithm has a wider spectral peak and a lower resolution under low signal-to-noise ratio because the orthogonality of the noise subspace and the array scanning vector obtained by eigendecomposition is destroyed. The technical solution of the present application uses Toeplitz reconstruction technology to improve the accuracy of eigendecomposition, thereby ensuring the orthogonality of the noise subspace and the array scanning vector. When calculating the MUSIC spatial spectrum, due to the strong orthogonality, the MUSIC spatial spectrum has a narrower spectral peak and a higher resolution.

[0176] It should be easily appreciated by those skilled in the art that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the present invention can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is performed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0177] The embodiment of the present invention can divide the functional modules of the detection device according to the above method example. For example, each functional module can be divided according to each function, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware or in the form of software functional modules. It should be noted that the division of modules in the embodiment of the present invention is schematic and is only a logical function division. There may be other division methods in actual implementation.

[0178] Although the present invention has been described in conjunction with specific features and embodiments thereof, it is apparent that various modifications and combinations may be made thereto without departing from the spirit and scope of the present invention. Accordingly, this specification and the accompanying drawings are merely exemplary illustrations of the present invention as defined by the appended claims and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present invention. Obviously, those skilled in the art may make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, the present invention is intended to include such modifications and variations if they fall within the scope of the claims of the present invention and their equivalents.

Claims

1. A target detection method for an under-ice environment, characterized in that: include: Perform sound pressure and velocity joint processing on the received signal of the vector circular array receiver to obtain the cross-covariance matrix of sound pressure and velocity at each frequency point; The cross-covariance matrix of the sound pressure and velocity at each frequency point at different scanning positions is transformed into the cross-covariance matrix of the sound pressure and velocity of the virtual linear array receiver in the modal domain; Toeplitz reconstruction and eigendecomposition are performed based on the sound pressure velocity cross-covariance matrix of the virtual linear array receiver to obtain the noise subspace signals corresponding to different scanning directions. Obtaining a MUSIC spatial spectrum corresponding to the received signal by using a MUSIC algorithm according to the noise subspace signal; Determine the target orientation according to the spectrum peak of the MUSIC spatial spectrum; The step of performing sound pressure and vibration velocity joint processing on the received signal of the vector circular array receiver comprises: Construct a broadband receiving signal model based on vector circular array under ice to obtain the sound pressure receiving signal and x, y vibration velocity receiving signals; The vibration velocity signal in the x direction and the vibration velocity signal in the y direction are combined to obtain a combined vibration velocity signal; The combined vibration velocity signal and the sound pressure signal at each frequency point are cross-correlated to obtain the cross-covariance matrix of the sound pressure and vibration velocity at each frequency point; The combined vibration velocity observation direction φ is traversed in all directions in the range of 0 to 360° to obtain the cross-covariance matrix of the sound pressure and combined vibration velocity in each scanning direction at different frequencies; The method of obtaining the MUSIC spatial spectrum corresponding to the received signal by using the MUSIC algorithm according to the noise subspace signal includes: obtaining the output power of the MUSIC algorithm according to the noise subspace under different scanning directions; and normalizing and taking the logarithm of the MUSIC output power of all scanning directions to obtain the MUSIC spatial spectrum.

2. The method according to claim 1, characterized in that The combined vibration velocity observation direction φ is consistent with the scanning direction of the vector circular array.

3. The method according to claim 1, characterized in that The method of transforming the sound pressure velocity cross-covariance matrix of each frequency point at different scanning positions into the sound pressure velocity cross-covariance matrix of the virtual linear array receiver in the modal domain includes: Perform Fourier transform processing on the received signal of the vector circular array; According to the signal after the spatial discrete Fourier transform, the modal transformation matrix is ​​determined; Based on the modal transformation matrix, the sound pressure velocity cross-covariance matrix at different frequency points and scanning directions is transformed in the modal domain, and the sound pressure velocity cross-covariance matrix of the virtual linear array receiver obtained by the modal domain transformation is output respectively.

4. The method according to claim 1, characterized in that: The Toeplitz reconstruction and eigendecomposition are performed according to the sound pressure velocity cross-covariance matrix of the virtual linear array receiver, including: The cross-covariance matrix of the sound pressure and velocity of the virtual linear array receiver is averaged over time to obtain the cross-covariance matrix of the sound pressure and velocity in each scanning direction at different frequencies. The cross-covariance matrices of different frequency points at each scanning direction are frequency-averaged in turn to obtain a frequency-averaged cross-covariance matrix; Perform Toeplitz reconstruction on the frequency-averaged cross-covariance matrix; The cross-covariance matrix obtained by Toeplitz reconstruction is subjected to eigendecomposition to obtain the noise subspace corresponding to different scanning orientations.

5. A target detection device for an under-ice environment, characterized in that: include: The joint processing unit is configured to perform sound pressure and velocity joint processing on the received signal of the vector circular array receiver to obtain the sound pressure and velocity cross-covariance matrix at each frequency point; the sound pressure and velocity joint processing on the received signal of the vector circular array receiver includes: Construct a broadband receiving signal model based on vector circular array under ice to obtain the sound pressure receiving signal and x, y vibration velocity receiving signals; The vibration velocity signal in the x direction and the vibration velocity signal in the y direction are combined to obtain a combined vibration velocity signal; The combined vibration velocity signal and the sound pressure signal at each frequency point are cross-correlated to obtain the cross-covariance matrix of the sound pressure and vibration velocity at each frequency point; The combined vibration velocity observation direction φ is traversed in all directions in the range of 0 to 360° to obtain the cross-covariance matrix of the sound pressure and combined vibration velocity in each scanning direction at different frequencies; A transformation unit is configured to transform the sound pressure velocity cross-covariance matrix of each frequency point at different scanning directions into the sound pressure velocity cross-covariance matrix of the virtual linear array receiver in the modal domain; A reconstruction unit is configured to perform Toeplitz reconstruction and eigendecomposition according to the sound pressure velocity cross-covariance matrix of the virtual linear array receiver to obtain noise subspace signals corresponding to different scanning orientations; The MUSIC processing unit is configured to obtain a MUSIC spatial spectrum corresponding to the received signal by using a MUSIC algorithm according to the noise subspace signal; the obtaining the MUSIC spatial spectrum corresponding to the received signal by using a MUSIC algorithm according to the noise subspace signal includes: obtaining an output power of the MUSIC algorithm according to the noise subspace under different scanning directions; normalizing and taking logarithms of the MUSIC output powers of all scanning directions to obtain the MUSIC spatial spectrum; The direction identification unit is configured to determine the target direction according to the spectrum peak of the MUSIC spatial spectrum.

6. A target detection device for an under-ice environment, characterized in that: include: A vector circular array receiver for receiving signals and a target detection device as claimed in claim 5.

7. A computer storage medium, characterized in that: The computer storage medium stores instructions, and when the instructions are executed, the target detection method for an under-ice environment as claimed in any one of claims 1 to 4 is implemented.

8. A computing device, characterized in that It comprises a processor and a communication interface coupled to the processor; the processor is used to run a computer program or instruction to implement the target detection method for an under-ice environment as claimed in any one of claims 1 to 4.

Images