Deep-sea large-aperture array deconvolution near-field high-precision direction finding method based on sound velocity correction

Abstract

The invention provides a deep-sea large-aperture array deconvolution near-field high-precision direction finding method based on sound velocity correction. According to the method, an equivalent sound velocity mapping table generated by a BELLHOP ray model is established, a sound ray bending effect is innovatively converted into a real-time table look-up correction mechanism, sound velocity correction time delay is injected in beam forming and deconvolution calculation, the problem of PSF distortion of a traditional linear model in a deep sea near field is thoroughly solved, high-precision direction finding and distance measuring are achieved, and the accuracy of direction finding and distance measuring is improved. And a subversive direction finding solution with physical accuracy, algorithm stability and engineering real-time performance is provided for the deep-sea large-aperture array.

Description

Technical Field

[0001] This invention relates to the field of underwater acoustic array signal processing technology, and in particular to a high-precision near-field direction finding method based on sound velocity correction for deconvolution of deep-sea large-aperture arrays. Background Technology

[0002] Large-aperture seabed arrays, with their advantages of high spatial gain and long detection range, have become core equipment for deep-sea target direction finding. However, for currently quiet underwater submarine targets, the detection range of large-aperture seabed arrays is almost entirely within the near-field region of the array. Therefore, to achieve high-precision direction finding of near-range deep-sea underwater targets, large-aperture seabed arrays must replace far-field beamforming with near-field focused beamforming. Conventional near-field focused beamforming (NFB) suffers from resolution limitations due to the Rayleigh criterion, high sidelobe levels, and, in the curved acoustic environment of the deep sea, time delay calculation errors based on the constant sound speed linear propagation model lead to main lobe shift, broadening, and sidelobe deterioration, severely limiting positioning accuracy. To overcome resolution limitations and suppress sidelobes, deconvolution beamforming (DBF) has been introduced into near-field processing (DB-NFB), which theoretically possesses super-resolution and low sidelobe advantages, making it a potential approach to achieve high-precision direction finding. However, the performance of deconvolution near-field focused beamforming is highly dependent on the accuracy of the point spread function (PSF). The bending of acoustic ray in the deep sea causes severe distortion of the propagation propagation flowform (PSF) based on a straight-line model, resulting in numerous artifacts, significant positioning errors, and failure of sidelobe suppression during the deconvolution process. This can even lead to performance inferior to conventional near-field focusing beamforming. The model mismatch caused by acoustic ray bending has become a key bottleneck hindering the high-precision potential of deconvolutional near-field focusing beamforming in the deep sea. Existing methods are insufficient to meet the stringent accuracy requirements of deconvolutional near-field focusing beamforming on the propagation model, necessitating a core solution that can accurately correct for the influence of sound velocity profiles and reconstruct a high-fidelity PSF. Summary of the Invention

[0003] The purpose of this invention is to solve the problem of severe performance degradation of deconvolution beamforming in deep-sea acoustic ray bending, and to propose a high-precision near-field direction finding method for deep-sea large-aperture array deconvolution based on sound velocity correction.

[0004] This invention is achieved through the following technical solution: This invention proposes a near-field high-precision direction finding method based on sound velocity correction using a large-aperture array deconvolution, the method comprising:

[0005] Step 1: Based on the measured sound velocity profile, the horizontal coordinates of the calibrated receiving array, and the deployment depth of the array, establish a sound velocity table within the horizontal distance range of the target of interest;

[0006] Step 2: Set the depth of the target of interest and pre-store the broadband point scattering function after sound velocity correction; divide the target into grids according to a certain frequency resolution within the frequency band where the target may appear; call the sound velocity table stored in Step 1 to calculate the sound velocity-corrected point scattering function corresponding to each frequency point and each sound source location; the point scattering function is a two-dimensional image, and the point scattering function stored throughout the entire measurement area is a four-dimensional function; convert the four-dimensional point scattering function into a two-dimensional point scattering function, and incoherently accumulate the point scattering functions of all frequencies at each depth to obtain the broadband point scattering function after sound velocity correction;

[0007] Step 3: Call the sound velocity table stored in Step 1 to calculate the one-dimensional broadband near-field focusing beamforming result after sound velocity correction; divide the target's possible frequency band into a grid with a certain frequency resolution, and keep the grid interval consistent with Step 2; at each frequency, convert the two-dimensional position grid of the possible sound source into a one-dimensional grid; call the sound velocity table stored in Step 1 to calculate the near-field focusing beamforming result after sound velocity correction at each sound source position; perform incoherent accumulation of the results for all frequency points to obtain the one-dimensional broadband near-field focusing beamforming result after sound velocity correction.

[0008] Step 4: Normalize and center the one-dimensional broadband near-field focusing beamforming results after sound speed correction.

[0009] Step 5: Call the pre-stored two-dimensional point scattering function after sound speed correction, and perform one-dimensional deconvolution on the one-dimensional broadband near-field focusing beamforming result after sound speed correction and normalization; perform two-dimensional reconstruction on the one-dimensional deconvolution result to obtain the two-dimensional broadband near-field deconvolution focusing beamforming result after sound speed correction.

[0010] Step 6: Obtain the peak point in the two-dimensional broadband near-field deconvolution focusing beamforming diagram after sound speed correction. The position of the peak point is the accurate estimate of the target's horizontal distance and horizontal orientation.

[0011] This invention also proposes a high-precision near-field direction finding system based on sound velocity correction and deconvolution of a large-aperture deep-sea array, the system comprising:

[0012] Sound velocity meter establishment module: Based on the measured sound velocity profile, the horizontal coordinates of the calibrated receiving array, and the deployment depth of the array, a sound velocity meter is established within the horizontal distance range of the target of interest.

[0013] Point scattering function pre-storage module: The depth of the target of interest is set, and a broadband point scattering function corrected for sound velocity is pre-stored; a grid is divided within the frequency band where the target may appear according to a certain frequency resolution; the sound velocity table stored in the sound velocity table establishment module is called, and the sound velocity-corrected point scattering function corresponding to each frequency point and each sound source location is calculated; the point scattering function is a two-dimensional image, and the point scattering function stored throughout the entire measurement area is a four-dimensional function; the four-dimensional point scattering function is converted into a two-dimensional point scattering function, and the point scattering functions of all frequencies at each depth are incoherently accumulated to obtain the broadband point scattering function corrected for sound velocity;

[0014] One-dimensional beamforming result acquisition module: This module calls the sound velocity table stored in the sound velocity table creation module to calculate the sound velocity-corrected one-dimensional broadband near-field focusing beamforming result; it divides the target's possible frequency band into a grid with a certain frequency resolution, maintaining the grid spacing consistent with the method used in the point scattering function pre-stored module; at each frequency, it converts the two-dimensional grid of possible sound source locations into a one-dimensional grid; it calls the sound velocity table stored in the sound velocity table creation module to calculate the sound velocity-corrected near-field focusing beamforming result at each sound source location; and it performs incoherent accumulation of the results for all frequency points to obtain the sound velocity-corrected one-dimensional broadband near-field focusing beamforming result.

[0015] Normalization and centering module: Normalizes and centers the one-dimensional broadband near-field focusing beamforming results after sound velocity correction;

[0016] Two-dimensional beamforming result acquisition module: calls the pre-stored two-dimensional point scattering function after sound speed correction, performs one-dimensional deconvolution on the one-dimensional broadband near-field focusing beamforming result after sound speed correction and normalization; performs two-dimensional reconstruction on the one-dimensional deconvolution result to obtain the two-dimensional broadband near-field deconvolution focusing beamforming result after sound speed correction.

[0017] Estimation module: Obtain the peak point in the two-dimensional broadband near-field deconvolution focused beamforming diagram after sound speed correction. The position of the peak point is the accurate estimate of the target's horizontal distance and horizontal orientation.

[0018] The present invention also proposes an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the sound speed correction-based deep-sea large aperture array deconvolution near-field high-precision direction finding method.

[0019] The present invention also proposes a computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the steps of the sound speed-corrected deep-sea large-aperture array deconvolution near-field high-precision direction finding method.

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0021] This invention addresses the problem of severely degraded direction finding performance of large-aperture seabed arrays for near-range deep-sea targets due to the bending of acoustic rays in the deep sea. It proposes a high-precision near-field direction finding method for large-aperture deep-sea arrays based on sound velocity correction and deconvolution. This method innovatively transforms the acoustic ray bending effect into a real-time lookup correction mechanism by establishing an equivalent sound velocity mapping table generated by the Bellhop ray model. Sound velocity correction delays are injected into beamforming and deconvolution calculations, completely resolving the PSF distortion problem of traditional linear models in the near-field deep sea. This achieves high-precision direction finding and ranging, providing a revolutionary direction finding solution for large-aperture deep-sea arrays that combines physical accuracy, algorithmic stability, and real-time engineering capabilities. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0023] Figure 1(a) shows the sound velocity gradient at the array deployment location;

[0024] Figure 1(b) shows the array's placement location;

[0025] Figure 1(c) shows the planar coordinate diagram after array calibration;

[0026] Figure 2 This is a flowchart of the near-field high-precision direction finding method based on sound velocity correction for deep-sea large aperture array deconvolution.

[0027] Figure 3 The diagram shows the relationship between the target and the array in the one-minute data processing results from 20230418213456 to 20230418213556; where (a), (b), and (c) are the situation diagram of the target and the array, the distance curve of the target relative to the array, and the azimuth curve of the target relative to the array, respectively.

[0028] Figure 4 Spatial spectrum diagrams for far-field conventional beamforming at 1s, 26s, and 48s, as well as a azimuth history diagram for 1 minute;

[0029] Figure 5(a) shows the two-dimensional and three-dimensional display diagrams of conventional near-field focusing beamforming without sound speed correction in the first second;

[0030] Figure 5(b) shows the two-dimensional and three-dimensional display diagrams of the deconvolution near-field focusing beamformation without sound speed correction in the first second;

[0031] Figure 6(a) shows the two-dimensional and three-dimensional display diagrams of conventional near-field focused beamforming after sound speed correction in the 1st second;

[0032] Figure 6(b) shows the two-dimensional and three-dimensional display of the deconvolution near-field focusing beamformation after sound speed correction in the 1st second;

[0033] Figure 7(a) shows the two-dimensional and three-dimensional display diagrams of conventional near-field focusing beamforming without sound speed correction at 26s;

[0034] Figure 7(b) shows the two-dimensional and three-dimensional display of the deconvolution near-field focusing beamformation without sound speed correction at 26s;

[0035] Figure 8(a) shows the two-dimensional and three-dimensional display diagrams of conventional near-field focused beamforming after sound speed correction at 26s;

[0036] Figure 8(b) shows the two-dimensional and three-dimensional display of the deconvolution near-field focusing beamform after sound speed correction at 26s;

[0037] Figure 9(a) shows the two-dimensional and three-dimensional display diagrams of conventional near-field focusing beamforming without sound speed correction at 48s;

[0038] Figure 9(b) shows the two-dimensional and three-dimensional display of the deconvolution near-field focusing beamformation without sound speed correction at 48s;

[0039] Figure 10(a) shows the two-dimensional and three-dimensional display diagrams of conventional near-field focused beamforming after sound speed correction at 48s;

[0040] Figure 10(b) shows the two-dimensional and three-dimensional display of the deconvolution near-field focusing beamform after sound speed correction at 48s;

[0041] Figure 11 The diagram shows the relationship between the target and the array in the one-minute data processing results from 20230418215456 to 20230418215556; where (a), (b), and (c) are the situation diagram of the target and the array, the distance curve of the target relative to the array, and the azimuth curve of the target relative to the array, respectively.

[0042] Figure 12 The spatial spectrum of the far-field conventional beamforming at the 1st second, 26th second, and 48th second, and the azimuth history map for one minute, in the one-minute data processing results of the experimental data from 20230418215456 to 20230418215556;

[0043] Figure 13(a) shows the two-dimensional and three-dimensional display of conventional near-field focused beamforming in the first second without sound velocity correction in the one-minute data processing results of experimental data 20230418215456~20230418215556.

[0044] Figure 13(b) shows the two-dimensional and three-dimensional display of the deconvolution near-field focusing beamformation in the first second of the one-minute data processing results from 20230418215456 to 20230418215556 without sound speed correction.

[0045] Figure 14(a) shows the two-dimensional and three-dimensional display of conventional near-field focused beamforming after sound speed correction in the first second of the one-minute data processing results of the experimental data 20230418215456~20230418215556.

[0046] Figure 14(b) shows the two-dimensional and three-dimensional display of the deconvolution near-field focusing beamform after sound speed correction in the first second of the one-minute data processing results of the experimental data 20230418215456~20230418215556.

[0047] Figure 15(a) shows the two-dimensional and three-dimensional display of conventional near-field focused beamforming without sound velocity correction in the 26th second of the one-minute data processing results from 20230418215456 to 20230418215556.

[0048] Figure 15(b) shows the two-dimensional and three-dimensional display of the deconvolution near-field focusing beamformation at the 26th second without sound speed correction in the one-minute data processing results of the experimental data 20230418215456~20230418215556.

[0049] Figure 16(a) shows the two-dimensional and three-dimensional display of conventional near-field focused beamforming after sound speed correction in the 26th second of the one-minute data processing results of the experimental data 20230418215456~20230418215556.

[0050] Figure 16(b) shows the two-dimensional and three-dimensional display of the deconvolution near-field focusing beamform after sound speed correction in the 26th second of the one-minute data processing results of the experimental data 20230418215456~20230418215556.

[0051] Figure 17(a) shows the two-dimensional and three-dimensional display of conventional near-field focused beamforming without sound velocity correction in the 48th second of the one-minute data processing results from 20230418215456 to 20230418215556.

[0052] Figure 17(b) shows the two-dimensional and three-dimensional display of the deconvolution near-field focusing beamformation at the 48th second without sound velocity correction in the one-minute data processing results of the experimental data 20230418215456~20230418215556.

[0053] Figure 18(a) shows the two-dimensional and three-dimensional display of conventional near-field focused beamforming after sound speed correction in the 48th second of the one-minute data processing results of the experimental data 20230418215456~20230418215556.

[0054] Figure 18(b) shows the two-dimensional and three-dimensional display of the deconvolution near-field focusing beamform after sound speed correction in the 48th second of the one-minute data processing results of the experimental data 20230418215456~20230418215556.

[0055] Figure 19 The figures show the estimation error curves of distance and azimuth for one minute of experimental data from 20230418213456 to 20230418213556 using the method of the present invention, without sound speed correction and with sound speed correction.

[0056] Figure 20 The figures show the estimation error curves of distance and azimuth for three minutes of experimental data from 20230418215256 to 20230418215556 using the method of the present invention, without sound speed correction and with sound speed correction. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. The essence of deconvolution beamforming is solving an inverse problem:

[0058]

[0059] Where p is the array pair point source (r p ,θ n The spatial response of sound. In the ideal straight-line sound propagation model:

[0060]

[0061] Among them, f l τ represents the l-th frequency point. m,pn Representative point source (r) p ,θ n The time delay to the m-th element, τ m Let q be the time delay from the scan point (r, θ) to the m-th array element, and q be the time delay from each target to each array element.

[0062] τ m,pn =r m,pn / c, τ m =r m / c

[0063] Where, r m,pn Point source (r) p ,θ n The straight-line distance from the m-th element to the m-th element; r m Let be the straight-line distance from the scan point (r, θ) to the m-th array element; and c be the speed of sound in water. However, in the curved acoustic environment of the deep sea, the actual propagation time delay deviates from the straight-line propagation model, causing severe distortion between the actual PSF and the model PSF.

[0064] p true ≠p ideal

[0065] Specifically, this manifests as follows:

[0066] Spatial migration: The position of the PSF main lobe peak deviates from the theoretical point (r p ,θ n );

[0067] Morphological distortion: asymmetric widening of the main lobe, disordered side lobe structure and increased voltage level;

[0068] Energy diffusion: Point source response energy leaks to non-real locations.

[0069] Taking the accelerated damping Richardson-Lucy (RL) algorithm as an example, its iterative process is as follows:

[0070]

[0071] Where q is the output of the near-field deconvolution; h r Let q be the direction vector. r Indicates the current iteration point. Let α represent the prediction point. r To accelerate the parameter calculation, the prediction step size is determined, p is the system point scattering function, and B is the original output signal. r Let B be the recovered signal for the r-th iteration. r The similarity between B and B indicates the convergence of the algorithm. When When the damping RL algorithm is changed to the original RL algorithm, N determines the flatness of the damping function, and only a slightly larger integer needs to be taken (N=10 in this embodiment). The value of T determines the noise suppression threshold, which is set according to the actual noise level.

[0072] Let the actual PSF be p ture The PSF with error is p err Error term δ p =p err -p ture .

[0073] Predicted output distortion: B r err =p err *q r =(p true +δ p )*q r ΔB r =||δ p *q r ||∝||δ p ||·||q r ||, when the target energy||q r The larger the ||, the greater the error ΔB r The more significant.

[0074] Energy function distortion: Damping factor failure: When u r err When >1, At this point, the damping effect is lost; when u r err <1 hour, At this point, negative damping causes divergence.

[0075] Step size calculation inaccuracy: g in previous iterations r Already contaminated by PSF, leading to alpha r It deviates from the optimal acceleration direction.

[0076] Direction vector deviation: Does not include gradient information, relies only on historical solutions. The error is further amplified in the difference.

[0077] Core iteration crash: Item, when Explosive growth; Item B r err Errors cause residual distortion, p err Prepending allows errors to propagate globally during convolution.

[0078] Sound velocity correction is essential for the stable operation of the acceleration-damped RL algorithm—an uncorrected PSF (Power Slip Function) will cause the iterative process to collapse. This invention achieves stable convergence of the originally unstable acceleration-damped RL algorithm in deep-sea scenarios through sound velocity correction, solving the iterative collapse problem caused by PSF errors.

[0079] Specifically, in combination Figure 2 This invention proposes a near-field high-precision direction finding method based on sound velocity correction for deep-sea large-aperture array deconvolution, the method comprising:

[0080] Step 1: Based on the measured sound velocity profile, the horizontal coordinates of the calibrated receiving array, and the deployment depth of the array, establish a sound velocity table within the horizontal distance range of the target of interest. Since the sound velocity tables at different frequencies do not differ much within a narrow frequency band, the center frequency of the frequency band can be selected to calculate the sound velocity table.

[0081] Step 2: Set the depth of the target of interest and pre-store the broadband point scattering function after sound velocity correction; divide the target into grids according to a certain frequency resolution within the frequency band where the target may appear; call the sound velocity table stored in Step 1 to calculate the sound velocity-corrected point scattering function corresponding to each frequency point and each sound source position (horizontal distance and horizontal azimuth); the point scattering function is a two-dimensional image, and the point scattering function stored throughout the entire measurement area is a four-dimensional function; convert the four-dimensional point scattering function into a two-dimensional point scattering function, and incoherently accumulate the point scattering functions of all frequencies at each depth to obtain the broadband point scattering function after sound velocity correction;

[0082] Step 3: Call the sound velocity table stored in Step 1 to calculate the one-dimensional broadband near-field focusing beamforming result after sound velocity correction; divide the target's possible frequency band into a grid with a certain frequency resolution, maintaining the same grid interval as in Step 2; at each frequency, convert the two-dimensional position (horizontal distance and horizontal azimuth) grid of the possible sound source location into a one-dimensional grid; call the sound velocity table stored in Step 1 to calculate the near-field focusing beamforming result after sound velocity correction at each sound source location; perform incoherent accumulation of the results for all frequency points to obtain the one-dimensional broadband near-field focusing beamforming result after sound velocity correction.

[0083] Step 4: Normalize and center the one-dimensional broadband near-field focusing beamforming results after sound speed correction.

[0084] Step 5: Call the pre-stored two-dimensional point scattering function after sound speed correction, and perform one-dimensional deconvolution on the one-dimensional broadband near-field focusing beamforming result after sound speed correction and normalization; perform two-dimensional reconstruction on the one-dimensional deconvolution result to obtain the two-dimensional broadband near-field deconvolution focusing beamforming result after sound speed correction.

[0085] Step 6: Obtain the peak point in the two-dimensional broadband near-field deconvolution focusing beamforming diagram after sound speed correction. The position of the peak point is the accurate estimate of the target's horizontal distance and horizontal orientation.

[0086] In step one, assume that a linear array with M elements is placed at a depth of z. r On the seabed, the coordinates of the m-th array element are (x... m ,y m ,z r ), fixed sound source depth z s Generate a horizontal distance sequence d within the target detection range. k =d min +k·Δd, for each d k Run Bellhop to calculate the direct sound delay τ k According to geometric relationships Calculate the slant range from the target to the receiver; calculate the slant range s. k The corresponding equivalent speed of sound c eq (s k ) = s k / τ k Generate discrete mapping tables Create tables to store the data.

[0087] In step two, assume that for depth z s The target's scan grid is P×N dimensional, where P distances r1,…r are defined. P N directions θ1,…θ N The target is at every location (r) p ,θ n Each of these points corresponds to a P×N dimensional point scattering function. The two-dimensional point scattering function at each position is transformed into a one-dimensional 1×PN function. All the point scattering functions at the PN positions form a PN×PN two-dimensional matrix.

[0088] Each target location (r) p ,θ n The point scattering function of ) satisfies:

[0089]

[0090] p(r,θ|r p ,θ n ) is (r p ,θ n The beam pattern at position (r,θ) is the value at the scan point, and each scan point corresponds to a p(r,θ|r) value. p ,θ n The values ​​of all scanned points constitute the point scattering function p of the near-field deconvolution; where τ m,pn For the target position (r) p ,θn The time delay to the m-th element, τ m This is the time delay compensated from the current scan point (r, θ) to the m-th array element; assuming s pn,1 ,s pn,2 ,…,s pn,M Representation position (r) p ,θ n The slant distance from the target at position ) to each array element, s1, s2, ..., s M This represents the slant distance from the current scan point (r, θ) to each array element;

[0091]

[0092] τ m The method for calculating the time delay value after sound speed correction is as follows: in the mapping table Find the nearest reference point Obtain the equivalent speed of sound Output correction delay τ m,pn The calculation steps for the time delay value after sound speed correction are the same as τ. m Similarly, the corrected point scattering function for sound speed can be calculated using the corrected time delay.

[0093] In step three, the near-field focusing beamforming result after sound speed correction is calculated at each sound source location; firstly, the received signal... Performing a discrete Fourier transform yields X(f1,…f L For any scanning point (r, θ) in the measurement area, frequency domain focusing beamforming is performed to obtain the near-field focused beam output as follows:

[0094]

[0095] Where, τ m This is the compensation delay from the current scan point to the m-th array element.

[0096] In step four, the calculated near-field focusing beamforming results after sound speed correction are normalized and centered:

[0097] B centered =B-μ

[0098]

[0099] Where B is the near-field focused beam output after sound speed correction; B is the average value of the beam output. centered For centralized data; B max and B min These are the minimum and maximum values ​​of the data, respectively; B normalizedThis is the result after normalization and centralization.

[0100] In step five, the one-dimensional broadband near-field focusing beamforming result after sound speed correction and normalization is deconvolved using the accelerated damping Richardson-Lucy method.

[0101] This invention also proposes a high-precision near-field direction finding system based on sound velocity correction and deconvolution of a large-aperture deep-sea array, the system comprising:

[0102] Sound velocity meter establishment module: Based on the measured sound velocity profile, the horizontal coordinates of the calibrated receiving array, and the deployment depth of the array, a sound velocity meter is established within the horizontal distance range of the target of interest.

[0103] Point scattering function pre-storage module: The depth of the target of interest is set, and a broadband point scattering function corrected for sound velocity is pre-stored; a grid is divided within the frequency band where the target may appear according to a certain frequency resolution; the sound velocity table stored in the sound velocity table establishment module is called, and the sound velocity-corrected point scattering function corresponding to each frequency point and each sound source location is calculated; the point scattering function is a two-dimensional image, and the point scattering function stored throughout the entire measurement area is a four-dimensional function; the four-dimensional point scattering function is converted into a two-dimensional point scattering function, and the point scattering functions of all frequencies at each depth are incoherently accumulated to obtain the broadband point scattering function corrected for sound velocity;

[0104] One-dimensional beamforming result acquisition module: This module calls the sound velocity table stored in the sound velocity table creation module to calculate the sound velocity-corrected one-dimensional broadband near-field focusing beamforming result; it divides the target's possible frequency band into a grid with a certain frequency resolution, maintaining the grid spacing consistent with the method used in the point scattering function pre-stored module; at each frequency, it converts the two-dimensional grid of possible sound source locations into a one-dimensional grid; it calls the sound velocity table stored in the sound velocity table creation module to calculate the sound velocity-corrected near-field focusing beamforming result at each sound source location; and it performs incoherent accumulation of the results for all frequency points to obtain the sound velocity-corrected one-dimensional broadband near-field focusing beamforming result.

[0105] Normalization and centering module: Normalizes and centers the one-dimensional broadband near-field focusing beamforming results after sound velocity correction;

[0106] Two-dimensional beamforming result acquisition module: calls the pre-stored two-dimensional point scattering function after sound speed correction, performs one-dimensional deconvolution on the one-dimensional broadband near-field focusing beamforming result after sound speed correction and normalization; performs two-dimensional reconstruction on the one-dimensional deconvolution result to obtain the two-dimensional broadband near-field deconvolution focusing beamforming result after sound speed correction.

[0107] Estimation module: Obtain the peak point in the two-dimensional broadband near-field deconvolution focused beamforming diagram after sound speed correction. The position of the peak point is the accurate estimate of the target's horizontal distance and horizontal orientation.

[0108] Example

[0109] Figures 1(a) to 1(c) The following information is required before sound speed correction. Figure 1(a) shows the sound speed gradient at the array deployment location. It can be seen from the figure that the sea depth at the deployment location of array 3 is approximately 1500m, and the sound speed varies between 1543m / s and 1483m / s. Figure 1(b) shows the latitude and longitude of the array deployment. Figure 1(c) shows the planar coordinates of the array after calibration. Calibration was performed using a cooperative target fishing vessel carrying a high-frequency sound source emitting 7kHz–8kHz LFM signals with a pulse width of 50ms and a period of 5s. The cooperative vessel had a draft of 5m and circled around the initial entry position marked by the array, radiating broadband noise signals during this period.

[0110] This invention proposes a near-field high-precision direction finding method based on sound velocity correction using deconvolution of a large-aperture deep-sea array. The method includes the following steps:

[0111] Step 1: Generate a sound velocity table corresponding to the center frequency of the band of interest using the Bellhop model. Specifically, the array is designed as a 128-element array as shown in Figure 1(c), with some curvature after calibration. The array deployment depth is based on the maximum depth of 1500m corresponding to the sound velocity gradient measured in Figure 1(a). The target depth is based on the draft of the target fishing vessel, which is 5m, and the center frequency is set to 250Hz (center frequency of 100Hz~400Hz). The horizontal receiving distance is set to 200m:10m:5000m. The direct sound delay corresponding to each horizontal receiving distance is calculated using the Bellhop model. The slant distance from the target to each horizontal receiving point is calculated based on the geometric relationship. The equivalent sound velocity c is calculated using the slant distance and the direct sound delay. eq Generate and store a sound velocity table.

[0112] Step Two: In this example, the target depth of interest is 5m. The point scattering function at each frequency point at a depth of 5m, after sound velocity correction, is pre-stored. In this example, the distance scanning grid is 200m:200m:5000m, a total of 25 distances; the azimuth scanning grid is 0°:1°:360°, a total of 361 azimuths; all distances and azimuths correspond to 9025 sound source locations. When calculating the point scattering function at each sound source location, first, based on the sound source location and the receiver location, the slant distance between the sound source and the receiver is calculated, and then mapped in a table. The system locates the nearest reference point, obtains the equivalent sound velocity, and outputs the corrected time delay. Using this corrected time delay, it calculates the sound velocity-corrected point scattering function. Each sound source location corresponds to a two-dimensional point scattering function. These functions are converted into one-dimensional vectors for storage, resulting in a 9025×9025 two-dimensional point scattering function matrix. The two-dimensional point scattering functions at different frequencies are then incoherently accumulated to obtain the sound velocity-corrected broadband point scattering function.

[0113] Step 3: Obtain the broadband spectrum of the near-field conventional focusing beamforming after sound speed correction. First, perform FFT on the simulated array received data to obtain the spectrum data of the array received signal, with a frequency resolution of 1Hz. Then, perform phase compensation on the obtained array received signal spectrum data. Phase compensation requires the sound speed-corrected time delay, and the calculation method for the sound speed-corrected time delay is similar to that in Step 2. The spectrum result after phase compensation using the sound speed-corrected time delay is incoherently accumulated within the processing frequency band to obtain the broadband spectrum. The calculation of the broadband spectrum of the near-field conventional focusing beamforming also requires dividing the range and azimuth grids, consistent with the division in Step 2. The range scan grid is from 200m:200m:5000m, a total of 25 ranges; the azimuth scan grid is from 0°:1°:360°, a total of 361 azimuths. All ranges and azimuths correspond to 9025 scan positions, resulting in a 1×9025 broadband spectrum vector.

[0114] Step 4: Normalize and center the 1×9025 broadband spectral vector obtained in Step 3;

[0115] Step 5: Call the two-dimensional broadband point scattering function with sound speed correction pre-stored in Step 2 to perform deconvolution calculation on the broadband spectrum vector with sound speed correction obtained in Step 4 to obtain a one-dimensional deconvolution result; and perform two-dimensional reconstruction on the one-dimensional deconvolution result to obtain the two-dimensional deconvolution focusing beamforming result with sound speed correction.

[0116] Step Six: Read the two-dimensional deconvolution focusing beamforming result after sound speed correction in Step Five to obtain the estimated values ​​of the target's azimuth and distance.

[0117] like Figure 3 As shown, within one minute from 20230418213456 to 20230418213556, the distance between the cooperative target fishing vessel and array 3 was between 1.305km and 1.325km, and the bearing was between 5.5° and 9.5°.

[0118] like Figure 4As shown, the beamforming azimuth spectrum under the far-field plane wave assumption is poor. It not only has high sidelobes, but the angle corresponding to the peak of the main lobe also differs significantly from the true angle of the target. The azimuth measured by far-field beamforming is 21 degrees, while the actual target azimuth is 5.46 degrees at 1 second, resulting in a direction-finding error of 15.54 degrees. At 26 seconds, the actual target azimuth is 7.02 degrees, with a direction-finding error of 13.98 degrees. At 48 seconds, the actual target azimuth is 8.44 degrees, with a direction-finding error of 12.56 degrees.

[0119] The first-second processing results are shown in Figure 5(a), which shows the result of conventional near-field focusing beamforming without sound speed correction, and Figure 5(b), which shows the result of deconvolution near-field focusing beamforming without sound speed correction. It can be seen from the figures that conventional focusing beamforming has high sidelobes and a less prominent main peak, with the highest peak positions at 1800m and 9 degrees. While deconvolution focusing beamforming has fewer sidelobes, it has more false peaks, and the distance and azimuth corresponding to the main peak differ significantly from the true distance and azimuth of the target. The second-second processing results are shown in Figure 6(a), which shows the result of conventional near-field focusing beamforming after sound speed correction, and Figure 6(b), which shows the result of deconvolution near-field focusing beamforming after sound speed correction. It can be seen from the figures that conventional focusing beamforming has slightly reduced sidelobes, but the main peak is still not prominent, with the highest peak positions at 1800m and 9 degrees. Deconvolution focusing beamforming shows a significant decrease in sidelobes, a significant reduction in false peaks, and a sharper main peak. The target azimuth and distance corresponding to the main peak are 9 degrees and 1800m, respectively. The direction-finding error was 3.54 degrees, and the ranging error was approximately 490 m, representing a significant improvement over the deconvolution near-field focusing beamforming results without sound speed correction. The 26-second processing results are shown in Figure 7(a) and Figure 7(b). The conventional near-field focusing beamforming without sound speed correction and the deconvolution near-field focusing beamforming without sound speed correction are shown. The figures show that the conventional focusing beamforming has high sidelobes and a less prominent main peak, with the highest peak position at 2600 m and 17 degrees. While the deconvolution focusing beamforming has fewer sidelobes, it has more false peaks, and the distance and azimuth corresponding to the main peak differ significantly from the true distance and azimuth of the target. The 26-second processing results are shown in Figure 8(a) and Figure 8(b). The conventional focusing beamforming has slightly reduced sidelobes, but the main peak remains indistinct, with the highest peak position at 1800 m and 10 degrees. The deconvolution focusing beamforming showed a significant decrease in sidelobes and a marked reduction in false peaks, with a sharp main peak. The target azimuth and distance corresponding to the main peak were 10 degrees and 1800m, respectively. The direction-finding error was 2.98 degrees, and the range error was approximately 480m, representing a significant improvement over the deconvolution near-field focusing beamforming results without sound velocity correction. The processing results at 48 seconds are shown in Figure 9(a), which shows the results of conventional near-field focusing beamforming without sound velocity correction, and Figure 9(b), which shows the results of deconvolution near-field focusing beamforming without sound velocity correction. The figures show that conventional focusing beamforming has high sidelobes, and the main peak is not prominent, with the highest peak position at 2000m and 13 degrees. Although the deconvolution focusing beamforming has fewer sidelobes, it has more false peaks, and the distance and azimuth corresponding to the main peak differ significantly from the true distance and azimuth of the target.The processing results at 48 seconds are shown in Figure 10(a), which shows the results of conventional near-field focusing beamforming after sound speed correction, and Figure 10(b), which shows the results of deconvolution near-field focusing beamforming after sound speed correction. The figures show that conventional focusing beamforming slightly reduces sidelobes, but the main peak remains indistinct, with the highest peak positions at 1800m and 11 degrees. Deconvolution focusing beamforming shows a significant reduction in sidelobes, a marked decrease in false peaks, and a sharper main peak. The target azimuth and distance corresponding to the main peak are 11 degrees and 1800m, respectively. The direction-finding error is 2.56 degrees, and the ranging error is approximately 480m, representing a significant improvement over the results of deconvolution near-field focusing beamforming without sound speed correction.

[0120] like Figure 11 As shown, within one minute from 20230418215456 to 20230418215556, the distance between the cooperative target fishing vessel and array 3 was between 1.76km and 1.85km, and the bearing was between 95° and 98°.

[0121] like Figure 12 As shown, the beamforming azimuth spectrum under the far-field plane wave assumption is poor. It not only has high sidelobes, but the angle corresponding to the peak of the main lobe also differs significantly from the true angle of the target. The azimuth measured by far-field beamforming is 87 degrees, while the actual target azimuth is 95.3 degrees at 1 second, resulting in a direction-finding error of 8.3 degrees. At 26 seconds, the actual target azimuth is 96.6 degrees, with a direction-finding error of 9.6 degrees. At 48 seconds, the actual target azimuth is 97.3 degrees, with a direction-finding error of 10.3 degrees.

[0122] The first-second processing results are shown in Figure 13(a), which shows the result of conventional near-field focusing beamforming without sound speed correction, and Figure 13(b), which shows the result of deconvolution near-field focusing beamforming without sound speed correction. The figures show that conventional focusing beamforming has high sidelobes and a less prominent main peak, with the highest peak position at 1800m and 95 degrees. While deconvolution focusing beamforming has fewer sidelobes, it has more false peaks, and the distance and azimuth corresponding to the main peak differ significantly from the true distance and azimuth of the target. The second-second processing results are shown in Figure 14(a), which shows the result of conventional near-field focusing beamforming after sound speed correction, and Figure 14(b), which shows the result of deconvolution near-field focusing beamforming after sound speed correction. The figures show that conventional focusing beamforming has slightly reduced sidelobes, but the main peak remains indistinct, with the highest peak position at 2000m and 96 degrees. Deconvolution focusing beamforming shows a significant decrease in sidelobes, no false peaks, and a sharp main peak. The target azimuth and distance corresponding to the main peak are 96 degrees and 2000m, respectively. The direction-finding error is 0.7 degrees, and the ranging error is approximately 234m, which is a significant improvement compared to the results of deconvolution near-field focusing beamforming without sound velocity correction. The processing results at 26 seconds are shown in Figure 15(a), which shows the results of conventional near-field focusing beamforming without sound velocity correction, and Figure 15(b), which shows the results of deconvolution near-field focusing beamforming without sound velocity correction. It can be seen from the figures that conventional focusing beamforming has high sidelobes, and the main peak is not obvious, with the highest peak position at 4200m and 233 degrees. Although deconvolution focusing beamforming has fewer sidelobes, it has more false peaks, and the distance and azimuth corresponding to the main peak differ significantly from the true distance and azimuth of the target. The processing results for the 26th second are shown in Figure 16(a), which shows the result of conventional near-field focusing beamforming after sound speed correction, and Figure 16(b), which shows the result of deconvolution near-field focusing beamforming after sound speed correction. The figures show that conventional focusing beamforming slightly reduces sidelobes, but the main peak remains indistinct, with the highest peak positions at 1800m and 96 degrees. Deconvolution focusing beamforming shows a significant reduction in sidelobes, a marked decrease in false peaks, and a sharper main peak. The target azimuth and distance corresponding to the main peak are 96 degrees and 1800m, respectively. The direction-finding error is 0.6 degrees, and the ranging error is approximately 5m, representing a significant improvement over the deconvolution near-field focusing beamforming result without sound speed correction. The processing results at 48 seconds are shown in Figure 17(a), which shows the results of conventional near-field focusing beamforming without sound velocity correction, and Figure 17(b), which shows the results of deconvolution near-field focusing beamforming without sound velocity correction. The figures show that conventional focusing beamforming has high sidelobes and a less prominent main peak, with the highest peak located at 1800m and 97 degrees. While deconvolution focusing beamforming has fewer sidelobes, it has more false peaks, and the distance and azimuth corresponding to the main peak differ significantly from the true distance and azimuth of the target.The processing results at 48 seconds are shown in Figure 18(a), which shows the results of conventional near-field focusing beamforming after sound speed correction, and Figure 18(b), which shows the results of deconvolution near-field focusing beamforming after sound speed correction. The figures show that conventional focusing beamforming slightly reduces sidelobes, but the main peak remains indistinct, with the highest peak positions at 1800m and 97 degrees. Deconvolution focusing beamforming shows a significant reduction in sidelobes, a marked decrease in false peaks, and a sharper main peak. The target azimuth and distance corresponding to the main peak are 97 degrees and 1800m, respectively. The direction-finding error is 0.3 degrees, and the ranging error is approximately 27m, representing a significant improvement over the results of deconvolution near-field focusing beamforming without sound speed correction.

[0123] like Figure 19 As shown, in the one-minute data set from 20230418213456 to 20230418213556, without sound velocity correction, the second-order moment error of the deconvolution near-field focusing beamforming for range estimation is 2726 m, and the second-order moment error for orientation estimation is 17.8°. After sound velocity correction, the second-order moment error of the deconvolution near-field focusing beamforming for range estimation is 462 m, and the second-order moment error for orientation estimation is 3.9°.

[0124] like Figure 20 As shown, in the three minutes of experimental data from 20230418215256 to 20230418215556, without sound velocity correction, the second-order origin moment error of deconvolution near-field focusing beamforming for range estimation is 871m, and the second-order origin moment error for orientation estimation is 11.2°. After sound velocity correction, the second-order origin moment error of deconvolution near-field focusing beamforming for range estimation is 141m, and the second-order origin moment error for orientation estimation is 1.1°.

[0125] In summary, this invention provides a high-precision near-field direction finding method for deep-sea large-aperture arrays based on sound velocity correction, belonging to the field of underwater acoustic array signal processing. The method aims to address the problem of severely degraded performance in near-range underwater target direction finding of large-aperture seabed arrays due to the bending of deep-sea acoustic rays. The proposed method innovatively transforms the acoustic ray bending effect into a real-time lookup correction mechanism by establishing an equivalent sound velocity mapping table generated by the Bellhop ray model. Sound velocity correction delays are injected into beamforming and deconvolution calculations, completely resolving the PSF distortion problem of traditional linear models in the near-field of deep sea, achieving high-precision direction finding and ranging. This provides a revolutionary direction finding solution for deep-sea large-aperture arrays that combines physical accuracy, algorithmic stability, and real-time engineering performance. Experimental data analysis results show that the accuracy of target distance and azimuth estimation is significantly improved by the deconvolution near-field focused beamforming after sound velocity correction, verifying the effectiveness and feasibility of this invention.

[0126] The present invention also proposes an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the sound speed correction-based deep-sea large aperture array deconvolution near-field high-precision direction finding method.

[0127] The present invention also proposes a computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the steps of the sound speed-corrected deep-sea large-aperture array deconvolution near-field high-precision direction finding method.

[0128] The memory in this application embodiment can be volatile memory or non-volatile memory, or it can include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct rambus RAM (DR RAM). It should be noted that the memory used in the methods described in this invention is intended to include, but is not limited to, these and any other suitable types of memory.

[0129] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., high-density digital video discs (DVDs)), or semiconductor media (e.g., solid-state disks (SSDs)).

[0130] In implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by instructions in software. The steps of the method disclosed in the embodiments of this application can be directly implemented by a hardware processor, or by a combination of hardware and software modules in the processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, detailed descriptions are omitted here.

[0131] The above provides a detailed description of the near-field high-precision direction finding method based on sound velocity correction for deep-sea large-aperture array deconvolution proposed in this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A high-precision near-field direction finding method based on sound velocity correction for deep-sea large-aperture array deconvolution, characterized in that, The method includes: Step 1: Based on the measured sound velocity profile, the horizontal coordinates of the calibrated receiving array, and the deployment depth of the array, establish a sound velocity table within the horizontal distance range of the target of interest; Step 2: Set the depth of the target of interest and pre-store the broadband point scattering function after sound velocity correction; divide the target into grids according to a certain frequency resolution within the frequency band where the target may appear; call the sound velocity table stored in Step 1 to calculate the sound velocity-corrected point scattering function corresponding to each frequency point and each sound source location; the point scattering function is a two-dimensional image, and the point scattering function stored throughout the entire measurement area is a four-dimensional function; convert the four-dimensional point scattering function into a two-dimensional point scattering function, and incoherently accumulate the point scattering functions of all frequencies at each depth to obtain the broadband point scattering function after sound velocity correction; Step 3: Call the sound velocity table stored in Step 1 to calculate the one-dimensional broadband near-field focusing beamforming result after sound velocity correction; divide the target's possible frequency band into a grid with a certain frequency resolution, and keep the grid interval consistent with Step 2; at each frequency, convert the two-dimensional position grid of the possible sound source into a one-dimensional grid; call the sound velocity table stored in Step 1 to calculate the near-field focusing beamforming result after sound velocity correction at each sound source position; perform incoherent accumulation of the results for all frequency points to obtain the one-dimensional broadband near-field focusing beamforming result after sound velocity correction. Step 4: Normalize and center the one-dimensional broadband near-field focusing beamforming results after sound speed correction. Step 5: Call the pre-stored two-dimensional point scattering function after sound speed correction, and perform one-dimensional deconvolution on the one-dimensional broadband near-field focusing beamforming result after sound speed correction and normalization; perform two-dimensional reconstruction on the one-dimensional deconvolution result to obtain the two-dimensional broadband near-field deconvolution focusing beamforming result after sound speed correction. Step 6: Obtain the peak point in the two-dimensional broadband near-field deconvolution focusing beamforming diagram after sound speed correction. The position of the peak point is the accurate estimate of the target's horizontal distance and horizontal orientation.

2. The method according to claim 1, characterized in that, In step one, assume that a linear array with M elements is placed at a depth of z. r On the seabed, the coordinates of the m-th array element are (x... m ,y m ,z r ), fixed sound source depth z s Generate a horizontal distance sequence d within the target detection range. k =d min +k·Δd, for each d k Run Bellhop to calculate the direct sound delay τ k According to geometric relationships Calculate the slant range from the target to the receiver; calculate the slant range s. k The corresponding equivalent speed of sound c eq (s k ) = s k / τ k Generate discrete mapping tables Create tables to store the data.

3. The method according to claim 2, characterized in that, In step two, assume that for depth z s The target's scan grid is P×N dimensional, where P distances r1,…r are defined. P N directions θ1,…θ N The target is at every location (r) p ,θ n Each of these points corresponds to a P×N dimensional point scattering function. The two-dimensional point scattering function at each position is transformed into a one-dimensional 1×PN function. All the point scattering functions at the PN positions form a PN×PN two-dimensional matrix. Each target location (r) p ,θ n The point scattering function of ) satisfies: p(r,θ|r p ,θ n ) is (r p ,θ n The beam pattern at position (r,θ) is the value at the scan point, and each scan point corresponds to a p(r,θ|r) value. p ,θ n The values ​​of all scanned points constitute the point scattering function p of the near-field deconvolution; where τ m,pn For the target position (r) p ,θ n The time delay to the m-th array element, τ m This is the time delay compensated from the current scan point (r, θ) to the m-th array element; assuming s pn,1 ,s pn,2 ,…,s pn,M Representation position (r) p ,θ n The slant distance from the target at position ) to each array element, s1, s2, ..., s M This represents the slant distance from the current scan point (r, θ) to each array element; τ m The method for calculating the time delay value after sound speed correction is as follows: in the mapping table Find the nearest reference point Obtain the equivalent speed of sound Output correction delay The corrected point scattering function can be calculated using the corrected time delay.

4. The method according to claim 3, characterized in that, In step three, the near-field focusing beamforming result after sound speed correction is calculated at each sound source location; firstly, the received signal... Performing a discrete Fourier transform yields X(f1,…f L For any scanning point (r, θ) in the measurement area, frequency domain focusing beamforming is performed to obtain the near-field focused beam output as follows: Where, τ m This is the compensation delay from the current scan point to the m-th array element.

5. The method according to claim 4, characterized in that, In step four, the calculated near-field focusing beamforming results after sound speed correction are normalized and centered: B centered =B-μ Where B is the near-field focused beam output after sound speed correction; It is the average value of the beam output; B centered For centralized data; B max and B min These are the minimum and maximum values ​​of the data, respectively; B normalized This is the result after normalization and centralization.

6. The method according to claim 5, characterized in that, In step five, the one-dimensional broadband near-field focusing beamforming result after sound speed correction and normalization is deconvolved using the accelerated damping Richardson-Lucy method.

7. The method according to claim 1, characterized in that, The depth of the target of interest is 5m.

8. A high-precision near-field direction finding system based on sound velocity correction and deconvolution of a large-aperture array in the deep sea, characterized in that, The system includes: Sound velocity meter establishment module: Based on the measured sound velocity profile, the horizontal coordinates of the calibrated receiving array, and the deployment depth of the array, a sound velocity meter is established within the horizontal distance range of the target of interest. Point scattering function pre-storage module: The depth of the target of interest is set, and a broadband point scattering function corrected for sound velocity is pre-stored; a grid is divided within the frequency band where the target may appear according to a certain frequency resolution; the sound velocity table stored in the sound velocity table establishment module is called, and the sound velocity-corrected point scattering function corresponding to each frequency point and each sound source location is calculated; the point scattering function is a two-dimensional image, and the point scattering function stored throughout the entire measurement area is a four-dimensional function; the four-dimensional point scattering function is converted into a two-dimensional point scattering function, and the point scattering functions of all frequencies at each depth are incoherently accumulated to obtain the broadband point scattering function corrected for sound velocity; One-dimensional beamforming result acquisition module: This module calls the sound velocity table stored in the sound velocity table creation module to calculate the sound velocity-corrected one-dimensional broadband near-field focusing beamforming result; it divides the target's possible frequency band into a grid with a certain frequency resolution, maintaining the grid spacing consistent with the method used in the point scattering function pre-stored module; at each frequency, it converts the two-dimensional grid of possible sound source locations into a one-dimensional grid; it calls the sound velocity table stored in the sound velocity table creation module to calculate the sound velocity-corrected near-field focusing beamforming result at each sound source location; and it performs incoherent accumulation of the results for all frequency points to obtain the sound velocity-corrected one-dimensional broadband near-field focusing beamforming result. Normalization and centering module: Normalizes and centers the one-dimensional broadband near-field focusing beamforming results after sound velocity correction; Two-dimensional beamforming result acquisition module: calls the pre-stored two-dimensional point scattering function after sound speed correction, performs one-dimensional deconvolution on the one-dimensional broadband near-field focusing beamforming result after sound speed correction and normalization; performs two-dimensional reconstruction on the one-dimensional deconvolution result to obtain the two-dimensional broadband near-field deconvolution focusing beamforming result after sound speed correction. Estimation module: Obtain the peak point in the two-dimensional broadband near-field deconvolution focused beamforming diagram after sound speed correction. The position of the peak point is the accurate estimate of the target's horizontal distance and horizontal orientation.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium for storing computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the steps of the method according to any one of claims 1-7.

Images