A target waveform recovery method based on a hybrid sparse Bayesian model against near-field blanking jamming

Abstract

The application relates to the technical field of underwater acoustic signal processing, in particular to a target waveform recovery method resisting near-field suppression interference based on a mixed sparse Bayesian model, which constructs a mixed sparse dictionary of a target and uncertain near-field interference, establishes a robust mixed sparse Bayesian model, unifies the near-field interference, the target and noise under the same model by three different sparse criteria, obtains the target waveform recovery under the near-field strong interference, and can realize high-fidelity recovery of weak target radiation signals in an underwater acoustic countermeasure environment.

Description

Technical Field

[0001] This invention relates to the field of underwater acoustic signal processing technology, specifically to a method for recovering target waveforms against near-field suppression interference based on a hybrid sparse Bayesian model. Background Technology

[0002] Waveform recovery is a key technology in passive sonar processing, and its output signal-to-noise ratio (SNR) and signal-to-interference ratio (SIR) directly affect subsequent recognition capabilities. In underwater acoustic countermeasures scenarios, when faced with strong near-field suppression interference, due to factors such as model mismatch, commonly used waveform recovery methods, whether conventional or high-performance adaptive, cannot reduce the impact of interference. This results in severe distortion between the recovered target waveform and the actual waveform, making it impossible to obtain high-fidelity target waveform information. In underwater acoustic countermeasures environments, although a series of anti-interference algorithms exist to reduce the impact of near-field interference and improve broadband detection performance, these methods struggle to overcome the mismatch of the interference array manifold and cannot significantly reduce the interference component during beamforming, thus making them difficult to directly utilize in waveform recovery methods. Sparse Bayesian methods have the advantage of separating targets during the iterative process, but the near-field target position is usually unknown during passive detection. Furthermore, targets, interference, and noise typically exhibit different sparsities, making it difficult for conventional sparse Bayesian methods to achieve waveform recovery for weak targets in the targeted countermeasures environments. Summary of the Invention

[0003] To address the shortcomings of existing technologies, the present invention aims to provide a method for recovering target waveforms against near-field suppression interference based on a hybrid sparse Bayesian model.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for recovering target waveforms against near-field suppression interference based on a hybrid sparse Bayesian model, the steps of which are as follows:

[0005] (1) Perform time-frequency transformation on the received time-domain data of each array element to obtain array element domain-frequency domain data;

[0006] (2) Using conventional beamforming methods, a conventional broadband warning space spectrum is obtained;

[0007] (3) Based on the conventional broadband warning space spectrum, roughly estimate the near-field interference azimuth and construct the uncertainty range of near-field target interference;

[0008] (4) Construct a hybrid sparse dictionary based on the detection range and the uncertain range of near-field target orientation;

[0009] (5) Based on the sparsity characteristics of near-field interference, far-field targets and noise, a hybrid sparse iterative criterion is constructed, and the target spatial spectrum estimation and reconstructed covariance matrix after interference suppression are obtained through iteration;

[0010] (6) Based on the results of the previous step, combined with the adaptive processing flow, a high-performance beamforming output is obtained;

[0011] (7) Based on the beamforming output of the previous step, the target waveform recovery result against recent suppression interference is obtained by time-frequency transformation.

[0012] In some embodiments, step (1) is specifically implemented as follows:

[0013] right The time-domain data received by the horizontal line array is subjected to time-frequency transformation to obtain... Individual frequency domain snapshots:

[0014]

[0015] In the formula, and All 3D matrices represent the element-domain time-domain data and the element-domain-frequency-domain data after Fourier transform, respectively. The Fourier transform length corresponds to Each time-domain sampling point and the transformed One frequency point.

[0016] In some embodiments, step (2) is specifically performed as follows:

[0017] Using conventional beamforming methods, the conventional broadband warning space spectrum is obtained:

[0018] a) Constructing the plane wave array manifold matrix:

[0019]

[0020] In the formula, Indicates direction The corresponding far-field plane wave array manifold,

[0021]

[0022] In the formula, Indicates frequency, Indicates the spacing between array elements. Indicates the speed of sound.

[0023] b) Obtain the conventional broadband surveillance spatial spectrum:

[0024]

[0025] In some embodiments, step (3) is specifically implemented as follows:

[0026] Based on the conventional broadband warning spatial spectrum, the location of near-field interference is roughly estimated:

[0027] a) Estimation of near-field interference location and uncertainty range:

[0028]

[0029] In the formula, The range of energy diffusion of a near-field target across the conventional broadband warning spectrum can be determined by the change in energy amplitude gradient.

[0030] b) Uncertain range of near-field interference location:

[0031]

[0032] In the formula, This is a small value added manually by someone, through adding... This ensures that the uncertainty range of the interference's azimuth can include the true azimuth of the interference, thereby reducing the impact of the interference's azimuth estimation error on subsequent processing procedures.

[0033] In some embodiments, step (4) is specifically implemented as follows:

[0034] Based on the detection range and the uncertain range of near-field target location, a hybrid sparse dictionary is constructed:

[0035]

[0036] In the formula, to To obtain a complete far-field plane wave array manifold vector, ; It indicates the first Several near-field array manifolds, among which... This indicates that the azimuth of the array manifold is within the range of uncertainty regarding the azimuth of the interference. , and These represent the nearest and farthest distances of near-field interference, respectively, and are set based on experience; The equivalent array manifold representing noise.

[0037] In some embodiments, step (5) is specifically implemented as follows:

[0038] The target space spectrum estimate and reconstructed covariance matrix after suppressing interference are obtained through hybrid sparse constraint iteration:

[0039] a) Initialization:

[0040] The mixed space spectrum is initialized based on the mixed sparse dictionary.

[0041]

[0042] in, Center front The first is used for initializing the far-field spatial spectrum, denoted as . , The following is the near-field spatial spectrum within the range of uncertainty in the direction of interference. Initialization of noise ;

[0043] b) Covariance matrix reconstruction:

[0044] ;

[0045] c) Mixed sparse space spectrum estimation:

[0046]

[0047]

[0048]

[0049]

[0050] In the formula, and These correspond to complete far-field array manifold dictionaries and near-field array manifold dictionaries containing uncertain interference orientations, respectively. , , These represent different sparse iterative algorithms for far-field target signals, near-field interference, and noise, respectively, to reconstruct three different types of data. Under the sparse Bayesian criterion, a factor is introduced to control the sparsity. ,get

[0051]

[0052]

[0053]

[0054] In the formula, , , They represent , , The One element; , , They are respectively , , The List;

[0055] d) Convergence judgment

[0056] If satisfied,

[0057]

[0058] When the iteration has converged, The parameters are used to determine convergence; otherwise, repeat steps (b)-(c), where, express Center front A vector consisting of 10 elements.

[0059] In some embodiments, step (6) is specifically implemented as follows:

[0060] Adaptive beamforming based on hybrid sparse spatial spectrum estimation and reconstructed covariance:

[0061] a) Constructing adaptive weights:

[0062]

[0063] b) Adaptive beamforming:

[0064]

[0065] In some embodiments, step (7) is specifically implemented as follows:

[0066] The final target waveform recovery output with resistance to near-field suppression interference is obtained:

[0067] By using inverse time-frequency transform, the time-domain signal after waveform reconstruction is obtained.

[0068]

[0069] In the formula, and These represent the minimum and maximum frequencies, respectively.

[0070] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention proposes a target waveform recovery method based on a hybrid sparse Bayesian model to resist strong near-field interference. By constructing a hybrid sparse dictionary of the target and uncertain near-field interference, a robust hybrid sparse Bayesian model is established. Near-field interference, target and noise are unified under the same model by three different sparsity criteria, and the target waveform recovery under strong near-field interference is obtained, realizing high-fidelity recovery of weak target radiation signals in underwater acoustic countermeasures environment.

[0071] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. The embodiments of this application will provide a detailed description and understanding of this application. Attached Figure Description

[0072] Figure 1 This is a flowchart of the processing of the present invention;

[0073] Figure 2 This is a simulation diagram showing the target and interference situation.

[0074] Figure 3 For standard BTR comparison charts;

[0075] Figure 4 This is a comparison chart of the spectrum processed in a conventional manner;

[0076] Figure 5 A comparison chart of waveform similarity after conventional waveform restoration processing;

[0077] Figure 6 The BTR diagram is shown for the proposed method under near-field suppression interference.

[0078] Figure 7 A comparison of waveform recovery between the proposed method and conventional methods under near-field suppression interference is shown.

[0079] Figure 8 A waveform similarity comparison between the proposed method and the conventional method under near-field suppression interference;

[0080] Figure 9 A comparison of the time-frequency waveforms of the recovered frequency-modulated signal under near-field suppression;

[0081] Figure 10 This is a comparison chart of the waveform similarity of the recovered FM signal under near-field suppression. Detailed Implementation

[0082] 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 some embodiments of the present invention, and not all embodiments. 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.

[0083] Please see Figure 1 This invention provides a technical solution: a method for recovering target waveforms against near-field suppression interference based on a hybrid sparse Bayesian model, the technical steps and specific implementation methods of which are as follows:

[0084] 1. To The time-domain data received by the horizontal line array is subjected to time-frequency transformation to obtain... Individual frequency domain snapshots:

[0085]

[0086] In the formula, and All 3D matrices represent the element-domain time-domain data and the element-domain-frequency-domain data after Fourier transform, respectively. The Fourier transform length corresponds to Each time-domain sampling point and the transformed One frequency point.

[0087] 2. Using conventional beamforming methods, the conventional broadband warning spatial spectrum is obtained:

[0088] a) Constructing the plane wave array manifold matrix:

[0089]

[0090] In the formula, Indicates direction The corresponding far-field plane wave array manifold,

[0091]

[0092] In the formula, Indicates frequency, Indicates the spacing between array elements. It indicates the speed of sound.

[0093] b) Obtain the conventional broadband surveillance spatial spectrum:

[0094]

[0095] 3. Based on the conventional broadband warning spatial spectrum, roughly estimate the azimuth of near-field interference:

[0096] a) Estimation of near-field interference location and uncertainty range:

[0097]

[0098] In the formula, The range of energy diffusion of a near-field target across the conventional broadband warning spectrum can be determined by the change in energy amplitude gradient.

[0099] b) Uncertain range of near-field interference location:

[0100]

[0101] In the formula, This is a small value added manually by one person. (Added via...) This ensures that the uncertainty range of the interference's azimuth can include the true azimuth of the interference, thereby reducing the impact of the interference's azimuth estimation error on subsequent processing procedures.

[0102] 4. Based on the detection range and the uncertain range of near-field target location, construct a hybrid sparse dictionary:

[0103]

[0104] In the formula, to To obtain a complete far-field plane wave array manifold vector, ; It indicates the first Several near-field array manifolds, among which... This indicates that the azimuth of the array manifold is within the range of uncertainty regarding the azimuth of the interference. , and These represent the nearest and farthest distances of near-field interference, respectively, and can be set based on experience; The equivalent array manifold representing noise.

[0105] 5. Obtain the target space spectrum estimate and reconstructed covariance matrix after suppressing disturbances through hybrid sparse constraint iteration:

[0106] a) Initialization:

[0107] The mixed space spectrum is initialized based on the mixed sparse dictionary.

[0108]

[0109] in, Center front The first is used for initializing the far-field spatial spectrum, denoted as . , The following is the near-field spatial spectrum within the range of uncertainty in the direction of interference. Initialization of noise .

[0110] b) Covariance matrix reconstruction:

[0111]

[0112] c) Mixed sparse space spectrum estimation:

[0113]

[0114]

[0115]

[0116]

[0117] In the formula, and They correspond to a complete far-field array manifold dictionary and a near-field array manifold dictionary containing uncertain interference orientations, respectively. , , These represent different sparse iterative algorithms for far-field target signals, near-field interference, and noise, respectively, to reconstruct three different types of data. Under the sparse Bayesian criterion, a factor controlling the sparsity is introduced. ,get

[0118]

[0119]

[0120]

[0121] In the formula, , , They represent , , The One element; , , They are respectively , , The List.

[0122] d) Convergence judgment

[0123] If satisfied,

[0124]

[0125] When the iteration has converged, The parameters are used to determine convergence; otherwise, iterate steps (b)-(c). Where, express Center front A vector consisting of 10 elements.

[0126] 6. Adaptive beamforming based on hybrid sparse spatial spectrum estimation and reconstruction covariance:

[0127] a) Constructing adaptive weights:

[0128]

[0129] b) Adaptive beamforming:

[0130]

[0131] 7. Finally, the target waveform recovery output with resistance to near-field suppression interference is obtained:

[0132] By using inverse time-frequency transform, the time-domain signal after waveform reconstruction is obtained.

[0133]

[0134] In the formula, and These represent the minimum and maximum frequencies, respectively.

[0135] The technical solution of this application is verified using simulation test data:

[0136] Assume there is a far-field target source and a near-field suppressive interference source in the sound field. Both the target source and the interference source radiate broadband signals in the 50-150Hz range, including narrowband line spectrum signals and broadband random signals. The target source contains 80, 101, and 122Hz line spectra, while the interference source contains 95 and 110Hz line spectra. The receiving array is a 512-element horizontal linear array with an element spacing of 1.2m. The situation of the target, interference, and receiving array is as follows: Figure 2 As shown, the origin of the receiving array is at the center of the coordinate system. The interference source is located at (0, 1000) meters, and the target source is located at (300, 1000) meters. Under this situation, assuming that the sound level of the interference source is 10 dB higher than that of the target source, it can be deduced that due to the close distance and high source level of the interference source, it has a suppressive effect on the target source. Figure 3 Broadband warning history spectra obtained by conventional wavenumber formation (CBF) are presented. The far-field plane wave model is used in the processing. It can be seen that when there is no suppression interference, the trajectory of the target in the spatial spectrum is clearly visible. However, when there is suppression interference, due to the strong suppression interference in the near field and the mismatch between the propagation model and the processing model, the interference source forms an extended band in the warning spectrum, which overwhelms the target.

[0137] Figure 4 A comparison of the spectra of the recovered waveforms after CBF waveform reconstruction with and without suppressed interference is presented, showing the results compared to the original signal waveforms. It can be seen that when suppressed interference is present, the spectrum of the waveform recovered by the CBF method is severely affected by the interference, and the interference line spectrum can almost be observed. Furthermore, Figure 5 The waveform similarity between the recovered waveform and the original waveform is compared under two different conditions. It can be seen that when near-field suppression interference is present, the waveform similarity decreases significantly, indicating a severe decline in the waveform recovery performance of the CBF method.

[0138] Figure 6 , Figure 7 , Figure 8The following results are presented: broadband warning history spectrum, spectrum comparison of the recovered waveform, and waveform similarity comparison after using the proposed hybrid sparse Bayesian waveform recovery method. It can be seen that, due to the effective removal of near-field interference components by the proposed method, the extended interference band in the broadband warning history spectrum is suppressed, and the target trajectory is highlighted. Simultaneously, the target line spectrum is recovered in the spectrum of the recovered waveform, while the interference line spectrum is effectively suppressed. These results demonstrate the effectiveness of the proposed method.

[0139] To further verify the proposed method, Figure 9 and Figure 10 The differences in waveform recovery results between the proposed method and the CBF method are presented when the target radiated frequency-modulated signal is encountered. Figure 9 It can be seen that when near-field suppression interference is present, the time-frequency structure of the target frequency-modulated signal is difficult to see in the time-frequency plot of the waveform recovered by conventional methods; however, when the proposed method is used, the frequency-modulated signal radiated by the target is effectively recovered, and its time-frequency plot structure is approximately consistent with the CBF when there is no interference; at the same time, Figure 10 The comparison of waveform similarity also demonstrates the effectiveness of the proposed method.

[0140] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

[0141] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for recovering target waveforms against near-field suppression interference based on a hybrid sparse Bayesian model, characterized in that: The steps are as follows: (1) Perform time-frequency transformation on the received time-domain data of each array element to obtain array element domain-frequency domain data; (2) Using conventional beamforming methods, a conventional broadband warning space spectrum is obtained; (3) Based on the conventional broadband warning space spectrum, roughly estimate the near-field interference azimuth and construct the uncertainty range of near-field target interference; (4) Construct a hybrid sparse dictionary based on the detection range and the uncertain range of near-field target orientation; (5) Based on the sparsity characteristics of near-field interference, far-field targets and noise, a hybrid sparse iterative criterion is constructed, and the target spatial spectrum estimation and reconstructed covariance matrix after interference suppression are obtained through iteration; (6) Based on the results of the previous step, combined with the adaptive processing flow, a high-performance beamforming output is obtained; (7) Based on the beamforming output of the previous step, the target waveform recovery result against recent suppression interference is obtained by time-frequency transformation; According to step (5), the specific method is as follows: The target space spectrum estimate and reconstructed covariance matrix after suppressing interference are obtained through hybrid sparse constraint iteration: a) Initialization: The mixed space spectrum is initialized based on the mixed sparse dictionary. in, Center front The first is used for initializing the far-field spatial spectrum, denoted as . , The following is the near-field spatial spectrum within the range of uncertainty in the direction of interference. Initialization of noise ; b) Covariance matrix reconstruction: ； c) Mixed sparse space spectrum estimation: In the formula, and These correspond to complete far-field array manifold dictionaries and near-field array manifold dictionaries containing uncertain interference orientations, respectively. , , These represent different sparse iterative algorithms for far-field target signals, near-field interference, and noise, respectively, to reconstruct three different types of data. Under the sparse Bayesian criterion, a factor is introduced to control the sparsity. ,get In the formula, , , They represent , , The One element; , , They are respectively , , The List; d) Convergence judgment If satisfied When the iteration has converged, The parameters are used to determine convergence; otherwise, repeat steps (b)-(c), where, express Center front A vector consisting of 10 elements.

2. The method for recovering target waveforms against near-field suppression interference based on a hybrid sparse Bayesian model according to claim 1, characterized in that: According to step (1), the specific method is as follows: right The time-domain data received by the horizontal line array is subjected to time-frequency transformation to obtain... Individual frequency domain snapshots: In the formula, and All 3D matrices represent the element-domain time-domain data and the element-domain-frequency-domain data after Fourier transform, respectively. The Fourier transform length corresponds to Each time-domain sampling point and the transformed One frequency point.

3. The method for recovering target waveforms against near-field suppression interference based on a hybrid sparse Bayesian model according to claim 2, characterized in that: According to step (2), the specific method is as follows: Using conventional beamforming methods, the conventional broadband warning space spectrum is obtained: a) Constructing the plane wave array manifold matrix: In the formula, Indicates direction The corresponding far-field plane wave array manifold, In the formula, Indicates frequency, Indicates the spacing between array elements. Indicates the speed of sound. b) Obtain the conventional broadband surveillance spatial spectrum: 。 4. The method for recovering target waveforms against near-field suppression interference based on a hybrid sparse Bayesian model according to claim 3, characterized in that: According to step (3), the specific method is as follows: Based on the conventional broadband warning spatial spectrum, the location of near-field interference is roughly estimated: a) Estimation of near-field interference location and uncertainty range: In the formula, The range of energy diffusion of a near-field target across the conventional broadband warning spectrum can be determined by the change in energy amplitude gradient. b) Uncertain range of near-field interference location: In the formula, This is a small value added manually by someone, through adding... This ensures that the uncertainty range of the interference's azimuth can include the true azimuth of the interference, thereby reducing the impact of the interference's azimuth estimation error on subsequent processing procedures.

5. The method for recovering target waveforms against near-field suppression interference based on a hybrid sparse Bayesian model according to claim 4, characterized in that: According to step (4), the specific method is as follows: Based on the detection range and the uncertain range of near-field target location, a hybrid sparse dictionary is constructed: In the formula, to To obtain a complete far-field plane wave array manifold vector, ; It indicates the first Several near-field array manifolds, among which... This indicates that the azimuth of the array manifold is within the range of uncertainty regarding the azimuth of the interference. , and These represent the nearest and farthest distances of near-field interference, respectively, and are set based on experience; The equivalent array manifold representing noise.

6. The method for recovering target waveforms against near-field suppression interference based on a hybrid sparse Bayesian model according to claim 1, characterized in that: According to step (6), the specific method is as follows: Adaptive beamforming based on hybrid sparse spatial spectrum estimation and reconstructed covariance: a) Constructing adaptive weights: b) Adaptive beamforming: 。 7. The method for recovering target waveforms against near-field suppression interference based on a hybrid sparse Bayesian model according to claim 6, characterized in that: According to step (7), the specific method is as follows: The final target waveform recovery output with resistance to near-field suppression interference is obtained: By using inverse time-frequency transform, the time-domain signal after waveform reconstruction is obtained. In the formula, and These represent the minimum and maximum frequencies, respectively.

Images