Deep sea underwater target tracking method based on frequency domain interference sound field

Abstract

The invention particularly relates to a deep sea underwater target tracking method based on a frequency domain interference sound field. The method comprises the following steps: firstly, reconstructing an array original broadband wave beam output by using a complementary integrated empirical mode decomposition algorithm and mapping the array original broadband wave beam output into a target glancing angle-depth ambiguity plane; and secondly, modeling ambiguity planes at multiple moments into a random finite set form, and filtering and outputting a grazing angle-depth trajectory of the underwater target by using a label multi-Bernoulli filter. And finally, converting the glancing angle of the target into a horizontal distance in combination with a sound field model, and outputting a distance-depth track of the underwater target. According to the method, a grazing angle-depth measurement random finite set of the underwater target is constructed by using frequency domain interference characteristics generated by the moving underwater target in a direct sound area range, and label multi-Bernoulli filtering is performed on the measurement set in combination with a target motion state equation and a sound field model, so that continuous tracking of the underwater moving target is realized.

Description

Technical Field

[0001] This invention relates to the field of underwater acoustic detection technology, specifically to a deep-sea underwater target tracking method based on frequency domain interferometric sound field, which uses frequency interferometric characteristic parameters generated by underwater moving targets in the deep-sea direct sound zone for tracking. Background Technology

[0002] On the one hand, as one of the important acoustic propagation channels for passive detection of deep-sea underwater acoustic targets, the direct sound zone has the characteristics of long propagation distance, no shadow zone at medium and short ranges, low propagation loss, and stable channel. On the other hand, the vertical linear array has the advantages of stable array attitude and easy deployment, making it the most important mode for passive detection of medium and short range deep-sea targets.

[0003] In recent years, passive target localization and tracking methods have been extensively studied in marine acoustics, mainly including methods based on matched field processing, multipath arrival structures, and acoustic field interference fringes. Although matched field processing was applied to the deep sea early on and has been sea-tested, its performance is quite sensitive to model errors and environmental mismatches, which limits its engineering applications. Furthermore, in the deep-sea environment, small-aperture arrays can be used to obtain multipath delay and angle of arrival information of strong target signals. Combined with acoustic field models, underwater target parameters can be estimated, but extracting multipath information of targets under low signal-to-noise ratio conditions is difficult. Multipath delay corresponds to the interference period in the frequency domain. By converting the array-received signal to the frequency domain, the periodic oscillation characteristics of the acoustic field frequency domain interference fringes can be used to achieve target localization.

[0004] In underwater target tracking, existing research mainly focuses on two techniques: DOA tracking and pure azimuth tracking. The former continuously tracks the azimuth of underwater targets based on the output of passive sonar arrays or the azimuth angle after DOA estimation; the latter utilizes multiple sonar arrays to estimate the DOA of the target simultaneously, thereby continuously tracking the target's planar position. These two techniques primarily target surface towed array detection scenarios, while research specifically on deep-sea vertical linear array target tracking is relatively limited. Against this background, this invention establishes a broadband acoustic source depth estimation method for complex deep-sea environments and verifies the effectiveness of the method through simulation.

[0005] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of the present invention, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] For scenarios involving passive detection using bottom-mounted vertical linear arrays in deep-sea direct acoustic zones, this invention provides a deep-sea underwater target tracking method based on frequency-domain interferometric acoustic field characteristics. This method combines a complementary ensemble empirical mode decomposition algorithm to reconstruct the array's broadband beam output power and transform it to a target grazing angle-depth ambiguity plane. Based on the ambiguity plane, a measurement random finite set is constructed and filtered using a tag-based multi-Bernoulli filter. Finally, combined with an acoustic field model, the filtered target grazing angle-depth trajectory is mapped to a horizontal range-depth track, thereby achieving the tracking of moving deep-sea targets.

[0007] Other features and advantages of the invention will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0008] According to a first aspect of the present invention, a deep-sea underwater target tracking method based on frequency domain interferometric acoustic field is provided, the method comprising: The acoustic signals acquired by the deep-sea bottom-mounted vertical linear array are processed by frequency domain beamforming to obtain the manifold matrix of the receiving array. Based on the manifold matrix of the receiving array, the MVDR spatial power spectrum, i.e. the original beam output matrix, is calculated. The original beam output matrix of the array is reconstructed using the complementary integrated empirical mode decomposition (CEEMD) algorithm to obtain a beam output matrix with enhanced interferometric features. A generalized Fourier transform is performed on the beam output matrix with enhanced interferometric features to convert it into a target grazing angle-depth ambiguity plane; The maximum points on the ambiguity plane are extracted to form a random finite set for grazing angle-depth measurement. A two-dimensional uniform motion state equation for the underwater target in the grazing angle-depth plane is established. The two-dimensional uniform motion state equation is predicted and updated iteratively using a tag-based multi-Bernoulli filter to track and output the grazing angle-depth trajectory of the target. Based on the acoustic field model simulation, the mapping relationship between the horizontal distance, depth and the arrival glancing angle of the underwater target is established, and the glancing-depth trajectory is mapped to the horizontal distance-depth track to complete the underwater target tracking.

[0009] In some exemplary embodiments, the implementation process of the Complementary Integrated Empirical Mode Decomposition (CEEMD) algorithm includes: Set the overall average frequency N; Add a small-amplitude white noise signal to the beam output sequence corresponding to each grazing angle; The Empirical Mode Decomposition (EMD) algorithm is used to decompose the noise-added sequence to obtain the intrinsic mode components and residual terms of each order. Average the results of N decompositions; The intrinsic mode components are sorted in descending order of variance contribution rate, and the first K components with a cumulative variance contribution rate greater than a preset threshold are summed. Repeat the above steps for the beam output sequence corresponding to all grazing angles to reconstruct the beam output matrix with enhanced interferometric features.

[0010] In some exemplary embodiments, the formula for calculating the MVDR spatial power spectrum is:

[0011] in, for The sampling covariance matrix at a given frequency point This indicates the conjugate transpose.

[0012] In some exemplary embodiments, the generalized Fourier transform of the beam output matrix enhanced by the interferometric features, converting it into a target grazing angle-depth ambiguity plane, specifically involves:

[0013] in, and These are the lower and upper frequency limits for broadband beamforming, respectively.

[0014] In some exemplary embodiments, the measurement random finite set is constructed as follows: The maxima on the normalized grazing angle-depth ambiguity plane are sorted in descending order of ambiguity value, and the grazing-depth coordinates corresponding to the top M maxima are selected to form the measurement set at time k. .

[0015] In some exemplary embodiments, the target motion state equation is:

[0016] in, This represents the random acceleration of the target in the two dimensions of glancing angle and depth. and This represents the target state transition matrix and the process noise matrix.

[0017] In some exemplary embodiments, the iterative process of the labeled multi-Bernoulli LMB filter includes: Prediction phase: Based on the multi-target state LMB parameter set at time k, combined with the new target LMB parameter set, the predicted LMB parameter set at time k+1 is obtained; Update phase: Convert the predicted LMB form to - In GLMB form, after multi-target state updates, it is approximated again in LMB form, and the updated target state parameters are output.

[0018] According to a second aspect of the present invention, a storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the deep-sea underwater target tracking method based on frequency domain interferometric acoustic field described in the first aspect.

[0019] According to a third aspect of the present invention, a computer program product is provided, on which a computer program is stored, wherein when the computer program is executed by a processor, the deep-sea underwater target tracking method based on frequency domain interference acoustic field described in the first aspect is implemented.

[0020] According to a fourth aspect of the present invention, an electronic device is provided, comprising: Processor; and Memory for storing the executable instructions of the processor; The processor is configured to implement the deep-sea underwater target tracking method based on frequency domain interferometric acoustic field described in the first aspect by executing the executable instructions.

[0021] The deep-sea underwater target tracking method based on frequency-domain interferometric acoustic field provided by the embodiments of the present invention, by combining complementary ensemble empirical mode decomposition algorithm to reconstruct the array beam output matrix, can enhance the interferometric characteristics of underwater targets in complex deep-sea environments and improve the estimation accuracy of the target's arrival grazing angle and depth. Simultaneously, by modeling the target motion equation and measurement set within a random finite set target tracking framework and iteratively filtering them using a labeled multi-Bernoulli filter, the complex data association process of traditional tracking algorithms is avoided, enabling accurate tracking of multiple underwater targets.

[0022] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0023] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention. It is obvious that the drawings described below are merely some embodiments of the invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0024] Figure 1 Implementation process of deep-sea underwater target tracking method based on frequency domain interferometric sound field; Figure 2 Typical deep-sea sound velocity profile; Figure 3 Schematic diagram showing the location of the underwater target and receiver array; Figure 4The array broadband beam output power in the presence of an underwater target, the reconstructed beam output power, and the target grazing angle-depth ambiguity plane obtained by mapping; Figure 5 Tag-based multi-Bernoulli filtering tracking results in an underwater single-target environment; Figure 6 The distribution of glancing angles of the target at different horizontal distances and depths in the model simulation; Figure 7 The actual distance, depth, and track output trajectory of a single underwater target during its motion; Figure 8 The actual distance, depth, and track output of each target in underwater multi-target motion scenarios. Detailed Implementation

[0025] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the invention will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0026] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0027] To address the shortcomings and deficiencies of existing technologies, this example embodiment provides a deep-sea underwater target tracking method based on frequency domain interferometric sound fields, referencing... Figure 1 As shown, the specific steps may include: The acoustic signals acquired by the deep-sea bottom-mounted vertical linear array are processed by frequency domain beamforming to obtain the manifold matrix of the receiving array. Based on the manifold matrix of the receiving array, the MVDR spatial power spectrum, i.e. the original beam output matrix, is calculated. The original beam output matrix of the array is reconstructed using the complementary integrated empirical mode decomposition (CEEMD) algorithm to obtain a beam output matrix with enhanced interferometric features. A generalized Fourier transform is performed on the beam output matrix with enhanced interferometric features to convert it into a target grazing angle-depth ambiguity plane; The maximum points on the ambiguity plane are extracted to form a random finite set for grazing angle-depth measurement. A two-dimensional uniform motion state equation for the underwater target in the grazing angle-depth plane is established. The two-dimensional uniform motion state equation is predicted and updated iteratively using a tag-based multi-Bernoulli filter to track and output the grazing angle-depth trajectory of the target. Based on the acoustic field model simulation, the mapping relationship between the horizontal distance, depth and the arrival glancing angle of the underwater target is established, and the glancing-depth trajectory is mapped to the horizontal distance-depth track to complete the underwater target tracking.

[0028] The steps in this exemplary embodiment will now be described in more detail with reference to the accompanying drawings and embodiments.

[0029] Example 1 Background information is as follows: The simulation environment is a typical deep-sea area in the South my country Sea, with a depth of 4000 m, and a seabed platform that satisfies the assumption of horizontal isotropy. The sound velocity profile is as follows. Figure 2 As shown, this is a typical deep-sea incomplete acoustic channel. The target sound source operates at depths shallower than 400 m, and the receiving array is a bottom-mounted 32-element vertical uniform linear array (array center depth and array spacing are 3936 m and 4 m, respectively). Deep-sea environmental noise is mainly considered as background noise from distant ships, and the sound field simulation calculations are all implemented using the Bellhop ray acoustic model.

[0030] The specific implementation process of the deep-sea underwater target tracking method based on frequency domain interferometric sound field is as follows: Step 1: Reconstruction of the wideband beam output matrix of the receiving array For a bottom-mounted vertical linear array deployed on the seabed directly in the deep sea acoustic zone, frequency-domain beamforming processing of the acoustic signals collected from each array element can distinguish the spatial distribution of signals with different frequency components. Under the far-field plane wave assumption, the flow matrix of the receiving array... for: (1) In the formula, For the frequency of the sound source, The direction of the incoming wave (grazing angle). and These are the number of array elements and the spacing between array elements, respectively. The speed of sound in ocean water.

[0031] Furthermore, the broadband spatial power spectrum of the minimum variance undistorted response (MVDR), i.e., the frequency-vertical angle of arrival (grazing angle) spectrum, can be calculated using the following formula: (2) In the formula, for The sampling covariance matrix at a given frequency point This indicates the conjugate transpose.

[0032] like Figure 3 As shown, within the deep-sea direct sound zone, the sound waves radiated by underwater broadband targets mainly travel via the direct sound path. and the sound path of a sea surface reflection Arrival at the array receiver The two acoustic waves, after coherent superposition, will appear as interference fringes on the broadband beam output of the array, which can be used for subsequent parameter estimation and tracking of underwater targets. For example... Figure 4 Figure a shows the distance to the underwater target. = 9 km, depth is The MVDR broadband beam output power of the receiving array at a grazing angle of 150 m can be seen. = 0.37 Underwater targets appear as alternating bright and dark frequency domain interference fringes, but in areas with strong marine environmental noise ( Under the influence of <0.3), the intensity of the interference fringes corresponding to the target is relatively weak, and the interference characteristics are not significant. Therefore, in order to further enhance the interference characteristics of the underwater target, the Complementary Integrated Empirical Mode Decomposition (CEEMD) algorithm is used to reconstruct the original beam output matrix of the array. The specific implementation steps are as follows: Step 1: Set the overall average frequency = 100.

[0033] Step 2: From Figure 4 The original array broadband beam output matrix shown in a) Extract ( = 1, 2,…, Sequence corresponding to grazing angle Adding a small amount of white noise to it according to the following formula can improve the stability of subsequent decomposition: (3) In the formula, For the first The added white noise sequence ( = 1, 2, …, ), This is the beam output sequence after adding noise.

[0034] Step 3: Use the Empirical Mode Decomposition (EMD) algorithm to... Decomposition yields the eigenmode components of each order: (4) in, and The decomposition results in the th The order intrinsic mode functions and residual terms.

[0035] Step 4: Regarding the above... The results of the decomposition are averaged: (5) Step 5: Sort all eigenmode functions of the above decomposition in descending order according to their variance contribution rates, and select those whose cumulative variance contribution rates are greater than a certain threshold. = 60% before Summing the eigenmode functions of order 1: (6) In the formula, This represents the intrinsic mode functions arranged in descending order. Indicates the first The variance of the intrinsic mode function components.

[0036] Step 6: For all glancing angles ( = 1, 2, …, The beam output sequence under the given conditions is processed using the above five steps, and the original beam output matrix is ​​reconstructed into an interferometric feature-enhanced beam output matrix according to the following formula: (7) like Figure 4 Figure b shows the reconstructed array broadband beam output power, which can be seen compared to... Figure 4 a, at the grazing angle The target frequency domain interference characteristics at 0.37 were further enhanced, while the impact of marine environmental noise was also reduced to some extent.

[0037] Step 2: Target grazing angle-depth ambiguity plane mapping Furthermore, the reconstructed beam output matrix is ​​calculated according to the following formula. Perform a generalized Fourier transform to convert the beam output matrix into a target grazing angle-depth ambiguity plane: (8) In the formula, = 100 Hz and = 300 Hz are the lower and upper limits of the frequency for broadband beamforming, respectively.

[0038] like Figure 4 c shows the (normalized) target grazing angle-depth ambiguity plane obtained through the above transformation. It can be seen that in ( = 0.37, There is an ambiguity peak at a location of 150 m, which corresponds to the true location of the underwater target.

[0039] Step 3: Target glancing angle-depth filtering tracking In the target grazing angle-depth ambiguity plane obtained through step 2, the location of the maximum ambiguity value usually corresponds to the arrival grazing angle and depth of the underwater target. However, under the influence of many practical factors such as strong marine environmental noise, severe sea states, target radiated noise fluctuations, and low signal-to-noise ratio, the maximum value on the ambiguity plane often does not correspond to a non-target location (called clutter), while the other second-largest values ​​have a certain probability of corresponding to the true location of the target. Although the quantity and state (grazing angle, depth) of clutter and underwater targets at each time point are random and uncertain, their quantity is finite and their state exists within a certain range. Therefore, the maximum points on the ambiguity plane at each time point can be formed into a set (i.e., a random finite set of measurements), and the elements (measurements) in the set contain relevant information about potential underwater targets. Assuming that in At any given time, all maxima points on the normalized ambiguity plane are sorted in descending order of ambiguity value, and the top ones are selected. = Grazing angle-depth coordinates corresponding to the 4 maxima ( and The set (representing the grazing angle and depth values ​​respectively) constitutes the following: Measurement set at time : (9) Furthermore, considering the underwater target undergoing two-dimensional uniform (CV) motion in the grazing angle-depth plane, then The target's state vector at time t is ,in and They represent The glancing angle of the target at time t and the rate of change of the glancing angle. and They represent The target depth and the rate of change in the depth direction at time t. Correspondingly, the target at the current time ( ) and the next moment ( The state transition equation between them is: (10) In the formula, This represents the random acceleration of the target in the two dimensions of glancing angle and depth. and Let represent the target state transition matrix and the process noise matrix, respectively, and their expressions are as follows: (11) (12) in, = 20 s is the sampling time interval.

[0040] After establishing the measurement random finite set and the target motion state equation based on equations (9) and (10) respectively, the grazing angle-depth of the underwater target can be further tracked using the tag-based Dobernouri (LMB) filter. The iterative filtering process of LMB is divided into two steps: LMB prediction and update. In the LMB prediction step, the target state equation is used to predict the target's state vector. Then, in the LMB update step, the measurement set is used to update (correct) the prediction result, as follows: Step 1: LMB Prediction Assuming in At any given moment, the underwater multi-target state can be determined using a random finite set. ( and Represent the target state space and the space defined by the labels, respectively. The discrete and distinct label space constitutes the corresponding parameter set. ( and These represent the tags as follows: The probability of the existence of the trajectory and the probability density distribution function of the target state). Accordingly, The probability density distribution function of the multi-objective state at time t can be written as: (13) in, (14) In addition, At this moment, the newly formed multi-objective state can also be represented by an LMB random finite set, with the corresponding parameter set being... , express The label space of newborn multi-objectives at the moment and ,but The probability density distribution function of the newly generated multi-objective state at time t can be written as: (15) in, (16) According to equations (13) and (15), The predicted multi-objective state probability density at time t also has a state space of t. Tag space is The LMB random finite set form, with the corresponding parameter set, can be written as: (17) In the formula, and They represent respectively to Surviving flight paths at that moment and their impact on At this moment, the new trajectory is Predictions at any given moment.

[0041] Accordingly, The multi-objective state probability density distribution function predicted at time t can be written as: (18) in: (19) (20) (twenty one) (twenty two) (twenty three) Step 2: LMB Update Although the LMB is closed during the prediction process, it is not closed during the update process. Therefore, an LMB approximation is needed for the multi-objective posterior probability density during the update process. The approximation process includes: expressing the predicted LMB form (multi-objective posterior probability density) as... -GLMB format, for -GLMB is being updated and the updated version is now available. -GLMB is re-represented as LMB. Specifically: According to the first step, The LMB multi-objective state probability density predicted at a given time can be obtained through the parameter set. It is expressed as follows, and its probability density distribution function is shown in equation (18). According to The relevant definition of -GLMB RFS can be expressed based on equation (18) as follows: it has only one component. -GLMB format: (twenty four) Furthermore, through The expression for the multi-objective probability density function after GLMB update is: (25) In the formula Indicates from the tag set arrive The mapping, that is, the association between the label set and the measurement set.

[0042] According to the relevant definition of LMB RFS, The multi-target probability density updated by GLMB (Equation (25)) is approximately expressed in LMB form, and is used with parameter sets. Let's represent it this way: the predicted probabilities of track existence and spatial distribution after LMB approximation are as follows: (26) (27) in: (28) (29) (30) (31) In equation (26) Indicates a contained function, if ,but = 1, otherwise = 0. (31) In the formula This indicates the probability of detecting the target; Let be the likelihood function, representing the state as , tag as Target generation measurement set The probability density; This represents the distribution density of clutter.

[0043] like Figure 5 The image shows the filtered tracking results of the tag's multi-Bernoulli filter output during the movement of an underwater target along a diameter within a 20 km radius centered on the vertical receiver array. It can be seen that when the target is at a relatively long distance (reaching a grazing angle...)... When <0.2), there are many clutter measurement points near the sea surface (<50 m). Figure 5 (b) It cannot output the target's true depth and grazing angle; when the target moves closer to the receiving array (20 min ~ 105 min), it can continuously output the underwater target's grazing angle and depth trajectory.

[0044] Step 4: Output the target trajectory for model matching By iteratively filtering the measurement random finite set composed of ambiguity plane coordinates at each time point using the LMB filter in step 3, the glancing angle-depth coordinate trajectory of the underwater target during its motion can be tracked and output. Based on this, a model matching method is used to map the target's glancing angle to a horizontal distance, thereby outputting the target's range-depth tracking track. The specific process is as follows: Using the Bellhop sound field model, underwater targets at different horizontal distances were simulated in advance. and depth Theoretical calculated value of the arrival grazing angle under the corresponding conditions And establish the following mapping relationship: (32) like Figure 6 The figure shows the distribution of the glancing angle of arrival of the target under different distance and depth conditions in the model simulation. It can be seen that the glancing angle of arrival is not sensitive to the change of the target depth, but it decreases exponentially with the increase of the target distance.

[0045] Furthermore, the target trajectory (grazing angle - depth) coordinates output in step 3 are... Substituting into the above formula, the horizontal distance to the target can be calculated and the final target tracking trajectory can be output. Figure 7 ): (33) To further verify the effectiveness of the proposed method in more complex environments, Figure 8 The method was validated in the multi-target scenario shown in Figure a, where target 1 (pink curve) has a depth of 150 m, a sound source level of 127 dB@200 Hz, and a speed of 13 knots; target 2 (green curve) has a depth of 100 m, a sound source level of 155 dB@200 Hz, and a speed of 8 knots; and target 3 (red curve) has a depth of 200 m, a sound source level of 143 dB@200 Hz, and a speed of 12 knots. Figure 8 The hollow and solid triangles in triangle a represent the starting and ending positions of the target, respectively. Figure 8 b and Figure 8 c presents the actual trajectories during the movement of multiple targets and the tracks tracked using the proposed method. It can be seen that because target 2 has a higher sound source level, its track can be continuously tracked for most of its movement. However, because target 2's initial position is far from the receiving array, the depth estimation error for target 2 is relatively large during the 20-40 minute period. Furthermore, target 3 can continuously output its track when it approaches the receiving array (<40 minutes), while target 1, due to its lower sound source level, can only have its track tracked for a short period when it is close to the receiving array.

[0046] This embodiment demonstrates that the method can be applied to underwater multi-target tracking in complex deep-sea environments. Addressing the problem of "blurred" target interferometric features due to deep-sea environmental noise, this invention proposes an array frequency domain beam output matrix reconstruction method based on complementary integrated empirical mode decomposition, which enhances the frequency domain interferometric features of underwater targets. Furthermore, by combining a tag-based multi-Bernoulli filter within a random finite set framework with a sound field theory model, it can achieve accurate tracking of underwater targets within a short-range (<20 km) area directly accessible to the sound field.

[0047] It should be noted that, as another aspect, this application also provides a storage medium, which may be included in an electronic device or may exist independently without being assembled into the electronic device. The storage medium carries one or more programs, which, when executed by an electronic device, cause the electronic device to perform the methods described in the following embodiments.

[0048] In one embodiment, this application provides a computer program product including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0049] Furthermore, the above figures are merely illustrative of the processes included in the method according to exemplary embodiments of the present invention, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Additionally, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0050] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention herein. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the claims.

[0051] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is defined only by the appended claims.

Claims

1. A deep-sea underwater target tracking method based on frequency domain interferometric sound field, characterized in that, The method includes: The acoustic signals acquired by the deep-sea bottom-mounted vertical linear array are processed by frequency domain beamforming to obtain the manifold matrix of the receiving array. Based on the manifold matrix of the receiving array, the MVDR spatial power spectrum, i.e. the original beam output matrix, is calculated. The original beam output matrix of the array is reconstructed using the complementary integrated empirical mode decomposition (CEEMD) algorithm to obtain a beam output matrix with enhanced interferometric features. A generalized Fourier transform is performed on the beam output matrix with enhanced interferometric features to convert it into a target grazing angle-depth ambiguity plane; The maximum points on the ambiguity plane are extracted to form a random finite set for grazing angle-depth measurement. A two-dimensional uniform motion state equation for the underwater target in the grazing angle-depth plane is established. The two-dimensional uniform motion state equation is predicted and updated iteratively using a tag-based multi-Bernoulli filter to track and output the grazing angle-depth trajectory of the target. Based on the acoustic field model simulation, the mapping relationship between the horizontal distance, depth and the arrival glancing angle of the underwater target is established, and the glancing-depth trajectory is mapped to the horizontal distance-depth track to complete the underwater target tracking.

2. The method according to claim 1, characterized in that, The implementation process of the complementary integrated empirical mode decomposition (CEEMD) algorithm includes: Set the overall average frequency N; Add a small-amplitude white noise signal to the beam output sequence corresponding to each grazing angle; The Empirical Mode Decomposition (EMD) algorithm is used to decompose the noise-added sequence to obtain the intrinsic mode components and residual terms of each order. Average the results of N decompositions; The intrinsic modal components are sorted in descending order of variance contribution rate, and the first K components with a cumulative variance contribution rate greater than a preset threshold are summed. Repeat the above steps for the beam output sequence corresponding to all grazing angles to reconstruct the beam output matrix with enhanced interferometric features.

3. The method according to claim 1, characterized in that, The formula for calculating the MVDR spatial power spectrum is as follows: in, for The sampling covariance matrix at a given frequency point This indicates the conjugate transpose.

4. The method according to claim 1, characterized in that, The generalized Fourier transform of the beam output matrix enhanced with interference features is performed to convert it into a target grazing angle-depth ambiguity plane, specifically as follows: in, and These are the lower and upper frequency limits for broadband beamforming, respectively.

5. The method according to claim 1, characterized in that, The method for constructing the measurement random finite set is as follows: The maxima on the normalized grazing angle-depth ambiguity plane are sorted in descending order of ambiguity value, and the grazing-depth coordinates corresponding to the top M maxima are selected to form the measurement set at time k. .

6. The method according to claim 1, characterized in that, The target motion state equation is: in, This represents the random acceleration of the target in the two dimensions of glancing angle and depth. and This represents the target state transition matrix and the process noise matrix.

7. The method according to claim 1, characterized in that, The iterative process of the labeled multi-Bernoulli LMB filter includes: Prediction phase: Based on the multi-target state LMB parameter set at time k, combined with the new target LMB parameter set, the predicted LMB parameter set at time k+1 is obtained; Update phase: Convert the predicted LMB form to - In GLMB form, after multi-target state updates, it is approximated again in LMB form, and the updated target state parameters are output.

8. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the deep-sea underwater target tracking method based on frequency domain interferometric acoustic field as described in any one of claims 1 to 7.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the deep-sea underwater target tracking method based on frequency domain interferometric sound field as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, include: processor; as well as Memory for storing the executable instructions of the processor; The processor is configured to execute the deep-sea underwater target tracking method based on frequency domain interferometric acoustic field as described in any one of claims 1 to 7 by executing the executable instructions.

Images