Deep sea broadband target passive positioning method based on sound pressure-vibration velocity combined observation

Abstract

The invention particularly relates to a deep sea broadband target passive positioning method based on sound pressure-vibration velocity combined observation, and belongs to the field of underwater acoustic detection. The method comprises the following steps: firstly, performing short-time Fourier transform on sound pressure and X / Y / Z three-direction vibration velocity signals to construct a frequency domain covariance matrix; a horizontal azimuth angle and a pitch angle of the target are obtained through MVDR two-dimensional azimuth spatial spectrum estimation, and the horizontal distance is calculated by combining the receiving depth; and finally, performing complementary integrated empirical mode decomposition on the sound pressure signal power spectrum density, extracting a maximum variance contribution rate eigenmode component, and performing generalized Fourier transform to obtain the target depth. The method does not need to depend on marine environment parameters, is high in environment adaptability, can highlight the frequency domain interference period, improves the depth estimation precision, is suitable for distance-depth passive positioning of a broadband target in a deep sea direct sound area, and is outstanding in engineering practicability.

Description

Technical Field

[0001] This invention relates to the field of underwater acoustic detection technology, specifically to a passive method for locating deep-sea broadband targets based on joint acoustic pressure and vibration velocity observation. Background Technology

[0002] In typical deep-sea environments, the low ambient noise (high signal-to-noise ratio) and good sound propagation characteristics of sensors or arrays deployed at great depths give them a certain detection advantage, thereby enabling passive detection and positioning of underwater targets.

[0003] In the early stages (before 2010), vertical linear arrays were widely used in passive target detection scenarios in the deep sea. They have advantages such as underwater attitude stability, high array gain, and ease of deployment. At the same time, passive positioning of underwater targets can be achieved by using mode-based signal processing techniques such as matched fields and matched modes, and their effectiveness has been verified by multiple sea trials. However, these processing techniques are highly dependent on the accuracy of environmental parameters and computational efficiency, and usually require a large array aperture to achieve more accurate target positioning results, which also increases the cost of engineering implementation to some extent. In recent years, with the increasing demand for networked and intelligent underwater warfare and observation from countries around the world, miniaturization, low power consumption, and high information capacity of detection platforms have become a major development trend. Vector hydrophones have also been gradually applied to deep-sea acoustic detection and positioning during this period. Compared with scalar (sound pressure) hydrophones mounted on traditional vertical linear arrays, they have richer particle velocity information. By using the coherence of the sound pressure and vibration velocity signals or the arrival angle of multipath signals, sound waves of different paths can be separated. Then, by estimating the time delay difference and angle of arrival difference between the deep-sea direct wave and the sea surface / seabed reflected wave, and combining geometric relationships, the distance and depth information of the target can be directly calculated.

[0004] Among the publicly available methods for passive target localization using sound pressure and vibration velocity components, horizontal direction finding based on the sound intensity meter method is relatively mature and widely used, providing stable and reliable target orientation information. However, target distance and depth estimation mainly relies on the unique physical characteristics of the deep-sea acoustic field, such as using waveguide invariant theory to invert spectral interferometric fringes to estimate distance, or using normal mode decomposition to perceive depth-related modal energy distribution. The accuracy of this method depends on precise knowledge of marine environmental parameters, but its robustness is insufficient under environmental mismatch conditions. Based on this background, this invention establishes a deep-sea broadband passive target localization method based on joint sound pressure and vibration velocity observations, and verifies the effectiveness of this method using sea trial data.

[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 deep-sea broadband target positioning scenarios, this invention provides a passive positioning method for deep-sea broadband targets based on joint acoustic pressure-vibration velocity observation. It aims to solve the problems of existing passive positioning methods for deep-sea broadband targets relying on marine environmental parameters, lacking robustness when there is environmental mismatch, and having limited positioning accuracy and high engineering implementation costs due to traditional array or single signal processing methods.

[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 passive localization method for deep-sea broadband targets based on joint acoustic pressure-vibration velocity observation is provided, the method comprising: Step 1: Construct the frequency domain covariance matrix of the sound pressure-velocity signal: Perform short-time Fourier transform on the time-synchronized sound pressure and vibration velocity time-domain signals in the X, Y, and Z directions to obtain the time spectrum of each component signal and construct the time spectrum matrix of the four component signals, and then calculate the frequency domain covariance matrix of the snapshot average. Step 2: MVDR two-dimensional azimuth spatial spectrum estimation: Construct a four-component signal steering vector containing the horizontal azimuth and elevation angles. Use the steering vector and the frequency domain covariance matrix to perform MVDR beamforming to obtain a narrowband two-dimensional azimuth spatial spectrum. Perform incoherent averaging on the narrowband two-dimensional azimuth spatial spectrum of each frequency point in the processing frequency band. Estimate the horizontal azimuth and elevation angles of the target signal based on the peak value of the averaged spatial spectrum, and estimate the horizontal distance of the target in combination with the receiving depth. Step 3: Depth estimation based on the frequency domain interference characteristics of the sound pressure signal: Calculate the power spectral density sequence of the sound pressure signal, perform complementary integrated empirical mode decomposition on the power spectral density sequence and extract the intrinsic mode components with the largest variance contribution rate, perform generalized Fourier transform on the intrinsic mode components, and obtain the depth ambiguity sequence after transformation to the depth domain. Estimate the target depth based on the peak value of the sequence.

[0009] In some exemplary embodiments, the short-time Fourier transform in step 1 divides each component long signal into multiple snapshots by segmenting and windowing, and performs Fourier transform on each snapshot to obtain the spectral information that changes over time.

[0010] In some exemplary embodiments, the formula for calculating the frequency domain covariance matrix is:

[0011]

[0012] in, The total number of snapshots for the short-time Fourier transform. This indicates the conjugate transpose. , , and They are respectively Time and The time-frequency spectrum of sound pressure, X-direction vibration velocity, Y-direction vibration velocity and Z-direction vibration velocity signals at a given frequency point.

[0013] In some exemplary embodiments, the configuration includes a four-component signal steering vector for the horizontal azimuth and elevation angles, specifically:

[0014] in, It is the horizontal azimuth. The pitch angle, and These represent the average density of seawater and the speed of sound, respectively.

[0015] In some exemplary embodiments, the step of performing incoherent averaging of the narrowband two-dimensional azimuth spatial spectrum at each frequency point within the processing frequency band, and estimating the horizontal azimuth and elevation angles of the target signal based on the peak values ​​of the averaged spatial spectrum, specifically involves:

[0016]

[0017] in, and These are the upper and lower limits of the processing frequency band, respectively. To handle the total number of frequency points, The maximum value corresponding to and This refers to the estimated horizontal azimuth and elevation angles of the target signal arrival.

[0018] In some exemplary embodiments, the step of estimating the horizontal distance of the target by combining the received depth specifically includes:

[0019] in, This refers to the receiving depth.

[0020] In some exemplary embodiments, the step of performing complementary integrated empirical mode decomposition on the power spectral density sequence and extracting the eigenmode components with the largest variance contribution rate specifically includes: power spectral density sequence of sound pressure signal A sequence of white noise with small amplitude applied:

[0021] in, For the first The white noise sequence added next time. = 1, 2, …, ; Using the empirical mode decomposition method to respectively and The decomposition yields its eigenmode components and residual terms of each order:

[0022] in, and From respectively and The first decomposition obtained from the middle eigenmode functions of order 1 and For the corresponding residual terms; right The eigenmode components of each order under the decomposition result are averaged: .

[0023] In some exemplary embodiments, the generalized Fourier transform of the intrinsic modulus components, after being transformed to the depth domain, yields a depth ambiguity sequence. The target depth is then estimated based on the peak values ​​of the sequence. Specifically, this involves:

[0024] in, For depth search sequences, The maximum value corresponds to the estimated depth of the underwater target.

[0025] The passive deep-sea target localization method based on joint acoustic pressure-velocity observation provided by the embodiments of the present invention can achieve passive localization of underwater broadband targets using acoustic pressure-velocity component signals without the need for environmental parameter information, and has good environmental adaptability. Simultaneously, by decomposing the power spectral density of the hydrophone acoustic pressure signal using a complementary integrated empirical mode decomposition algorithm, the frequency domain interference period can be further highlighted, achieving more accurate target depth estimation.

[0026] Compared with the prior art, the present invention has the following beneficial effects: 1. It does not depend on environmental parameters and has strong environmental adaptability. Existing passive target localization methods for deep-sea targets (such as matched field, waveguide invariant inversion, normal mode decomposition, etc.) are highly dependent on the accurate knowledge of marine environmental parameters (sound velocity profile, ground acoustic parameters, water depth, etc.). Once environmental mismatch occurs (such as deviation between the actual environment and model parameters), the localization accuracy drops significantly, and the robustness is insufficient. In contrast, this invention utilizes only the joint observation signal of the four components of sound pressure and vibration velocity to achieve localization through the time-domain, frequency-domain characteristics and geometric relationships of the signal itself. It does not require the prior acquisition or calibration of marine environmental parameters, can adapt to the complex and variable characteristics of the deep-sea environment, and can function stably in different sea areas and under different environmental conditions, reducing the pre-requisite requirements for environmental detection.

[0027] 2. High positioning accuracy and more accurate depth estimation In terms of distance estimation: By equating the sound pressure-velocity signal to a four-component array and combining it with MVDR two-dimensional azimuth spatial spectrum estimation technology, the horizontal azimuth and elevation angles of the target are accurately extracted, and then the horizontal distance is calculated through the geometric relationship between the transmitting and receiving positions. Sea trials show that the estimation error rate is only 6.9% at a target horizontal distance of 2.35km, 18.5% at 3.93km, and the overall average error rate is 13.2%, demonstrating excellent distance estimation accuracy in broadband target positioning scenarios.

[0028] In terms of depth estimation: An innovative Complementary Integrated Empirical Mode Decomposition (CEEMD) method is employed to process the power spectral density of the acoustic pressure signal, effectively avoiding mode aliasing and extracting the eigenmode components with the largest variance contribution rate. This significantly highlights the frequency domain interference period (the "spiculation" interference of the original power spectral density is eliminated, exhibiting clear periodic oscillation characteristics). Then, a generalized Fourier transform is used to map the frequency domain information to the depth domain, achieving accurate depth calculation. In sea trials at target depths of 50m and 100m, the average estimation error rate was only 5.8%, with an estimation error rate as low as 3.42% for the actual depth of 49.7m. This solves the problems of blurred perception of depth-related mode energy distribution and difficulty in extracting the interference period in existing methods.

[0029] 3. Adapts to the development trend of detection platforms and has strong engineering practicality. With the upgrading demands for networked and intelligent underwater combat and observation, detection platforms are developing towards miniaturization, low power consumption, and high information content. This invention is based on the joint observation of sound pressure and vibration velocity using vector hydrophones. Compared with traditional vertical linear arrays (which require large array apertures and high deployment costs), vector hydrophones are smaller, consume less power, and can provide richer particle vibration velocity information. They do not require complex array deployment and maintenance, are suitable for the application scenarios of miniaturized detection platforms, reduce engineering implementation costs, and the four-component signal processing flow is simple and efficient, making it easy to implement in engineering.

[0030] 4. Clearly defined scope of application and excellent positioning stability. This invention is applicable to broadband underwater targets in the 200-400Hz frequency band within the deep-sea direct sound range, covering the core application range of deep-sea passive detection. Verification was conducted using 13 sea trial samples at different distances (2.35km, 3.93km) and depths (50m, 100m). All samples achieved effective positioning, with distance and depth estimation error rates consistently controlled at low levels. No positioning failures due to changes in target position were observed, demonstrating good scenario adaptability and result stability.

[0031] 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

[0032] 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.

[0033] Figure 1 Flowchart for the implementation of a passive deep-sea target localization method based on joint acoustic pressure-vibration velocity observation signals; Figure 2 A geometrical diagram illustrating the arrival of acoustic signals directly at the target and reflected from the sea surface. Figure 3 To measure the density and sound velocity profiles of the deep sea; Figure 4 To receive sound pressure and three-directional particle velocity signals; Figure 5 The power spectral density of the received sound pressure signal and the eigenmode components with the maximum variance contribution rate after decomposition; Figure 6 This is a two-dimensional (horizontal azimuth-elevation) azimuth spectrum for MVDR. Figure 7 The target depth ambiguity vector; Figure 8 The results show the passive localization of the target (a) distance and (b) depth under multiple samples. Detailed Implementation

[0034] 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.

[0035] 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.

[0036] To address the shortcomings and deficiencies of existing technologies, this example embodiment provides a passive deep-sea target localization method based on joint acoustic pressure-velocity (ASP) and vibration velocity (VVV) observation signals. It constructs a frequency domain covariance matrix using the four components of ASP and VVV signals and obtains the target signal's vertical angle of arrival (UHVOA) through MVDR two-dimensional azimuth spectrum estimation. Furthermore, it performs complementary integrated empirical mode decomposition (CIMD) on the power spectral density sequence of the ASP signal and extracts the intrinsic mode components with the largest variance contribution rate. The target depth is then calculated by performing a generalized Fourier transform on the intrinsic mode components, thereby achieving range-depth estimation of underwater broadband targets.

[0037] refer to Figure 1 As shown, the specific steps may include: Step 1: Construct the frequency domain covariance matrix of the sound pressure-velocity signal: Perform short-time Fourier transform on the time-synchronized sound pressure and vibration velocity time-domain signals in the X, Y, and Z directions to obtain the time spectrum of each component signal and construct the time spectrum matrix of the four component signals, and then calculate the frequency domain covariance matrix of the snapshot average. Step 2: MVDR two-dimensional azimuth spatial spectrum estimation: Construct a four-component signal steering vector containing the horizontal azimuth and elevation angles. Use the steering vector and the frequency domain covariance matrix to perform MVDR beamforming to obtain a narrowband two-dimensional azimuth spatial spectrum. Perform incoherent averaging on the narrowband two-dimensional azimuth spatial spectrum of each frequency point in the processing frequency band. Estimate the horizontal azimuth and elevation angles of the target signal based on the peak value of the averaged spatial spectrum, and estimate the horizontal distance of the target in combination with the receiving depth. Step 3: Depth estimation based on the frequency domain interference characteristics of the sound pressure signal: Calculate the power spectral density sequence of the sound pressure signal, perform complementary integrated empirical mode decomposition on the power spectral density sequence and extract the intrinsic mode components with the largest variance contribution rate, perform generalized Fourier transform on the intrinsic mode components, and obtain the depth ambiguity sequence after transformation to the depth domain. Estimate the target depth based on the peak value of the sequence.

[0038] The steps in this exemplary embodiment will now be described in more detail with reference to the accompanying drawings and embodiments.

[0039] Step 1: Construction of the frequency domain covariance matrix of sound pressure-vibration velocity signals For time-synchronized sound pressure-velocity components, they can be equivalently represented as a four-component array. Considering that the received acoustic signals in actual deep-sea channels are usually non-stationary, short-time Fourier transforms are performed on each component signal. By segmenting and windowing, each component signal is divided into multiple snapshots, and the Fourier transform of each snapshot is calculated to obtain the spectral information of each component signal as a function of time, as follows: (1) In the formula , , and These are the time-domain signals of sound pressure, X-direction vibration velocity, Y-direction vibration velocity, and Z-direction vibration velocity, respectively. For window functions; , , and They are respectively Time and The time-frequency spectrum of sound pressure, X-direction vibration velocity, Y-direction vibration velocity and Z-direction vibration velocity signals at a given frequency point.

[0040] Furthermore, the time-frequency spectra of each component signal are arranged in the following matrix form: (2) Finally, the four-component signal is calculated according to the following formula. The average covariance matrix of snapshots at a given frequency (matrix size 4×4): (3) In the formula The total number of snapshots for the short-time Fourier transform. This indicates the conjugate transpose.

[0041] Step 2: MVDR Two-Dimensional Azimuth Spatial Spectrum Estimation Similar to the spatial spectrum estimation of array signals, the aforementioned four-component signal is equivalent to a four-channel array signal. By constructing a steering vector containing two-dimensional azimuth angles (horizontal azimuth and elevation angles), and using an azimuth spatial spectrum estimation algorithm, the estimated direction of arrival of the target wave can be obtained. The specific implementation process is as follows: Step 1: Construct the steering vector of the four-component signal as follows: (4) In the formula This is the horizontal azimuth angle (0° corresponds to the X direction, and counterclockwise is the positive direction); This is the pitch angle (0° corresponds to the horizontal direction, and upward is the positive direction); and These represent the average density of seawater and the speed of sound, respectively.

[0042] Step 2: Using the covariance matrix from Step 1 and the constructed guiding vector MVDR (Constrained Minimum Variance) beamforming was performed, and the narrowband MVDR two-dimensional azimuth space spectrum was obtained as follows: (5) Step 3: Perform incoherent averaging of the narrowband MVDR spatial spectrum calculated at each frequency point within the processing frequency band according to the following formula: (6) In the formula and These are the upper and lower limits of the processing frequency band, respectively. To process the total number of frequency points. The maximum value corresponding to and This refers to the estimated horizontal azimuth and elevation angles of the target signal arrival.

[0043] Finally, based on the geometric relationship between the transmitting and receiving positions, the horizontal distance from the target to the receiving position can be calculated using the following approximation: (7) In the formula This refers to the receiving depth.

[0044] Step 3: Depth estimation based on the frequency domain interferometry characteristics of the sound pressure signal For underwater broadband targets within the direct acoustic range, their radiated sound waves reach the receiving position primarily via a direct path and a first reflection path from the sea surface. The two sound wave signals have a path difference, resulting in coherent superposition at the receiving point. This leads to an interference pattern with periodically varying intensity in the received signal's spectrum, and the interference period is modulated by the target's depth. Therefore, by estimating the frequency domain interference period of the sound pressure signal, the target's depth information can be calculated. The specific process is as follows: Step 1: Transform the sound pressure time-domain signal to the frequency domain, and calculate the power spectral density of the received sound pressure component signal according to the following formula: (8) In the formula This is the discrete form of the sound pressure signal. and These represent the number of discrete time points and the sampling rate, respectively.

[0045] Step 2: Power spectral density sequence of sound pressure component signal Complementary integrated empirical mode decomposition is performed, and the intrinsic mode (IMF) component with the largest variance contribution rate is extracted. The decomposition process is as follows: (1) Set the overall average frequency .

[0046] (2) To avoid mode aliasing during the decomposition process, the power spectral density sequence of the sound pressure signal is analyzed. A sequence of white noise with small amplitude applied: (9) In the formula For the first The added white noise sequence ( = 1, 2, …, ).

[0047] (3) Using the empirical mode decomposition method to respectively analyze the... and The decomposition yields its intrinsic eigenmodes (IMF) components and residual terms of each order: (10) in and From respectively and The first decomposition obtained from the middle eigenmode functions of order 1 and The corresponding residual term represents the average trend of the sound pressure signal power spectral density sequence.

[0048] (4) The eigenmode components of each order under the decomposition result are averaged: (11) After completing the above decomposition process, the variance contribution rate of each order intrinsic modulus component is further calculated, and the component with the largest variance contribution rate is extracted. The intrinsic mode component can further highlight the interference period compared to the original power spectral density component.

[0049] Step 3: Analyze the intrinsic modulus components according to the following formula. Perform a generalized Fourier transform to transform it from the frequency domain to the depth domain, obtaining the ambiguity sequence of the target depth: (12) In the formula For depth search sequences, The maximum value corresponds to the estimated depth of the underwater target.

[0050] Example 1: Background information is as follows: The test scenario is as follows: Figure 2As shown, the near-shore test vessel suspended a sound source and emitted multiple sets of 200-400 Hz linear frequency modulated signals at fixed points at different horizontal distances and depths. The receiving sensor depth was 3916 m. The measured density and sound velocity profiles of the test sea area are shown below. Figure 3 As shown, the local water depth is approximately 4200 m, and the acoustic axial depth is approximately 1100 m.

[0051] The specific implementation process of the deep-sea broadband target passive localization method is as follows: Step 1: Construction of the frequency domain covariance matrix of sound pressure-vibration velocity signals like Figure 4 The image shows the time-domain signals of the target sound pressure and particle velocities in the X, Y, and Z directions received by the sensor. Since the target signals received in the actual deep sea are non-stationary, short-time Fourier transforms are required for each component signal. By segmenting and windowing, each long signal component is divided into multiple snapshots, and the Fourier transform of each snapshot is calculated to obtain the spectral information of each component signal over time, as detailed below: (13) In the formula, , , and These are the time-domain signals of sound pressure, X-direction vibration velocity, Y-direction vibration velocity, and Z-direction vibration velocity, respectively. For window functions, we choose the Hamming window here; , , and They are respectively Time and The time-frequency spectrum of sound pressure, X-direction vibration velocity, Y-direction vibration velocity and Z-direction vibration velocity signals at a given frequency point.

[0052] Furthermore, the sound pressure-velocity component signal can be equivalently represented as a four-channel array, with the time-frequency spectra of each component signal arranged in the following matrix form: (14) Finally, the four-component signal is calculated according to the following formula. The average covariance matrix of snapshots at a given frequency (matrix size 4×4): (15) In the formula, The total number of snapshots divided for the short-time Fourier transform (each snapshot lasts 1 second, and the snapshot overlap rate is 75%). This indicates the conjugate transpose.

[0053] Step 2: MVDR Two-Dimensional Azimuth Spatial Spectrum Estimation Similar to the spatial spectrum estimation of array signals, by constructing a steering vector containing two-dimensional azimuth angles (horizontal azimuth and elevation angles) and using an azimuth spatial spectrum estimation algorithm, the estimated direction of arrival of the target wave can be obtained. The specific implementation process is as follows: Step 1: Construct the steering vector of the four-component signal as follows: (16) In the formula, Horizontal azimuth ( 0° corresponds to the X direction, and counterclockwise is the positive direction. pitch angle ( (0° corresponds to the horizontal direction, and upward is the positive direction). = 1025 kg / m 3 and = 1500 m / s, representing the average density of seawater and the speed of sound, respectively.

[0054] Step 2: Using the covariance matrix from Step 1 and the constructed guiding vector MVDR (Constrained Minimum Variance) beamforming was performed, and the narrowband MVDR two-dimensional azimuth space spectrum was obtained as follows: (17) Step 3: Perform incoherent averaging of the narrowband MVDR spatial spectrum calculated at each frequency point within the processing frequency band according to the following formula: (18) In the formula, = 200 Hz and = 400 Hz represents the upper and lower limits of the processing frequency band. = 201 Total number of frequency points processed.

[0055] like Figure 5 The figure shows the two-dimensional azimuth spectrum estimation results after broadband incoherent averaging (spectral values ​​have been normalized), where Maximum value ( Figure 5 The white cross in the middle corresponds to = 110° and = 58° is the estimated horizontal azimuth and elevation angle of the target signal arrival.

[0056] Finally, according to Figure 2 The geometric relationship between the transmitting and receiving positions shown can be used to calculate the horizontal distance between the target and the receiving position using the following approximation: (19) In the formula, = 3916 m is the receiving depth.

[0057] The horizontal distance to the target estimated through the above steps is 2.45 km (the actual horizontal distance to the target is 2.35 km), and the distance estimation error rate calculated according to the following formula is 4.20%.

[0058] (20) In the formula, and These represent the estimated and measured values ​​of the target location parameters, respectively.

[0059] Step 3: Depth estimation based on the frequency domain interferometry characteristics of the sound pressure signal like Figure 2 As shown, for underwater broadband targets within the direct sound zone... Its radiated sound waves mainly travel through a direct path. First reflection path from the sea surface Arrive at the receiving location The two acoustic signals have a path difference, resulting in coherent superposition at the receiver. According to ray acoustics and virtual source theory, the vertical angles of arrival (pitch angles) of the two signals are relatively close, and can be approximated as follows: direction reached The path signal, simultaneously exhibiting interference characteristics with periodically varying intensity in the received signal's spectrum, has its interference period modulated by the target depth. Therefore, by estimating the frequency domain interference period of the sound pressure signal, the target depth information can be calculated. The specific process is as follows: Step 1: Transform the sound pressure time-domain signal to the frequency domain, and calculate the power spectral density of the received sound pressure component signal according to the following formula: (twenty one) In the formula, This is the discrete form of the sound pressure signal. and These represent the number of discrete time points and the sampling rate, respectively.

[0060] Step 2: Power spectral density sequence of sound pressure component signal Complementary integrated empirical mode decomposition is performed, and the intrinsic mode (IMF) component with the largest variance contribution rate is extracted. The decomposition process is as follows: (1) Set the overall average frequency = 500.

[0061] (2) To avoid mode aliasing during the decomposition process, the power spectral density sequence of the sound pressure signal is analyzed. A sequence of white noise with small amplitude applied: (twenty two) In the formula, For the first The added white noise sequence ( = 1, 2, …, ).

[0062] (3) Using the empirical mode decomposition method to respectively analyze the... and The decomposition yields its intrinsic eigenmodes (IMF) components and residual terms of each order: (twenty three) in, and From respectively and The first decomposition obtained from the middle eigenmode functions of order 1 and The corresponding residual term represents the average trend of the sound pressure signal power spectral density sequence.

[0063] (4) The eigenmode components of each order under the decomposition result are averaged: (twenty four) After completing the above decomposition process, the variance contribution rate of each order intrinsic modulus component is further calculated, and the component with the largest variance contribution rate is extracted. The intrinsic mode component can further highlight the interference period compared to the original power spectral density component.

[0064] like Figure 6 The figure shows the power spectral density of the received sound pressure signal and the eigenmode components with the maximum variance contribution rate obtained after decomposition. It can be seen that the original power spectral density ( Figure 6 The red curve (in the middle) shows an overall decreasing trend with frequency, accompanied by several violent oscillations. The power spectral density curve exhibits numerous spikes, and the frequency domain interference period is relatively indistinct. After the above decomposition process, the maximum variance contribution rate of the eigenmode ( Figure 6 The green component is relatively "smooth" and exhibits a periodic oscillation characteristic in the form of a cosine, corresponding to the frequency domain interference period of the target signal.

[0065] Step 3: Decompose the intrinsic mode components obtained from the decomposition. The estimated vertical angle of arrival (pitch angle) of the target signal obtained in step 2 is substituted into the following formula for generalized Fourier transform, which transforms it from the frequency domain to the depth domain, yielding the target depth ambiguity sequence: (25) In the formula For depth search sequences, The maximum value corresponds to the estimated depth of the underwater target.

[0066] like Figure 7 The image shows the normalized depth ambiguity sequence, with its maximum value. = 48 m is the estimated value of the target depth (the actual target depth is 49.7 m). According to formula (20), the error rate of the target depth estimation is 3.42%.

[0067] To further verify the effectiveness of the proposed method, a total of 13 sample signals emitted by sound sources at different horizontal distances and depths during the experiment were used for validation. The sound source distances included 2.35 km and 3.93 km, and the sound source depths included 50 m and 100 m. A comparison of the measured and method estimation results is provided below. Figure 8 As shown, when the target's horizontal distance is 3.93 km, the average error rates for target distance and depth estimation by the proposed method are 18.5% and 7.1%, respectively; when the target's horizontal distance is 2.35 km, the average error rates for target distance and depth estimation are 6.9% and 4.3%, respectively. The average error rates for target distance and depth estimation after averaging all samples are 13.2% and 5.8%, respectively, indicating that the proposed method can achieve effective passive localization of broadband underwater targets in deep water.

[0068] This embodiment demonstrates that the method can be applied to passive localization of broadband underwater targets within typical deep-sea acoustic zones. The proposed method can achieve passive localization of broadband underwater targets using only the four components of sound pressure and vibration velocity signals without requiring environmental parameter information, exhibiting good environmental adaptability. Furthermore, by combining the complementary ensemble empirical mode decomposition algorithm to decompose the power spectral density of the sound pressure signal, the frequency domain interference period can be further highlighted, resulting in more accurate target depth estimation.

[0069] 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. Furthermore, it is readily understood that these processes may, for example, be executed synchronously or asynchronously in multiple modules.

[0070] 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.

[0071] 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 passive localization method for deep-sea broadband targets based on joint acoustic pressure-vibration velocity observation, characterized in that, The method includes: Step 1: Construct the frequency domain covariance matrix of the sound pressure-velocity signal: Perform short-time Fourier transform on the time-synchronized sound pressure and vibration velocity time-domain signals in the X, Y, and Z directions to obtain the time spectrum of each component signal and construct the time spectrum matrix of the four component signals, and then calculate the frequency domain covariance matrix of the snapshot average. Step 2: MVDR two-dimensional azimuth spatial spectrum estimation: Construct a four-component signal steering vector containing the horizontal azimuth and elevation angles. Use the steering vector and the frequency domain covariance matrix to perform MVDR beamforming to obtain a narrowband two-dimensional azimuth spatial spectrum. Perform incoherent averaging on the narrowband two-dimensional azimuth spatial spectrum at each frequency point within the processing frequency band. Estimate the horizontal azimuth and elevation angles of the target signal based on the peak value of the averaged spatial spectrum, and estimate the horizontal distance of the target by combining the receiving depth. Step 3: Depth estimation based on the frequency domain interference characteristics of the sound pressure signal: Calculate the power spectral density sequence of the sound pressure signal, perform complementary integrated empirical mode decomposition on the power spectral density sequence and extract the intrinsic mode components with the largest variance contribution rate, perform generalized Fourier transform on the intrinsic mode components, and obtain the depth ambiguity sequence after transformation to the depth domain. Estimate the target depth based on the peak value of the sequence.

2. The method according to claim 1, characterized in that, The short-time Fourier transform described in step 1 divides the long signal of each component into multiple snapshots by segmenting and windowing. Each snapshot is then subjected to a Fourier transform to obtain the spectral information that changes over time.

3. The method according to claim 1, characterized in that, The formula for calculating the frequency domain covariance matrix is ​​as follows: in, The total number of snapshots for the short-time Fourier transform. This indicates the conjugate transpose. , , and They are respectively Time and The time-frequency spectrum of sound pressure, X-direction vibration velocity, Y-direction vibration velocity and Z-direction vibration velocity signals at a given frequency point.

4. The method according to claim 3, characterized in that, The construction includes a four-component signal steering vector for the horizontal azimuth and elevation angles, specifically: in, It is the horizontal azimuth. The pitch angle, and These represent the average density of seawater and the speed of sound, respectively.

5. The method according to claim 4, characterized in that, The process involves incoherently averaging the narrowband two-dimensional azimuth spatial spectrum at each frequency point within the processing frequency band, and estimating the horizontal azimuth and elevation angles of the target signal based on the peak values ​​of the averaged spatial spectrum. Specifically: in, and These are the upper and lower limits of the processing frequency band, respectively. To handle the total number of frequency points, The maximum value corresponding to and This refers to the estimated horizontal azimuth and elevation angles of the target signal arrival.

6. The method according to claim 5, characterized in that, The method of estimating the horizontal distance of the target by combining the received depth is specifically as follows: in, This refers to the receiving depth.

7. The method according to claim 6, characterized in that, The process of performing complementary integrated empirical mode decomposition on the power spectral density sequence and extracting the eigenmode components with the largest variance contribution rate specifically involves: power spectral density sequence of sound pressure signal A sequence of white noise with small amplitude applied: in, For the first The white noise sequence added next time. = 1, 2, …, ; Using the empirical mode decomposition method to respectively and The decomposition yields its eigenmode components and residual terms of each order: in, and From respectively and The first decomposition obtained from the middle eigenmode functions of order 1 and For the corresponding residual terms; right The eigenmode components of each order under the decomposition result are averaged: 。 8. The method according to claim 7, characterized in that, The generalized Fourier transform of the intrinsic mode components, after being transformed to the depth domain, yields a depth ambiguity sequence. The target depth is estimated based on the peak values ​​of the sequence. Specifically: in, For depth search sequences, The maximum value corresponds to the estimated depth of the underwater target.

Images