Robust discrimination method for vertical array line spectrum target depth attribute

Abstract

The present application belongs to the field of underwater target recognition, and proposes a robust discrimination method for the depth attribute of line spectrum target of vertical array. The method increases sampling points in the depth direction of shallow water waveguide, filters the internal coupled mode of the received sound pressure amplitude of vertical array by using the mode orthogonality formed by vertical array sampling, and extracts the robust line spectrum sound pressure amplitude fluctuation feature in the horizontal distance dimension of sound source. The sound pressure amplitude of vertical array is obtained by fusing the array element data. The mode orthogonality makes the sound pressure amplitude of vertical array eliminate the influence of the coherent term of normal mode related to the distance of sound source and the receiving depth on the line spectrum amplitude fluctuation feature. The present application uses a small amount of sound pressure amplitude data of vertical array to construct the discrimination feature quantity sound pressure amplitude fluctuation index, realizes efficient discrimination, and has small calculation amount. At the same time, the discrimination performance of the depth attribute of sound source does not decrease with the change of the relative horizontal distance of sound source, and has robustness.

Description

Technical Field

[0001] This invention belongs to the field of underwater target recognition based on underwater acoustics and surface, and specifically relates to a robust method for determining the depth attributes of line spectrum targets for vertical arrays. Background Technology

[0002] Determining the depth attributes of sound sources in shallow marine environments is one of the key challenges and areas of expertise in passive sonar detection. Accurately identifying whether a non-cooperative sound source is located on the surface or underwater is of significant practical importance for ensuring safety in underwater environments.

[0003] When a sound source moves in the ocean, its depth fluctuates due to waves and the inhomogeneity of the seawater. This fluctuation in depth causes varying degrees of amplitude variation in the line spectrum noise radiated by surface and underwater sound sources at the receiving point, with surface sound sources exhibiting stronger line spectrum fluctuations than underwater sources. This difference in fluctuation characteristics can be used to distinguish between surface and underwater sound sources. Existing research indicates that the essence of line spectrum fluctuations is the fluctuation of different orders of normal modes excited by the sound source at the surface and underwater. Therefore, researchers have successively constructed the Modal Scintillation Index (MSI), the Modified-MSI (MMSI), and the Fused-MSI (FMSI) to distinguish surface and underwater line spectrum targets by utilizing the differences in the fluctuations of different orders of modes at the surface and underwater. However, the above methods rely on separating each order of modes from the signal, which places high demands on computation, often proving difficult to meet.

[0004] To achieve more efficient sound source depth attribute discrimination, the sound pressure amplitude at the receiving point of surface and underwater line spectrum sound sources is directly used to construct discrimination features. Existing technology discloses a method for discriminating the depth attributes of line spectrum targets in horizontally towed arrays ([RAWagstaff 1997] The AWSUM Filter: A 20-dB Gain Fluctuation-based Processor). This method constructs normalized discrimination features with a relatively small amount of data, enabling target depth attribute discrimination. However, when the horizontal distance of the sound source changes, the difference in the fluctuation characteristics of surface and underwater sound pressure amplitude acquired by a passive sonar system at a fixed operating depth weakens. This leads to a decrease in the discrimination accuracy of this method.

[0005] The invention patents with application number 2020102920511, entitled "A Single Hydrophone Target Recognition Method Based on Fusion Modal Scintillation Index", and application number 2020106130377, entitled "A Target Depth Identification Method and System Based on Modal Scintillation Index Matching Analysis", both utilize fluctuating line spectra and separate modes from signals to qualitatively identify surface or underwater targets. However, they involve a large amount of computation. The former requires a long sound source motion time for identification, and the latter uses a matching processing method, which makes the stability worse when the environment changes.

[0006] In summary, existing methods for determining the depth attributes of targets with undulating line spectra cannot simultaneously achieve high discrimination efficiency and good discrimination performance. Therefore, there is an urgent need to utilize new discrimination methods that can achieve robust target depth attribute determination. Summary of the Invention

[0007] To overcome the shortcomings of existing target depth attribute discrimination methods based on fluctuating line spectrum amplitudes, which struggle to achieve both discrimination efficiency and performance, this invention proposes a robust discrimination method for shallow-sea line spectrum target depth attributes using vertical arrays. This method increases sampling points along the depth direction of the shallow-sea waveguide and utilizes the modal orthogonality formed by vertical array sampling to filter the inherent coupling modes of the received sound pressure amplitude from the vertical array, extracting robust line spectrum sound pressure amplitude fluctuation features in the horizontal distance dimension of the sound source. The vertical array metadata is then fused to obtain the vertical array sound pressure amplitude. Modal orthogonality eliminates the influence of normal mode coherence terms, which are related to the sound source distance and receiving depth, on the line spectrum amplitude fluctuation features. This invention utilizes a small amount of vertical array sound pressure amplitude data to construct a discrimination feature quantity, the sound pressure amplitude fluctuation index, achieving efficient discrimination. Simultaneously, the sound source depth attribute discrimination performance does not decrease with changes in the relative horizontal distance of the sound source, demonstrating robustness.

[0008] The technical solution of this invention is: a robust method for determining the depth attribute of line spectrum targets in a vertical array, comprising the following steps:

[0009] Step 1: Extract the line spectrum sound pressure amplitude from the time-domain signal received by the vertical array elements;

[0010] Step 2: Filter out the horizontally distance-dependent normal mode coherence terms using modal orthogonality to obtain the vertical array acoustic pressure amplitude;

[0011] Step 3: Calculate the normalized depth attribute discrimination feature quantity related to the fluctuation of the vertical array sound pressure amplitude;

[0012] Step 4: Preset the discrimination threshold based on the waveguide environment information, and compare the size of the discrimination feature quantity with the discrimination threshold value.

[0013] Furthermore, step 1 includes the following sub-steps:

[0014] Step 1.1: In a shallow sea environment with a water depth of D, the target radiation line spectrum signal is received using an N-element vertical array. The frequency domain sound pressure field of the line spectrum received by the nth element is expressed as:

[0015]

[0016] z n z s Here, r represents the receiving depth of the nth element of the vertical array and the sound source depth, respectively, and r is the relative horizontal distance between the sound source and the vertical array. With k rm Let represent the mode function and horizontal wavenumber of the m-th mode excited by the sound source, respectively. Let Q represent the sound pressure amplitude correlation term. The target moves away from the vertical array, and the fluctuation of the motion depth is Δz. s M is an M-order normal wave excited by a sound source, and N is greater than M;

[0017] Step 1.2: Define the receiving depth of the nth array element as z. n The nth time-domain signal acquired by the vertical array within time t is denoted as x. n (t). The nth time-domain signal within time t is divided into L time-domain sub-bands, resulting in a total of N×L time-domain sub-bands x of the time-domain signal received by the vertical array. n (t l For each frame of the time-domain signal, perform a Fourier transform to the frequency domain; determine the target line spectrum frequency, and extract the N×L dimensional sound pressure matrix of the target frequency point, denoted as Y:

[0018] Y = [p1, p2, ..., p L ]

[0019] Where, p l This represents the column vector composed of the frequency domain sound pressures of all elements of the vertical array within the l-th frame;

[0020] Without considering changes in the horizontal distance of the sound source, the sound pressure amplitude of the l-th frame frequency domain line spectrum received by the vertical array can be expressed in vector form using equation (1).

[0021] p l =[p(r,z0,z)] s ),p(r,z1,z s ),,p(r,z N-1 ,z s )] T

[0022] =a T ψ

[0023] Where the superscript T denotes matrix transpose, a represents the modal matrix corresponding to the vertical matrix, and ψ represents the sound source location matrix, and the expressions are as follows:

[0024]

[0025]

[0026] Furthermore, step 2 includes the following sub-steps:

[0027] Step 2.1: Calculate the sum of the frequency domain acoustic power received by the N vertical array elements in the l-th frame.

[0028]

[0029] Step 2.2: Combining with Step 2.1, further obtain the vertical array acoustic pressure amplitude of the l-th frame received by the vertical array:

[0030]

[0031] The vector composed of the vertical array sound pressure amplitudes of the L frames is denoted as...

[0032]

[0033] Where, diag represents taking the diagonal elements;

[0034] 4. The robust discrimination method for line spectrum target depth attributes for vertical arrays as described in claim 1, characterized in that, in step 3, the L-frame vertical array received sound pressure amplitude obtained in step 2 is used to calculate a discrimination feature quantity. The discrimination feature quantity is named the Amplitude Fluctuation Index (AFI), specifically written as...

[0035]

[0036] Where, |p l | represents the vertical array sound pressure level in the l-th frame, and the denominator is the harmonic average of the vertical array received sound pressure levels in the L frames.

[0037] Furthermore, step 4 includes the following sub-steps:

[0038] Step 4.1: At different sound source depths, use the acoustic toolbox to calculate the result of equation (13) corresponding to the line spectrum frequency to be judged, and draw the depth distribution curve of the sound pressure amplitude fluctuation index;

[0039] Step 4.2: Determine the critical depth for distinguishing between surface and underwater targets, and preset the discrimination threshold AFIh based on the sound pressure amplitude fluctuation index curve;

[0040] Step 4.3: Compare the sound pressure amplitude fluctuation index AFI calculated in Step 3 with the discrimination threshold AFIh to determine whether the target is located on the water surface or underwater. If AFI < AFIh, the target is an underwater target; otherwise, the target is a surface target.

[0041] Invention Effects

[0042] The technical advantages of this invention are as follows: The basic principles and implementation schemes of this invention have been verified by computer numerical simulations. The results show that this invention directly utilizes the line spectrum sound pressure amplitude received by the vertical array to construct discriminative features, requiring minimal computation. Simultaneously, the method proposed in this invention fully utilizes the modal orthogonality formed by the vertical array to filter the sound source horizontal distance term that affects the differences in the fluctuation characteristics of the line spectrum amplitude at the water surface and underwater, thus exhibiting superior sound source depth attribute discrimination performance compared to methods relying solely on reception depth. This invention possesses both advantages in discrimination efficiency and performance, enabling robust line spectrum target depth attribute discrimination. Attached Figure Description

[0043] Figure 1 A coordinate schematic diagram of the discrimination method proposed in this invention;

[0044] Figure 2 The flowchart of this invention

[0045] Figure 3 The process for processing signals acquired by the vertical array to determine the target depth attribute;

[0046] Figure 4(a) shows the vertical array sound pressure amplitude corresponding to the 5m and 70m sound sources, and Figure 4(b) shows the discrimination results of the depth attributes of the 5m and 70m sound sources.

[0047] Figure 5 The sound pressure amplitude fluctuation index at different sound source distances;

[0048] Figure 6 The existing discrimination method (single reception depth) and the method proposed in this invention have discrimination-false alarm probability curves. Detailed Implementation

[0049] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0050] See Figures 1-6 This invention proposes a robust method for discriminating the depth attributes of shallow-sea line spectrum targets using a vertical array. This method increases sampling points along the depth direction of the shallow-sea waveguide and utilizes the modal orthogonality formed by the vertical array sampling to filter the inherent coupling modes of the received sound pressure amplitude from the vertical array, extracting robust line spectrum sound pressure amplitude fluctuation features in the horizontal distance dimension of the sound source. The vertical array metadata is then fused to obtain the vertical array sound pressure amplitude. Modal orthogonality eliminates the influence of normal mode coherence terms, which are related to the sound source distance and receiving depth, on the line spectrum amplitude fluctuation features. This invention utilizes a small amount of vertical array sound pressure amplitude data to construct a discriminative feature quantity, the sound pressure amplitude fluctuation index, achieving efficient discrimination. Furthermore, the sound source depth attribute discrimination performance does not decrease with changes in the relative horizontal distance of the sound source, demonstrating robustness.

[0051] The main contents of this invention are:

[0052] 1. In shallow sea environments, a vertical linear array is used to receive the target radiation line spectrum noise signal. The vertical array covers the entire water depth and is arranged at equal intervals. The number of array elements in the vertical array is greater than the normal mode order excited by the sound source in shallow sea. The number of array elements in the vertical array is increased along the depth direction of the sound source to fully sample the modes excited by the sound source. The inherent coupling modes of the sound pressure amplitude received by the vertical array are filtered using modal orthogonality to extract the robust line spectrum fluctuation characteristics of the sound source in the horizontal distance dimension.

[0053] 2. Extracting sound pressure amplitude fluctuation characteristics from the vertical array acquisition signal. Divide the time-domain signal of the N array element channels into N×L time-domain sub-bands and perform a Fourier transform to extract the N×L-dimensional sound pressure matrix of the target frequency point of the line spectrum. Multiply the conjugate transpose of the frequency-domain sound pressure matrix by itself, and extract all diagonal elements from the resulting L×L-dimensional square matrix to obtain the L×1-dimensional vertical array element sound power and vector after modal filtering. Take the square root of the sound power and vector to obtain the L-frame vertical array sound pressure amplitude with stable fluctuation characteristics, defined as the vertical array sound pressure amplitude. Calculate the harmonic mean and arithmetic mean of the L-frame vertical array sound pressure amplitude to obtain the normalized characteristic quantity characterizing the sound pressure amplitude fluctuation, defined as the sound pressure amplitude fluctuation index.

[0054] 3. Plot the sound pressure amplitude fluctuation index curve based on environmental information and preset the discrimination threshold. Compare the sound pressure amplitude fluctuation index with the discrimination threshold value to obtain the depth attribute discrimination result of the sound source.

[0055] 4. The feasibility of the discrimination method proposed in this invention is demonstrated through computer numerical simulation. The accuracy of sound source depth attribute discrimination is compared with that of the horizontal array method with fixed reception depth. This proves that the method proposed in this invention can achieve more robust sound source depth attribute discrimination performance.

[0056] like Figure 2As shown in the flowchart, the technical solution adopted by this invention to solve the existing problems can be divided into the following four steps:

[0057] 1) In shallow sea environments, an N-element vertical array (N greater than the normal mode order excited by the sound source) is used to receive the target radiation line spectrum noise signal. Time-domain signals where the sound source's movement distance is much smaller than its horizontal distance are extracted within a given time period, dividing the time-domain signal into L time-domain sub-bands. Each frame of the time-domain signal is Fourier transformed to the frequency domain, and the N×L dimensional sound pressure matrix of the target frequency points in the line spectrum is extracted.

[0058] 2) Calculate the product of the conjugate transpose of the N×L dimensional sound pressure matrix from 1) and itself to obtain an L×L dimensional square matrix. Extract all diagonal elements from the L×L dimensional square matrix to obtain an L×1 dimensional vector composed of the sum of the acoustic power of the vertical array elements. Due to the modal orthogonality formed by the vertical array, the vector elements filter out the normal mode coherence terms related to the horizontal distance from the sound source. Calculate the square root of the vector elements to obtain the L-frame vertical array sound pressure amplitude. The vertical array sound pressure amplitude exhibits robust fluctuation characteristics in the horizontal distance dimension from the sound source.

[0059] 3) Calculate the harmonic mean and arithmetic mean of the sound pressure amplitude of the vertical array in frame L in 2) to obtain the normalized sound pressure amplitude fluctuation characteristic, i.e., the sound pressure amplitude fluctuation index.

[0060] 4) Obtain water depth, sound velocity profiles, and seabed data. Use the acoustic toolbox to plot the depth distribution curve of the sound pressure amplitude fluctuation index in line spectrum 1). Determine the critical depth for distinguishing between surface and underwater targets, and preset a discrimination threshold on the sound pressure amplitude fluctuation index curve. In a given shallow sea environment, the threshold is independent of the horizontal distance from the sound source and is stable. Compare the sound pressure amplitude fluctuation index in 3) with the discrimination threshold. If the discrimination characteristic is less than the threshold, the target is underwater; otherwise, the target is at the surface.

[0061] Each step of the present invention will be described in detail below:

[0062] The specific content involved in step 1) is as follows:

[0063] like Figure 1 As shown, in a shallow sea environment with a water depth of D, an N-element vertical array is used to receive the target radiation line spectrum signal. The nth element receives signals at a depth of z. n The nth time-domain signal acquired by the vertical array within time t is denoted as x. n (t). The sound source excites an M-order normal wave. Where N is greater than M. According to the normal wave propagation theory, the line spectrum frequency domain sound pressure field received by the nth array element can be expressed as:

[0064]

[0065] Among them, z n zs Here, r represents the receiving depth of the nth element of the vertical array and the sound source depth, respectively, and r is the relative horizontal distance between the sound source and the vertical array. With k rm Let represent the mode function and horizontal wavenumber of the m-th mode excited by the sound source, respectively, and Q represent the sound pressure amplitude correlation term. The target moves away from the vertical array, and the fluctuation in the depth of motion is Δz. s .

[0066] like Figure 3 As shown in the flowchart, the time-domain signal is divided into N×L frame time-domain subbands x n (t l In a set of signals, the sound source depth changes once per frame. A Fourier transform is performed on each frame of the time-domain signal to the frequency domain. The target line spectrum frequency is determined, and the N×L dimensional sound pressure matrix Y of the target frequency point is extracted. The sound pressure matrix Y consists of L frequency-domain sound pressure column vectors, denoted as...

[0067] Y = [p1, p2, ..., p L (2)

[0068] Where, p l This represents the column vector consisting of the frequency domain sound pressures of all elements of the vertical array within the l-th frame.

[0069] The spectral fluctuations are a rapidly changing characteristic; the distance the sound source travels within L frames is much smaller than the relative distance between the sound source and the receiver. Here, the change in the horizontal distance of the sound source is not considered. Combining equation (1), the frequency domain sound pressure vector of the l-th frame received by the vertical array is expressed as:

[0070]

[0071] Where the superscript T denotes matrix transpose. The modal matrix and source position matrix corresponding to the vertical matrix are respectively expressed as follows:

[0072]

[0073] The specific content involved in step 2) is as follows:

[0074] Calculate the sum of the N frequency domain acoustic powers in each frame, and then take the square root of the result to obtain the sound pressure level received by the vertical array. Define the modal correlation matrix R = aa H Then the sum of the N frequency domain acoustic powers in the l-th frame It can be represented as

[0075]

[0076] The superscript H indicates conjugate transpose.

[0077] A vertical array covering the entire water depth can fully sample information from all modes. In this case, the mode functions satisfy an orthogonal relationship, i.e.

[0078]

[0079] Where ρ(z) represents the water density at different depths, δ mn Let be the Dirac function. From equation (7), it can be seen that when the water density is uniform, the modal correlation matrix obtained by sampling a vertical array with a finite number of elements is approximately a diagonal matrix with equal diagonal elements, while the off-diagonal elements are approximately 0. Let the water density be 1; then, the average acoustic power in equation (6) is approximately...

[0080]

[0081] The vertical array acoustic pressure amplitude of the l-th frame received by the vertical array is further obtained.

[0082]

[0083] The vector composed of the vertical array sound pressure amplitudes of L frames is denoted as... The matrix is ​​formed by multiplying the sound pressure matrix of equation (2) with its conjugate transpose and then taking the square root of the diagonal elements. Specifically, it is written as...

[0084]

[0085] Here, diag represents taking the diagonal elements.

[0086] Equation (9) shows that by superimposing the acoustic power received by all array elements, the horizontal distance factor of the sound source coupled with the mode in the normal mode coherence term is filtered out. The sound pressure amplitude is simply the superposition of the mode functions of each order. Taking the derivative of Equation (9) in the direction of sound source depth, the fluctuation of the sound pressure amplitude between adjacent frames can be obtained, i.e.

[0087]

[0088] At water surface depth, the modal function and its derivative are always positive. Equation (11) satisfies the inequality at water surface depth.

[0089]

[0090] At depths far from the water surface, according to the Cauchy-Schwarz inequality, equation (11) satisfies the inequality.

[0091]

[0092] The right side of equations (12) and (13) has the characteristic that the water surface is higher than the underwater surface. Furthermore, it can be seen that the vertical array sound pressure amplitude obtained by equation (9) has a stable fluctuation characteristic difference between the water surface and the underwater surface.

[0093] The specific content involved in step 3) is as follows:

[0094] The discrimination feature is calculated using the L-frame vertical array received sound pressure amplitude obtained in step 2). This discrimination feature is named the Amplitude Fluctuation Index (AFI), and specifically...

[0095]

[0096] Where, |p l | represents the vertical array sound pressure amplitude in the l-th frame. The denominator of equation (13) is the harmonic mean of the vertical array received sound pressure amplitude in the L-th frame. Equation (13) reflects the degree of deviation of the vertical array sound pressure amplitude from the mean, that is, the degree of fluctuation of the L-th frame sound pressure amplitude described in equation (11). In addition, the sound pressure amplitude fluctuation index is independent of the horizontal distance of the sound source, the sound source level, and the number of array elements.

[0097] The specific content involved in step 4) is as follows:

[0098] Obtain rough water depth, sound velocity profiles, and bottom sediment data. As shown in step 2), the modal orthogonality of the vertical array makes the sound pressure amplitude of the vertical array simply a superposition of the modal functions of each order. Therefore, the spectral fluctuation characteristics are not sensitive to changes in the sound velocity profile and bottom sediment. At different sound source depths, use the acoustic toolbox to calculate the result of equation (13) corresponding to the spectral frequency to be distinguished, and plot the depth distribution curve of the sound pressure amplitude fluctuation index. Determine the critical depth for distinguishing surface and underwater targets, and preset the discrimination threshold AFI based on the sound pressure amplitude fluctuation index curve. h Since the sound pressure amplitude fluctuation index depth distribution curve is stable in the horizontal distance dimension from the sound source, the threshold value is also reliable. Compare the discrimination feature quantity AFI calculated in step 3) with the discrimination threshold value AFI. h This allows us to determine whether the target is located on the water surface or underwater. Specifically, if AFI < AFI h If the target is underwater, then it is an underwater target; otherwise, the target is a surface target.

[0099] Taking the discrimination of line spectrum target depth attributes in shallow sea environments as an example, an implementation example of the present invention is given.

[0100] Basic simulation parameters:

[0101] In a shallow sea environment with isosonic speeds, at a depth of 200m, the speed of sound in seawater is 1500m / s, and the density of seawater is 1.0g / cm³. 3 The speed of sound at the bottom of the sea is 1800 m / s, and the density of seawater is 1.5 g / cm³. 3 The absorption coefficient at the seabed is 0.2 dB·km / Hz.

[0102] Simulation 1: The line spectrum sound source frequency is 30Hz, and a 5th-order normal mode wave is excited in the simulation environment. An 8-element vertical array is set up to cover a water layer of 1-200m. Sound sources at two depths are considered: surface and underwater, with depths of 5m and 70m respectively. The sound source is 5km away from the vertical array. The critical depth for determining the target's depth attribute is set to 15m. The sound source depth fluctuation follows a standard normal distribution with a mean of 0 and a variance of 1.

[0103] Based on steps 1) and 2), the frequency domain sound pressure was calculated using the acoustic toolbox for five changes in sound source depth (five frames). In step 2), the normalized comparison results of the vertical array sound pressure amplitudes corresponding to the surface and underwater sound sources are shown in Figure 4(a). Based on step 3), the sound pressure amplitude fluctuation index (AFI) for the two sound sources (surface and underwater) was calculated. Based on step 4), after inputting the target line spectrum frequency, water depth, sound velocity profile, and seabed data into the acoustic toolbox, the depth distribution curve of the sound pressure amplitude fluctuation index was plotted. A discrimination threshold was preset based on the critical depth. The discrimination results are shown in Figure 4(b).

[0104] In Figure 4(a), there is a difference in the fluctuation of the sound pressure amplitude received by the vertical array on the water surface and underwater. Specifically, the fluctuation of the sound pressure amplitude on the water surface, represented by the 5m sound source, is significantly stronger than that underwater, represented by the 70m sound source. Based on this characteristic difference, the sound pressure replication fluctuation index (AFI) of the two sets of vertical array sound pressure amplitudes is calculated. In Figure 4(b), the sound pressure amplitude fluctuation index corresponding to the 5m sound source is higher than the discrimination threshold corresponding to a depth of 15m, while that of the 70m sound source is lower than the discrimination threshold. Furthermore, this discrimination threshold completely distinguishes targets with depth attributes above and below 15m. If the sound source is located at other depths, the depth attribute can also be determined by the discrimination criterion of this invention. This indicates that this invention can clearly distinguish the depth attribute of line spectrum targets.

[0105] Simulation 2: The line spectrum sound source frequency is 30Hz, and the relative distance between the sound source and the vertical array is set to 2-15km. Other parameters are the same as in Simulation 1. The frequency domain sound pressure is taken at the sound source depth for 5 times (5 frames), and the sound pressure amplitude fluctuation index of the sound source moving in water depths of 1-200m is calculated using steps 2) and 3). The influence of the horizontal distance of the sound source on the depth attribute discrimination feature is determined by... Figure 5 Provided.

[0106] Figure 5 In the diagram, the red dashed line represents the contour line corresponding to the -30dB discrimination threshold at a critical depth of 15m. The sound pressure amplitude fluctuation index on both sides of the contour line shows a significant difference. This indicates that when the same sound source is located at different horizontal distances, the characteristic quantity of this invention can characterize stable differences in sound pressure amplitude fluctuations at the water surface and underwater, demonstrating the robustness of the discrimination method. Furthermore, when the sound source is located at different horizontal distances, the discrimination threshold preset based on environmental information exhibits stability.

[0107] Simulation 3: Simulation parameters are the same as Simulation 2. The sound sources are located 5km and 6km from the receiver, respectively. The signal-to-noise ratio within the line spectrum band on the vertical array element is set to 15dB. 200 sound sources ranging from 1-200m are set up. The accurate discrimination probability and false alarm probability (target located on the water surface but judged as underwater) for each sound source depth are calculated based on the results of step 4), and ROC curves are plotted. In addition, ROC curves for sound source depth attribute discrimination are plotted for hydrophones at receiving depths of 1m and 50m. 100 Monte Carlo experiments are performed for each case. Statistical proof of the robustness of sound source depth attribute discrimination of this invention compared to existing technologies (single receiving depth) is presented. Results are provided by… Figure 6 Provided.

[0108] Figure 6 In this study, for sound sources at the same horizontal distance, the performance of hydrophones with different receiving depths in determining the sound source depth attributes varies significantly. The present invention achieves a discrimination accuracy of 0.99 with a false alarm probability of 0.1, which is superior to existing horizontal array (single receiving depth) methods. When the horizontal distance of the sound source changes, the discrimination performance of the present invention remains unchanged, while the sound source depth attribute discrimination accuracy of single receiving depth methods decreases. Statistical results show that the present invention is more robust than the depth attribute discrimination method based on single receiving depth.

Claims

1. A robust method for determining the depth attribute of line spectrum targets in a vertical array, characterized in that, Includes the following steps: Step 1: Extract the line spectrum sound pressure amplitude from the time-domain signal received by the vertical array elements; in Includes the following sub-steps: Step 1.1: In a shallow sea environment with a water depth of D, the target radiation line spectrum signal is received using an N-element vertical array. The frequency domain sound pressure field of the line spectrum received by the nth element is expressed as: z n z s Here, r represents the receiving depth of the nth element of the vertical array and the sound source depth, respectively, and r is the relative horizontal distance between the sound source and the vertical array. With k rm Let N and Q represent the mode function and horizontal wavenumber of the m-th mode excited by the sound source, respectively. Q represents the sound pressure amplitude correlation term. The target moves away from the vertical array. M is the m-th normal mode excited by the sound source, and N is greater than M. Step 1.2: Define the receiving depth of the nth array element as z. n The nth time-domain signal acquired by the vertical array within time t is denoted as x. n (t); Divide the nth time-domain signal within time t into L time-domain sub-bands, and divide the time-domain signal received by the vertical array into a total of N×L time-domain sub-bands x. n (t l For each frame of the time-domain signal, perform a Fourier transform to the frequency domain; determine the target line spectrum frequency, and extract the N×L dimensional sound pressure matrix of the target frequency point, denoted as Y: Where, p l This represents the column vector composed of the frequency domain sound pressures of all elements of the vertical array within the l-th frame; Without considering changes in the horizontal distance of the sound source, and combining the formula in step 1.1, the frequency domain line spectrum sound pressure amplitude of the l-th frame received by the vertical array is expressed in vector form as follows: Where the superscript T denotes matrix transpose, This represents the mode matrix corresponding to the vertical matrix. The sound source location matrix is ​​represented by the following expressions: Step 2: Filter out the horizontally distance-dependent normal mode coherence terms using modal orthogonality to obtain the vertical array acoustic pressure amplitude; Step 3: Calculate the normalized depth attribute discrimination feature quantity related to the fluctuation of the vertical array sound pressure amplitude; Step 4: Preset the discrimination threshold based on the waveguide environment information, and compare the size of the discrimination feature quantity with the discrimination threshold value.

2. The robust method for determining the depth attribute of line spectrum targets for vertical arrays as described in claim 1, characterized in that, Step 2 includes the following sub-steps: Step 2.1: Calculate the sum of the frequency domain acoustic power received by the N vertical array elements in the l-th frame. : Step 2.2: Combining with Step 2.1, further obtain the vertical array acoustic pressure amplitude of the l-th frame received by the vertical array: The vector composed of the vertical array sound pressure amplitudes of the L frames is denoted as... Here, diag represents taking the diagonal elements.

3. The robust method for determining the depth attribute of line spectrum targets for vertical arrays as described in claim 1, characterized in that, In step 3, the L-frame vertical array received sound pressure amplitude obtained in step 2 is used to calculate the discrimination feature quantity; the discrimination feature quantity is named the sound pressure amplitude fluctuation index, specifically written as: Where, |p l | represents the vertical array sound pressure level in the l-th frame, and the denominator is the harmonic average of the vertical array received sound pressure levels in the L frames.

4. The robust method for determining the depth attribute of line spectrum targets for vertical arrays as described in claim 1, characterized in that, Step 4 includes the following sub-steps: Step 4.1: At different sound source depths, use the acoustic toolbox to calculate the formula corresponding to the line spectrum frequency to be discriminated. The results were obtained, and the depth distribution curve of the sound pressure amplitude fluctuation index was plotted. Step 4.2: Determine the critical depth for distinguishing between surface and underwater targets, and preset the discrimination threshold AFIh based on the sound pressure amplitude fluctuation index curve; Step 4.3: Compare the sound pressure amplitude fluctuation index AFI calculated in Step 3 with the discrimination threshold AFIh to determine whether the target is located on the water surface or underwater. If AFI < AFIh, then the target is an underwater target. Otherwise, the target is a surface target.

Images