A Passive Localization Method for Low-Frequency Sound Sources of a Deep-Sea Small-Scale Vertical Array

Abstract

The present invention discloses a method for passive localization of low-frequency sound sources by a deep-sea small-scale vertical array, belonging to the field of ocean engineering technology, and comprising the following steps: S1, setting initial parameters; S2, generating signals and isotropic noise in combination with the sound field; S3, positioning by using the sparse reconstruction matched field localization method; S4, changing the conditions of the number of array elements and the array element spacing, and looping S2 and S3 to obtain the root mean square error curves of the distance and depth of the SR-MF method under different conditions; With the present invention, the distance and depth information of low-frequency passive sound sources can be obtained under deep-sea conditions by the SR-MF method, effectively avoiding the estimation of the number and angle of multi-paths, and being able to obtain better positioning estimation results under a smaller array scale, effectively avoiding the influence of the sound field on the positioning performance of large-scale arrays and the inclination problem existing after the deployment of large-scale arrays, and having a lower array cost and more convenient deployment and recovery.

Description

Technical Field

[0001] The present invention belongs to the technical field of ocean engineering, and particularly relates to a method for passive localization of low-frequency sound sources of a deep-sea small-scale vertical array. Background Art

[0002] Passive localization of low-frequency targets in the deep-sea environment is of great significance for fields such as ocean resource development and underwater acoustic detection; for low-frequency targets, large-scale arrays are often used for distance estimation and depth estimation; however, with the increase in the array scale, the tilt and distortion of the array shape and the decrease in the correlation of the signals received by the array will lead to a reduction in the localization performance, and a too large scale will result in a substantial increase in cost and more difficult deployment and recovery.

[0003] As mentioned in the literature "Research on reliable acoustic path_Physical properties and asource localization method, Chinese Physics B 21(12)2012", a passive target localization method applicable to reliable acoustic paths is described. This method has good localization performance, but has high requirements for the array scale and will fail when using a small-scale array for localization;

[0004] As disclosed in a Chinese patent (Publication No.: CN105629220B), a deep-sea underwater acoustic passive ranging method based on a single hydrophone is disclosed. This method can only estimate the target distance and requires matching the frequency-domain interval of each time segment fringe obtained by the receiving hydrophone with the frequency-domain interference period of the sound field at different distances. This method has high requirements for the signal-to-noise ratio;

[0005] As disclosed in a Chinese patent (Publication No.: CN116540220A), a three-dimensional passive target localization method and device are disclosed. This method uses a large-scale vector vertical array, which has high costs and is difficult to deploy and recover;

[0006] Therefore, a method for passive localization of low-frequency sound sources of a deep-sea small-scale vertical array is needed to solve the above problems. Summary of the Invention

[0007] The purpose of the present invention is to provide a method for passive localization of low-frequency sound sources of a deep-sea small-scale vertical array to solve the problems raised in the above background art.

[0008] To achieve the above purpose, the present invention provides the following technical solution: A method for passive localization of low-frequency sound sources of a deep-sea small-scale vertical array, comprising the following steps:

[0009] S1. Initial parameter setting;

[0010] S2. Combining the sound field to generate signals and isotropic noise;

[0011] S3. Locate using the sparse reconstruction matched field localization method;

[0012] S4. Change the conditions of the number of array elements and the element spacing, and loop S2 and S3 to obtain the root mean square error curves of distance and depth of the SR-MF method under different conditions.

[0013] This method uses a small-scale array to receive signals, realizes the estimation of the distance and depth of low-frequency passive sound sources, breaks through the limitation of the traditional deep-sea scalar array on the array scale, realizes better positioning effects under the condition of a smaller array scale, can effectively avoid the reduction of signal correlation affected by the sound field of a large-scale array and the positioning error caused by the array tilt in actual situations. The positioning error of the disclosed method is small, the manufacturing cost is low, and the deployment and recovery are convenient.

[0014] As a preferred solution, in "S1", set the parameters of the small-scale vertical array, the number of array elements M, the hydrophone is non-directional, and set the array manifold vector P(θ, φ), where θ is the pitch angle, φ is the horizontal angle, k = 2π / λ, λ represents the wavelength of the incident plane wave, and p m is the coordinate position of the m-th array element.

[0015] As a preferred solution, in "S2", combined with the sound field model, use the ray method to calculate the multi-path arrival structure of each hydrophone of the vertical array, only consider the direct wave and the first-order reflection wave from the sea surface, and ignore the rest of the multi-path arrival structures;

[0016] Use the multi-path arrival structure simulated by each hydrophone to correct the array manifold vector. At this time, the simulated single-frequency target signal S(t) = [s 1 (t), s 2 (t), …, s M (t)], and the signal amplitude is the signal amplitude of the multi-path arrival structure corresponding to each hydrophone. Then the received data of the vertical array is X(t) = P(θ m , φ 1 )S(t) + N(t),

[0017] As a preferred solution, where P(θ m , φ 1 ) represents the array manifold vector received by the m-th array element with a known horizontal angle, where θ m represents the received pitch angle of the m-th array element, φ 1 represents the horizontal angle of the target arriving at the receiving array, t represents any moment, and N(t) is the isotropic noise uncorrelated with the signal.

[0018] As a preferred solution, in "S3", the array received signal X(t) is subjected to singular value decomposition and dimensionality reduction processing to obtain the M×K matrix X SR (t), where K is the number of signal sources, and X SR (t) can be expressed as X SR (t) = ULD K , where U and L are the left unitary matrix and diagonal matrix obtained by singular value decomposition of the array received signal X(t), and D K The expression is D K = [I K , 0 K*(T-K) T , I K is the K×K identity matrix, 0 K×(T-K) is the zero matrix with dimensions K×(T - K), T is the number of snapshots, and X SR (t) can be deformed into X SR (t) = P(θ m , φ 1 )×X SR + N SR , where, S SR = SVD K , N SR = NVD K , S represents the received signal matrix, N represents the noise matrix, V is the right unitary matrix obtained by singular value decomposition of the array received signal X, and using the joint minimization model, the azimuth spectrum solution formula of the SR algorithm is In the formula, ||·|| F represents the matrix F-norm, μ is the regularization parameter, and the final azimuth spectrum obtained by the SR algorithm is l 2 represents the l 2 norm.

[0019] Combined with the idea of the matched field passive localization method, the SR-MF method is proposed for passive localization of the target. Assume the target position is L h = [z, r], where z is the target depth from the sea surface and r is the horizontal distance of the target from the array. The multi-path arrival angle of the signal is θ h = g(L h ), where g(·) is determined by the environment and can be calculated using the ray model. At this time, substituting θ h = g(L h ) gives Define the ambiguity plane of SR-MF and perform normalization processing to obtain the target localization estimation result.

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

[0021] ​In the present invention, the SR-MF method can obtain the distance and depth information of low-frequency passive sound sources under deep-sea conditions, effectively avoiding the estimation of the number and angle of multi-paths, and being able to obtain better positioning estimation results with a smaller array scale. It can effectively avoid the influence of the sound field on the positioning performance of large-scale arrays and the tilting problem existing after the deployment of large-scale arrays, and the array has a lower cost and is more convenient for deployment and recovery. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 is a flow chart of the present invention;

[0023] Figure 2 is a diagram showing the variation of sound speed with sea depth in the simulation scenario of the present invention;

[0024] Figure 3 is a propagation loss diagram in the simulated deep-sea environment of the present invention;

[0025] Figure 4 is a SR-MF positioning result diagram of the present invention.

[0026] Figure 5 is for the present invention and Figure 3 is a WSF-MF positioning result diagram under the same conditions.

[0027] Figure 6 is a diagram showing the root mean square error results of the distances of SR-MF and WSF-MF under different signal-to-noise ratio conditions in 100 independent repeated experiments of the present invention.

[0028] Figure 7 is a diagram showing the root mean square error results of the distances of SR-MF and WSF-MF under different element spacings in 100 independent repeated experiments of the present invention.

[0029] Figure 8 is a diagram showing the root mean square error results of the depths of SR-MF and WSF-MF under different signal-to-noise ratio conditions in 100 independent repeated experiments of the present invention.

[0030] Figure 9 is a diagram showing the root mean square error results of the depths of SR-MF and WSF-MF under different element spacings in 100 independent repeated experiments of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0031] The present invention will be further described below in conjunction with embodiments.

[0032] The following embodiments are used to illustrate the present invention, but cannot be used to limit the protection scope of the present invention. The conditions in the embodiments can be further adjusted according to specific conditions, and simple improvements to the method of the present invention under the premise of the concept of the present invention all belong to the scope required to be protected by the present invention.

[0033] Please refer to Figures 1-9 , the present invention provides a method for passive localization of low-frequency sound sources by a small-scale vertical array in deep sea, including the following steps:

[0034] S1. Initial parameter setting;

[0035] Set the parameters of the small-scale vertical array. In the simulation, a small-scale vertical array is used, with the number of array elements M = 8, the element spacing 0.3λ, the array scale 36m, and the signal incident direction is The signal is a 100Hz line spectrum, the signal-to-noise ratio is 30dB, the sound source depth is 70m, and the distance between the sound source and the receiving array is 10km;

[0036] S2. Generate signals and isotropic noise in combination with the sound field;

[0037] Use the ray model to obtain the multi-path arrival structure, divide the distance grid [6:0.05:15] (unit: km), divide the sound source depth grid [30:10:1200] (unit: m), the receiving array depth is 4200m, the sea depth is 5000m, adopt the Munk sound speed profile, and the sound speed c of the sediment layer p = 1650m / s, c s is ignored, ρ = 1.8g / cm 3 , α = 0.1dB / λ, the sediment layer thickness is 50m, the seabed consists of the sediment layer and the basement layer, and only the direct wave and the first sea surface reflection wave are considered in the multi-path arrival structure, and the rest of the multi-paths are ignored;

[0038] Ensure that the signal model has sparsity. Let the scanning grid be θ Ω , then there is Ω >> M >> K, where K is the number of signal sources. In combination with the sound field model, use the ray method to calculate the multi-path arrival structure of each hydrophone in the vertical array, only consider the direct wave and the first sea surface reflection wave, and ignore the rest of the multi-path arrival structures. Use the multi-path arrival structure simulated by each hydrophone to correct the array manifold vector. At this time, the simulated single-frequency target signal S(t) = [s 1 (t), s 2 (t), …, s M (t)], and the signal amplitude is the signal amplitude of the multi-path arrival structure corresponding to each hydrophone. Then the received data of the vertical array is X(t) = P(θ m , φ 1 )S(t) + N(t), where P(θ m , φ 1 ) represents the array manifold vector received by the m-th array element with a known horizontal angle, where θ m represents the receiving elevation angle of the m-th array element, φ 1 represents the horizontal angle of the target arriving at the receiving array, t represents any moment, and N(t) is the isotropic noise uncorrelated with the signal.

[0039] S3. Locate using the sparse reconstruction matched field localization method;

[0040] Perform singular value decomposition and dimensionality reduction on the array received signal X(t) to obtain the M×K matrix X SR (t), where X SR (t) can be expressed as X SR (t) = ULD K , where U and L are the left unitary matrix and diagonal matrix obtained by singular value decomposition of the array received signal X(t), and D K The expression is D K = [I K , 0 K*(T-K) T , I K is the K×K identity matrix, 0 K×(T-K) is the zero matrix with dimensions K×(T - K), and T is the number of snapshots. At this time, X SV (t) can be deformed into X SR (t) = P(θ m , φ 1 )×S SR + N SR , where S SR = SVD K , N SR = NVD K , S represents the received signal matrix, N represents the noise matrix, and V is the right unitary matrix obtained by singular value decomposition of the array received signal Y. At this time, use the sparse signal reconstruction algorithm to obtain the azimuth spectrum solution formula of the SR algorithm In the formula, ||·|| F represents the matrix F - norm, and μ is the regularization parameter. The final azimuth spectrum obtained by the SR algorithm is l 2 represents the l2 norm.

[0041] Combined with the idea of the matched field passive localization method, the SR - MF method is proposed for target passive localization. Assume the target position is L h = [z, r], where z is the target depth from the sea surface and r is the horizontal distance of the target from the array. The multi - path arrival angle of the signal is a function of the target position, which is θ h = g(L h ), where g(·) is determined by the environment and can be calculated using the ray model. At this time, substituting θ h = g(L h ) gives Obtain the target position L by searching the divided distance - depth grid 0 , define the ambiguity plane of SR - MF and perform normalization processing: ​At this time, draw the blurred plane of distance and depth, as Figure 4 shown, the position corresponding to the maximum value in E N is the target positioning estimation result;

[0042] S4. Performance analysis of the SR-MF method;

[0043] Use the root mean square error of distance and the root mean square error of depth to analyze the distance estimation error of the SR-MF passive positioning method under different conditions.

[0044]

[0045] where z i represents the depth estimation result of the SR-MF passive positioning method in the i-th independent experiment, and r i represents the distance estimation result of the SR-MF passive positioning method in the i-th independent experiment. The units of the depth and distance estimation results are both meters.

[0046] In Figure 4 , Figure 5 the white asterisk represents the true position of the sound source;

[0047] In Figure 6 , the array element spacing is sampled at 0.3λ, the receiving distance is 10 km, the sound source depth is 70 m, and an RMSE less than 1000 m is considered to be able to accurately locate (10% error).

[0048] In Figure 7 , the array element spacing is sampled at 0.3λ, the receiving distance is 10 km, the sound source depth is 70 m, and an RMSE less than 1000 m is considered to be able to accurately locate (10% error).

[0049] In Figure 8 , the array element spacing is sampled at 0.3λ, the receiving distance is 10 km, the sound source depth is 70 m, and an RMSE less than 50 m is considered to be able to accurately locate.

[0050] In Figure 9 , the array element spacing is sampled at 0.3λ, the receiving distance is 10 km, the sound source depth is 70 m, and an RMSE less than 50 m is considered to be able to accurately locate.

[0051] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for passively locating low-frequency sound sources using a deep-sea small-scale vertical array, characterized in that: The following steps are involved: S1, initial parameter setting; S2, combining the sound field to generate a signal and isotropic noise; S3, positioning using sparse reconstruction matching field positioning method; S4. Change the number of array elements and the array element spacing conditions, repeat S2 and S3, and obtain the distance and depth root mean square error curves of the SR-MF method under different conditions.

2. The method for passively locating low-frequency sound sources of a deep-sea small-scale vertical array according to claim 1 is characterized in that: In "S1", set the small-scale vertical array parameters, the number of array elements M, the hydrophone is non-directional, and the array manifold vector P(θ, φ). In the formula θ is the pitch angle, φ is the horizontal angle, k = 2π / λ, λ is the wavelength of the incident plane wave, p m is the coordinate position of the mth array element.

3. The method for passively locating low-frequency sound sources of a deep-sea small-scale vertical array according to claim 1 is characterized in that: In "S2", combined with the acoustic field model, the ray method is used to calculate the multipath arrival structure of each hydrophone in the vertical array, only the direct wave and the first reflection wave of the sea surface are considered, and the rest of the multipath arrival structure is ignored; The array manifold vector is modified by using the multipath arrival structure obtained by simulating each hydrophone. At this time, the simulated single-frequency target signal S(t) = [s1(t), s2(t), ..., s M (t)], the signal amplitude is the signal amplitude of the multipath arrival structure corresponding to each hydrophone, then the vertical array receiving data is X(t) = P(θ m ,φ1)S(t)+N(t), where P(θ m , φ1) represents the array flow vector received by the mth array element with a known horizontal angle, where θ m represents the receiving elevation angle of the mth array element, φ1 represents the horizontal angle of the target reaching the receiving array, t represents any time, and N(t) is the isotropic noise unrelated to the signal.

4. The method for passively locating low-frequency sound sources of a deep-sea small-scale vertical array according to claim 1 is characterized in that: In "S3", the array received signal X(t) is subjected to singular value decomposition and dimensionality reduction to obtain an M×K matrix XS R (t), where K is the number of information sources, X SR (t) can be expressed as X SR (t)=ULD K , where U and L are the left unitary matrix and diagonal matrix obtained by singular value decomposition of the array received signal X(t), D K The expression is D K =[I K , 0 K*(T-K) ] T , I K is the K×K identity matrix, 0 K×(T-K) is a zero matrix of dimension K×(TK), where T is the number of snapshots, X SR (t) can be transformed into X SR (t) = P(θ m ,φ1)×X SR +N SR , where S SR =SVD K , N SR =NVD K , S represents the received signal matrix, N represents the noise matrix, V is the right unitary matrix obtained by singular value decomposition of the array received signal X, and using the joint minimization model, the azimuth spectrum solution formula of the SR algorithm is obtained as follows: In the formula ||·|| F represents the matrix F-norm, μ is the regularization parameter, and the azimuth spectrum obtained by the final SR algorithm is l2 denotes the l2 norm.

Images