A deep sea acoustic target distance estimation method based on graph Fourier transform

Abstract

The invention discloses a deep sea acoustic target distance estimation method based on graph Fourier transform, comprising the following steps: step 1, establishing an acoustic field model based on a ray model BELLHOP, calculating the target signal arrival angle and the target distance corresponding to the arrival angle; step 2, constructing a graph structure based on the array form of a horizontal array; step 3, calculating the phase difference between adjacent array elements based on the graph structure; step 4, establishing a sparse directed adjacency matrix based on the graph structure and the phase difference between adjacent array elements; step 5, calculating the graph Fourier transform result of the array element phase difference adjacency matrix, matching the graph Fourier transform result with the original data, and estimating the target signal arrival angle; step 6, matching the target signal arrival angle calculated in step 5 with the corresponding relationship between the target signal arrival angle and the target distance in step 1, and obtaining the estimated distance of the target. The method of the invention has good accuracy and can detect a longer distance.

Description

Technical Field

[0001] The invention belongs to the field of signal processing, relates to theories of ocean acoustics, signal processing, etc., and specifically relates to a deep sea water acoustic target distance estimation method based on graph Fourier transform. Background Art

[0002] Deep-sea target ranging has always been a hot research topic. There are four main traditional target ranging methods: one is matching field positioning, which constructs an acoustic field calculation model by inputting ocean environment parameters to obtain a copy field, and matches the copy field sound pressure with the measurement field sound pressure to estimate the target's position information. However, this method requires accurate ocean environment parameters, has poor tolerance, and has a large amount of calculation; the second is to use the waveguide invariant method to measure distance. In shallow sea waveguides, the sound field interferes due to the superposition of simple normal waves of different orders. The waveguide invariant method can extract distance information from the slope of the sound field interference fringes. This method has better tolerance, but this method is not suitable for deep-sea environments; the third is multi-path time delay method ranging, which uses acoustic signals in the deep sea to measure distance. There are two main propagation models: direct sound propagation and seabed reflected sound propagation. The time delay between array elements is calculated, and the target distance is estimated based on the geometric structure of the sound signal propagation path. This method has a small amount of calculation for ranging, but the sound propagation mode needs to be determined before ranging, and the accuracy of the distance estimate is low. The fourth is the matching target arrival angle ranging method. It only needs to calculate the sound field model once to construct the relationship between the target distance and the simulated target arrival angle, and then calculate the actual signal arrival angle by receiving the array signal. The target distance can be quickly estimated by matching the actual signal arrival angle with the simulated signal arrival angle. However, the signal arrival angle is currently mainly obtained through conventional beamforming calculations, and the main lobe is wide, making it difficult to accurately estimate the distance. Summary of the invention

[0003] In order to solve the problem of deep-sea target ranging, the present invention provides a target ranging method based on graph Fourier transform method, which can realize deep-sea target ranging.

[0004] To achieve the above object, the embodiments of the present invention adopt the following technical solutions:

[0005] A deep sea water acoustic target distance estimation method based on graph Fourier transform specifically comprises the following steps:

[0006] Step 1: Establish a sound field model based on the BELLHOP ray model, and calculate the target signal arrival angle and the target distance corresponding to the arrival angle;

[0007] Step 2: Construct a graph structure based on the array form of the horizontal array;

[0008] Step 3: Calculate the phase difference between adjacent array elements based on the graph structure. The calculation method of the phase difference is:

[0009] w(n)=exp(ikndsinθ) (1)

[0010] Where n represents the number of array element spacings to be calculated, i represents an imaginary number, θ is the steering angle, k = 2πf / c is the signal wave number, f is the signal frequency, c is the sound speed near the receiving array, d is the array element spacing, and exp(·) is the exponential operator;

[0011] Step 4: The sparse directed adjacency matrix represents the independence and similarity relationship between adjacent vertices. The sparse directed adjacency matrix is ​​established based on the graph structure and the phase difference between adjacent array elements. The sparse directed adjacency matrix is ​​constructed as follows:

[0012]

[0013] The sparse directed adjacency matrix is ​​a diagonal matrix, and the element at the position (a, b) 0<a, b<=N represents the phase difference between the a-th array element and the b-th array element;

[0014] Step 5: Calculate the graph Fourier transform result of the array element phase difference adjacency matrix, match the graph Fourier transform result with the original data, and estimate the arrival angle of the target signal;

[0015] Among them, the graph Fourier transform method of the adjacency matrix is:

[0016] Perform eigendecomposition on the adjacency matrix A:

[0017] A=TΛT -1 (3)

[0018] In the formula, the matrix The column vector t n is the eigenvector of A, Λ is the diagonal matrix containing the corresponding eigenvalues, and the Fourier transform of the graph is defined as the inverse of the matrix T, F = T -1 ;

[0019] Among them, the matching coefficient M is defined as:

[0020]

[0021] In the formula, M represents the matching degree, represents the estimated target signal arrival angle, F(θ) is the Fourier transform of the signal arrival angle θ (F = T -1 ), The actual signal arrival angle is expressed as θ 0 The data measured by the horizontal array, [·] - To remove the maximum absolute value operator, is the square operator of the L2 norm, and the signal arrival angle is estimated by searching for the peak value of the matching coefficient M;

[0022] Step 6: Match the target signal arrival angle calculated by the graph Fourier transform method with the corresponding relationship between the target signal arrival angle and the target distance in step 1 to obtain the estimated distance of the target.

[0023] During the graph structure construction process of step 2, adjacent array elements of the N-element horizontal array are associated with each other, and the first and last array elements are also associated with each other.

[0024] Compared with the prior art, the deep sea water acoustic target distance estimation method based on graph Fourier transform provided by the present invention has better accuracy and can detect farther distances. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art or ordinary technicians, other drawings can be obtained based on these drawings without paying creative work.

[0026] Figure 1 It is the processing flow of deep sea water acoustic target distance estimation method based on graph Fourier transform method;

[0027] Figure 2 It is a graph structure based on an N-element horizontal matrix;

[0028] Figure 3 The sound velocity profile adopted for the simulation environment;

[0029] Figure 4 This is a diagram showing the relationship between the simulated target signal arrival angle and the target distance;

[0030] Figure 5 The relationship between the arrival angle and the target distance obtained by the conventional beamforming method;

[0031] Figure 6 Figure 2 is a graph showing the relationship between the angle of arrival and the target distance obtained using the Fourier transform method. DETAILED DESCRIPTION

[0032] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention is further described below in conjunction with implementation modes and drawings.

[0033] like Figure 1 As shown, the present invention provides a deep sea water acoustic target distance estimation method based on graph Fourier transform, which specifically includes the following steps:

[0034] Step 1: Establish a sound field model based on the BELLHOP ray model, and calculate the target signal arrival angle and the target distance corresponding to the arrival angle;

[0035] Step 2: Construct a graph structure based on the array form of the horizontal array;

[0036] Step 3: Calculate the phase difference between adjacent array elements based on the graph structure. The calculation method of the phase difference is:

[0037] w(n)=exp(ikndsinθ) (1)

[0038] Where n represents the number of array element spacings to be calculated, i represents an imaginary number, θ is the steering angle, k = 2πf / c is the signal wave number, f is the signal frequency, c is the sound speed near the receiving array, d is the array element spacing, and exp(·) is the exponential operator;

[0039] Step 4: The sparse directed adjacency matrix represents the independence and similarity relationship between adjacent vertices. The sparse directed adjacency matrix is ​​established based on the graph structure and the phase difference between adjacent array elements. The sparse directed adjacency matrix is ​​constructed as follows:

[0040]

[0041] The sparse directed adjacency matrix is ​​a diagonal matrix, and the element at the position (a, b) 0<a, b<=N represents the phase difference between the a-th array element and the b-th array element;

[0042] Step 5: Calculate the graph Fourier transform result of the array element phase difference adjacency matrix, match the graph Fourier transform result with the original data, and estimate the arrival angle of the target signal;

[0043] Among them, the graph Fourier transform method of the adjacency matrix is:

[0044] Perform eigendecomposition on the adjacency matrix A:

[0045] A=TΛT -1 (3)

[0046] In the formula, the matrix The column vector t n is the eigenvector of A, Λ is the diagonal matrix containing the corresponding eigenvalues, and the Fourier transform of the graph is defined as the inverse of the matrix T, F = T -1 ;

[0047] Among them, the matching coefficient M is defined as:

[0048]

[0049] In the formula, M represents the matching degree, represents the estimated target signal arrival angle, F(θ) is the Fourier transform of the signal arrival angle θ (F = T -1 ), Expressed as the corresponding actual signal arrival angle θ 0The data measured by the horizontal array, [·] - To remove the maximum absolute value operator, is the square operator of the L2 norm, and the signal arrival angle is estimated by searching for the peak value of the matching coefficient M;

[0050] Step 6: Match the target signal arrival angle calculated by the graph Fourier transform method with the corresponding relationship between the target signal arrival angle and the target distance in step 1 to obtain the estimated distance of the target.

[0051] In the process of constructing the graph structure in step 2, the graph of the N-element horizontal array can be Figure 2 In the form of , adjacent array elements of the N-element horizontal array are associated, and the first and last array elements are also associated.

[0052] The following is a comparison between the relationship between the arrival angle and the target distance obtained by the method of the present invention and the conventional beamforming method. Figure 3 It is the sound velocity profile adopted in the simulation environment; the simulation condition is the deep sea with a depth of 5000m, the receiving array is an 11-element horizontal array, the array element spacing is 10m, the target depth is 50m, the receiving array depth is 100m, and the frequency is 250Hz.

[0053] Figure 4 This is a diagram showing the relationship between the simulated target signal arrival angle and the target distance. By inputting the simulation environment parameters to establish a ray acoustic field model, the corresponding signal arrival angle at different target distances is obtained, such as Figure 4 As shown, it can be seen that the target distance and the signal arrival angle are in a one-to-one correspondence, so the target distance can be estimated by calculating the signal arrival angle.

[0054] Figure 5 The relationship between the arrival angle and the target distance obtained by the conventional beamforming method is provided, where the black dotted line represents the relationship between the actual target arrival angle and the target distance. The conventional beamforming method calculates the signal arrival angle beam output of the target signal at different distances to the receiving array, and normalizes it to form Figure 5 As can be seen from the figure, the main lobe of the signal arrival angle calculated by the conventional beamforming method is relatively wide, and it is difficult to accurately estimate the target distance through the calculated signal arrival angle, especially when the target distance is greater than 20km, the main lobe width of the signal arrival angle is relatively large, making it difficult to estimate the distance. Under simulation conditions, the maximum ranging distance is about 20km.

[0055] Figure 6 The relationship between the arrival angle and the target distance obtained by the Fourier transform method is provided, where the black dotted line represents the relationship between the actual target arrival angle and the target distance. The Fourier transform method is used to calculate the signal arrival angle beam output of the target signal at different distances to the receiving array, and then normalized to form Figure 6As can be seen from the figure, compared with the conventional beamforming method, the main lobe of the signal arrival angle calculated by the Fourier transform method is narrower. Within a distance of 40km, the target distance can be accurately estimated by the calculated signal arrival angle. The maximum ranging distance is improved by 20km compared with the traditional method.

[0056] It should be noted that the above embodiments are merely preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. All equivalent modifications made on the basis of the above embodiments belong to the protection scope of the present invention.

Claims

1. A deep sea water acoustic target distance estimation method based on graph Fourier transform, characterized in that: The following steps are involved: Step 1: Establish a sound field model based on the BELLHOP ray model, and calculate the target signal arrival angle and the target distance corresponding to the arrival angle; Step 2: Construct a graph structure based on the array form of the horizontal array; Step 3: Calculate the phase difference between adjacent array elements based on the graph structure. The calculation method of the phase difference is: w(n)=exp(ikndsinθ) (1) Where n represents the number of array element spacings to be calculated, i represents an imaginary number, θ is the steering angle, k = 2πf / c is the signal wave number, f is the signal frequency, c is the sound speed near the receiving array, d is the array element spacing, and exp(·) is the exponential operator; Step 4: The sparse directed adjacency matrix represents the independence and similarity relationship between adjacent vertices. The sparse directed adjacency matrix is ​​established based on the graph structure and the phase difference between adjacent array elements. The sparse directed adjacency matrix is ​​constructed as follows: The sparse directed adjacency matrix is ​​a diagonal matrix, and the element at the position (a, b) 0<a, b<=N represents the phase difference between the a-th array element and the b-th array element; Step 5: Calculate the graph Fourier transform result of the array element phase difference adjacency matrix, match the graph Fourier transform result with the original data, and estimate the arrival angle of the target signal; Among them, the graph Fourier transform method of the adjacency matrix is: Perform eigendecomposition on the adjacency matrix A: A=TΛT -1 (3) In the formula, the matrix The column vector t n is the eigenvector of A, Λ is the diagonal matrix containing the corresponding eigenvalues, and the Fourier transform of the graph is defined as the inverse of the matrix T, F = T -1 ; Among them, the matching coefficient M is defined as: In the formula, M represents the matching degree, represents the estimated target signal arrival angle, F(θ) is the Fourier transform of the corresponding signal arrival angle θ, It is expressed as the data measured by the horizontal array with the actual signal arrival angle θ0, [·] - To remove the maximum absolute value operator, is the square operator of the L2 norm, and the signal arrival angle is estimated by searching for the peak value of the matching coefficient M; Step 6: Match the target signal arrival angle calculated by the graph Fourier transform method with the corresponding relationship between the target signal arrival angle and the target distance in step 1 to obtain the estimated distance of the target.

2. A deep sea water acoustic target distance estimation method based on graph Fourier transform as claimed in claim 1, characterized in that: During the graph structure construction process of step 2, adjacent array elements of the N-element horizontal array are associated with each other, and the first and last array elements are also associated with each other.

Images