Uniform circular array wideband beamforming method and target distance and azimuth detection method

Abstract

The application discloses a kind of uniform circular array wideband beam forming method and target distance, azimuth detection method, and uniform circular array wideband beam forming includes: first, establish uniform circular array wideband signal processing model, then combine empirical mode decomposition timescale filter with uniform circular array, introduce empirical mode decomposition into uniform circular array, carry out empirical mode decomposition to each subband signal, obtain multiple groups of solid eigenfunctions, then construct new modal field data matrix, carry out array beam forming in modal field, improve weak target detection effect in interference environment.The application utilizes the characteristics that empirical mode decomposition can strengthen signal local instantaneous characteristics and effectively decompose different signal characteristic scales, which helps to improve the performance of traditional adaptive beam forming method in signal direction vector mismatch environment and the problem that conventional beam forming resolution is limited by "Rayleigh limit".

Description

Technical Field

[0001] This invention relates to the field of array signal processing and target detection technology, specifically to a method for forming a uniform circular array broadband beam and a method for detecting target distance and orientation. Background Technology

[0002] Beamforming technology is an important research area in array signal processing, and it has been widely applied in fields such as sonar, radar, communications, semiconductors, radio astronomy, and biomedical engineering. However, the underwater acoustic environment is complex and variable. Changes in seawater temperature, waves, and other environmental conditions can affect the array structure of underwater acoustic arrays, thus posing challenges to beamforming and direction-of-arrival estimation of underwater acoustic signals.

[0003] Uniform circular array (UCA) is a widely used sonar array configuration, featuring two-dimensional scanning and omnidirectional horizontal coverage, and has considerable application prospects. However, UCA suffers from high sidelobe levels and relatively shallow null depths. Conventional beamforming (CBF), commonly used in UCA, has advantages such as simplicity and stable performance, but it suffers from drawbacks such as high sidelobe levels and weak noise suppression capabilities.

[0004] Minimum variance distortionless response beamforming (MVDR) can effectively overcome the above-mentioned shortcomings and break through the Rayleigh limit of conventional beamforming resolution. However, when there are model errors such as steering vector error and array element correction error, the inversion of the signal covariance matrix required by the MVDR method is prone to distortion. Therefore, there is an urgent need for a beamforming method suitable for signal steering vector mismatch environments. Summary of the Invention

[0005] To address the aforementioned issues, the inventors provide a uniform circular array broadband beamforming method and a target range and azimuth detection method. The broadband signal is decomposed into sub-bands corresponding to different frequency bands. Then, empirical mode decomposition (EMD) is performed on each sub-band signal to obtain a set of IMFs. Combined with the beamforming method, an empirical mode uniform array processing structure is proposed to form multiple beams and further obtain the range and azimuth of weak targets.

[0006] According to a first aspect, the present invention provides a method for forming a uniform circular array broadband beam, comprising:

[0007] Step 1: Establish a uniform circular array broadband signal model;

[0008] Define an xy-plane containing... A uniform circular array of elements, where O is the center of the array and r is the radius. The hydrophones are located at... d represents the distance between any two adjacent elements of the circular array. Generally, ,in It is the signal wavelength corresponding to the center frequency of the underwater acoustic signal.

[0009] Because the array has With a given array of elements, all evenly distributed, the angle between any two adjacent elements and the center of the circle is calculated as follows: Based on the trigonometric relationship, the relationship between the radius r of a uniform circular array and the distance d between adjacent elements can be derived as follows:

[0010]

[0011] If the angle at which the signal is incident on the array is Then the first The time delay of the signal received by each element relative to the center can be expressed as:

[0012]

[0013] In the formula, Let be the speed of sound in water. The corresponding phase difference can be expressed as:

[0014]

[0015] In the formula, Indicates the angular frequency of the signal. Indicates the signal frequency.

[0016] Assume that a total of M signals are incident on a uniform circular array, and the incident azimuth angle is denoted as . The complex envelope of the signal is denoted as The signal center frequencies are respectively ,in The target signal to be extracted is the remaining signal. This is an interference signal. Therefore, the array output vector energy can be expressed as:

[0017]

[0018] In the formula, For the first The steering vector of a signal, It is additive white Gaussian noise, and: In the formula, This indicates the matrix transpose.

[0019] If the incident signal is a broadband signal, then dividing each broadband signal into Q frequency bands, the above formula can be written as:

[0020]

[0021] Considering the centrally symmetric array characteristics of a uniform circular array, the array normal direction can be arbitrary. Therefore, in use, the array normal direction can be adjusted according to requirements to make the direction of the desired signal consistent with the normal direction. That is, the incident angle of the desired signal is set to... ,Right now And the effect of frequency variation on beamforming is ignored. In this case, the steering vector of the desired signal from the incident array is simplified as... The above formula can also be rewritten as

[0022]

[0023] Assuming a uniform circular array for receiving The sampled data of each snapshot, the above formula is converted into discrete form as follows:

[0024]

[0025] For a uniform circular array with L elements to receive sampled data from K snapshots, its original data matrix... for:

[0026]

[0027] The desired signal, interference signal, and noise are all zero-mean, wide-range stationary random processes and are uncorrelated. Therefore, the output of the beamformer can be expressed as:

[0028]

[0029] in, For the corresponding number The weight vector of the beamformer at each frequency point This indicates the conjugate transpose.

[0030] Step 2: Array element reception of a uniform circular array The original data matrix is ​​obtained from the sampling data of each snapshot. ;

[0031] Step 3: Process the original data matrix Sampling data for each row Perform empirical mode decomposition to obtain One IMF component;

[0032] Step 4: Each row of sampled data The IMF components of the same order are extracted and combined into an empirical mode function matrix;

[0033] Step 5: Replace the original data matrix with the empirical mode function matrix. The output of the broadband beamformer is obtained by inputting a uniform circular array broadband signal model.

[0034] Furthermore, in step 3, the process of processing the original data matrix... Sampling data for each row The steps for performing empirical mode decomposition include:

[0035] Step 31: Find The maximum and minimum points are then identified, and cubic spline curves are used to fit these points to obtain the upper and lower envelopes, denoted as: and ;

[0036] Step 32: Calculate the envelope mean based on the upper and lower envelopes. ; .

[0037] Step 33: Mean the envelope from Remove from the middle to obtain the residual components. ;

[0038] Step 34: Determine if the residual component is an IMF component. If not, use the residual component. As Repeat steps 31, 32, and 33 until the residual component is reached. As an IMF component, the final result is IMF components and a low-order polynomial residue with a non-zero mean .Right now:

[0039]

[0040] in, For the first IMF, denoted as EMD remaining amount, and N as the number of IMFs obtained from EMD.

[0041] The present invention also provides a target distance and azimuth detection method, which involves segmenting the output of the broadband beamformer and the sample signal to improve the sample matching accuracy, then summing the correlation results of each segment and finding the maximum value. The position corresponding to the maximum value is the distance of the target from the circular array; the direction pointed to by the main lobe of the beam is the target azimuth.

[0042] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0043] (1) This invention utilizes the characteristics of empirical mode decomposition to enhance the local transient features of a signal and effectively decompose different signal feature scales. It introduces empirical mode decomposition into the uniform circular array broadband beamforming to construct a new mode domain data matrix. Array beamforming is then performed in the mode domain, which improves the detection effect of weak targets in the interference environment.

[0044] (2) Compared with existing conventional beamforming technologies, it can improve the performance degradation of traditional beamforming methods under signal steering vector mismatch environment and the problem that the resolution of conventional beamforming is limited by the Rayleigh limit. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the uniform circular array in this invention;

[0046] Figure 2 This is a schematic diagram of the 9th array element receiving raw data in Example 2;

[0047] Figure 3 This is a schematic diagram of the beamforming signal in Example 2;

[0048] Figure 4 This is a schematic diagram of the matching results in Example 2;

[0049] Figure 5 The results are from the static distance test of 16 meters in Example 2;

[0050] Figure 6 The results are from the static distance test of 33 meters in Example 2;

[0051] Figure 7 The results are from the static distance test of 42 meters in Example 2;

[0052] Figure 8 The results are from the vertical motion test in Example 2. Detailed Implementation

[0053] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings.

[0054] Example 1

[0055] This invention provides a target distance and orientation detection method, comprising:

[0056] Step 1: Establish a uniform circular array broadband signal model.

[0057] like Figure 1 As shown, a plane containing xy plane is defined. A uniform circular array of elements, where O is the center of the array and r is the radius. The hydrophones are located at... d represents the distance between any two adjacent elements of the circular array. Generally, ,in It is the signal wavelength corresponding to the center frequency of the underwater acoustic signal.

[0058] Because the array has With a given array of elements, all evenly distributed, the angle between any two adjacent elements and the center of the circle is calculated as follows: Based on the trigonometric relationship, the relationship between the radius r of a uniform circular array and the distance d between adjacent elements can be derived as follows:

[0059]

[0060] If the angle at which the signal is incident on the array is Then the first The time delay of the signal received by each element relative to the center can be expressed as:

[0061]

[0062] In the formula, Let be the speed of sound in water. The corresponding phase difference can be expressed as:

[0063]

[0064] In the formula, Indicates the angular frequency of the signal. Indicates the signal frequency.

[0065] Assume that a total of M signals are incident on a uniform circular array, and the incident azimuth angle is denoted as . The complex envelope of the signal is denoted as The signal center frequencies are respectively ,in The target signal to be extracted is the remaining signal. This is an interference signal. Therefore, the array output vector energy can be expressed as:

[0066]

[0067] In the formula, For the first The steering vector of a signal, It is additive white Gaussian noise, and: In the formula, This indicates the matrix transpose.

[0068] If the incident signal is a broadband signal, then dividing each broadband signal into Q frequency bands, the above formula can be written as:

[0069]

[0070] Considering the centrally symmetric array characteristics of a uniform circular array, the array normal direction can be arbitrary. Therefore, in use, the array normal direction can be adjusted according to requirements to make the direction of the desired signal consistent with the normal direction. That is, the incident angle of the desired signal is set to... ,Right now And the effect of frequency variation on beamforming is ignored. In this case, the steering vector of the desired signal from the incident array is simplified as... The above formula can also be rewritten as

[0071]

[0072] Assuming a uniform circular array for receiving The sampled data of each snapshot, the above formula is converted into discrete form as follows:

[0073]

[0074] For a uniform circular array with L elements to receive sampled data from K snapshots, its original data matrix... for:

[0075]

[0076] The desired signal, interference signal, and noise are all zero-mean, wide-range stationary random processes and are uncorrelated. Therefore, the output of the beamformer can be expressed as:

[0077]

[0078] in, For the corresponding number The weight vector of the beamformer at each frequency point This indicates the conjugate transpose.

[0079] Step 2: Array element reception of a uniform circular array The original data matrix is ​​obtained from the sampling data of each snapshot. ;

[0080] Step 3: Process the original data matrix Sampling data for each row Perform empirical mode decomposition to obtain One IMF component. Specifically:

[0081] Step 31: Find The maximum and minimum points are then identified, and cubic spline curves are used to fit these points to obtain the upper and lower envelopes, denoted as: and ;

[0082] Step 32: Calculate the envelope mean based on the upper and lower envelopes. ; .

[0083] Step 33: Mean the envelope from Remove from the middle to obtain the residual components. ;

[0084] Step 34: Determine if the residual component is an IMF component. If not, use the residual component. As Repeat steps 31, 32, and 33 until the residual component is reached. As an IMF component, the final result is IMF components and a low-order polynomial residue with a non-zero mean .Right now:

[0085]

[0086] right By performing empirical mode decomposition on each row of sampled data, we can obtain...

[0087]

[0088] in, For the first IMF, denoted as EMD remaining amount, and N as the number of IMFs obtained from EMD.

[0089] Step 4: Each row of sampled data Extract the IMF components of the same order and combine them into an empirical mode function matrix. .

[0090]

[0091] Step 5: Use the empirical mode function matrix Replace the original data matrix To remove most of the noise, By inputting a uniform circular array broadband signal model, the output of the broadband beamformer can be obtained.

[0092] The output of the broadband beamformer is correlated with the sample signal, and the position corresponding to the maximum value is the distance of the target from the circular array; the direction pointed to by the main lobe of the beam is the azimuth of the target.

[0093] Example 2

[0094] A uniform circular array of 96 elements with a radius r of 0.13 meters was set up, and the elements were distributed in a counterclockwise direction. The bandwidth of the broadband signal to be processed generated in the experiment was [10kHz, 20kHz], the bandwidth was 10kHz, and the sampling rate was 100kHz.

[0095] A. Water Tank Test. In the test, the uniform circular array was positioned such that the 9th element was closest to the water tank wall, approximately 1.5 meters away; the 21st element was furthest away, approximately 3.35 meters away; and the 33rd element was approximately 3 meters away. The beamforming method was first verified.

[0096] Based on analysis and calculations, and according to the known distribution of the circular array during measurement, the 9th element should correspond to the position with the shortest distance from the pool wall (approximately 1.5 meters). Assuming the angle corresponding to the 9th element is... The original data was received using the 9th element extracted by MATLAB, as shown below. Figure 2 As shown, the horizontal axis represents time, and the vertical axis represents signal strength.

[0097] exist Figure 2 In the range [0,1], the pulses are the transmitted pulses, the pulses are the direct echoes of the transmitted signal, and the pulses are the reflected echoes.

[0098] Take the incident signal Beamforming is performed on the array elements within the range of the 9th element to obtain a beamforming signal with the main lobe direction of the 9th element as shown in the figure. Figure 3 As shown in the figure, the correlation results between the sample signal and the beamforming signal are illustrated in the diagram. Figure 4 As shown. By Figure 4 As can be seen, the distance is 1.568 meters, which is consistent with the actual situation, indicating that the beamforming result is correct.

[0099] B. Lake test: Static distance-azimuth test.

[0100] The uniform circular array used in the experiment was the same as above. The spherical target deployed in the lake was at the same underwater depth as the uniform circular array, and the straight-line distance between the center of the sphere and the 33rd channel (90° azimuth) of the uniform circular array was... , , The test data corresponding to different distances were recorded and the distances were calculated. The static distance test results are attached. Figures 5-7 As shown.

[0101] As shown in the figure, the calculated distances at the 90° azimuth are 15.9 meters, 33.12 meters, and 43.02 meters, which are basically consistent with the actual target distances of 16 meters, 32.5 meters, and 42.5 meters. However, the target echo intensity is low when the distance is 42.5 meters.

[0102] The experimental procedure for static target orientation testing involves keeping the target position unchanged, with... Using this as a baseline, the sonar is rotated clockwise and the rotation angle is recorded. The data is then analyzed and processed to locate the target and calculate the channel corresponding to the maximum target intensity. This value is then compared with the recorded rotation angle to verify the target's location.

[0103] The test results are as follows: (1) When the beam direction of the 33rd channel of the image sonar is set to 90 degrees when facing the target, the image sonar is rotated clockwise by about 20 degrees and the target is scanned. After analyzing the data, it is found that the channel corresponding to the maximum target intensity is the 39th channel. The angle interval between the 39th and 33rd channels is calculated to be 6*3.75=22.5 degrees. Considering that the angle resolution between the two channels is 3.75 degrees, it is considered that the actual rotation angle (20 degrees) of the image sonar is basically consistent with the change angle (22.5 degrees) obtained by data processing. (2) Continue to rotate the image sonar clockwise. After rotating about 28 degrees, the scan data is recorded. After data analysis, it is found that the channel facing the target at this time is the 47th channel. The angle interval between the 47th and 39th channels is 8*3.75=30 degrees. The error of 2 degrees is within the range of the resolution of the two channels of 3.75 degrees. Therefore, it is considered that the actual rotation angle (28 degrees) of the image sonar is basically consistent with the change angle (30 degrees) obtained by data processing.

[0104] C. Lake test: conduct dynamic distance-orientation test.

[0105] The vertical motion distance test procedure involves fixing the target's azimuth relative to the sonar, and then gradually moving the target away from the sonar from that azimuth, while measuring the distance between the sonar and the target. To facilitate observation and reproduce the target's vertical motion process, the timeframes used for distance calculation should cover as wide a range as possible. Therefore, motion data at 0.43s, 2.42s, 14.78s, 33.17s, 44.17s, and 59.17s were selected for distance calculation, and the vertical motion test results are shown in the attached figure. Figure 8 As shown in the figure. The radius of the sector represents the actual distance, and the angle between the sector radius and the horizontal axis represents the target's bearing. Since the target's bearing is fixed at 90 degrees during vertical movement, the y-value of the coordinate point in the figure is the actual distance between the sonar and the target.

[0106] The above examples illustrate the present invention only to aid in understanding it and are not intended to limit the scope of the invention. Those skilled in the art can make various simple deductions, modifications, or substitutions based on the principles of this invention.

Claims

1. A uniform circular array wideband beamforming method, characterized in that, The method comprises the following steps: Step 1: a uniform circular array containing L elements is set, and a uniform circular array wideband signal model is established; Step 2: The array elements of the uniform circular array receive K snapshots of sampling data, and the original data matrix is X L×K ; Step 3: Perform empirical mode decomposition on each row of the original data matrix X L×K to obtain n IMF components l×K to obtain n IMF components Step 4: X L×K The same order IMF components of each row of sampling data x l×K are taken out and combined into an empirical mode function matrix; Step 5: Replace the original data matrix X with the matrix of empirical mode functions L×K and input the uniform circular array wideband signal model to obtain the wideband beamformer output.

2. The method of claim 1, wherein, The uniform circular array wideband signal model is as follows: wherein, is a weight vector of the beamformer corresponding to the qth frequency bin, H denotes the conjugate transpose.

3. The method of claim 2, wherein, In step 2, the original data matrix X L×K is:

4. The method according to any one of claims 1 to 3, characterized in that, In step 3, the pair of original data matrices X L×K The sampled data x l×K The step of performing empirical mode decomposition comprises: Step 31: find the maximum and minimum points of the sampling data x l×K , then use a cubic spline curve to fit the maximum and minimum points to obtain the upper envelope and the lower envelope; Step 32: envelope mean values are calculated based on the upper envelope line and the lower envelope line; Step 33: The envelope mean is removed from x l×K to obtain the residual component; Step 34: judging whether the residual component is an IMF component, if not, taking the residual component as a new x l×K , repeating step 31, step 32 and step 33 until the residual component is an IMF component, and finally obtaining n IMF components and 1 non-zero mean low-order polynomial residual.

5. A method of target range and bearing detection, characterized by The method comprises the following steps: The output of the wideband beamformer obtained by the method is correlated with sample signal segments, then the maximum value of the accumulated results of each segment correlation result is obtained, and the position corresponding to the maximum value is the distance of the target from the circular array; the direction indicated by the beam main lobe is the target azimuth.

Images