Highly stable underwater multi-target direction finding method and system based on characteristic spectrum tracking using single vector hydrophone

Abstract

The invention discloses a method and system for underwater multi-target high-stability direction finding based on characteristic spectrum tracking by a single vector hydrophone. The method comprises: 1. picking up data and intercepting ultra-low frequency signals; sampling and processing, and dividing N frames of data; 2. performing FFT analysis on discrete signals, and performing instantaneous phase difference variance weighted processing to obtain a spectrum; 3. performing cross-spectrum azimuth estimation on sound pressure and vibration velocity information, performing multi-target detection and azimuth estimation based on acoustic energy flow, and locking and tracking multi-target characteristic spectrum information; 4. performing secondary azimuth estimation on multi-targets based on characteristic spectrum, calculating characteristic spectrum azimuth information of each target, performing azimuth acoustic energy flow clustering processing, and obtaining target azimuth and characteristic spectrum information; 5. repeating steps 2 to 4, updating and iterating to obtain target azimuth and characteristic spectrum information data sets, fusing the data sets of the previous and next time periods and performing Euclidean distance calculation, and performing matching processing through the metric space threshold; 6. eliminating wild values ​​in the azimuth estimation result, and outputting multi-target high-stability azimuth information.

Description

Technical Field

[0001] The present invention belongs to the field of signal processing technology, and relates to a technical solution for multi-target direction estimation, and in particular to a method and system for underwater multi-target high-stability direction finding using a single vector hydrophone based on characteristic spectrum tracking. Background Art

[0002] Passive multi-target direction estimation is a key technology in underwater acoustic signal processing. It primarily utilizes hydrophones to receive the radiated noise generated by underwater sound sources and obtain direction information through related signal processing techniques. Existing pressure hydrophones can only measure scalar sound pressure information and are unable to determine target orientation or estimate direction. Only arrays of multiple hydrophones can possess spatial directivity and spatial anti-interference capabilities. However, with the advancement of vibration and noise reduction technologies, the radiated noise from underwater targets such as ships is becoming increasingly lower. The noise levels of some quiet submarines in the frequency band above 200 Hz are no higher than those of the ocean environment. This places higher demands on the low-frequency detection capabilities of hydrophones. Therefore, ultra-low-frequency vector hydrophones and their signal processing technologies have become a research hotspot.

[0003] The vector hydrophone is a new type of underwater acoustic signal detection device. Combining a traditional pressure sensor with multiple particle velocity sensors, it can capture more comprehensive underwater sound field information. Compared to pressure hydrophones, vector hydrophones possess both frequency-independent dipole directivity and robustness against isotropic noise, providing more comprehensive sound field information. Currently, direction-finding technologies based on vector hydrophones can be divided into two categories based on their characteristics. The first utilizes the array flow characteristics of vector hydrophones, analogizing each channel to a multi-element array and employing existing array signal processing methods for direction-finding. The second utilizes combined pressure-velocity signal processing, which can suppress isotropic noise and resolve multiple targets within the same beam. Compared to other direction-finding methods, combined pressure-velocity signal processing offers superior robustness, ease of implementation, and minimal computational complexity, resulting in superior performance under ideal conditions. However, the real ocean often contains interfering targets such as surface vessels, as well as natural noise generated by wind and other factors, and biological noise from fish schooling, resulting in a highly anisotropic ocean background sound field. The above factors will cause serious interference to the target sound source signal, which leads to problems such as poor azimuth estimation accuracy and estimation value jump in the existing underwater multi-target direction-finding technology based on single vector hydrophone in engineering applications, and even causes algorithm failure, which seriously restricts the application and promotion of vector hydrophones. Summary of the Invention

[0004] In response to the technical problems in the prior art of conventional cross-spectrum direction finding methods under low signal-to-noise ratio conditions, where the estimated value deviates greatly from the true value and the observability of multiple targets is poor, the present invention proposes a high-stability underwater multi-target direction finding method and system using a single vector hydrophone based on characteristic spectrum tracking. The present invention obtains sound field information through a vector hydrophone, utilizes the line spectrum instantaneous phase difference stability weighting technology to improve the line spectrum signal-to-noise ratio gain and effectively suppress the energy interference of environmental noise, combines target detection and azimuth estimation based on acoustic energy flow, completes characteristic spectrum locking and tracking, eliminates wild values ​​in the azimuth estimation results, and thus achieves high-precision and stable direction finding for multiple targets. Through simulation analysis and experimental verification, it can be seen that the accuracy of multi-target direction finding under low signal-to-noise ratio conditions of the present invention is improved by an average of 24% compared with conventional methods, and the azimuth history is highly consistent with the true azimuth, can significantly eliminate interference signals, and has strong direction finding stability.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] A high-stability underwater multi-target direction finding method using a single vector hydrophone based on characteristic spectrum tracking comprises the following steps:

[0007] Step 1: The vector hydrophone picks up data and intercepts the ultra-low frequency signal; samples and processes it and divides it into N frames of data;

[0008] Step 2: Perform FFT analysis on the discrete signal and perform instantaneous phase difference variance weighting to obtain the spectrum;

[0009] Step 3: Perform cross-spectral azimuth estimation on the sound pressure and vibration velocity information, perform multi-target detection and azimuth estimation based on the acoustic energy flow, and lock and track the multi-target characteristic spectrum information;

[0010] Step 4: Perform secondary azimuth estimation on multiple targets based on the characteristic spectrum, calculate the characteristic spectrum azimuth information of each target, perform azimuth acoustic energy flow clustering processing, and obtain the target azimuth and characteristic spectrum information;

[0011] Step 5: Repeat steps 2 to 4 N times, update and iterate to obtain the target position and characteristic spectrum information dataset, fuse the datasets of the previous and next time periods, calculate the Euclidean distance, and perform matching processing through the metric space threshold;

[0012] Step 6: Eliminate outliers in the azimuth estimation results and output multi-target highly stable azimuth information.

[0013] Based on the above-mentioned problems existing in the prior art, the present invention is dedicated to solving the problem of high-precision direction finding for multiple underwater targets. First, the present invention studies the problem of poor azimuth estimation under low signal-to-noise ratio conditions and achieves high-precision direction finding for targets by weighting the stability of the instantaneous phase difference of the line spectrum. Then, based on the accurate target azimuth, the invention combines characteristic spectrum tracking technology to effectively eliminate outliers in the azimuth estimation results, thereby improving the observability of the multi-target azimuths. This high-precision multi-target direction finding technology has broad application prospects in the field of underwater acoustics and has important theoretical significance and engineering application value.

[0014] As a preference, in step 1, the received signal is first filtered and amplified to extract the 15Hz to 200Hz low-frequency information, and then sampled by the A / D circuit with a sampling rate of f s ; and divide the sampled data into N frames of equal length.

[0015] As a preferred embodiment, step 2 is as follows: the single discrete signal obtained in step 1 is divided into M segments for FFT analysis. Assuming that the single sampling signal is p(t), the obtained characteristic spectrum is recorded as P(f k ), k = 1, 2, ..., K; where K is the number of frequency units obtained by FFT analysis; the instantaneous phase difference value under each frequency unit is calculated according to formula (11),

[0016]

[0017] Where, 1≤m'≤M-1, 1≤m≤M, 1≤k≤K;

[0018] The variance of the instantaneous phase difference of all frequency units is weighted; it is calculated according to formula (12), which is recorded as

[0019]

[0020] In the formula, 1≤m'≤M-1, 1≤k≤K.

[0021] As a preferred method, step 3 is as follows: cross-spectral azimuth estimation is performed on the spectrum of sound pressure and velocity information, and multi-target detection is performed based on acoustic energy flow according to the following three principles: (1) the boundary width is less than 8; (2) the beam slope exceeds 3; (3) the amplitude exceeds 3 times the average energy in the beam domain; the target acoustic energy flow is automatically extracted, and the target characteristic spectrum information is combined with the target positioning and tracking matrix, as shown in formula (13):

[0022] If the current data frame is the initial data frame m=0, the target spectrum is expressed as:

[0023]

[0024] If the current data frame m>0, the instantaneous frequency corresponding to the target in the data frame is fg , g=1,2,...,G, the corresponding instantaneous frequency under interference noise is f y , y=1,2,...,Y, update the characteristic spectrum matrix according to formula (14):

[0025]

[0026] Where: a1 and a2 are the weight changes of the target instantaneous frequency and the noise instantaneous frequency in each iteration, ε1 is the target instantaneous frequency width, g is the frequency unit corresponding to the target instantaneous frequency at the current moment, and y is the frequency unit corresponding to the noise instantaneous frequency at the current moment;

[0027] The target spectrum is expressed as:

[0028]

[0029] In the above formula, 1≤k≤K.

[0030] As a preferred embodiment, step 4 is as follows: perform conjugate cross-spectrum on the multi-target characteristic spectra, calculate the quadratic azimuth estimation of each frequency point azimuth information according to formula (16), and perform cluster statistics to obtain the acoustic energy flow target azimuth E(n,f g ) and characteristic spectrum Information dataset:

[0031]

[0032] In the above formula, and is the final acoustic energy flow component obtained by the single discrete signal characteristic spectrum information locking and tracking method, 1≤m≤M, g=1,2,...,G.

[0033] As a preferred method, step 5 is as follows: repeat steps 2 to 4 N times, update and iterate to obtain N rows of orientation and characteristic spectrum information data sets, match and fuse the information of the previous and next time periods in sequence; count the fusion results in sequence starting from the second frame data, and calculate the Euclidean distance of the characteristic spectrum and orientation information according to formula (17) to obtain ρ f (n',g) and ρ e (n',g),

[0034]

[0035] Where: f g (n) is the frequency unit of the current characteristic spectrum corresponding to the current target, I(n,f g ) is the peak amplitude of the characteristic spectrum corresponding to the current target, θ(n) is the current target direction information, E(n,θ) is the current target acoustic energy flow energy, f g is the instantaneous frequency unit corresponding to the target at the current moment, g=1,2,3,…,G;

[0036] According to formula (18), the threshold value of the characteristic spectrum and target orientation metric space is set to D f and D e :

[0037]

[0038] The present invention also discloses a system based on the above-mentioned single vector hydrophone underwater multi-target high-stability direction finding method, comprising the following modules:

[0039] Data acquisition module: vector hydrophone picks up data and intercepts ultra-low frequency signals; performs sampling processing and divides N frames of data;

[0040] Phase difference variance weighted processing module: performs FFT analysis on discrete signals and performs instantaneous phase difference variance weighted processing to obtain the spectrum;

[0041] Multi-target characteristic spectrum information locking and tracking module: performs cross-spectral azimuth estimation of sound pressure and vibration velocity information, performs multi-target detection and azimuth estimation based on acoustic energy flow, and locks and tracks multi-target characteristic spectrum information;

[0042] Target direction and characteristic spectrum information acquisition module: performs secondary direction estimation on multiple targets based on characteristic spectrum, calculates the characteristic spectrum direction information of each target, performs azimuth acoustic energy flow clustering processing, and obtains target direction and characteristic spectrum information;

[0043] Iteration module: updates and iterates to obtain the target position and characteristic spectrum information dataset, fuses the datasets of the previous and next time periods, calculates the Euclidean distance, and performs matching processing through the metric space threshold;

[0044] Multi-target highly stable azimuth information output module: removes outliers in the azimuth estimation results and outputs multi-target highly stable azimuth information.

[0045] The present invention has the following beneficial effects:

[0046] 1. The present invention utilizes the characteristics that the instantaneous phase difference corresponding to the target line spectrum frequency unit is relatively stable, while the instantaneous phase difference corresponding to the background noise frequency unit varies randomly. By performing variance weighting on the instantaneous phase difference, the energy interference of the background noise can be further suppressed, the line spectrum signal-to-noise ratio gain can be enhanced, and high-precision direction finding for underwater target azimuth estimation can be achieved.

[0047] 2. This invention locks and tracks the characteristic spectrum information of multiple targets, further enhancing the correlation between target position and characteristic spectrum information within each time period, improving the observability of multiple targets. Simultaneously, the characteristic data sets for each time period are fused, and outliers in the position estimation results are eliminated through metric space matching, achieving highly stable position estimation for multiple underwater targets.

[0048] 3. The present invention has no requirements on the bandwidth and frequency of the received signal and has good robustness and engineering practical value. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is the principle block diagram of cross-spectral method direction estimation.

[0050] Figure 2 This is a flowchart of multi-target detection and direction estimation based on acoustic energy flow.

[0051] Figure 3 It is the flow chart of multi-target signature spectrum locking and tracking.

[0052] Figure 4 It is a flow chart of the underwater multi-target high-stability direction finding method.

[0053] Figure 5 Schematic diagram of the metric space matching simulation experiment results in the preferred embodiment.

[0054] Figure 6 Schematic diagram of Monte Carlo simulation experiment results in a preferred embodiment.

[0055] Figure 7 This is a schematic diagram of the positions of two moving targets and observation stations in the simulation experiment.

[0056] Figure 8 It is the LOFAR diagram based on the characteristic spectrum tracking method in the simulation experiment.

[0057] Figure 9 It is a time-orientation history diagram based on the characteristic spectrum tracking method in the simulation experiment.

[0058] FIG10(a) and FIG10(b) are schematic diagrams of the deployment of the signal collection system in the preferred embodiment.

[0059] FIG11( a ) and FIG11 ( b ) are respectively a LOFAR graph based on a conventional method and a LOFAR graph based on a characteristic spectrum tracking method in a preferred embodiment.

[0060] FIG12(a) and FIG12(b) are respectively a time-orientation history diagram based on a conventional method and a time-orientation history diagram based on a characteristic spectrum tracking method in a preferred embodiment.

[0061] Figure 13 This is a system block diagram of the highly stable underwater multi-target direction finding method based on a single vector hydrophone according to the present invention. DETAILED DESCRIPTION

[0062] The present invention will be described in further detail below with reference to the accompanying drawings.

[0063] The present invention relates to a method and system for highly stable underwater multi-target direction finding using a vector hydrophone based on characteristic spectrum tracking. The present invention utilizes a single vector hydrophone to pick up more comprehensive sound field information, and improves the line spectrum signal-to-noise ratio gain and effectively suppresses ocean background noise through instantaneous phase difference weighting. Combined with target detection and azimuth estimation technology using acoustic energy flow, the method completes multi-target characteristic spectrum locking and tracking, fuses and matches the obtained information data set, and eliminates outliers in the azimuth estimation results, thereby achieving highly precise and stable direction finding for multiple targets.

[0064] 1. Cross-spectrum direction finding method based on single vector hydrophone

[0065] For point source signals in the far field, the three-dimensional space can be approximately simplified to a two-dimensional case. At this time, the sound pressure information collected by the vector hydrophone and the vibration velocity information of each channel can be expressed as follows:

[0066]

[0067] Where p(t) is the sound pressure signal picked up by the vector hydrophone, v x (t) is the X-axis velocity signal picked up by the vector hydrophone, v y (t) is the Y-axis velocity signal picked up by the vector hydrophone, α is the azimuth angle of the target in the horizontal direction, n p (t), n vx (t), n vy (t) are the interference noise of sound pressure and vibration velocity in the X-axis and Y-axis directions respectively.

[0068] The cross-spectral components are defined as:

[0069]

[0070] Where: ω represents frequency; superscript * represents complex conjugate; P(f), V x (f) and V y (f) are p(t), v x (t) and v y (t) is obtained by Fourier transform. According to the basic characteristics of Fourier transform, the energy of two in-phase inputs is concentrated in the real part of the cross spectrum. Therefore, taking the real part of the result of the operation of formula (2) will give the acoustic energy flow component I of each frequency. x (f) and I y (f), and thus the modulus R(f) of the acoustic energy flow at each frequency point is obtained:

[0071]

[0072] The estimated value of the sound source direction at each frequency point is:

[0073]

[0074] The principle block diagram of cross-spectral method direction estimation is as follows Figure 1 shown.

[0075] 2. Highly stable underwater multi-target direction finding method based on vector hydrophone and characteristic spectrum tracking

[0076] 1.1 Multi-target detection and orientation estimation based on acoustic energy flow

[0077] Since the cross-spectral method can obtain the azimuth corresponding to each frequency point, it is impossible to intuitively know the direction of each target without prior knowledge such as the target distribution frequency band and the number of signal sources. Therefore, it is necessary to use the energy statistics method to count the azimuth corresponding to each frequency point.

[0078] Divide [0°, 360°] into several intervals Δθ, and the total number of intervals N allowed for spatial angles is:

[0079]

[0080] From the above formula, we can get the range of horizontal azimuth angle: 0~Δθ, Δθ~2Δθ, 2Δθ~3Δθ, ..., (N-1)Δθ~NΔθ. Scan the entire frequency band to get the corresponding azimuth angle. Assume that the number of estimated azimuths of each frequency point falling into each statistical angle interval is m1, m2, m3, ..., m N , then calculate the following statistic S(θ):

[0081]

[0082] Where, In order to obtain the modulus of the acoustic energy flow that conforms to the angle range of the statistical interval, the distribution of the acoustic energy flow in each statistical interval is obtained by formula (6), and the acoustic energy flow beam diagram can be obtained.

[0083] According to the fact that the line spectrum of underwater target radiation noise carries a large amount of azimuth information, combined with the characteristics of the acoustic energy flow beam diagram that uses the frequency point acoustic energy flow as a weight for clustering statistics, three criteria for automatic target azimuth extraction can be obtained: 1) The target beam must contain left and right boundaries, and the boundary width should meet a certain threshold value; 2) The absolute value of the target beam slope should exceed the corresponding threshold value; 3) The target acoustic energy flow amplitude should exceed a certain threshold value and be the maximum value within the interval. Through these three basic principles, the detection and azimuth estimation of multi-target azimuth information in the histogram can be achieved. The program flow chart is as follows: Figure 2 As shown:

[0084] The peak height threshold is closely related to the background radiation noise. for:

[0085]

[0086] Where B is the bandwidth of the bandwidth signal bs(t), σ bs 2 is the variance of the bandwidth signal bs(t).

[0087] As can be seen from the above formula, under normal circumstances, the spectral level of the target radiation signal is much higher than the average spectral level of the noise signal. Because the background noise acoustic energy flow fluctuates greatly, the final target acoustic energy flow is significantly affected by the noise. However, the more times the accumulation (averaging) is performed, the smoother the background noise becomes, and the peak height threshold can be set relatively small. The peak height threshold of the present invention is 3 times the average energy value of the acoustic energy flow. In actual engineering applications, the target extraction threshold can be set according to the application requirements.

[0088] 1.2. Multi-target signature spectrum locking and tracking technology

[0089] In view of the problems of poor multi-target observability and strong instantaneous narrowband signal interference during multi-target tracking and direction finding, a preferred embodiment of the present invention proposes a multi-target characteristic spectrum locking and tracking method. This embodiment uses a single vector hydrophone to collect data and perform time-frequency conversion, obtains the acoustic energy flow information of each frequency point through cross-spectral azimuth estimation, combines the acoustic energy flow multi-target detection and azimuth estimation, performs spectrum extraction and spectrum locking, and obtains accurate characteristic spectrum information of the target to be measured. The specific process is shown in Figure 3 .

[0090] Based on the characteristics of the target radiation noise line spectrum being strong and stable and carrying a large amount of target azimuth information, the target azimuth and characteristic spectrum information locking is mainly achieved based on a matrix composed of multiple groups of narrow-band rectangular windows with adjustable center frequencies. This has advantages that fixed parameter filters do not have. In the processing of target noise, due to the change in target distance, the signal strength varies randomly and is not a steady process. In addition, the ocean environment noise also fluctuates, so the time for single data processing cannot be too long. Assuming the input signal x(t), the obtained spectrum is recorded as X(f k ), k = 1, 2, ..., K; where K is the number of frequency units obtained by FFT analysis. The characteristic spectrum tracking and locking matrix can be established as W(k), k = 1, 2, ..., K. If the current time is the initial time, the target spectrum can be expressed as:

[0091]

[0092] The azimuth angle corresponding to each frequency unit is calculated by formula (4). The instantaneous frequency corresponding to the target azimuth is obtained by processing multi-target detection and azimuth estimation based on the acoustic energy flow: g , g=1,2,...,G, the corresponding instantaneous frequency under interference noise is f y , y=1,2,...,Y. Based on this, the tracking matrix is ​​locked as the weight at the next moment. The iterative process is as follows:

[0093]

[0094] Where: a1 and a2 are the weight changes of the target instantaneous frequency and the noise instantaneous frequency in each iteration, ε1 is the target instantaneous frequency window width, g is the frequency unit corresponding to the target instantaneous frequency at the current moment, and y is the frequency unit corresponding to the noise instantaneous frequency at the current moment.

[0095] If the current time is not the starting time, t≥1, as shown in formula (10):

[0096]

[0097] Where k = 1, 2, ..., K.

[0098] The current multi-target characteristic spectrum is extracted and locked using the tracking matrix obtained at the previous moment. In a single extraction process, the main interference is the pseudo-peak caused by the continuous spectrum of the signal. Since the instantaneous frequency of the ocean interference signal changes randomly, its amplitude will tend to zero after multiple iterations. Ultimately, the precise instantaneous frequency and target direction information of the target in the ultra-low frequency characteristic spectrum can be obtained. 1.3. High-stability underwater multi-target direction finding method based on vector hydrophone characteristic spectrum tracking

[0099] Assume that during the observation time, multiple targets move at a constant speed in a fixed direction. Each target's radiation noise contains a component with a frequency of f0, and there may be multiple components. Due to the instability of the amplitude and frequency of the target radiation characteristic spectrum, the method of the present invention uses an iterative tracking matrix to match the target characteristic spectrum with azimuth frequency points, thereby obtaining the target's characteristic spectrum history for subsequent azimuth estimation, and performing fusion matching processing on the data set to eliminate outliers in the azimuth estimation results, thereby realizing a highly stable underwater multi-target direction finding method using a vector hydrophone with characteristic spectrum tracking. The process is as follows: Figure 4 As shown, the specific implementation method is as follows:

[0100] Step 1: Taking the sound pressure signal received by the two-dimensional vector hydrophone as an example, the received signal is first filtered and amplified to extract its 15Hz to 200Hz low-frequency information, and then sampled through the A / D circuit with a sampling rate of f s ; and divide the sampled data into N frames of equal length.

[0101] Step 2: Divide the single discrete signal obtained in step 1 into M segments for FFT analysis. Assuming that the single sampling signal is p(t), the obtained characteristic spectrum is recorded as P(f k ), k=1,2,…,K; where K is the number of frequency units obtained by FFT analysis. According to formula (11), the instantaneous phase difference value under each frequency unit is calculated.

[0102]

[0103] In the formula, 1≤m'≤M-1, 1≤m≤M, 1≤k≤K.

[0104] The variance of the instantaneous phase difference is weighted for all frequency units. Calculated according to formula (12), it is expressed as

[0105]

[0106] In the formula, 1≤m'≤M-1, 1≤k≤K.

[0107] Step 3: Perform cross-spectral azimuth estimation on the above-mentioned sound pressure and velocity information spectra. Based on the multi-target detection of acoustic energy flow, according to the following three principles: (1) the boundary width should be less than 8; (2) the beam slope should be greater than 3; (3) the amplitude should be greater than 3 times the average energy in the beam domain. These three principles can realize the automatic extraction of target acoustic energy flow and establish the target characteristic spectrum information joint target positioning and tracking matrix, as shown in formula (13):

[0108] If the current data frame is the initial data frame m=0, the target spectrum can be expressed as:

[0109]

[0110] If the current data frame m>0, the instantaneous frequency corresponding to the target in the data frame is f g , g=1,2,...,G, the corresponding instantaneous frequency under interference noise is f y , y=1,2,...,Y, update the characteristic spectrum matrix according to formula (14):

[0111]

[0112] Where: a1 and a2 are the weight changes of the target instantaneous frequency and the noise instantaneous frequency in each iteration, ε1 is the target instantaneous frequency width, g is the frequency unit corresponding to the target instantaneous frequency at the current moment, and y is the frequency unit corresponding to the noise instantaneous frequency at the current moment.

[0113] The target spectrum can be expressed as:

[0114]

[0115] In the above formula, 1≤k≤K.

[0116] Step 4: Conjugate cross-spectrum is performed on the above multi-target characteristic spectra, and the azimuth information of each frequency point is obtained by calculating the quadratic azimuth estimation according to formula (16), and the acoustic energy flow target azimuth E(n,f g ) and characteristic spectrum Information dataset:

[0117]

[0118] In the above formula, and is the final acoustic energy flow component obtained by the single discrete signal characteristic spectrum information locking and tracking method, 1≤m≤M, g=1,2,...,G.

[0119] Step 5: Repeat steps 2 to 4 N times, update and iterate to obtain N rows of orientation and characteristic spectrum information data sets, match and fuse the information of the previous and next time periods in sequence; count the fusion results in sequence starting from the second frame data, and calculate the Euclidean distance of the characteristic spectrum and orientation information according to formula (17) to obtain ρ f (n',g) and ρ e (n',g).

[0120]

[0121] Where: f g (n) is the frequency unit of the current characteristic spectrum corresponding to the current target, I(n,f g ) is the peak amplitude of the characteristic spectrum corresponding to the current target, θ(n) is the current target direction information, E(n,θ) is the current target acoustic energy flow energy, f g is the instantaneous frequency unit corresponding to the target at the current moment, g=1,2,3,…,G.

[0122] According to formula (18), the threshold value of the characteristic spectrum and target orientation metric space is set to D f and D e :

[0123]

[0124] Since the background noise energy flow fluctuates greatly, the final target measurement space is greatly affected by the noise. In the experimental verification analysis of the present invention, the threshold value is selected as 1 / 10 times the energy of the collected data. In actual engineering applications, the measurement space threshold value can be set according to the actual environment.

[0125] Step 6: Eliminate wild-value information in the orientation estimation according to the above-mentioned metric space threshold value, and output highly stable orientation information of multiple targets.

[0126] 3. Performance Analysis

[0127] Due to the fluctuation of background noise energy and the random change of noise frequency, the final metric space threshold is greatly affected by the ocean background noise. The results of the azimuth root mean square error (RMSE) under different metric space threshold coefficients are compared through simulation analysis. Figure 5As shown in the figure, the ocean background noise is white noise, the signal frequency is uniformly distributed within the range of 15Hz to 200Hz, the single sample time is 3 minutes, the sampling frequency is 512Hz, the multi-target horizontal incident angle is uniformly distributed within the range of 0° to 360°, the signal-to-noise ratio is set to 0dB, and the calculation results are statistical data of 2000 independent experiments. Figure 5 As is known, when the threshold coefficient is 10, which is 1 / 10 times the current energy sum, the direction estimation gain effect is the best and the direction finding accuracy is significantly improved.

[0128] To further validate the performance of a highly stable underwater multi-target direction-finding method based on characteristic spectrum tracking, broadband continuous noise, obtained through a FIR linear filter, was used as a broadband continuous spectrum signal within the underwater target radiated noise. A narrowband strong line spectrum signal was superimposed on it, with the average spectral level ratio between the narrowband line spectrum and the nearby broadband continuous spectrum ranging from 10 to 25 dB. The azimuth root mean square error (RMSE) of the characteristic spectrum tracking direction-finding method was compared with that of a conventional cross-spectrum method and a line spectrum instantaneous phase difference stability weighting method at signal-to-noise ratios of -20 to 30 dB. 10,000 independent simulations were performed at each signal-to-noise ratio for statistical analysis. The single data sampling time was set to 3 minutes, and the sampled data was divided into 36 segments, each 5 seconds long, for data analysis. The frequencies of the narrowband line spectra of the multi-target sound sources in each sample were uniformly distributed between 15 and 200 Hz, and the horizontal incidence angles of the target sound sources were uniformly distributed between 0° and 360°. Gaussian white noise and noise of different directions and intensities are added to the above signal to simulate the ocean environment, and strong narrowband line spectrum interference signals are added at random time periods in the ocean environment. The simulation results are as follows Figure 6 As shown, the root mean square error (RMS) of the estimated values ​​using the proposed method is significantly reduced below SNR = 0dB. Above SNR = 0dB, the azimuth estimation accuracy remains within 1°. The azimuth RMS error obtained by the proposed method is reduced by an average of 24% compared to that obtained using a conventional complex intensometer. When the signal-to-noise ratio (SNR) is above 10dB, the performance of the two methods is similar.

[0129] The above simulation experiments can fully prove that the present invention has a good effect on strong line spectrum interference. In order to further verify the effectiveness of the algorithm, on the basis of randomly adding strong line spectrum interference, the target sound source is simulated to simulate the real ship azimuth motion trajectory. Assume that there are two underwater targets in the sea area with uniform linear motion. The initial position, initial azimuth and observation station position of the target are as follows: Figure 7 The line spectrum frequencies of target 1 are 50 and 90, and the line spectrum frequencies of target 2 are 100 and 150. The simulation experiment tracks the target for 30 minutes, divides the data into 360 frames, and the interval between each frame is 5 seconds. Figure 8 The LOFAR diagram based on the characteristic spectrum tracking method and Figure 9The time-azimuth history diagram based on the characteristic spectrum tracking method is shown. By analyzing the LOFAR results based on the characteristic spectrum tracking and the time-azimuth results of the characteristic spectrum tracking, it can be seen that when there is an intersection in the target azimuth history, there is no mutual influence that causes a large deviation between the estimated value and the true value, and it has a good suppression effect on some strong line spectrum interference.

[0130] To verify the effectiveness of the present invention, relevant sea trial data was processed. This data was collected and stored during relevant tests conducted in the South China Sea in November 2017. The water depth at the time was approximately 100 meters, and the seabed was flat. A schematic diagram of the receiving equipment used in the test is shown in Figure 10(a). The vector hydrophone connected to the buoy was deployed at a depth of approximately 70 meters. A schematic diagram of the target orientation of the low-frequency sound source used in the test is shown in Figure 10(b). The low-frequency target sound source with a frequency of 48 Hz was moving approximately 6 km behind the buoy, 40 meters from the water surface. The low-frequency target sound source with a frequency of 92 Hz was moving approximately 7 km to the right of the buoy, 30 meters from the water surface. The sampling rate of the acquisition equipment connected to the hydrophone was 512 Hz, and the acquisition time duration was approximately 3 minutes. The acquired data was divided into 35 segments and subjected to FFT analysis. The final LOFAR map and time-to-direction history map were then generated using conventional methods and the present invention's underwater multi-target high-stability direction-finding method based on characteristic spectrum tracking. The LOFAR comparison diagram is shown in Figure 11 (a) and Figure 11 (b), and the time-azimuth comparison diagram is shown in Figure 12 (a) and Figure 12 (b). Comparing Figures 11 and 12, it can be seen that under low signal-to-noise ratio, the multi-target azimuth angles calculated using the conventional complex sound intensity method have large deviations due to the interference of ocean background noise, making it difficult to observe multiple targets. Relatively speaking, the spectrum obtained by the instantaneous phase difference weighted calculation based on the method of the present invention is more stable, and combined with the multi-target characteristic spectrum locking and tracking method, the stability of the multi-target azimuth angle has been improved to a certain extent, effectively eliminating outliers in the azimuth estimation results, making the time-azimuth history diagram clearer.

[0131] like Figure 13 As shown, the present invention also discloses a system based on the above-mentioned single vector hydrophone underwater multi-target high-stability direction finding method, which includes the following modules:

[0132] Data acquisition module: vector hydrophone picks up data and intercepts ultra-low frequency signals; performs sampling processing and divides N frames of data;

[0133] Phase difference variance weighted processing module: performs FFT analysis on discrete signals and performs instantaneous phase difference variance weighted processing to obtain the spectrum;

[0134] Multi-target characteristic spectrum information locking and tracking module: performs cross-spectral azimuth estimation of sound pressure and vibration velocity information, performs multi-target detection and azimuth estimation based on acoustic energy flow, and locks and tracks multi-target characteristic spectrum information;

[0135] Target direction and characteristic spectrum information acquisition module: performs secondary direction estimation on multiple targets based on characteristic spectrum, calculates the characteristic spectrum direction information of each target, performs azimuth acoustic energy flow clustering processing, and obtains target direction and characteristic spectrum information;

[0136] Iteration module: updates and iterates to obtain the target position and characteristic spectrum information dataset, fuses the datasets of the previous and next time periods, calculates the Euclidean distance, and performs matching processing through the metric space threshold;

[0137] Multi-target highly stable azimuth information output module: removes outliers in the azimuth estimation results and outputs multi-target highly stable azimuth information.

[0138] The present invention discloses a method and system for highly stable underwater multi-target direction finding using a vector hydrophone based on characteristic spectrum tracking. The present invention utilizes a single vector hydrophone to detect underwater sound field information, improves the line spectrum signal-to-noise ratio gain through instantaneous phase difference weighting, and effectively suppresses the energy interference of ocean background noise, thereby improving the accuracy of azimuth estimation. At the same time, the present invention completes multi-target characteristic spectrum information locking and tracking through target detection and azimuth estimation based on acoustic energy flow, which can further enhance the correlation between target azimuth and characteristic spectrum in each time period, thereby solving the problem of poor observability of multiple targets. The method combines information data set fusion matching processing to eliminate outliers in the azimuth estimation results, ultimately achieving highly stable direction finding for multiple targets. Simulation analysis shows that the multi-target direction finding accuracy of the present invention method under low signal-to-noise ratio conditions is improved by an average of 24% compared with conventional methods. The azimuth history obtained from the sea trial data processing results is highly consistent with the true azimuth, can significantly eliminate interference signals, and has strong direction finding stability, demonstrating the effectiveness and good engineering practicality of the processing method of the present invention.

[0139] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that a person skilled in the art can easily conceive within the technical scope disclosed in the present application should be included within the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A single vector hydrophone underwater multi-target high-stability direction finding method based on characteristic spectrum tracking, characterized by: The following steps are involved: Step 1: The vector hydrophone picks up data and intercepts ultra-low frequency signals; performs sampling processing and divides N frames of data; Step 2, perform FFT analysis on the discrete signal, and perform instantaneous phase difference variance weighting processing to obtain the spectrum; Step 3: perform cross-spectral azimuth estimation on the sound pressure and vibration velocity information, perform multi-target detection and azimuth estimation based on the acoustic energy flow, and lock and track the multi-target characteristic spectrum information; Step 4: perform secondary azimuth estimation on multiple targets based on characteristic spectra, calculate the characteristic spectrum azimuth information of each target, perform azimuth acoustic energy flow clustering processing, and obtain the target azimuth and characteristic spectrum information; Step 5, repeat steps 2 to 4 N times, update and iterate to obtain the target position and characteristic spectrum information data set, fuse the previous and next time period data sets and perform Euclidean distance calculation, and perform matching processing through the metric space threshold; Step 6: Eliminate outliers in the azimuth estimation results and output multi-target highly stable azimuth information; In step 1, the received signal is first filtered and amplified to extract the low-frequency information of 15Hz to 200Hz, and then sampled by the A / D circuit with a sampling rate of f s ; and divide the sampled data into N frames of data of equal length; Step 2 is as follows: divide the single discrete signal obtained in step 1 into M segments for FFT analysis. Let the single sampling signal be p(t), and the obtained characteristic spectrum be recorded as P(f k ), k = 1, 2, ..., K; where f k is the frequency unit corresponding to the FFT analysis, K is the number of frequency units obtained by the FFT analysis; the instantaneous phase difference value under each frequency unit is calculated according to formula (11), Where, 1≤m'≤M-1, 1≤m≤M, 1≤k≤K; The variance of the instantaneous phase difference of all frequency units is weighted; the average value of the instantaneous phase difference of M segments is calculated according to formula (12) The variance of the instantaneous phase difference of segment M is denoted as δ(f k ), the final characteristic spectrum is recorded as In the formula, 1≤m'≤M-1, 1≤k≤K.

2. The single vector hydrophone underwater multi-target high-stability direction finding method based on characteristic spectrum tracking as claimed in claim 1 is characterized in that: Step 3 is as follows: cross-spectral azimuth estimation is performed on the sound pressure and vibration velocity information spectra, and multi-target detection is performed based on the acoustic energy flow according to the following three principles: (1) the boundary width is less than 8; (2) the beam slope exceeds 3; (3) the amplitude exceeds 3 times the average energy in the beam domain; the target acoustic energy flow is automatically extracted, and the target characteristic spectrum information joint target positioning and tracking matrix is ​​established, as shown in formula (13): If the current data frame is the initial data frame m=0, the target spectrum is expressed as: If the current data frame m>0, the instantaneous frequency corresponding to the target in the data frame is f g , g=1,2,...,G, the corresponding instantaneous frequency under interference noise is f y , y = 1, 2, ..., Y, update the characteristic spectrum matrix according to formula (14): Where: a1, a2 are the weight changes of the target instantaneous frequency and the noise instantaneous frequency in each iteration, ε1 is the target instantaneous frequency width, g is the frequency unit corresponding to the target instantaneous frequency at the current moment, and y is the frequency unit corresponding to the noise instantaneous frequency at the current moment; The target spectrum is expressed as: In the above formula, 1≤k≤K.

3. The single vector hydrophone underwater multi-target high-stability direction finding method based on characteristic spectrum tracking as claimed in claim 2 is characterized in that: Step 4 is as follows: perform conjugate cross-spectrum on the multi-target characteristic spectra, calculate the azimuth information of each frequency point at the second-order azimuth estimation according to formula (16), and perform cluster statistics to obtain the acoustic energy flow target azimuth E(n,f g ) and characteristic spectrum Information dataset: In the above formula, I x (m,f g ) and I y (m,f g ) is the final acoustic energy flow component obtained by the single discrete signal characteristic spectrum information locking and tracking method, 1≤m≤M, g=1,2,...,G.

4. The method for underwater multi-target high-stability direction finding using a single vector hydrophone based on characteristic spectrum tracking as claimed in claim 3 is characterized in that: Step 5 is as follows: repeat steps 2 to 4 N times, update and iterate to obtain N rows of azimuth and characteristic spectrum information data sets, match and fuse the information of the previous and next time periods in sequence; count the fusion results in sequence starting from the second frame data, and calculate the Euclidean distance of the characteristic spectrum and azimuth information according to formula (17) to obtain ρ f (n',g) and ρ e (n',θ), Where: f g (n) is the frequency unit of the current characteristic spectrum corresponding to the current target, I(n,f g ) is the peak amplitude of the characteristic spectrum corresponding to the current target, θ(n) is the current target direction information, E(n,θ) is the energy of the current target acoustic energy flow, f g is the instantaneous frequency unit corresponding to the target at the current moment, g = 1, 2, 3, ..., G; According to formula (18), the threshold value of the characteristic spectrum and target orientation metric space is set to D f and D e :

5. A system based on the single vector hydrophone underwater multi-target high-stability direction finding method according to any one of claims 1 to 4, characterized in that: Includes the following modules: Data acquisition module: vector hydrophone picks up data and intercepts ultra-low frequency signals; performs sampling and processing and divides N frames of data; Phase difference variance weighted processing module: perform FFT analysis on discrete signals and perform instantaneous phase difference variance weighted processing to obtain the spectrum; Multi-target characteristic spectrum information locking and tracking module: cross-spectral azimuth estimation of sound pressure and vibration velocity information, multi-target detection and azimuth estimation based on acoustic energy flow, and locking and tracking of multi-target characteristic spectrum information; Target direction and characteristic spectrum information acquisition module: perform secondary direction estimation on multiple targets based on characteristic spectrum, calculate the characteristic spectrum direction information of each target, perform azimuth acoustic energy flow clustering processing, and obtain target direction and characteristic spectrum information; Iteration module: Updates and iterates to obtain the target position and characteristic spectrum information data set, fuses the previous and next time period data sets and performs Euclidean distance calculation, and performs matching processing through the metric space threshold; Multi-target highly stable azimuth information output module: removes outliers in the azimuth estimation results and outputs multi-target highly stable azimuth information.

Images