Analysis of detection performance of single vector hydrophone histogram direction finding algorithm

The histogram algorithm of a single vector hydrophone has good robustness and target azimuth estimation performance. This article analyzes and summarizes the target detection performance of the histogram algorithm, and proposes an autonomous detection of underwater targets based on target azimuth estimation. Tracking algorithm, this algorithm can achieve autonomous detection of the presence or absence of targets in the water. Simulation and anechoic pool test results show that the signal-to-noise ratio required by the histogram algorithm to achieve autonomous target tracking needs to be greater than −7 dB. At this time, the direction finding error is about 8◦, and the −3 dB azimuth spectrum width is about 20◦. The analysis of marine test data shows that the histogram algorithm can achieve full target detection and tracking within a distance of 13.8 km for a surface vessel with a speed of 8.4 kn, with an optimal direction finding error of 5◦, and a −3 dB bearing at a distance of 2 km. Spectral width can reach 10◦

The vector channel of the vector hydrophone sensor has frequency-independent dipole directivity, and has the ability to resist isotropic noise interference.A vector hydrophone can achieve full-space blur-free orientation, which provides a solution for target detection on small underwater platforms equipped with underwater acoustic sensors.

its advantage of space. In recent years, with the continuous improvement of vector hydrophone technology, vector signal processing technology is also being applied powerfully.Driven by demand, it has developed rapidly. Compared with conventional sound pressure hydrophones, vector hydrophones provide more comprehensive sound field information.Only the scalar of the sound field can be measured, and the vector characteristics of the sound field can also be obtained, which greatly broadens the signal processing space. There are many target azimuth estimation algorithms based on single vector hydrophones, in general, they can be divided into two categories according to the principle of direction finding: one is azimuth estimation based on sound energy flow; the other is to regard each channel of the vector hydrophone. It is a multi-element array, each element is approximately at the same position in space, and the existing array signal processing method is applied to the single vector hydrophone by using the characteristics of the array flow pattern of the single vector hydrophone itself. Various target direction finding algorithms of vector hydrophone have their own advantages and disadvantages.Compared with other algorithms, the medium histogram algorithm has better robustness and target orientation estimation performance, and has the ability to suppress narrowband and strong line spectrum interference, which is especially suitable for engineering applications. This paper analyzes and summarizes the histogram direction finding algorithm based on a single vector hydrophone, and proposes an autonomous detection and tracking algorithm for underwater targets based on target orientation estimation, using computer simulation, anechoic pool measurement data and marine experiments data analyzed histogram and graph algorithm target detection performance.

1 Theoretical algorithm

1.1 Histogram direction finding algorithm

The histogram algorithm needs to first calculate the target azimuth estimates at different frequency points, and the calculation expression is

θ(f) = arctan Re ⟨P∗w(f) × Vyw(f)⟩ Re ⟨P∗w(f) × Vxw(f)⟩ = arctan ⟨Iy(i, f)

⟨Ix(i, f)⟩, (1) In formula (1), θ(f) represents the target azimuth calculated at different frequencies f, and Pw, Vxw, and Vyw represent the sound pressure of the vector hydrophone in P and the vibration in the x direction, respectively. The speed channel and the y-direction vibration speed channel collect signal spectrum values, and Ix and Iy represent the acoustic energy flow in the x-direction and y-direction, respectively. It can be seen from equation (1) that the target azimuth calculated by equation (1) is related to the frequency f, and the target azimuth estimates at different frequency points are different. The method of estimating the target azimuth through the histogram can be used to calculate the target azimuth in the environment. Narrowband interference and strong line spectrum interference suppression, but when there are multiple targets in the environment .When the radiated noise frequencies overlap each other, the histogram method cannot get the true azimuth of each target, only the sound energy flow of each target.

The combined orientation will be biased towards the more intense target orientation. The histogram azimuth statistics is to count the target estimated azimuth θ(f) in the corresponding azimuth interval according to the number of frequency points. If the azimuth interval is divided by 1◦, then k = [θ(f) × 180/π], φ (k) = φ(k) + 1, (2) In formula (2), [] represents the rounding operation, k is the value obtained by rounding θ(f), such as θ(f) 60, then θ( f) = θ(f)+ 360◦, so that the estimated azimuth of the target falls on the interval [0◦ 360◦), φ is the frequency of the azimuth estimation at each angle, and the angle value corresponding to the maximum value is the estimated azimuth of the target.

1.2

An Algorithm for Autonomous Target Detection and Tracking

The autonomous detection and tracking algorithm for underwater targets based on target orientation estimation. The basic idea is to perform statistical analysis on the target orientation estimated by the histogram algorithm, and compare the orientation statistics with preset thresholds, which can finally realize autonomous detection of underwater targets And tracking. The flowchart of autonomous target detection and tracking includes the following five steps: (1) First, use the single vector hydrophone histogram algorithm to scan the entire spatial direction to obtain the estimated azimuth Ag of the received signal; (2) Use constant virtual The alarm detector (CA-CFAR detector) performs constant false alarm processing on the target orientation obtained in step (1); (3) If the CA-CFAR detector judges Ag as the target signal orientation, the Ag value is assigned to the matrix AgT[ i], otherwise, assign −1 to the matrix AgT[i] (i = 1, 2, ·, N); (4) If the number of values of the matrix AgT = −1 is greater than AT (AT is the preset threshold , AT <N, N is the number of elements of the matrix AgT), then perform statistical analysis on the matrix AgT, otherwise repeat steps (1)∼(3); (5) For the matrix AgT .

Calculating the root mean square error StdAT, if StdAT is less than the threshold StdDT, it is judged that there is a target, and the target position is tracked, otherwise repeat steps (1) ∼ (4). Through the above 5 steps, autonomous detection and tracking of underwater targets can be achieved. The principle of CA-CFAR processing is that when detecting and tracking a certain azimuth target, due to the non-stationary nature of the marine environment, the false alarm probability is unstable near a certain detection probability, and the real-time tracking of the environmental noise level is setting a time-varying threshold can achieve a constant detection effect with a constant false alarm probability for the azimuth target. In general, the threshold is a function of the detection probability and the false alarm probability. CA-CFAR processing technology of  is a signal processing algorithm that provides the detection threshold in the automatic detection system and minimizes the influence of noise and interference on the false alarm probability of the detection system. In the CA-CFAR processing technology, when a specific unit needs to be tested, the tested unit is called the test unit (Cell under test, CUT), and the sample unit used to extract the noise power around the test unit is called the reference unit (Reference). cells, RC). In order to prevent the target signal from leaking into the reference unit, which will adversely affect the noise power estimation, a part of the sample should be reserved as a guard cell (GC) between the reference unit and the test unit. The relationship between the test unit, reference unit and protection unit is given.

2 Target detection performance analysis

This section will give the computer simulation results of the target detection performance of the histogram algorithm, and use the anechoic pool and sea test data to analyze the

Algorithm target direction finding and autonomous tracking performance. For the sake of simplicity, this article only analyzes the single target situation.

2.1 Simulation analysis

The simulation conditions are as follows: considering that a broadband target signal is incident on a single vector hydrophone with an incident azimuth of 100◦, and the signal-to-noise ratio (Signal to noise ratio (SNR)) in the same frequency band is set to −20 ∼ 16 dB, with 2 dB intervals, the additional noise is Gaussian white noise that is not related to the incident signal, and the sampling frequency is 20 kHz. The data length of each calculation process is 5 s, and 75% of the data is reproduced in the time window.

The stack rate is subdivided into 17 pieces of data with a length of 1 s, and 32768-point Fast Fourier Transform (Fast Fourier Transform) is performed on each piece of data.form, FFT) calculation, the processing frequency band is 200 Hz ∼3 kHz, 17 groups of sound intensity spectra are calculated and averaged, and then the histogram algorithm is used for the purpose.

Estimated standard orientation. Figure 3 shows the azimuth estimation results of the histogram algorithm using the above simulation conditions as a function of the signal-to-noise ratio (that is, the normalized azimuth spectrum varies with the signal.The noise ratio changes, and the azimuth spectrum is the amplitude in different azimuths), and 200 independent Monte Carlo simulation experiments are performed under each signal-to-noise ratio.It can be seen that the estimated azimuth history gradually becomes clear as the signal-to-noise ratio increases. In order to quantitatively describe the target orientation estimation performance of the histogram algorithm, Figure 4 and Figure 5 .The curve of direction finding error and −3 dB azimuth spectrum width versus SNR are respectively given. It can be seen that when the signal to noise ratio is −7 dB, the direction finding .The error is about 8◦, and the −3 dB azimuth spectrum width is about 19◦; when the signal-to-noise ratio is greater than 0 dB, the direction finding error and −3 dB azimuth spectrum width are respectively less than 3◦ and 7◦

Figure 6 is the curve of the target autonomous tracking flag with the signal-to-noise ratio according to the target autonomous detection and tracking algorithm proposed in Section 1. The target tracking flag 1 represents that the algorithm achieves target tracking, and 0 means that the target tracking is not achieved. It can be seen from Figure 6 that when the signal-to-noise ratio is greater than −7 dB.Time histogram algorithm can achieve autonomous target .

2.2 Tank test analysis

In order to master the target detection performance of the single-vector hydrophone histogram algorithm, a single-vector hydrophone target detection performance verification test was carried out in an anechoic pool. UW350 was used as the sound source target during the test, and the depth was used for 3 m underwater. The signal used in the test is the width of the signal source output.With Gaussian white noise, the output peak-to-peak value is set to 10 mV, 20 mV, 25 mV, 50 mV, 100 mV, 1 V, and 10 V respectively. The transmission time of each signal is 60 s, and the sound source level of the small signal emission passes the formula 20 lg (A1/A2) is calculated, where A1 and A2 are the peak-to-peak values of the signal source settings. From the signal emission sound source level, the signal-to-noise ratio of each channel of the vector hydrophone can be calculated based on the distance between the vector hydrophone and the sound source. Table 1 shows the results of the broadband average signal-to-noise ratio of the sound source signal received by each channel of the vector hydrophone, and gives the average value of the signal-to-noise ratio of each channel under different sound source emission intensities. It can be seen that the peak-to-peak value of the signal source output is respectively At 10 mV, 20mV, 25mV, 50 mV, 100 mV, 1 V and 10 V, the broadband acoustic transducer average signal-to-noise ratio of the sound source signal received by the vector hydrophone is −13 dB, −7 dB, −5 dB, 1 dB, 7 dB, 27 dB and 47 dB.

The seven signal-to-noise ratio signals are processed separately using the histogram algorithm. The calculated azimuth estimation results change with time as shown in Figure 7. The figure also marks the peak-to-peak value of the signal output and the vector hydrophone in each time period. Receiver signal-to-noise ratio. It can be seen from Figure 7 that the estimated azimuth of the sound source target gradually stabilizes as the received signal-to-noise ratio increases and basically coincides with the true azimuth. Figure 8 and Figure 9 respectively show the azimuth estimation error and −3 dB azimuth spectrum width of the signal-to-noise ratio signals emitted by the seven sound sources by the histogram algorithm. The ratio increases and gradually decreases. The direction finding error increases when the sound source emits a 10 V peak-to-peak noise signal compared to 1V peak-to-peak. This is because the sound source emits a high sound source level signal.

The pool has incomplete noise reduction in the low frequency band and there is strong interface reflection; when the signal-to-noise ratio is −7 dB, the direction-finding error is about 8°, and the −3 dB azimuth spectrum width is about 23°; and when the signal-to-noise ratio is greater than At 1 dB, the direction-finding error and −3 dB azimuth spectrum width are less than 4◦ and 19◦, respectively. Figure 10 is the curve of target tracking mark with the intensity of the sound source emission signal calculated according to the target autonomous detection and tracking algorithm. It can be seen that when the signal-to-noise ratio is −7 dB, the histogram algorithm can achieve autonomous tracking of the sound source target.

2.3 Marine test analysis

Using data from the underwater acoustic buoy target detection performance verification test data carried out in the northern waters of the South China Sea in August 2019, the single-vector hydrophone histogram algorithm was used to analyze the detection performance of maritime targets. The depth of the test sea area is about 1500 m. During the test period, the weather conditions are good and the wind.

The speed is about level 2. The measurement results of the ship-borne abandonment thermosalt depth instrument show that the sound velocity profile is a uniform layer within a depth of 40 m, and the sound velocity main catastrophic layer is within a depth of 40 ∼ 200 m, and the sound channel axis is at 1000 m. Near the depth. During the test day from 12:33-14:02, a surface vessel with a length of 42 m, a width of 6 m, and a speed of 8.4 kn passed near the underwater acoustic buoy at a heading of 301°. During the period, the surface vessel and the underwater acoustics The distance of the buoy is about 2 km at the shortest time and 13.8 km at the farthest time. A comparison chart of the target azimuth estimation result calculated by the histogram algorithm and the real azimuth of the surface ship is given, and it can be seen that the histogram algorithm is in the entire 12:33-14:02 time.

Figure 13 and Figure 14 respectively show the histogram algorithm to the surface ship target direction finding error and −3 dB azimuth spectrum width change curve with time in the time period of 12:33-14:02. It can be seen that the direction finding error is the best It can reach within 5°, and the −3 dB azimuth spectrum width can reach about 10°near the close location point; in addition, due to the deviation of the underwater estimated position of the underwater acoustic buoy, the distance between the surface ship and the buoy platform is closer The error of direction finding at time increases. Figure 15 is the curve of the target tracking mark over time calculated by the target autonomous detection and tracking algorithm. It can be seen that the algorithm can achieve autonomous target tracking throughout the entire range for a surface vessel with a speed of 8.4 kn within a distance of 13.8 km.

3 Conclusion

Aiming at the engineering application requirements of single vector hydrophones on underwater unmanned platforms, this paper proposes an autonomous detection and tracking of underwater targets.Tracing method, and use simulation calculation, anechoic tank test and sea test analysis to summarize the histogram algorithm based on single vector hydrophone.Standard detection performance. The results of computer simulation and anechoic tank test data show that the histogram algorithm achieves the signal-to-noise ratio required for autonomous tracking.If it is greater than −7 dB, the direction finding error is about 8°, and the azimuth spectrum width of −3 dB is about 20°. The sea test data shows that the deep sea is good hydrological conditions, the histogram algorithm can achieve full target detection and tracking for a surface vessel with a speed of 8.4 kn within a distance of 13.8 km.The best direction-finding error can reach 5◦, and the −3 dB azimuth spectrum width can reach 10◦ near the close position.