A passive target detection distance evaluation method based on difference characteristics under strong interference

Abstract

The present application relates to the technical field of sonar signal processing, and particularly relates to a passive target detection distance evaluation method under strong interference based on difference characteristics. The method is used for evaluating the influence of strong interference on targets. In a traditional sonar equation, an interference level, an interference suppression gain and a feature detection threshold are introduced to quantize the influence of strong interference on passive sonar prediction distance. Firstly, a propagation loss curve is calculated according to hydrological conditions. Then, the interference level, the interference suppression gain, the interference propagation loss, the feature detection index obtained by a multi-dimensional feature joint detection method and the feature detection threshold are calculated according to the set parameters of the strong interference and the target. The traditional passive sonar equation is optimized to obtain an optimized quality factor. In combination with the propagation loss curve, a corrected passive sonar prediction distance is obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of sonar signal processing, and particularly relates to a passive target detection distance evaluation method under strong interference based on difference characteristics. BACKGROUND

[0002] In modern marine tactical research, the research on sonar is paid more and more attention. The action distance of sonar is the most concerned index. If the action distance of the sonar of the own side can be known in advance in a specific sea area, the action of the sonar of the own side can be greatly increased. However, in a strong interference environment, the traditional sonar equation has the following problems: 1) only the uniform isotropic noise is considered, but the existence of strong interference changes the background noise of the sonar operation from the uniform isotropic noise to the strong directional background noise; 2) only the energy detector under the constant false alarm rate is considered, and the processing gain brought by the feature detector in the strong interference environment is not considered. The two aspects lead to the problem that the performance of the action distance prediction of the traditional sonar equation is insufficient in the strong interference environment, and even invalid. SUMMARY

[0003] The purpose of the present application is to provide a passive target detection distance evaluation method under strong interference based on difference characteristics. The method proposes a passive sonar action distance evaluation method under strong interference for the passive target detection distance evaluation under strong interference. The method covers the traditional elements such as the sound source level, the propagation loss and the new elements such as the interference level, the interference suppression gain and the feature detection threshold, and actually improves the sonar equipment performance evaluation ability in the strong interference environment.

[0004] To achieve the above purpose, the present application provides the following technical scheme: a passive target detection distance evaluation method under strong interference based on difference characteristics, comprising the following steps:

[0005] S1, inputting the strong interference spectrum level SL of a specified frequency under a strong interference environment, a target spectrum level SL and a relative distance between the strong interference and a receiving array; j

[0006] S2, calculating the ocean environmental noise level NL under the specified frequency, wherein the NL decays by 6 dB / octave with the frequency, and the initial value at 1 kHz is 64 dB;

[0007] S3, obtaining the temperature, depth and salinity parameters of the sea area by a temperature-depth salinity instrument, calculating the sound velocity gradient, and generating a propagation loss curve of the specified frequency by using a sound propagation model in combination with the seabed medium parameters;

[0008] S4, calculating the strong interference propagation loss TL according to the propagation loss curve and the strong interference distance; j

[0009] ​​S5, calculate the receiving directivity index DI, when the receiving array has no distortion or model mismatch, DI=10lgM, where M is the number of array elements;

[0010] S6, based on the multi-dimensional difference feature joint detection method, calculate the feature detection threshold DT c , specifically including:

[0011] (a) extract the high and low frequency energy proportion, narrowband line spectrum and spatial energy distribution difference features of the target and background noise;

[0012] (b) construct the probability density distribution of the joint detection statistic, obtain the mean μ s and variance σ s of the target and the mean μ n and variance σ n of the background noise respectively;

[0013] (c) generate the ROC curve by the detection probability and false alarm probability under different signal-to-noise ratios, determine the feature detection index , and calculate

[0014]

[0015] Where, B is the processing bandwidth, and T is the integration time;

[0016] S7, calculate the interference suppression gain IG(θ) of the wideband detection algorithm, satisfying

[0017]

[0018] Where, and are the beam energy of the interference energy at the target direction before and after suppression respectively;

[0019] S8, obtain the energy of the strong interference and the target direction in the background energy distribution graph when the strong interference exists and ;

[0020] S9, optimize the passive sonar equation, generate the high-quality factor

[0021]

[0022] Where, BL j (θ) is the beam domain interference and noise level, and its expression is

[0023]

[0024] Combine the propagation loss curve to calculate the target detection prediction distance.

[0025] Preferably, the sound propagation model in S3 is the Kraken model, and the propagation loss curve is generated by jointly calculating the sound velocity gradient and the seabed medium parameters.

[0026] Preferably, the multidimensional differential feature joint detection method in S6 includes feature fusion and optimized threshold selection under the maximum a posteriori probability criterion.

[0027] Preferably, the calculation of the interference suppression gain in S7 employs a super-resolution algorithm. and The energy difference between conventional beamforming and super-resolution beamforming is determined respectively.

[0028] Preferably, the calculation of the quality factor FOM in S9 further includes the ambient noise level NL, the receiver directivity index DI, and the interference propagation loss TL. j The coupling correction term.

[0029] Preferably, the detection index d of the multidimensional differential feature joint detection method j The false alarm probability threshold and detection probability threshold are correlated with the ROC curve, and when the false alarm probability ≤ 0.1 and the detection probability ≥ 0.9, d j The value range is from 1.2 to 1.6.

[0030] Preferably, the calculation of the marine environmental noise level NL includes a towed platform self-noise correction term with a correction amount of 3dB.

[0031] Preferably, the calculation of the receiver directivity index DI includes an additional attenuation term caused by array distortion or model mismatch, and the attenuation amount is determined by actual measurement or simulation.

[0032] Compared with the prior art, the beneficial effects of the present invention are:

[0033] This method specifically assesses the impact of strong interference on targets and analyzes the effects of signal processing methods with interference suppression capabilities on interference and target detection. It introduces interference level, interference suppression gain, and feature detection threshold into the traditional sonar equations to characterize the impact of interference on detection and the suppression effect of signal processing methods on interference. This optimizes the traditional sonar equations to meet the performance prediction requirements for range under strong interference. Specifically, the interference level characterizes the new noise background formed when strong interference propagates through the underwater acoustic channel and reaches the detection platform, combined with the ambient noise at the detection platform. The interference suppression gain characterizes the spatial suppression capability of the interference suppression method against strong interference. The feature detection threshold characterizes the detection performance of a feature detection method based on the combined multidimensional differences between strong interference and target features. Using these elements, the impact of strong interference on the traditional sonar equations can be effectively quantified, thereby effectively estimating the range of passive sonar targets under strong interference. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 Flow chart of passive sonar detection range evaluation method under strong interference

[0035] Figure 2 Schematic diagram of test trial route design for verifying the performance improvement ability of passive sonar equation prediction under strong interference

[0036] Figure 3 In (a), the sound speed gradient is obtained by using the hydrological environment measurement of the temperature-depth salinity instrument; in (b), the propagation loss curve at 240 Hz is calculated by using the Kraken model;

[0037] Figure 4 In (a), the target signal amplitude is obtained; in (b), the target signal source level is obtained; in (c), the detection result of the conventional method is obtained; in (d), the detection result of the feature detection method is obtained;

[0038] Figure 5 Propagation loss curve of target signal

[0039] Figure 6 In (a), the mean μ s and variance σ s of the joint detection target are calculated; in (b), the performance curve of multi-dimensional feature joint detection is obtained; n n

[0040] Figure 7 Simulation result of interference suppression gain

[0041] Figure 8 In the background energy distribution graph when strong interference exists, the energy at the strong interference and target azimuth

[0042] Figure 9 Prediction distance results of the traditional sonar equation and the passive sonar detection range evaluation method under strong interference. DETAILED DESCRIPTION

[0043] The specific embodiments of the present application are described in detail below with reference to the accompanying drawings, so that those skilled in the art can more clearly understand how to practice the present application. Although the present application is described in conjunction with its preferred specific embodiments, these embodiments are merely illustrative, and do not limit the scope of the present application.

[0044] (I) Implementation process

[0045] The present application provides a passive target detection range evaluation method under strong interference based on differential features, which combines Figure 1 The method comprises the following steps: ​​

[0046] S1. Under strong interference environment, the strong interference spectrum level SL at a specified frequency is input. j The target spectral level SL and the distance between the strong interference and the receiving array.

[0047] S2. Calculate the marine ambient noise level NL at a specified frequency: The marine ambient noise level can be approximated as NL=64dB@1kHz, with attenuation every 6dB / octave.

[0048] S3. Calculate the sound velocity gradient by measuring parameters such as temperature, depth, and salinity of the sea area using a temperature, depth, and salinity meter, and then use the Kraken model to calculate the propagation loss curve at a specified frequency using the sound velocity gradient and parameters such as the selected seabed medium.

[0049] S4. Calculate the strong interference propagation loss TL using the propagation loss curve and the distance between the strong interference source and the receiving array. j .

[0050] S5. Calculate the receiver directivity index. : When there is no array distortion / model mismatch, ,in Indicates the number of array elements.

[0051] S6, Calculate the feature detection threshold DT c First, the probability density distribution of the optimal test metric for the joint detection method based on multi-dimensional difference features such as high- and low-frequency energy ratio characteristics, narrowband line spectrum characteristics, and spatial energy distribution difference characteristics under the maximum a posteriori probability criterion is calculated, with a mean of μ and a variance of σ. Then, the target mean μ of the joint detection is obtained. s and variance σ s The mean μ of the background noise, respectively n and variance σ n Furthermore, the joint detection performance of multidimensional features under different signal-to-noise ratios is calculated. Given different thresholds T, the feature detection index is calculated:

[0052]

[0053] The detection probability and false alarm probability curves under different signal-to-noise ratios can be obtained, i.e., the multi-dimensional feature joint detection performance curve (ROC curve); and then the feature detection threshold can be obtained.

[0054]

[0055] Where d j It is related to the detection probability and false alarm probability, where B is the processing bandwidth and T is the integration time.

[0056] S7. Calculate the interference suppression gain of the broadband detection algorithm used. :

[0057]

[0058] wherein and are the beam energy of the interference energy at the target bearing before interference suppression and the beam energy remaining at the target bearing after interference suppression, respectively. For the case of using super resolution algorithm for wideband detection, can be approximated as the difference between the interference energy and the energy of the interference at the target bearing when using conventional beamforming, can be approximated as the difference between the interference energy and the energy of the interference at the target bearing when using the super resolution algorithm.

[0059] S8, calculate the energy at the strong interference and the target bearing in the background energy distribution map when the strong interference exists and .

[0060] S9, by adding the interference level, interference suppression gain, feature detection threshold and other parameters obtained through the above steps to the traditional sonar equation, obtain the updated quality factor of the passive sonar equation, and obtain the passive target detection prediction distance based on the method combined with the propagation loss curve: the quality factor of the optimized passive sonar equation is

[0061]

[0062] wherein is the beam domain interference and noise level, and its expression is

[0063] .

[0064] (II) Verify the feasibility of the method based on "(I) implementation process"

[0065] Figures 2 to 9 As shown in the figure, the verification of the method needs the cooperation of the ship, the target ship and the interference ship. The ship sails from point A to point A' at a speed of 3kn, and the passive detection mode is turned on during the period; the interference ship drifts at point B without power, and the interference equipment emits suppressive strong interference signals; the target ship drifts at point C without power, and emits simulated target radiation noise signals of different source levels according to the instructions; the ship uses an equidistant towed array with an array spacing of 1.5m;

[0066] According to the propagation loss curve and the strong interference distance, the strong interference propagation loss TL is calculated j : attached Figure 4 is the signal source level change and the detection results of the two algorithms in this test, wherein Figure 4 (a) in (a) is the target signal amplitude, Figure 4 (b) in (b) is the target signal source level, Figure 4 (c) in (c) is the detection result of the conventional method,Figure 4 (d) represents the detection result of the feature detection method. The high-resolution spatial spectrum estimation algorithm used in this example is the feature detection algorithm. It can be seen that time 1 is the moment when the target is just suppressed by strong interference under the conventional beamforming detection method. At this time, the signal source level is reduced by 10dB, and the feature detection method can still detect the target. Time 2 is the moment when the target can be detected by the feature detection method after the strong interference and the target cross. At this time, the signal source level is increased by 10dB. Time 3 is the moment when the feature detection method can no longer detect the target. At this time, the strong interference azimuth is 83.7°, and the target azimuth is 86.2°. Before time 1, the target spectral level at 240Hz is set to SL=125.1dB@240Hz. From time 1 to time 2, the target spectral level is set to SL=115.1dB@240Hz. The marine environmental noise level is NL=79.8dB@240Hz (including 3dB of self-noise from the towing vessel).

[0067] Receive Directivity Index (6dB loss due to array distortion / model mismatch); the interference spectral level remained consistently at SL. j =130dB@240Hz; the processing bandwidth was set to 140Hz, the center frequency to 240Hz, and the integration time to 8s; based on the GPS information of the test vessel, at time 1, the target vessel was X2=8.80km away from the towed vessel, and at time 3, the target vessel was Y2=8.78km away from the towed vessel, while the interfering vessel was 3.2km away from the towed vessel; from the attached... Figure 5 It can be seen that the propagation loss TL of the strong interference at time 3 is... j =58.06dB@240Hz;

[0068] Based on a multidimensional differential feature joint detection method, the feature detection threshold DT is calculated. c Appendix Figure 6 This is a schematic diagram of the feature detection index, where (a) is the mean μ of the jointly detected targets obtained by calculation. s and variance σ s The mean μ of the background noise, respectively n and variance σ n The schematic diagram shows that by specifying different thresholds T, the detection probability and false alarm probability curves under different signal-to-noise ratios can be obtained.

[0069] For the problem of predicting detection performance under strong interference, it can be based on Figure 6 The ROC curve shown in (b) yields a range of selectable detection indices under specified maximum false alarm probability and minimum detection probability conditions; when the false alarm probability is 0.1 and the detection probability is 0.9, the corresponding feature detection index d j =1.4; For the traditional sonar equation, the detection index d=12 under the same false alarm probability and detection probability can be obtained by looking up the table; thus, the feature detection threshold is obtained:

[0070] ,

[0071] Detection threshold of traditional sonar equation:

[0072] ;

[0073] Figure 7 CBF represents the conventional beamforming algorithm, TZ represents the feature detection algorithm, and IG(θ) represents the interference suppression gain of the wideband detection algorithm, i.e.

[0074] ;

[0075] The energy at the strong interference and target bearing in the background energy distribution map when the strong interference exists is calculated respectively and ;

[0076] The figure of merit of the optimized passive sonar equation is obtained by calculation:

[0077]

[0078] The figure of merit of the traditional passive sonar equation:

[0079]

[0080] Combined with the propagation loss curve, the passive target detection prediction distance Y1=12.2km under strong interference, and the traditional passive sonar equation prediction distance X1=34.0km, so the passive target detection action distance prediction error under strong interference is reduced to

[0081] .

[0082] The experimental data analysis results show that: in the case of strong interference distance 3.2km, spectral level 130dB@240Hz, and bearing difference between strong interference and target 2.5°, the action distance prediction error of this method is reduced by 86.43% compared with the traditional sonar equation prediction error, which can support the passive sonar target action distance prediction under strong interference environment.

Claims

1. A method for evaluating the detection range of passive targets under strong interference based on differential characteristics, characterized in that, Includes the following steps: S1, Strong interference spectral level SL at a specified frequency under strong interference environment. j The target spectral level (SL) and the relative distance between strong interference and the receiving array; S2. Calculate the marine ambient noise level NL at a specified frequency, where NL decreases with frequency at 6dB / octave, and has an initial value of 64dB at 1kHz. S3. Obtain the temperature, depth and salinity parameters of the sea area using a temperature, depth and salinity meter, calculate the sound velocity gradient, and generate a propagation loss curve at a specified frequency using a sound propagation model in combination with seabed medium parameters. S4. Calculate the strong interference propagation loss TL based on the propagation loss curve and the strong interference distance. j ; S5. Calculate the receiver directivity index DI. When the receiver array is undistorted or the model is mismatched, DI = 10lgM, where M is the number of array elements. S6. Based on a multi-dimensional differential feature joint detection method, calculate the feature detection threshold DT. c ; S7. Calculate the interference suppression gain IG(θ) of the broadband detection algorithm, satisfying... in, and These represent the beam energy at the target azimuth before and after suppression of interference energy; S8. Obtain the energy distribution map of the strong interference and the target's location in the background energy distribution map when strong interference exists. and ; S9. Optimize the passive sonar equations to generate high-quality factors. Among them BL j (θ) represents the beam domain interference and noise level, and its expression is: The target detection prediction distance is calculated by combining the propagation loss curve.

2. The passive target detection range evaluation method based on differential characteristics under strong interference according to claim 1, characterized in that, The sound propagation model in S3 is the Kraken model, and the propagation loss curve is generated by jointly calculating the sound velocity gradient and the parameters of the seabed medium.

3. The passive target detection range evaluation method based on differential characteristics under strong interference according to claim 1, characterized in that, The multidimensional differential feature joint detection method in S6 includes feature fusion and optimized threshold selection under the maximum a posteriori probability criterion.

4. The passive target detection range evaluation method based on differential characteristics under strong interference according to claim 1, characterized in that, The calculation of the interference suppression gain in S7 uses a super-resolution algorithm. and The energy difference between conventional beamforming and super-resolution beamforming is determined respectively.

5. The method for evaluating the detection range of passive targets under strong interference based on differential characteristics according to claim 1, characterized in that, The calculation of the factor of excellence (FOM) in S9 further includes the ambient noise level (NL), receiver directivity index (DI), and interference propagation loss (TL). j The coupling correction term.

6. The passive target detection range evaluation method based on differential characteristics under strong interference according to claim 1, characterized in that, In S6, the feature detection threshold DT is calculated. c Specifically, it includes: (a) Extract the high and low frequency energy ratios, narrowband line spectra, and spatial energy distribution differences between the target and background noise; (b) Construct the probability density distribution of the joint detection statistic to obtain the target mean μ. s and variance σ s The mean μ of the background noise, respectively n and variance σ n ; (c) Generate ROC curves by using the detection probability and false alarm probability under different signal-to-noise ratios to determine the feature detection index. and calculate Where B is the processing bandwidth and T is the integration time.

7. The passive target detection range evaluation method based on differential characteristics under strong interference according to claim 6, characterized in that, The detection index d of the multidimensional differential feature joint detection method j The false alarm probability threshold and detection probability threshold are correlated with the ROC curve, and when the false alarm probability ≤ 0.1 and the detection probability ≥ 0.9, d j The value range is from 1.2 to 1.

6.

8. The method for evaluating the detection range of passive targets under strong interference based on differential characteristics according to claim 1, characterized in that, The calculation of the marine environmental noise level NL includes a towed platform self-noise correction term, with a correction amount of 3dB.

9. The method for evaluating the detection range of passive targets under strong interference based on differential characteristics according to claim 1, characterized in that, The calculation of the receiver directivity index DI includes an additional attenuation term caused by array distortion or model mismatch, and the attenuation amount is determined by actual measurement or simulation.

Images