A broadband sound source DOA estimation method based on shallow water channel multipath characteristics

Abstract

The application belongs to the field of underwater acoustic signal processing and analysis, and discloses a wideband sound source DOA estimation method based on the multipath characteristics of a shallow sea channel. The method first combines a wideband pulse signal with a frequency domain sound pressure wideband, and then performs Fourier transform on the combined signal to perform time-frequency analysis. Then, the signal is subjected to Warping transformation to separate and extract each order signal of the normal mode wave, and the time-domain signal is obtained. The obtained time-domain signal is then passed through a vector hydrophone to obtain a complex sound intensity vector corresponding to the signal in the frequency domain. Finally, the obtained complex sound intensity vector is subjected to DOA estimation to obtain the average value of the direction angle. Compared with the traditional method, the DOA estimation method effectively suppresses the interference caused by the multipath characteristics and improves the accuracy of DOA estimation.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of underwater acoustic signal processing and analysis, and particularly relates to a wideband sound source DOA estimation method based on the multi-path characteristics of a shallow sea channel. BACKGROUND

[0002] The received signal of a wideband pulse sound source after propagation in a shallow sea waveguide is a superposition of multi-order normal mode signals, each of which carries a large amount of ocean environment information and is widely used in DOA estimation of underwater sound sources and inversion of environmental parameters. However, in actual application, various uncertain factors exist in the shallow sea environment, and the multi-path characteristics and noise are also important environmental factors that cannot be ignored. Moreover, due to the influence of the frequency dispersion characteristics of the shallow sea, different orders of normal mode signals have different characteristics and contain different ocean environment information. Therefore, when the DOA of a shallow sea underwater acoustic target is estimated, the accuracy of the DOA estimation is often affected by the above-mentioned environmental parameters. SUMMARY

[0003] In view of the problems caused by the multi-path characteristics of the shallow sea in the prior art, the application provides a wideband sound source DOA estimation method based on the multi-path characteristics of a shallow sea channel, which improves the accuracy of DOA estimation and reduces the interference caused by the multi-path characteristics.

[0004] In order to achieve the above-mentioned purpose, the application is implemented by the following technical scheme:

[0005] The application is a wideband sound source DOA estimation method based on the multi-path characteristics of a shallow sea channel, which comprises the following steps:

[0006] Step a: a wideband pulse sound source h(t) is emitted in a shallow sea waveguide, the seabed condition of the shallow sea waveguide is a liquid semi-infinite seabed, and after a sound propagation distance r0, a time-domain received signal p(t0, r0) is obtained at a receiving point;

[0007] Step b: the received time-domain signal p(t0, r0) is subjected to short-time Fourier transform for time-frequency analysis processing, and then the normal mode signals with unknown accurate mode orders that already exist in the received signal are separated and extracted by warping transformation and narrow-band filtering in the frequency domain;

[0008] Step c: the data of each order of normal mode signal separated and extracted by warping transformation are subjected to spectral peak search and angle estimation by a complex sound intensity DOA estimation method through a two-dimensional vector hydrophone model in three channels of sound pressure P and vibration velocity X, Y;

[0009] Step d: the DOA estimations of each order of normal mode signal are respectively denoted as θ E1 、θ E2 ...θ Enθn is the azimuth angle of the nth order normal mode, and θ is the azimuth angle of the target. En The weighted sum is denoted as p(θ), and the azimuth angle is finally obtained.

[0010] Further, in step a, the frequency bandwidth of the wideband pulse sound source h(t) is 100-300 Hz, and the sound source propagation environment is a shallow sea waveguide, and the specific environmental conditions are that the waveguide characteristics do not change in the horizontal direction, the waveguide is horizontally stratified, and has a liquid semi-infinite high sound speed seabed, the sound source position is unknown, the receiving is a single sound pressure hydrophone receiving, and the receiving point is only one, the receiving position is located in seawater, and the average sound speed in seawater is The propagation distance is r0, and the sea depth is H.

[0011] Further, in step b, the Fourier transform formula and the warping transform formula in the time domain are respectively: The narrowband filtering is filtered by a zero-phase shift filter, the normal mode is an arbitrary class normal mode on the sea surface, and the order of the normal mode in the received signal is arbitrary, wherein f cn is the cutoff frequency of the nth order normal mode, r0 is the distance of the sound source propagation, c is the average sound speed in the waveguide, A n (t) is the instantaneous amplitude of the nth order normal mode.

[0012] Further, in step c, the formula for performing spectral peak search by using the complex sound intensity DOA estimation method is as follows:

[0013]

[0014] In the above formula, p s (t) is the target signal, θ is the target water azimuth angle, n p (t) is the noise component of the sound pressure, v xn (t) and v yn (t) are the isotropic noise components of the seismic velocity, and are independent of p s (t).

[0015] Therefore, the average sound intensity formulas in the x direction and the y direction are respectively:

[0016]

[0017] The Fourier transforms of the average sound intensities in the x direction and the y direction in the time domain in formula (3) are respectively performed, then the results are respectively multiplied by the conjugate of the velocity component, and finally the DOA estimation formula of the complex sound intensity is obtained as:

[0018]

[0019] In the formula, P(f) is the result of the Fourier transform of the sound pressure, are the results of the Fourier transform of the vibration velocity in the x and y directions, respectively, and p s (f) is the Fourier transform of the signal.

[0020] Therefore, according to (4), the DOA estimation angle calculation formula in the frequency domain is:

[0021]

[0022] Further, in step d, the azimuth angle calculation formula obtained by using the first-order separated normal wave is:

[0023]

[0024] Therefore, the azimuth angle calculation formula of the nth-order separated normal wave is:

[0025]

[0026] In summary, the DOA estimation angle calculation formula of the nth-order normal wave is:

[0027]

[0028] where θ E1 , θ En are the angle values obtained by DOA estimation of the first-order and nth-order separated normal waves, respectively, I x1 (f), I y1 (f), I xn (f), I yn (f) are the complex sound intensity of the first-order and nth-order normal waves in the x and y directions, respectively, n = 1, 2,..., N, and n is the order of the normal wave.

[0029] The present application has the following advantages:

[0030] The present application requires less prior knowledge of the marine environment, effectively reducing the error caused by parameter correlation;

[0031] The present application has good results in DOA estimation under high and low signal-to-noise ratios.

[0032] The present application uses the DOA estimation of the separated and extracted normal waves after warping transformation and the weighted summation calculation method, which can effectively suppress the interference angle in different directions and improve the resolution of DOA estimation. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 The flowchart of the DOA estimation method in the embodiment of the present application.

[0034] Figure 2 An application example in marine environment of the present application.

[0035] Figure 3 A received time-domain signal and a spectrum diagram in an embodiment of the present application.

[0036] Figure 4 A time-frequency analysis result of a received time-domain signal in an embodiment of the present application.

[0037] Figure 5 A DOA estimation result under different signal-to-noise ratios in an embodiment of the present application.

[0038] Figure 6 A DOA estimation result of suppressing different direction interference angles in an embodiment of the present application. DETAILED DESCRIPTION

[0039] Embodiments of the present application will be described below with reference to drawings. Many practical details will be described in the following description for the purpose of clear illustration. However, it should be understood that these practical details should not be used to limit the present application. That is, these practical details are not necessary in some embodiments of the present application.

[0040] As shown in Figure 1 the present application is a wideband sound source DOA estimation method based on the multi-path characteristics of a shallow sea channel. The method first synthesizes a wideband pulse signal with a frequency-domain sound pressure wideband, performs Fourier transform on the synthesized received signal p(t0, r0) through the channel to perform time-frequency analysis, then separates the normal mode signals that already exist in the received signal but whose accurate modal orders are unknown, extracts each order normal mode signal of the P channel, the X channel and the Y channel respectively, obtains the signal under the time-domain condition, then obtains the complex sound intensity vector corresponding to the signal under the frequency domain by passing the obtained time-domain signal through a vector hydrophone respectively, and finally performs DOA estimation on the obtained complex sound intensity vector respectively, and obtains the direction angle by taking the average value.

[0041] Specifically, the wideband sound source DOA estimation method based on the multi-path characteristics of a shallow sea channel of the present application includes the following steps:

[0042] Step 1: a wideband pulse sound source h(t) is transmitted in a shallow sea waveguide, the frequency bandwidth is 100-300 Hz, and a time-domain received signal p(t0, r0) is obtained at the receiving point after the sound propagation distance r0;

[0043] The marine environment of the sound signal propagation of the wideband pulse sound source h(t) is a horizontally stratified shallow sea waveguide with a liquid semi-infinite seabed, and the specific environmental conditions are: the waveguide characteristics do not change in the horizontal direction, the sound source position is unknown, the receiving is a single sound pressure hydrophone, there is only one receiving point, the average sound speed in seawater is The propagation distance is r0, and the sea depth is H.

[0044] Step 2: The time-domain received signal p(t0, r0) obtained in step 1 is subjected to short-time Fourier transform, time-frequency analysis processing, and then the normal mode signals already existing in the received signal but unknown in the accurate mode order are separated and extracted through warping transformation and frequency-domain narrowband filtering.

[0045] The Fourier transform formula in the time domain is as follows:

[0046]

[0047] Wherein, f(t) is a time function, F(w) is a frequency density, j is an imaginary unit, and w is an angular frequency.

[0048] The warping transformation formula is as follows:

[0049]

[0050] Wherein: f cn is the cutoff frequency of the nth-order normal mode, r0 is the distance of the sound source propagation, c is the average sound speed in the waveguide, A n (t) is the instantaneous amplitude of the nth-order normal mode, N represents the normal mode order, e is an index, and j is an imaginary number.

[0051] The narrowband filtering is filtering by a zero-phase shift filter, the normal mode is an arbitrary normal mode on the sea surface, and the order of the normal mode in the received signal is arbitrary.

[0052] Step 3: The normal mode data obtained through warping transformation separation and extraction are subjected to spectrum peak search and angle estimation through the complex sound intensity DOA estimation method in the sound pressure P and the vibration speed X, vibration speed Y three channels through the two-dimensional vector hydrophone signal model.

[0053] The spectrum peak search formula is as follows:

[0054] p(t) = p s (t) + n p (t)

[0055] v x (t) = p s (t)cosθ + v xn (t)

[0056] v y (t) = p s (t)sinθ + v yn (t)

[0057] In the formula, ps (t) is the target signal, θ is the target azimuth angle, n p (t) is the noise component of sound pressure, v xn (t) and v yn (t) are the x-direction and y-direction isotropic noise components of vibration velocity respectively, and are independent of p s (t), p(t), v x (t), v y (t) are the sound pressure and vibration velocity signals picked up by the vector hydrophone respectively.

[0058] The average sound intensity of the normal mode data in the x-direction and y-direction respectively is:

[0059]

[0060]

[0061] In the above formula, p(t) is the sound pressure signal, θ is the incident angle, v x (t), v y (t) are the vibration velocities in the x-direction and y-direction respectively, p s (t) is the target signal.

[0062] The Fourier transform of the average sound intensity in the x-direction and y-direction respectively in the time domain is performed, and then the results are respectively multiplied by the conjugate of the vibration velocity component, and finally the DOA estimation formula of the complex sound intensity is obtained:

[0063]

[0064]

[0065] In the above formula, P(f) is the result of the Fourier transform of the sound pressure, are the results of the Fourier transform of the vibration velocities in the x-direction and y-direction respectively, p s (f) is the Fourier transform of the signal.

[0066] Step 4: The DOA estimation of each order normal mode obtained in step 3 is respectively denoted as θ E1 , θ E2 ... θ En , and then the θ En estimated by each order normal mode estimation is weighted and summed, denoted as p(θ), and finally the azimuth angle is obtained.

[0067] The angle estimation calculation formula is:

[0068]

[0069] where, I x (f), Ix (f) are the complex sound intensity in x and y direction in frequency domain respectively.

[0070] The azimuth angle calculation formula of the first-order separated normal mode is:

[0071]

[0072] The azimuth angle calculation formula of the nth-order separated normal mode is:

[0073]

[0074] The θ En The DOA estimation angle calculation formula of the nth-order normal mode is obtained by weighted summation:

[0075]

[0076] Where, θ E1 , θ En are the angle values obtained by DOA estimation of the first-order and nth-order separated normal modes respectively, I x1 (f), I y1 (f), I xn (f), I yn (f) are the complex sound intensity in x and y direction of the first-order and nth-order normal modes respectively, n = 1, 2... N, and n is the order of the normal mode.

[0077] The specific process of the DOA estimation method of the application is as follows by taking the linear frequency modulation signal and the seventh-order normal mode as an example:

[0078] It is assumed that the receiving point is located on the seabed, a broadband pulse sound source is emitted by the sound source transmitter, the sea depth is H = 100, the propagation distance r0 = 10 km, the average sound speed in seawater is 1500 m / s, and the sound source positions are respectively the node (sound pressure is 0) z s = 28 m and the seabed.

[0079] As shown in FIG. 4, the time-domain receiving signal and the frequency spectrum of the linear frequency modulation broadband pulse sound source signal after propagation through the shallow sea waveguide are shown. Figure 3

[0080] As shown in FIG. 5, the time-frequency analysis result of the receiving time-domain signal under the sound source position is shown. Figure 4

[0081] As shown in FIG. 6, the time-frequency analysis result of the receiving time-domain signal under the sound source position is shown.​​Figure 5 Fig. 1 shows the DOA estimation graph under different signal-to-noise ratios, wherein (a), (b), (c) and (d) are DOA estimation graphs under different signal-to-noise ratios, respectively. Figure 5 It can be clearly obtained from the figures that the interference signal source is at about 10°, and the sidelobe is always generated when the original algorithm is used for DOA estimation under different signal-to-noise ratios, and the interference cannot be effectively suppressed, while the method of the present application can effectively suppress the interference angle; meanwhile, in terms of estimation accuracy and resolution, the accuracy of the method of the present application is higher than that of the original DOA estimation method.

[0082] As shown in Fig. 2, (a), (b), (c) and (d) are DOA estimation graphs under different interference direction angles, respectively. Figure 6 It can be clearly obtained from the figures that under the same signal-to-noise ratio and different interference direction angles, the original DOA estimation method cannot effectively suppress the interference when performing DOA estimation, resulting in the sidelobe in the experiment, which seriously affects the accuracy of DOA estimation. Figure 6 Meanwhile, under the same conditions, the method of the present application can effectively suppress the interference angle and improve the resolution of DOA estimation, so as to make the estimation effect more accurate.

[0083] In combination with the above embodiments, it is further evidenced that the method of the present application can effectively suppress the interference and improve the accuracy of the bearing estimation when the multipath characteristics of the shallow water channel are used for the bearing estimation of the wideband pulse signal.

[0084] The above only describes the preferred embodiments of the present application, and the protection scope of the present application is not limited to the above embodiments, and any equivalent modification or change made by the ordinary skilled in the art according to the disclosed content of the present application shall be included in the protection scope recorded in the claims.

Claims

1. A method for wideband sound source DOA estimation based on the multipath characteristics of a shallow water channel, characterized in that: The wideband sound source DOA estimation method comprises the following steps: Step 1: a wideband pulse sound source h(t) is emitted in a shallow sea waveguide, and after a sound propagation distance r0, a time-domain received signal p(t0, r0) is obtained at a receiving point; Step 2: the time-domain received signal p(t0, r0) obtained in step 1 is subjected to short-time Fourier transform, time-frequency analysis processing, and then the normal mode signals already existing in the received signal but unknown in the accurate mode order are separated and extracted through warping transformation and frequency-domain narrowband filtering; Step 3: the normal mode data of each order separated and extracted through warping transformation are subjected to angle estimation, specifically: the normal mode data of each order separated and extracted through warping transformation are respectively subjected to spectral peak search through the complex sound intensity DOA estimation method through a two-dimensional vector hydrophone model under sound pressure P and vibration velocity X, vibration velocity Y three channels, angle estimation; Step 4: Denote the DOA estimates obtained from each normal mode in Step 3 as θ. E1 θ E2 ...θ En Then, θ is estimated from each normal mode. En We perform a weighted summation, denoted as p(θ), to obtain the azimuth angle. 2.The wideband acoustic source DOA estimation method based on the channel multi-path characteristics of shallow sea according to claim 1, characterized in that: In the step 1, the frequency bandwidth of the wideband pulse sound source h(t) is 100-300 Hz, and the sound source propagation environment is a shallow sea waveguide.

3. The method of claim 2, wherein the method is characterized by: The sound source propagation environment is specifically that the waveguide feature does not change in the horizontal direction, the waveguide is horizontally layered, and has a liquid semi-infinite high sound speed seabed, the sound source position is unknown, and the receiving is single sound pressure hydrophone receiving, the receiving point is only one, the receiving position is located in seawater, and the average sound speed in seawater is The propagation distance is r0, and the sea depth is H.

4. The wideband acoustic source DOA estimation method based on the channel multi-path characteristics of shallow sea according to claim 3, characterized in that: In the step 2, the frequency-domain narrowband filtering adopts zero-phase shift filter for filtering, the normal mode type is arbitrary, and the order of the normal mode in the received signal is arbitrary.

5. The method of claim 1, wherein the method is characterized by: Step 4 is specifically: θn=arctan (Im (Rn) / Re (Rn) ) (1) En The n-order normal wave DOA estimation angle calculation formula is obtained by weighted summation: where θ E1 , θ En are the angle values obtained by DOA estimation of the separated 1st order and n th order normal modes, respectively.

Images