An anti-interference direction finding and automatic ambiguity resolution method for underwater narrowband signals

Abstract

An anti-interference direction finding and automatic ambiguity resolution method for underwater narrowband signals, which is used to realize underwater target direction finding and phase ambiguity resolution of a uniform linear array based on narrowband sound source signals. The power spectral density analysis is performed on the array data to determine the peak frequency and regard it as the frequency of the sound source signal; the frequency-domain broadband beamforming method based on the fast Fourier transform is used and applied to the direction finding process of the narrowband sound source signal. In the frequency-domain broadband beamforming method, a frequency band range centered on the peak frequency is set to effectively suppress the interference outside the sound source signal frequency band and reduce the calculation amount at the same time. Aiming at the problem of false spectral peaks in the beamforming power spectrum when the array element spacing is greater than half of the carrier wavelength, the normalized zero-delay cross-correlation of different carrier frequency power spectral functions is performed to realize the automatic extraction of real spectral peaks. The present invention can realize high-precision underwater target direction finding estimation and automatic extraction of real spectral peaks, and the recognition rate of real spectral peaks can reach 100%.

Description

Technical Field

[0001] The present invention relates to a beamforming direction finding and automatic deblurring method for underwater narrowband signals, and particularly to an anti-interference direction finding and automatic deblurring method for underwater narrowband signals. Background Art

[0002] With the continuous development of array sensors, beamforming technology has been widely applied in fields such as communication, radar, sonar, and seismic exploration. This technology uses a spatially distributed sensor array to collect sound field data and performs linear combination processing on the received array data to obtain a scalar beam output. As an important research direction in underwater array signal processing, the main purpose of beamforming is to complete the estimation of the direction of arrival (DOA) of signals.

[0003] As the signal environment becomes increasingly complex and the signal frequency distribution range continues to widen, the limitations of traditional narrowband beamforming technology have gradually emerged. Processing broadband signals will cause problems such as beam pattern distortion and reduced effective resolution. Broadband signal processing can extract far more detection and positioning information than narrowband processing and obtain a higher processing gain. The DOA estimation of broadband signals has become a development trend.

[0004] Although the sound source emits narrowband signals, when there is noise, the signal-to-noise ratio of the array signal is low. Using narrowband beamforming technology cannot effectively resist noise interference, resulting in low beamforming direction finding accuracy or high sidelobe energy in the power spectrum, causing low resolution of the direction finding result. In addition, when the carrier half-wavelength of the sound source signal is less than the element spacing, there is a multi-valued phase ambiguity in the beamforming result. For the same incident direction, the position of the true spectral peak is always unchanged under different carrier frequencies, while the position of the false spectral peak will change with the change of the carrier frequency. Based on this principle, for the multi-valued ambiguity problem caused by the element spacing exceeding the sound source carrier half-wavelength, beamforming can be performed separately on multiple carrier frequency data and then the true spectral peak can be extracted. In existing methods, mostly the power spectra of beamforming under different carrier frequencies are overlapped and placed, and the true spectral peak is extracted by manual identification. This not only wastes time and labor costs but also may increase the probability of misidentification. Summary of the Invention

[0005] The purpose of the present invention is to overcome the deficiencies of the prior art and provide an anti-interference direction finding and automatic deblurring method for underwater narrowband signals, which can effectively suppress the noise interference in the array received data during underwater narrowband carrier emission and the automatic extraction of the true spectral peak in multi-valued phase ambiguity.

[0006] The technical solution of the present invention is as follows:

[0007] An anti-interference direction finding and automatic ambiguity resolution method for underwater narrowband signals, comprising the following steps:

[0008] 1) Receiving far-field narrowband incident source signals to form array data;

[0009] 2) Determining the peak frequency according to the array data and regarding it as the frequency of the narrowband sound source signal;

[0010] 3) Segmenting the array data and performing FFT transformation to obtain the array data in the frequency domain;

[0011] 4) Setting a frequency band range centered on the peak frequency, and performing direction finding processing of the narrowband sound source signal within this frequency band range to obtain the power spectral function corresponding to the frequency of the sound source signal;

[0012] 5) Extracting the true incoming wave direction of the sound source according to the obtained power spectral function.

[0013] An M-element uniform linear array is used to receive underwater far-field narrowband incident source signals to form array data, the array pitch is d, and M≥1.

[0014] In step 2), by performing power spectral density analysis on the array data, the peak frequency is extracted from the power spectral density curve as the carrier frequency fs of the narrowband incident source.

[0015] In step 3), the array data is divided into K sub-segments in the time domain, and FFT transformation is performed on the sampled data of each segment. Specifically, the following is done: a frequency band range [fs - f1, fs + f2] centered on the peak frequency fs is set, f1 + f2 is the width of this frequency band range, 0≤f1≤f s , f2≥0. Within this frequency band range, the broadband signal is divided into J sub-bands in the frequency domain, and for each effective narrowband frequency f j , 1≤j≤J, K frequency domain sampled data are obtained.

[0016] In the set frequency band range [fs - f1, fs + f2], the direction vector a(f j ,θ j ) and the estimated value of the frequency domain autocorrelation matrix are calculated for each effective narrowband frequency f i ) The frequency domain broadband beamforming method based on the fast Fourier transform is used for the direction finding processing of the narrowband sound source signal to obtain the power spectral function P(f j ,θ).

[0017] The calculation process of the direction vector a(f j ,θ j ,θ i ) for each effective narrowband frequency f where θ iis the incident direction of the i-th incident source, τ i represents the time delay between the signal arriving at the i-th array element and the signal arriving at the reference array element.

[0018] For each effective narrowband frequency f j the estimated value of the frequency-domain autocorrelation matrix is calculated as follows: where x(f j ,t i ) represents the array data with frequency f i in the t-th sub-segment, and x j (f H (f j ,t i ) represents the transpose of x(f j ,t i ).

[0019] The calculation process of the power spectral function P(f j ,θ) corresponding to the sound source signal frequency is as follows: where θ is the spatial scanning angle; a H (f j ,θ) represents the transpose of a(f j ,θ i ).

[0020] In step 5), based on the obtained power spectral function and the relationship between the half-wavelength corresponding to the sound source carrier frequency and the array element spacing, the direct output or automatic extraction of the true incoming wave direction of the underwater sound source is completed.

[0021] The process of step 5) to complete the direct output or automatic extraction of the true incoming wave direction of the underwater sound source based on the obtained power spectral function and the relationship between the half-wavelength corresponding to the sound source carrier frequency and the array element spacing includes:

[0022] Judge the size relationship between the half-wavelength corresponding to the sound source carrier frequency and the array element spacing d; if In the two-dimensional power spectral function P(f j ,θ) formed by beam output for each effective narrowband frequency f j , screen the maximum energy value. The horizontal and vertical coordinates of the position where the maximum value is located represent the sound signal frequency and the incoming wave angle; if It is judged that there is a problem of false spectral peaks in the beamforming power spectrum P(f j ,θ), then select another sound source carrier frequency different from fs, repeat steps 1)-4), obtain the power spectral function P'(f j ',θ) corresponding to another sound source carrier frequency fs', perform normalized zero-delay cross-correlation on the beamforming power spectral functions corresponding to the two carrier frequencies, screen the maximum cross-correlation value, realize the automatic extraction of the true spectral peak, and obtain the corresponding DOA estimation result.

[0023] The advantages of the present invention compared with the prior art are as follows:

[0024] (1) Based on the power spectral density analysis and broadband frequency-domain beamforming method, the present invention realizes the orientation of underwater narrowband signal targets, which can effectively suppress the influence of noise in the array signal on orientation and improve the orientation accuracy.

[0025] (2) The anti-interference orientation method for underwater narrowband signals based on power spectral density analysis and broadband frequency-domain beamforming provided by the present invention extracts the carrier frequency of the emitted sound source signal through power spectral density analysis, and performs beamforming calculation by restricting the frequency band range centered on this carrier frequency in the frequency domain, reducing the calculation amount and improving the calculation efficiency.

[0026] (3) The automatic deblurring method provided by the present invention does not require manual identification and extraction of real spectral peaks, saving time and labor costs, and improving the real spectral peak recognition rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is the system block diagram of the anti-interference orientation and automatic deblurring method for underwater narrowband signals based on power spectral density analysis and broadband frequency-domain beamforming of the present invention;

[0028] Figure 2 is the waveform data and power spectral density curve of element 1;

[0029] Figure 3 is the power spectral function curve corresponding to the carrier frequency 800 Hz of the sound source signal;

[0030] Figure 4 is the power spectral function curve corresponding to the carrier frequency 1400 Hz of the sound source signal;

[0031] Figure 5 is the power spectral function curve corresponding to the carrier frequency 1800 Hz of the sound source signal;

[0032] Figure 6 is the normalized zero-delay cross-correlation curve of power spectral functions with different carrier frequencies. DETAILED DESCRIPTION OF THE INVENTION

[0033] An anti-interference orientation and automatic deblurring method for underwater narrowband signals of the present invention includes the following steps:

[0034] 1) Receive far-field narrowband incident source signals to form array data;

[0035] 2) Determine the peak frequency according to the array data and regard it as the frequency of the narrowband sound source signal;

[0036] 3) Segment the array data and perform FFT transformation to obtain the array data in the frequency domain;

[0037] 4) Set a frequency band range centered on the peak frequency, and perform directional processing on the narrowband sound source signal within this frequency band range to obtain the power spectral function corresponding to the frequency of the sound source signal;

[0038] 5) Extract the true incident direction of the sound source based on the obtained power spectral function.

[0039] Use a uniform linear array with M elements to receive the underwater far-field narrowband incident source signal to form array data. The array spacing is d, and M≥1.

[0040] In step 2), by performing power spectral density analysis on the array data, extract the peak frequency in the power spectral density curve as the carrier frequency fs of the narrowband incident source.

[0041] In step 3), divide the array data into K sub-segments in the time domain, and perform FFT transformation on each segment of the sampled data. Specifically, set a frequency band range [fs - f1, fs + f2] centered on the peak frequency fs. The width of this frequency band range is f1 + f2, 0≤f1≤f s , f2≥0. Within this frequency band range, divide the broadband signal into J sub-bands in the frequency domain. For each effective narrowband frequency f j , 1≤j≤J, K frequency domain sampled data are obtained.

[0042] In the set frequency band range [fs - f1, fs + f2], calculate the direction vector a(f j ,θ j ) and the estimated value of the frequency domain autocorrelation matrix i for each effective narrowband frequency f Use the frequency domain broadband beamforming method based on the fast Fourier transform for the directional processing of the narrowband sound source signal to obtain the power spectral function P(f j ,θ) corresponding to the frequency of the sound source signal.

[0043] The calculation process of the direction vector a(f j ,θ j ) for each effective narrowband frequency f i is as follows: where θ i is the incident direction of the i-th incident source, and τ i represents the time delay between the signal arriving at the i-th array element and the signal arriving at the reference array element.

[0044] The calculation process of the estimated value j of the frequency domain autocorrelation matrix for each effective narrowband frequency f is as follows: where, x(f j ,t i ) represents the t iThe sub - segment frequency is f j of the array data, x H (f j , t i ) represents the transpose of x(f j , t i ).

[0045] The calculation process of the power spectral function P(f j , θ) corresponding to the sound source signal frequency is as follows: where θ is the spatial scanning angle; a H (f j , θ) represents the transpose of a(f j , θ i ).

[0046] In the said step 5), according to the obtained power spectral function and the relationship between the half - wavelength corresponding to the sound source carrier frequency and the array element spacing, the direct output or automatic extraction of the true incoming wave direction of the underwater sound source is completed.

[0047] The process of the said step 5) to complete the direct output or automatic extraction of the true incoming wave direction of the underwater sound source according to the obtained power spectral function and the relationship between the half - wavelength corresponding to the sound source carrier frequency and the array element spacing includes:

[0048] Judge the size relationship between the half - wavelength corresponding to the sound source carrier frequency and the array element spacing d; if At each effective narrow - band frequency f j in the two - dimensional power spectral function P(f j , θ) formed by beam output, screen the maximum energy value, and the horizontal and vertical coordinates of the position where the maximum value is located represent the sound signal frequency and the incoming wave angle; if it is judged that there is a problem of false spectral peaks in the beam - formed power spectrum P(f j , θ), then select another sound source carrier frequency different from fs, repeat steps 1) - 4), obtain the power spectral function P'(f j ', θ) corresponding to another sound source carrier frequency fs', perform normalized zero - delay cross - correlation on the beam - formed power spectral functions corresponding to the two carrier frequencies, screen the maximum value of the cross - correlation, realize the automatic extraction of the true spectral peak, and obtain the corresponding DOA estimation result.

[0049] The present invention will be described in more detail below with reference to the accompanying drawings and actual experimental data.

[0050] Figure 1 is a flowchart of an underwater narrow - band signal anti - interference direction - finding and automatic de - ambiguity method in an embodiment of the present invention. Referring to Figure 1 , the underwater narrow - band signal anti - interference direction - finding and automatic de - ambiguity method based on power spectral density analysis and broadband frequency - domain beamforming provided in this embodiment may specifically include the following steps:

[0051] First, perform power spectral density analysis on the array data to determine the peak frequency and regard it as the carrier frequency of the sound source signal;

[0052] Segment the array received data in the time domain and perform FFT transformation;

[0053] Set the frequency band range centered on the peak frequency. In this frequency band range, use the frequency-domain wideband beamforming method based on the fast Fourier transform for the directional processing of narrowband sound source signals to obtain the power spectral function corresponding to the frequency of the sound source signal;

[0054] Judge the size relationship between the half-wavelength corresponding to the sound source carrier frequency and the element spacing;

[0055] When the carrier half-wavelength is greater than or equal to the element spacing, search for the spectral peak in the beamforming power spectral curve, that is, the true incoming wave direction of the sound source and directly output it;

[0056] When the carrier half-wavelength is less than the element spacing, there is a problem of false spectral peaks in the beamforming power spectrum. Perform normalized zero-delay cross-correlation on the power spectral functions of different carrier frequencies to automatically extract the true spectral peak.

[0057] The 16-element uniform linear array receives the far-field narrowband incident source signal to form array data, and the array spacing is d = 0.75m. Taking element 1 as an example, perform power spectral density analysis on the data it receives, as Figure 2 shown. The frequency corresponding to the maximum amplitude in its power spectral density curve is approximately 800Hz, and this frequency is extracted as the carrier frequency fs of the narrowband incident source.

[0058] Divide the array data received by the 16 elements into 195 sub-segments in the time domain. Each sub-segment has 1024 sampling points, and perform FFT transformation on each segment of the sampled data; Set the frequency band range [650Hz, 1000Hz] centered on the peak frequency fs = 800Hz. In this frequency band range, divide the broadband signal into 19 sub-bands in the frequency domain, and for each effective narrowband frequency f j (1 ≤ j ≤ J), 195 frequency-domain sampling data can be obtained.

[0059] In the set frequency band range [650Hz, 1000Hz], for each effective narrowband frequency f j (1 ≤ j ≤ J), calculate the direction vector corresponding to the frequency:

[0060]

[0061] where θ is the incident direction of the sound source, A(f j , θ) = [a(f j , θ), a(f j, θ),..., a(f j , θ)] constitutes an M×1 dimensional array direction matrix;

[0062] Calculate the frequency-domain autocorrelation matrix estimate based on the array received data obtained from 195 frequency-domain samplings. The formula is as follows:

[0063]

[0064] The power spectral function corresponding to the frequency is obtained as

[0065]

[0066] where θ is the spatial scanning angle.

[0067] Use the above frequency-domain wideband beamforming method based on the fast Fourier transform for the directional processing of narrowband sound source signals, and obtain the power spectral function corresponding to the carrier frequency of the sound source signal, as Figure 3 shown.

[0068] According to the relationship between the half-wavelength corresponding to the sound source carrier frequency and the element spacing, directly output or automatically extract the true incident direction of the sound source: The carrier half-wavelength is The element spacing is d = 0.75m, satisfying Therefore, the spectral peak in the beamforming power spectral curve can be directly searched, that is, 25.3°, and this is used as the true incident direction of the sound source. According to the sound source direction recorded by GPS is 26.9°, and the directional error is 1.6°. Therefore, the accuracy of the sound source directional result obtained based on this method is relatively high.

[0069] Figure 4 、 Figure 5 are the power spectral function results obtained by using the frequency-domain wideband beamforming method for the sound source signals with carrier frequencies of 1400Hz and 1800Hz respectively. The half-wavelengths corresponding to the two carrier frequencies are 0.53m and 0.41m respectively. At this time, The problem of false spectral peaks appears in the beamforming power spectrum. There are two peaks in the power spectral curve. When the carrier frequency is 1400Hz, the DOA angles corresponding to the power spectral peaks are -47° and 42°. When the carrier frequency is 1800Hz, the DOA angles corresponding to the power spectral peaks are -47° and 21° respectively, resulting in multi-valued phase ambiguity. Based on the frequency-domain wideband beamforming processing of the data of these two carrier frequencies, as Figure 6 shown, perform normalized zero-delay cross-correlation on the power spectral functions of these two carrier frequencies to extract the true spectral peak:

[0070]

[0071] By screening the maximum value of the cross-correlation, the automatic extraction of the true spectral peak is realized, and the corresponding DOA estimation result is obtained. The DOA at the maximum value of the cross-correlation curve can be obtained as -47°. According to the GPS position information, the incident azimuth of the active sound source is calculated to be -48°. The error between the automatically extracted true spectral peak value and the incident direction of the actual sound source signal is 1°, and the accuracy is relatively high.

Claims

1. An anti-interference direction finding and automatic ambiguity resolution method for underwater narrowband signals, characterized in that Including: Receiving far-field narrowband incident source signals to form array data; Determining the peak frequency based on the array data and regarding it as the frequency of the narrowband sound source signal; Segmenting the array data and performing FFT transformation to obtain the array data in the frequency domain; Setting a frequency band range centered on the peak frequency, performing directional processing of the narrowband sound source signal within this frequency band range, and obtaining the power spectral function corresponding to the frequency of the sound source signal; Completing the extraction of the true incident direction of the sound source based on the obtained power spectral function; Using a uniform linear array with M elements to receive underwater far-field narrowband incident source signals to form array data, the array pitch is d, and M≥1; The specific process of determining the peak frequency and regarding it as the frequency of the narrowband sound source signal includes: extracting the peak frequency in the power spectral density curve as the carrier frequency fs of the narrowband incident source through power spectral density analysis of the array data; The step of segmenting the array data and performing FFT transformation to obtain the array data in the frequency domain specifically includes: dividing the array data into K sub-segments in the time domain, and performing FFT transformation on the sampled data of each segment. Specifically as follows: setting the frequency band range [fs - f1, fs + f2] centered on the peak frequency fs, where f1 + f2 is the width of the frequency band range, 0 ≤ f1 ≤ f s , f2 ≥ 0. Within this frequency band range, dividing the wideband signal into J sub-bands in the frequency domain, and for each effective narrowband frequency f j , 1 ≤ j ≤ J, K frequency domain sampled data are obtained; In the set frequency band range [fs - f1, fs + f2], calculate each effective narrowband frequency f j of the direction vector a(f j , θ i ) and the estimated value of the frequency domain autocorrelation matrix Use the frequency domain wideband beamforming method based on the fast Fourier transform for the directional processing of narrowband sound source signals to obtain the power spectral function P(f j , θ); The specific process of completing the extraction of the true incident direction of the sound source based on the obtained power spectral function includes: directly outputting or automatically extracting the true incident direction of the underwater sound source according to the obtained power spectral function and the relationship between the half-wavelength corresponding to the sound source carrier frequency and the array element pitch; The process of directly outputting or automatically extracting the true incident direction of the underwater sound source according to the obtained power spectral function and the relationship between the half-wavelength corresponding to the sound source carrier frequency and the array element pitch includes: Determine the sound source carrier frequency corresponding to half wavelength The relationship between the size of the array element spacing d; if At each effective narrowband frequency f j The two-dimensional power spectrum function P(f j ,θ) to select the maximum energy value, the horizontal and vertical coordinates of the maximum value position represent the frequency and angle of the sound signal; if It is judged as the beamforming power spectrum P(f j ,θ), then select another sound source carrier frequency different from fs to obtain the power spectrum function P'(f j ',θ), perform normalized zero-delay cross-correlation on the beamforming power spectrum functions corresponding to the two carrier frequencies, select the maximum value of the cross-correlation, realize the automatic extraction of the real spectrum peak, and obtain the corresponding DOA estimation result.

2. An underwater narrowband signal anti-interference direction finding and automatic ambiguity resolution method according to claim 1, characterized in that: Each valid narrowband frequency f j The calculation process of the direction vector a(f j , θ i ) is as follows: where θ i is the incident direction of the i-th incident source, and τ i represents the time delay between the signal arriving at the i-th array element and the signal arriving at the reference array element.

3. An underwater narrowband signal anti-interference direction finding and automatic ambiguity resolution method according to claim 1, characterized in that: The estimated value of the frequency-domain autocorrelation matrix for each effective narrowband frequency f j is calculated as follows: The calculation process is: where x(f j , ti) represents the array data of the t i -th sub-segment with frequency f j , and x H (f j , t i ) represents the transpose of x(f j , t i ).

4. An anti-interference direction finding and automatic deblurring method for underwater narrowband signals according to claim 1, characterized in that: The calculation process of the power spectrum function P(f j , θ) corresponding to the sound source signal frequency is as follows: where θ is the spatial scanning angle; a H (f j , θ) represents the transpose of a(f j , θ i ).

Images