A time domain waveform extraction method based on deconvolution beamforming
By using deconvolution beamforming technology in water acoustic array signal processing, the time domain waveform of the transmitted signal of the water acoustic target is extracted, and the problems of low resolution and insufficient extraction accuracy in the prior art are solved, and higher resolution is achieved and higher precision time domain waveform extraction effect.
Patent Information
- Application Number
- CN202211285445.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-10-20
- Publication Date
- 2025-02-21
- Estimated Expiration
- 2042-10-20
AI Technical Summary
Existing water acoustic array signal processing methods are difficult to accurately extract the time domain waveform of the transmitted signal of the water acoustic target in complex marine environments. Especially when multiple water acoustic targets with similar orientations exist, the resolution of conventional beamforming methods is low. , it is impossible to effectively distinguish and extract various signals.
The time domain waveform extraction method based on deconvolution beam formation is adopted. The receiving array receives the original time domain waveform and performs beamforming processing to obtain the frequency azimuth spatial spectrum output matrix and the frequency azimuth beam output matrix, and then performs deconvolution operation to extract The amplitude spectrum and phase spectrum of the signal are finally obtained through the inverse Fourier transform to obtain the time domain waveform of the transmitted signal.
It realizes higher resolution and higher accuracy time domain waveform extraction, which can accurately distinguish multiple water acoustic targets with similar orientations, and can extract the time domain waveforms of the transmitted signals of each target separately within the same beam coverage range. It has strong anti-interference ability.
Smart Images
Figure CN115616483B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of marine technology, and in particular to a time domain waveform extraction method based on deconvolution beamforming. Background Art
[0002] In underwater acoustic array signal processing, due to the complexity of the ocean environment, the signals of underwater acoustic targets are often covered by strong interference or background noise, which increases the difficulty of underwater acoustic target recognition. Therefore, how to accurately restore the time domain waveform of the transmitted signal of the underwater acoustic target has become an important research branch, which is conducive to improving the effectiveness and reliability of underwater acoustic target recognition and classification.
[0003] Various conventional beamforming methods are currently the main methods used to extract the time domain waveform of the transmitted signal of the hydroacoustic target. Conventional beamforming methods can obtain the azimuth, amplitude and phase of the signal. The time domain waveform of the signal can be extracted by inverse Fourier transforming the amplitude and phase information of the known signal. However, due to the limitation of Rayleigh limit, the resolution of conventional beamforming methods is often low, and it is impossible to distinguish the signals in the same beam. When the hydroacoustic target and the interferer exist at the same time and the azimuth is close, or when there are multiple hydroacoustic targets with close azimuths, it is easy to have the same beam, especially when there are multiple targets in the same main lobe. At this time, the signal of each hydroacoustic target will be interfered by the signal of the interferer or other hydroacoustic targets. Therefore, it is difficult to accurately extract the amplitude and phase information of each signal using conventional beamforming methods, and the error is large.
[0004] Although there are some high-resolution beamforming algorithms currently, these high-resolution beamforming algorithms can only obtain the direction of the signal or the direction and amplitude of the signal, but cannot obtain the phase information of the signal. Therefore, it is difficult to use them to extract the time domain waveform of the signal. Summary of the invention
[0005] In view of the above problems and technical requirements, the applicant proposes a time domain waveform extraction method based on deconvolution beamforming. The technical solution of this application is as follows:
[0006] A time domain waveform extraction method based on deconvolution beamforming, the method comprising:
[0007] Receiving the original time domain waveform respectively through each array element in the receiving array;
[0008] The received original time domain waveform is processed based on the beamforming method to obtain the first signal direction θ k , frequency-azimuth spatial spectrum output matrix P and frequency-azimuth beam output matrix Y, the frequency-azimuth spatial spectrum output matrix P includes spatial spectra P at different frequencies f and different beam pointing angles θ CBF(f,θ), the frequency azimuth beam output matrix Y includes the frequency domain signals Y(f,θ) at different frequencies f output by beams with different beam pointing angles θ;
[0009] The frequency azimuth spatial spectrum output matrix P and the frequency azimuth beam pattern matrix B are deconvolved to obtain the second signal azimuth θ d And the frequency azimuth spatial spectrum output matrix S, the frequency azimuth beam pattern matrix B includes the natural directivity function of the beams of different beam pointing angles θ of the receiving array at different frequencies f, and the frequency azimuth spatial spectrum output matrix S includes the spatial spectrum P at different frequencies f and different beam pointing angles θ dcv (f,θ);
[0010] Extract the second signal position θ from the frequency position spatial spectrum output matrix S d The corresponding amplitude spectrum P dcv (f,θ d ), extract the first signal azimuth θ from the frequency azimuth beam output matrix Y k The corresponding phase spectrum Y CBF (f,θ k ), based on the extracted amplitude spectrum P dcv (f,θ d ) and phase spectrum Y CBF (f,θ k ) extracts the time domain waveform of the transmitted signal.
[0011] A further technical solution is to perform a deconvolution operation on the frequency-azimuth spatial spectrum output matrix P and the frequency-azimuth beam pattern matrix B to obtain a plurality of second signal azimuths θ d , then based on the extracted amplitude spectrum P dcv (f,θ d ) and phase spectrum Y CBF (f,θ k ) extracts the time domain waveforms of multiple transmitted signals.
[0012] A further technical solution is that the time domain waveforms of the extracted multiple transmission signals are located within the same beam coverage of the receiving array.
[0013] A further technical solution is to extract the amplitude spectrum P dcv (f,θ d ) and phase spectrum Y CBF (f,θ k ) extracts the time domain waveform of the transmitted signal, including:
[0014] The amplitude spectrum P of the processing frequency band dcv (f,θ d ) and phase spectrum Y CBF (f,θk ) is converted from 0 Hz to the Fourier transform point number, and then an inverse Fourier transform is performed to obtain the time domain waveform of the transmitted signal.
[0015] A further technical solution is that a method for performing a deconvolution operation on a frequency-azimuth spatial spectrum output matrix P and a frequency-azimuth beam pattern matrix B includes:
[0016] The RL algorithm is used to perform deconvolution operations on the frequency-azimuth spatial spectrum output matrix P and the frequency-azimuth beam pattern matrix B.
[0017] A further technical solution is to obtain the first signal azimuth θ k The method of outputting the frequency-azimuth spatial spectrum matrix P includes:
[0018] Processing the received original time domain waveform to obtain the total covariance matrix R of different frequencies of the original time domain waveform;
[0019] The total covariance matrix R is used to perform frequency domain beamforming on each frequency f within the working frequency band bandwidth to obtain the first signal azimuth θ k And the frequency-azimuth spatial spectrum output matrix P.
[0020] A further technical solution is to obtain the spatial spectrum P at any frequency f and beam pointing angle θ in the frequency azimuth spatial spectrum output matrix P. CBF (f,θ) is:
[0021] P CBF (f,θ)=a H (f,θ)R(f)a(f,θ);
[0022] Among them, a(f,θ) is the steering vector corresponding to frequency f, a H (f,θ) represents the conjugate transpose of a(f,θ), and R(f) represents the signal of the total covariance matrix R at frequency f.
[0023] A further technical solution is that a method for obtaining a total covariance matrix R of different frequencies of the original time domain waveform includes:
[0024] Perform sliding averaging on the original time domain waveforms received by multiple array elements of the receiving array to obtain M segments of overlapping sub-time domain signals;
[0025] For each sub-time domain signal, perform Fourier transform on the sub-time domain signal to obtain the corresponding sub-frequency domain signal, and multiply the sub-frequency domain signal by the conjugate transpose of the sub-frequency domain signal to obtain the covariance matrix R of each processing frequency point of the sub-time domain signal m ;
[0026] The covariance matrix R of the M-segment time domain signal mBy adding the corresponding processing frequency points, the total covariance matrix R of different frequencies of the original time domain waveform is obtained.
[0027] A further technical solution is that a method for obtaining a frequency azimuth beam output matrix Y includes:
[0028] Perform Fourier transform on the received original time domain waveform to obtain the original frequency domain signal;
[0029] Based on the weighted beamforming method, the frequency-azimuth beam output matrix Y is obtained using the original frequency domain signal.
[0030] A further technical solution is that the frequency domain signal Y(f,θ) at any frequency f of the beam output with the beam pointing angle θ in the obtained frequency azimuth beam output matrix Y is:
[0031] Y(f,θ)=w H (f,θ)·X(f);
[0032] Among them, w H (f,θ)=[w1(f,θ),w2(f,θ),…,w N (f,θ)] T , w n (f,θ) is the weighting coefficient of the nth array element at frequency f that generates a beam with a beam pointing angle θ, n∈[1,N], N is the number of array elements included in the receiving array, and X(f) is the original frequency domain signal corresponding to frequency f.
[0033] The beneficial technical effects of this application are:
[0034] The present application discloses a time domain waveform extraction method based on deconvolution beamforming. The method utilizes deconvolution beamforming technology to estimate the amplitude spectrum of a signal, and combines the phase spectrum estimated by the beamforming method to accurately extract the time domain waveform of the signal. The present application provides a new time domain waveform extraction method, and compared with the traditional beamforming method, the method can achieve higher resolution and higher accuracy time domain waveform extraction.
[0035] Since this method can achieve higher resolution and higher precision time domain waveform extraction, when there are multiple targets with similar orientations, even if these multiple targets are within the coverage of the same beam, this method can also accurately extract the time domain waveform of the transmitted signal of each target separately, with high precision, wide applicability and high practical value.
[0036] Since this method has the robustness of deconvolution technology, when there is an element position error in the receiving array or the signal-to-noise ratio is low, this method can also accurately realize time domain waveform extraction and has strong anti-interference ability. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] Figure 1 It is a calculation flow chart of the time domain waveform extraction method in one embodiment of the present application.
[0038] Figure 2 It is a computational flow chart for time domain waveform extraction using conventional beamforming methods.
[0039] Figure 3 This is an example of using Figure 2 The spatial spectrum frequency orientation output matrix diagram obtained by performing time domain waveform extraction using the conventional beamforming method shown.
[0040] Figure 4 is with Figure 3 Using the same instance Figure 1 The method in the illustrated embodiment performs time domain waveform extraction to obtain a spatial spectrum frequency orientation output matrix diagram.
[0041] Figure 5 yes Figure 3 The example shown is a correlation diagram of the extracted time domain waveform and the simulated signal emitted by the target.
[0042] Figure 6 yes Figure 4 The example shown is a correlation diagram between one of the time domain waveforms extracted and the corresponding simulated signal emitted by the target. DETAILED DESCRIPTION
[0043] The specific implementation of the present application is further described below in conjunction with the accompanying drawings.
[0044] This application discloses a time domain waveform extraction method based on deconvolution beamforming, please refer to Figure 1 The calculation information flow diagram shown in FIG. 1 includes the following steps:
[0045] Step 120: Receive the original time domain waveform x(t) through each array element in the receiving array.
[0046] Step 140: Process the received original time domain waveform based on the beamforming method to obtain the first signal direction θ k , frequency-azimuth spatial spectrum output matrix P and frequency-azimuth beam output matrix Y.
[0047] (1) The frequency-azimuth spatial spectrum output matrix P includes the spatial spectrum P at different frequencies f and different beam pointing angles θ. CBF (f,θ). Methods for determination include:
[0048] The received original time domain waveform is processed to obtain a total covariance matrix R of different frequencies of the original time domain waveform. In one embodiment, a sliding average is performed on the original time domain waveform received by multiple array elements of the receiving array to obtain M overlapping sub-time domain signals. For each sub-time domain signal, a Fourier transform is performed on the sub-time domain signal to obtain a corresponding sub-frequency domain signal, and the sub-frequency domain signal is multiplied by the conjugate transpose of the sub-frequency domain signal to obtain a covariance matrix R of each processing frequency point of the sub-time domain signal. m . The covariance matrix R of the M-segment time domain signal m By adding the corresponding processing frequency points, the total covariance matrix R of different frequencies of the original time domain waveform is obtained.
[0049] Then, the total covariance matrix R is used to perform frequency domain beamforming on each frequency f within the working frequency band bandwidth to obtain the first signal direction θ k And the frequency-azimuth spatial spectrum output matrix P. The spatial spectrum P at any frequency f and beam pointing angle θ in the obtained frequency-azimuth spatial spectrum output matrix P CBF (f,θ) is:
[0050] P CBF (f,θ)=a H (f,θ)R(f)a(f,θ);
[0051] Among them, a(f,θ) is the steering vector corresponding to frequency f, a H (f,θ) represents the conjugate transpose of a(f,θ), and R(f) represents the signal of the total covariance matrix R at frequency f.
[0052] (2) The frequency azimuth beam output matrix Y includes frequency domain signals Y(f,θ) at different frequencies f of beam outputs with different beam pointing angles θ. The determination method includes:
[0053] The original time domain waveform received is Fourier transformed to obtain the original frequency domain signal, and then the original frequency domain signal is used based on the weighted beamforming method to obtain the frequency azimuth beam output matrix Y. The frequency domain signal Y(f,θ) at any frequency f of the beam output with the beam pointing angle θ in the obtained frequency azimuth beam output matrix Y is:
[0054] Y(f,θ)=w H (f,θ)·X(f);
[0055] Among them, w H (f,θ)=[w1(f,θ),w2(f,θ),…,w N (f,θ)] T , w n(f,θ) is the weighting coefficient of the nth array element at frequency f that generates a beam with a beam pointing angle θ, n∈[1,N], N is the number of array elements included in the receiving array, and X(f) is the original frequency domain signal corresponding to frequency f.
[0056] Step 160: Deconvolve the frequency-azimuth spatial spectrum output matrix P and the frequency-azimuth beam pattern matrix B to obtain the second signal azimuth θ d And the frequency-azimuth spatial spectrum output matrix S.
[0057] Among them, the frequency-azimuth beam pattern matrix B includes the natural directivity function B(f,θ) of the receiving array at different frequencies f, which is used to indicate that the signal is concentrated on the beam pointing angle θ. The natural directivity function B(f,θ) of the receiving array is also the beam pattern, also known as the point spread function of the receiving array, which can be obtained based on the parameters of the receiving array and will not be elaborated in this application.
[0058] Since the frequency-azimuth spatial spectrum output matrix P of conventional beamforming can be regarded as the convolution of the frequency-azimuth beam pattern matrix B and the signal target strength, the present application uses the frequency-azimuth spatial spectrum output matrix P and the frequency-azimuth beam pattern matrix B to apply a deconvolution algorithm to obtain the second signal azimuth θ d And the frequency-azimuth spatial spectrum output matrix S. The frequency-azimuth spatial spectrum output matrix S includes the spatial spectrum P at different frequencies f and different beam pointing angles θ dcv (f,θ).
[0059] In the present application, the beam pattern of the receiving array is independent of the angle, that is, the beam pattern pointing to a certain angle can be regarded as a shift of the natural directivity function, so the point spread function of the system satisfies the shift-invariant condition. Therefore, in one embodiment, the RL algorithm can be used to perform deconvolution operations on the frequency-azimuth spatial spectrum output matrix P and the frequency-azimuth beam pattern matrix B.
[0060] Step 180: extract the second signal direction θ from the frequency-direction spatial spectrum output matrix S d The corresponding amplitude spectrum P dcv (f,θ d ), extract the first signal azimuth θ from the frequency azimuth beam output matrix Y k The corresponding phase spectrum Y CBF (f,θ k ), based on the extracted amplitude spectrum P dcv (f,θ d ) and phase spectrum Y CBF (f,θ k ) extracts the time domain waveform s(t) of the transmitted signal.
[0061] Since the amplitude spectrum P dcv (f,θd ) and phase spectrum Y CBF (f,θ k ) are all in the processing frequency band range, so they cannot be directly used for P dcv (f,θ d ) and Y CBF (f,θ k ) to perform inverse Fourier transform, but firstly transform the amplitude spectrum P of the processing frequency band dcv (f,θ d ) and phase spectrum Y CBF (f,θ k ) is converted from 0 Hz to the Fourier transform point number, and then an inverse Fourier transform is performed to obtain the time domain waveform of the transmitted signal.
[0062] Please compare Figure 2 The conventional beamforming extraction method shown in FIG. 1 is used to extract the first signal azimuth θ from the frequency azimuth spatial spectrum output matrix P. k The corresponding amplitude spectrum P CBF (f,θ k ), then the amplitude spectrum P CBF (f,θ k ) and phase spectrum Y CBF (f,θ k ) is inverse Fourier transformed to obtain the time domain waveform s′(t) of the transmitted signal. dcv (f,θ d ) is obtained through deconvolution operation, which can achieve higher resolution main peak extraction, thereby achieving higher resolution and higher accuracy time domain waveform extraction.
[0063] In the above step 160, one or more second signal directions θ can be obtained by performing a deconvolution operation on the frequency-azimuth spatial spectrum output matrix P and the frequency-azimuth beam pattern matrix B. d , when multiple second signal positions θ are obtained d This step is based on the extracted amplitude spectrum P dcv (f,θ d ) and phase spectrum Y CBF (f,θ k ) extracts the time domain waveforms of multiple transmission signals. In practical applications, a typical case is that the time domain waveforms of the multiple transmission signals extracted are located within the same beam coverage of the receiving array.
[0064] Therefore, when there are multiple transmission signals within the coverage of a beam according to the conventional beamforming method, more commonly, within the coverage of the main lobe, only one time domain waveform can be extracted according to the conventional beamforming, while the present application can identify multiple second signal azimuths θ by performing a deconvolution operation in step 160.d , and then extract the time domain waveforms of multiple transmitted signals to achieve higher accuracy and higher resolution time domain waveform extraction.
[0065] In one example, it is assumed that the receiving array is a 16-element array with an array element spacing of 1.1538m (650Hz half-wave spacing), the operating frequency band bandwidth ranges from 500 to 800Hz, the FFT frequency resolution is 1Hz, and the integration time is 1s. It is assumed that the azimuth of target 1 is 73.21° (cosine value is 0.2889), the azimuth of target 2 is 78.14° (cosine value is 0.2056), the signal-to-noise ratio of target 1 is 20dB, and the signal-to-noise ratio of target 2 is 20dB.
[0066] according to Figure 2 The conventional beamforming method shown in the figure extracts the time domain waveform. Figure 3 As shown, according to Figure 1 The results of extracting the time domain waveform using the method of the present application are as follows: Figure 4 As shown, Figure 3 and 4 The horizontal axis is the direction and the vertical axis is the frequency. Figure 3 and Figure 4 It can be seen that Figure 3 The conventional beamforming method can only extract a time domain waveform and cannot distinguish target 1 and target 2. However, the method of the present application can be used as follows: Figure 4 Two time domain waveforms are extracted and target 1 and target 2 are successfully distinguished with higher resolution.
[0067] A time domain waveform extracted by conventional beamforming method is cross-correlated with the simulation signal of target 1. The result is as follows: Figure 5 As shown, the correlation between the two is about 0.65, which is poor. This is because the time domain waveform extracted by the conventional beamforming method is the result of the signals of target 1 and target 2 being mixed together, so the error of the time domain waveform is large.
[0068] The correlation between a time domain waveform extracted by the method of the present application and the simulation signal of target 1 is about 0.93. Figure 6 As shown, it is basically completely correlated with the simulation signal of target 1.
[0069] The above is only a preferred embodiment of the present application, and the present application is not limited to the above embodiments. It is understood that other improvements and changes directly derived or associated by those skilled in the art without departing from the spirit and concept of the present application should be considered to be included in the protection scope of the present application.
Claims
1. A time domain waveform extraction method based on deconvolution beamforming, characterized in that: The method comprises: Receiving the original time domain waveform respectively through each array element in the receiving array; The received original time domain waveform is processed based on a beamforming method to obtain a first signal direction , frequency-azimuth spatial spectrum output matrix And the frequency-azimuth beam output matrix , the frequency-azimuth spatial spectrum output matrix Including different frequencies and different beam pointing angles The spatial spectrum , the frequency-azimuth beam output matrix Including different beam pointing angles The beam output has different frequencies The frequency domain signal ; The frequency-azimuth spatial spectrum output matrix and frequency-azimuth beam pattern matrix Perform deconvolution operation to obtain multiple second signal positions And the frequency-azimuth spatial spectrum output matrix , the frequency-azimuth beam pattern matrix Different beam pointing angles of the receiving array are included The beams at different frequencies Under the natural directivity function, the frequency-azimuth spatial spectrum output matrix Including different frequencies and different beam pointing angles The spatial spectrum ; The frequency-azimuth spatial spectrum output matrix Extracting a plurality of the second signal positions The corresponding amplitude spectrum , from the frequency-azimuth beam output matrix Extract the first signal direction The corresponding phase spectrum , based on the extracted amplitude spectrum and phase spectrum The time domain waveforms of the multiple transmission signals are extracted, and the time domain waveforms of the multiple transmission signals are located within the same beam coverage of the receiving array.
2. The method according to claim 1, characterized in that The amplitude spectrum obtained based on the extraction and phase spectrum The time domain waveform of the transmitted signal is extracted, including: The amplitude spectrum of the frequency band will be processed and phase spectrum After converting from 0 Hz to the Fourier transform points, an inverse Fourier transform is performed to obtain the time domain waveform of the transmitted signal.
3. The method according to claim 1, characterized in that The frequency-azimuth spatial spectrum output matrix and frequency-azimuth beam pattern matrix Methods for performing deconvolution operations include: The RL algorithm is used to output the frequency azimuth spatial spectrum matrix and frequency-azimuth beam pattern matrix Perform deconvolution operation.
4. The method according to claim 1, characterized in that: Get the first signal position And the frequency-azimuth spatial spectrum output matrix The methods include: The received original time domain waveform is processed to obtain the total covariance matrix of different frequencies of the original time domain waveform ; With the total covariance matrix Each frequency within the working frequency band bandwidth Perform frequency domain beamforming respectively to obtain the first signal direction And the frequency-azimuth spatial spectrum output matrix .
5. The method according to claim 4, characterized in that The frequency-azimuth spatial spectrum output matrix obtained Any frequency and beam pointing angle The spatial spectrum for: ； in, is the frequency The corresponding steering vector, express The conjugate transpose of Denotes the total covariance matrix In frequency The signal at.
6. The method according to claim 4, characterized in that Get the total covariance matrix of different frequencies of the original time domain waveform The methods include: Perform sliding averaging on the original time domain waveform received by multiple array elements of the receiving array to obtain The segments have overlapping sub-time domain signals; For each sub-time domain signal, perform Fourier transform on the sub-time domain signal to obtain the corresponding sub-frequency domain signal, and multiply the sub-frequency domain signal by the conjugate transpose of the sub-frequency domain signal to obtain the covariance matrix of each processing frequency point of the sub-time domain signal ; Will The covariance matrix of the time domain signal According to the corresponding addition of the processing frequency points, the total covariance matrix of different frequencies of the original time domain waveform is obtained. .
7. The method according to claim 1, characterized in that Get the frequency azimuth beam output matrix The methods include: Performing Fourier transform on the received original time domain waveform to obtain an original frequency domain signal; The frequency azimuth beam output matrix is obtained by using the original frequency domain signal based on a weighted beam forming method. .
8. The method according to claim 7, characterized in that The frequency azimuth beam output matrix obtained is Medium beam pointing angle Any frequency of beam output The frequency domain signal for: ； in, , is the beam pointing angle The beam Array element at frequency The weighting coefficient under , is the number of elements contained in the receiving array, is the frequency The corresponding original frequency domain signal.