Frequency domain weighted phase shift reduction line spectrum enhancement method
Through the frequency domain weighted phase shift regulation line spectrum enhancement method, the problem of signal-to-noise ratio reduction and insufficient gain of vertical array signal processing in the detection of water acoustic target radiation noise is solved, and effective line spectrum enhancement of water acoustic target radiation noise is achieved and improve detection capabilities.
Patent Information
- Application Number
- CN202410977582.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-12
AI Technical Summary
The existing technology has problems of signal-to-noise ratio reduction and insufficient gain of vertical array signal processing in the detection of water acoustic target radiation noise. Especially after the widespread application of ship stealth technology and vibration reduction and noise reduction technology, the radiated noise of water acoustic targets is carried out. It is gradually submerged in the noise of the marine environment, making it difficult to effectively detect and identify.
The frequency domain weighted phase shift regulation linear spectrum enhancement method is adopted. By grouping the received signals of each hydrophone vertical array, calculating the average power spectrum and phase difference, performing low-pass filtering fit and nonlinear mapping, and calculating the frequency domain complex weighting. The spectrum sequence of the received signal is weighted and standardized to improve the signal-to-noise ratio of the line spectrum.
The signal-to-noise ratio gain of the linear spectrum is significantly improved, the linear spectrum enhancement is effectively realized, and the spatial processing gain of the vertical array is improved, which enhances the detection ability of water acoustic target radiation noise, which is suitable for the detection of weak target radiation noise.
Smart Images

Figure CN118938202A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of underwater acoustic target radiation noise line spectrum feature extraction and recognition, in particular to a frequency domain weighted phase shift reduction line spectrum enhancement method. Background Art
[0002] Hydroacoustic target detection is a method of detecting targets by using the information received by sonar. It is an important support for underwater information acquisition and underwater information confrontation, and is one of the important research directions in the field of hydroacoustics. According to the way of obtaining information, hydroacoustic target detection can be divided into active detection and passive detection. The passive detection process does not require the emission of sound wave signals. The existence of the target is determined by the hydroacoustic target radiation noise signal received by the passive sonar. The hydroacoustic target radiation noise signal is the noise radiated by hydroacoustic targets such as ships through the ship's hull during navigation. It contains components such as broadband continuous spectrum, narrowband discrete line spectrum and modulation spectrum. Among them, the line spectrum component is closely related to the working status of the ship's mechanical equipment and carries important characteristic information of the hydroacoustic target. The low-frequency discrete line spectrum has a long propagation distance, strong stability, and high signal-to-noise ratio. It is an important detection object in the passive detection process.
[0003] Nowadays, the line spectrum detection technology for the radiation noise of underwater acoustic targets still faces many difficulties. With the widespread application of ship stealth technology and vibration reduction and noise reduction technology, the radiation noise spectrum level of underwater acoustic targets is decreasing day by day, and its radiation noise is gradually submerged in the ocean environment noise, making it difficult to detect and identify, making it more difficult to extract the characteristics of underwater acoustic targets. In addition, vertical arrays have become a common array form for underwater unmanned platforms such as buoys and submersibles due to their convenience in deployment and recovery, and are widely used in underwater target detection. However, due to the influence of the multipath effect of the sound velocity profile and the reflection of the seabed and sea surface, the spatial correlation of the target signal between the array elements and the array signal processing gain are lower than those of the horizontal line array, and the detection ability of weak target radiation noise is not strong. Therefore, how to improve the spatial processing gain of the noise signal through the information of multiple array elements of the vertical array, so as to effectively use the vertical array for line spectrum detection of target radiation noise is a major technical difficulty in underwater acoustic signal processing.
[0004] The present invention proposes a frequency domain weighted phase shift reduction line spectrum enhancement method, the input signal is the underwater acoustic target radiation noise data, and the output signal is the noise signal after line spectrum enhancement. Compared with the traditional beamforming processing method, this method significantly improves the signal-to-noise ratio gain of the line spectrum, effectively realizes line spectrum enhancement, and is beneficial to line spectrum detection and target recognition in the subsequent process.
[0005] The technical differences between this application and the prior art are as follows:
[0006] Technical comparison with patent CN107886068A "A method for detecting unknown line spectrum targets based on spatiotemporal line spectrum enhancer"
[0007] 1. Patent CN107886068A adopts space-time two-dimensional line spectrum enhancement processing, and each new array data obtained by processing is segmented according to the set parameters and discrete Fourier transform is performed, and the discrete Fourier transform output value of each new array element at the same segment number is taken out, and the discrete Fourier transform output value at the same frequency point is subjected to frequency domain beamforming, and the frequency domain beamforming result is integrated and summed in the frequency domain and time, and finally the azimuth spectrum output result is obtained. However, the present invention does not use the traditional frequency domain beamforming and time-frequency domain integration and summation method to enhance the line spectrum, but adopts the low-pass filter fitting method to perform background equalization processing on the average power spectrum of two adjacent array elements, and performs nonlinear mapping on the logarithmic difference spectrum amplitude after background equalization, so as to calculate the frequency domain complex weighted value when the two adjacent array elements are reduced, and the spectrum sequence of the received signal is weighted and reduced according to the complex weighted value, and finally the enhanced line spectrum is obtained.
[0008] Second, in patent CN107886068A, the data obtained by the line spectrum enhancer is divided into Q segments, each segment has N data points, and then a discrete Fourier transform is performed on each segment of data, and then subsequent frequency domain beamforming and time-frequency domain integral summation are performed. However, the present invention uses a two-by-two grouping method to process the radiation noise signals received by two adjacent array elements, calculate the average power spectrum and phase difference of each group of two received signals, and then perform subsequent weighted reduction processing.
[0009] Technical comparison with patent CN116755097A "A multi-channel frequency domain line spectrum enhancement method and system for manta ray-like aircraft"
[0010] 1. Patent CN116755097A carries two linear arrays with an angle of α degrees on the two wings of the manta ray-like aircraft platform, with a total of M hydrophone array elements, and collects data through the hydrophone array during the sliding process of the aircraft. However, the present invention mainly focuses on underwater vertical arrays rather than linear arrays, aiming to improve the spatial processing gain of noise signals.
[0011] 2. In patent CN116755097A, after low-pass filtering, downsampling and data block processing are performed on the collected data, the mutual spectra of the same block data of two adjacent line arrays are averaged and normalized, and the prior frequency domain weight vector of the line spectrum enhancement module is calculated. Then, the frequency domain weight vector of the previous array element is combined to obtain the updated frequency domain weight vector, and finally, the block data of each array element is input into the line spectrum enhancement module, and the result of multi-channel line spectrum enhancement is obtained by directly multiplying the frequency domain weight and the block data. In the present invention, after the received signals are grouped in pairs, the average power spectrum of each group of two received signals is calculated instead of the mutual spectrum of the two, and then the average power spectrum is processed by background equalization, nonlinear mapping, etc., so as to obtain the complex weighted value in the frequency domain. In addition, the present invention performs phase shift superposition on two adjacent array elements in pairs according to the Hadamard product operation instead of direct multiplication, until the line spectrum enhancement data of a single channel is finally output.
[0012] Technical comparison with patent CN109799495A "A broadband delay estimation method for high-fidelity array processing"
[0013] 1. Patent CN109799495A calculates beam energy diagram, pre-reconstructs source signal, calculates frequency domain signal-to-noise ratio and phase difference, and performs weighted fusion to finally realize broadband delay vector estimation and secondary reconstruction of source signal. The present invention calculates the average power spectrum and phase difference of each group of two received signals, background equalization, nonlinear mapping to obtain frequency domain complex weighting, and then performs other steps, and finally performs weighted reduction processing on the spectrum sequence of the received signal to realize line spectrum enhancement and signal detection. The present invention does not need to calculate pre-beam data of different orientations of array acquisition data, nor does it need to pre-reconstruct the source signal.
[0014] Second, patent CN109799495A emphasizes the reconstruction of source signals with enhanced fidelity under conditions of array distortion and near-field sources, and mainly processes broadband signals, with the emphasis on delay estimation, in order to obtain reconstructed source signals with enhanced fidelity. The present invention mainly processes narrowband signals under far-field source conditions, with the emphasis on improving the signal-to-noise ratio gain of the line spectrum, in order to enhance the line spectrum of weak target lines. Summary of the invention
[0015] Aiming at the problem of extracting line spectrum features of underwater acoustic targets under a vertical array with low spatial correlation, the present invention provides a frequency domain weighted phase shift reduction line spectrum enhancement method. The method groups the received signals of each hydrophone in the vertical array into groups of two, calculates the average power spectrum of each group of two-way received signals and the phase difference at different frequencies of the received signals of two adjacent array elements, and corrects the obtained phase difference to the main value interval; performs background equalization processing on the average power spectrum using a low-pass filter fitting method, performs nonlinear mapping on the amplitude of the logarithmic difference spectrum after background equalization, and calculates the frequency domain complex weighted value of the reduced processing of two adjacent array elements; performs weighted reduction processing on the spectrum sequence of the received signal until a 1-way spectrum sequence is obtained, the line spectrum enhancement process is completed, and the output data can be further used for line spectrum detection.
[0016] To achieve the above object, the technical solution adopted by the present invention is:
[0017] A frequency domain weighted phase shift rule line spectrum enhancement method, characterized in that it comprises the following steps:
[0018] (1) Initialization parameters;
[0019] (2) Read in the sampling sequence of the radiated noise received by the hydrophone and perform discrete Fourier transform on the sampling sequence;
[0020] (3) Grouping the received signals of each hydrophone into groups of two, calculating the average power spectrum of the two received signals in each group and the phase difference of the two received signals in each group at different frequencies, and correcting the obtained phase difference to the main value interval;
[0021] (4) Perform background equalization on the obtained average power spectrum using a low-pass filter fitting method, perform nonlinear mapping on the amplitude of the logarithmic difference spectrum after background equalization, and calculate the frequency domain complex weighted value of the two adjacent array elements during the reduction processing;
[0022] (5) The spectrum sequence of the received signal is weighted and reduced until a single spectrum sequence is obtained. The line spectrum enhancement process is completed and the output data can be further used for line spectrum detection and extraction.
[0023] As a further improvement of the present invention, step (1) specifically comprises the following steps:
[0024] (1.1) Initialize parameters and set sampling frequency f s , number of sampling points N, calculation resolution
[0025] As a further improvement of the present invention, step (2) specifically comprises the following steps:
[0026] (2.1) Read in the sampling sequence x of the radiated noise received by each of the M hydrophones m (n), m=1,2,…,M, n=0,1,…,N-1;
[0027] (2.2) The sampling sequence x of the radiated noise of the underwater acoustic target m (n) Perform discrete Fourier transform to obtain M sequences X' of length N m (k), k=0,1,…,N-1.
[0028] As a further improvement of the present invention, step (3) specifically comprises the following steps:
[0029] (3.1) The received signals of each hydrophone are grouped into two groups, and the average power spectrum of the two received signals in each group is calculated to obtain The average power spectrum sequence D of length N i '(k):
[0030]
[0031] in,
[0032] (3.2) Let the discrete Fourier transform of the N-point sampling sequence of two adjacent array elements be X' m (k) and X' m+1 (k), X' m (k) and X' m+1 (k) is written in complex exponential form as:
[0033]
[0034] in, and They represent the phases of the two array element signals at the frequency point k. The phases take the main value interval [-π,π] and are calculated by the following formula: and
[0035]
[0036] Among them, Im(·) represents the imaginary part of the complex number, and Re(·) represents the imaginary part of the complex number;
[0037] (3.3) Calculate the phase difference between two adjacent array elements at frequency point k:
[0038]
[0039] (3.4) The phase difference between the two array elements at the frequency point k is adjusted to the main value interval [-π,π] by adding or subtracting 2π:
[0040]
[0041] As a further improvement of the present invention, step (4) specifically comprises the following steps:
[0042] (4.1) Using the low-pass filter fitting method, the average power spectrum sequence D i '(k) Perform background equalization processing, and record the average power spectrum sequence after fitting as D i ”(k), then the processed logarithmic difference spectrum amplitude sequence D i (k) is:
[0043]
[0044] Among them, D i The mean of (k) is μ i , with variance σ i 2 ;
[0045] (4.2) The parameter λ is determined by the following formula i :
[0046]
[0047] Among them, η i is the noise suppression parameter of the algorithm, ln represents the natural logarithm operation;
[0048] (4.3) The average power spectrum sequence D i (k) The nonlinear mapping is α i (k):
[0049]
[0050] (4.4) The phase shift ψ is calculated by the following formula i (k):
[0051]
[0052] (4.5) From the phase shift ψ i (k) Calculate the complex weighted value w in the frequency domain i (k):
[0053]
[0054] w i (k) is written in vector form:
[0055] w i =[w i (0),w i (1),...,w i (N-1)] T .
[0056] As a further improvement of the present invention, step (5) specifically comprises the following steps:
[0057] (5.1) The sequence X' after discrete Fourier transform of M array elements and N points is obtained m (k) multiplied by the scaling factor Let the new sequence be X m (k), the frequency domain vector of the new sequence is recorded as X m , that is, satisfying:
[0058]
[0059] (5.2) The two adjacent array elements are phase-shifted and superimposed according to the following formula, and the frequency domain output vector is recorded as Y i :
[0060]
[0061] In the above formula, X 2 i- 1 and X 2 i is the frequency domain vector of the sampling sequences of two array elements, and operation (e) represents the Hadamard product;
[0062] (5.3) If the output frequency domain vector Y i If the number of is greater than 1, then Y i As new array data, return to step (2) for calculation, continue to superimpose two by two until the single channel data is finally output, and record the final output frequency domain vector as Y. At this point, the line spectrum enhancement process is completed, and the output data can be further used for line spectrum detection and extraction.
[0063] Beneficial effects:
[0064] Compared with the prior art, the method disclosed in the present invention has the following advantages: compared with the traditional beamforming processing method, the line spectrum enhancement method based on the multi-array element frequency domain weighted phase shift rule is used to more effectively obtain the spatial gain of the vertical array, improve the signal-to-noise ratio of the line spectrum, and make the line spectrum enhancement effect significant. BRIEF DESCRIPTION OF THE DRAWINGS
[0065] Figure 1 It is a flow chart for implementing the method of the present invention.
[0066] Figure 2 This is a processing block diagram of the phase shift reduction algorithm in the method of the present invention.
[0067] Figure 3 It is the logarithmic difference spectrum amplitude sequence after background equalization in Example 1.
[0068] Figure 4 is the α after nonlinear mapping in Example 1 i (k) Sequence value.
[0069] Figure 5is the average power spectrum in Example 1.
[0070] Figure 6 This is the power spectrum after using the method of the present invention in Example 1. DETAILED DESCRIPTION
[0071] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments:
[0072] The present invention enhances the line spectrum of an underwater acoustic target in a vertical array space, calculates it in the frequency domain through a multi-array element frequency domain weighted phase shift reduction algorithm, and obtains a signal after enhancing the line spectrum and suppressing noise.
[0073] Embodiment 1:
[0074] The present invention is further explained below in conjunction with the accompanying drawings and specific implementation methods.
[0075] A frequency domain weighted phase shift reduction line spectrum enhancement method, such as Figure 1 , Figure 2 As shown, the following steps are included:
[0076] Step 1 is as follows:
[0077] (1.1) Initialization parameters: number of sensors M = 22, number of sampling points N = 6000, sampling frequency f s =1500Hz, resolution
[0078] Step 2 is as follows:
[0079] (2.1) Read in the sampling sequence x of the measured radiated noise at sea received by each of the M hydrophones m (n), m=1,2,…,M, n=0,1,…,N-1.
[0080] (2.2) The sampling sequence x of the radiated noise of the underwater acoustic target m (n) Perform discrete Fourier transform to obtain M sequences X' of length N m (k), k=0,1,…,N-1.
[0081] Step 3 is as follows:
[0082] (3.1) The received signals of each hydrophone are grouped into two groups, and the average power spectrum of the two received signals in each group is calculated to obtain The average power spectrum sequence D of length N i '(k):
[0083]
[0084] in,
[0085] (3.2) Let the discrete Fourier transform of the N-point sampling sequence of two adjacent array elements be X' m (k) and X' m+1 (k), X' m (k) and X' m+1 (k) is written in complex exponential form as:
[0086]
[0087] in, and They represent the phases of the two array element signals at the frequency point k. The phases take the main value interval [-π,π] and are calculated by the following formula: and
[0088]
[0089] (3.3) Calculate the phase difference between two adjacent array elements at frequency point k:
[0090]
[0091] (3.4) The phase difference between the two array elements at the frequency point k is adjusted to the main value interval [-π,π] by adding or subtracting 2π:
[0092]
[0093] Step 4 is as follows:
[0094] (4.1) Using the low-pass filter fitting method, the average power spectrum sequence D i '(k) Perform background equalization processing, and record the average power spectrum sequence after fitting as D i ”(k), then the processed logarithmic difference spectrum amplitude sequence D i (k) is:
[0095]
[0096] Among them, D i (k) The mean is μ i , with variance σ i 2 The logarithmic difference spectrum amplitude sequence D after background equalization i (k) Figure 3 shown.
[0097] (4.2) The parameter λ is determined by the following formula i :
[0098]
[0099] Among them, ηi is the noise suppression parameter of the algorithm, which affects the algorithm's ability to suppress noise. Here, η is taken i =0.5, and we get λ i =0.0061.
[0100] (4.3) The average power spectrum sequence D i (k) The nonlinear mapping is α i (k):
[0101]
[0102] α after nonlinear mapping i (k) The sequence value is as follows Figure 4 shown.
[0103] (4.4) The phase shift ψ is calculated by the following formula i (k):
[0104]
[0105] (4.5) From the phase shift ψ i (k) Calculate the complex weighted value w in the frequency domain i (k):
[0106]
[0107] w i (k) is written in vector form:
[0108] w i =[w i (0),w i (1),...,w i (N-1)] T
[0109] Step 5 is as follows:
[0110] (5.1) Sample the M array elements at N points and perform discrete Fourier transform on the resulting sequence X' m (k) multiplied by the scaling factor Let the new sequence be X m (k), the frequency domain vector of the new sequence is recorded as X m , that is, satisfying:
[0111]
[0112] (5.2) The two adjacent array elements are phase-shifted and superimposed according to the following formula, and the frequency domain output vector is recorded as Y i :
[0113]
[0114] In the above formula, X 2i-1 and X 2i is the frequency domain vector of the two array element sampling sequences, operation (e) represents the Hadamard product; the coefficient 1 / 2 is to keep the output power unchanged.
[0115] (5.3) The frequency domain vector Y output at this time i The number of Y is 11. i If the number of is greater than 1, then Y i As new array data, return to step (2) for calculation, continue to superimpose two by two until the single channel data is finally output, and record the final output frequency domain vector as Y. At this point, the line spectrum enhancement process is completed, and the output data can be further used for line spectrum detection and extraction.
[0116] Figure 5 is the average power spectrum in Example 1, Figure 6 is the power spectrum after using the method of the present invention.
[0117] The above embodiments show that the method of the present invention can effectively enhance the low-frequency line spectrum of the underwater acoustic target radiation noise, and still has good noise suppression capability in the vertical array space, which is beneficial to the subsequent line spectrum detection and extraction work.
[0118] The above description is only a preferred embodiment of the present invention and does not constitute any other form of limitation to the present invention. Any modification or equivalent change made based on the technical essence of the present invention still falls within the scope of protection required by the present invention.
Claims
1. A frequency domain weighted phase shift reduction line spectrum enhancement method, Features: The following steps are involved: (1) Initialization parameters; (2) Read in the sampling sequence of the radiated noise received by the hydrophone and perform discrete Fourier transform on the sampling sequence; (3) Grouping the received signals of each hydrophone into groups of two, calculating the average power spectrum of the two received signals in each group and the phase difference of the two received signals in each group at different frequencies, and correcting the obtained phase difference to the main value interval; (4) Perform background equalization on the obtained average power spectrum using a low-pass filter fitting method, perform nonlinear mapping on the amplitude of the logarithmic difference spectrum after background equalization, and calculate the frequency domain complex weighted value of the two adjacent array elements during the reduction processing; (5) The spectrum sequence of the received signal is weighted and reduced until a single spectrum sequence is obtained. The line spectrum enhancement process is completed and the output data can be further used for line spectrum detection and extraction.
2. A frequency domain weighted phase shift reduction line spectrum enhancement method according to claim 1, Features: Step (1) specifically includes the following steps: (1.1) Initialize parameters and set sampling frequency f s , number of sampling points N, calculation resolution 3. The method for enhancing line spectrum by frequency domain weighted phase shift reduction according to claim 1, Features: Step (2) specifically includes the following steps: (2.1) Read in the sampling sequence x of the radiated noise received by each of the M hydrophones m (n), m=1,2,…,M, n=0,1,…,N-1; (2.2) The sampling sequence x of the radiated noise of the underwater acoustic target m (n) Perform discrete Fourier transform to obtain M sequences X' of length N m (k), k=0,1,…,N-1.
4. The method for enhancing line spectrum by frequency domain weighted phase shift reduction according to claim 1, Features: Step (3) specifically includes the following steps: (3.1) The received signals of each hydrophone are grouped into two groups, and the average power spectrum of the two received signals in each group is calculated to obtain The average power spectrum sequence D of length N i '(k): in, (3.2) Let the discrete Fourier transform of the N-point sampling sequence of two adjacent array elements be X' m (k) and X' m+1 (k), X' m (k) and X' m+1 (k) is written in complex exponential form as: in, and They represent the phases of the two array element signals at the frequency point k. The phases take the main value interval [-π,π] and are calculated by the following formula: and Among them, Im(·) represents the imaginary part of the complex number, and Re(·) represents the imaginary part of the complex number; (3.3) Calculate the phase difference between two adjacent array elements at frequency point k: (3.4) The phase difference between the two array elements at the frequency point k is adjusted to the main value interval [-π,π] by adding or subtracting 2π:
5. The method for enhancing line spectrum by frequency domain weighted phase shift reduction according to claim 1, Features: Step (4) specifically includes the following steps: (4.1) Using the low-pass filter fitting method, the average power spectrum sequence D i '(k) Perform background equalization processing, and record the average power spectrum sequence after fitting as D i ”(k), then the processed logarithmic difference spectrum amplitude sequence D i (k) is: Among them, D i The mean of (k) is μ i , with variance σ i 2 ; (4.2) The parameter λ is determined by the following formula i : Among them, η i is the noise suppression parameter of the algorithm, ln represents the natural logarithm operation; (4.3) The average power spectrum sequence D i (k) The nonlinear mapping is α i (k): (4.4) The phase shift ψ is calculated by the following formula i (k): (4.5) From the phase shift ψ i (k) Calculate the complex weighted value w in the frequency domain i (k): w i (k) is written in vector form: w i =[w i (0),w i (1),...,w i (N-1)] T 。 6. The method for enhancing line spectrum by frequency domain weighted phase shift reduction according to claim 1, Features: Step (5) specifically includes the following steps: (5.1) The sequence X' after discrete Fourier transform of M array elements and N points is obtained m (k) multiplied by the scaling factor Let the new sequence be X m (k), the frequency domain vector of the new sequence is recorded as X m , that is, satisfying: (5.2) The two adjacent array elements are phase-shifted and superimposed according to the following formula, and the frequency domain output vector is recorded as Y i : In the above formula, X 2i-1 and X 2i is the frequency domain vector of the two array element sampling sequences, and operation (e) represents the Hadamard product; (5.3) If the output frequency domain vector Y i If the number of is greater than 1, then Y i As new array data, return to step (2) to perform the calculation, and continue to superimpose two by two until a single channel data is finally output. The frequency domain vector of the final output is recorded as Y. At this point, the line spectrum enhancement process is completed, and the output data can be further used for line spectrum detection and extraction.
