A multi-channel frequency-domain line-spectrum enhancement method and system for manta ray vehicle

Abstract

This invention discloses a multi-channel frequency domain line spectrum enhancement method and system for manta ray-inspired vehicles, belonging to the field of underwater acoustic target line spectrum detection technology for biomimetic vehicles. The method comprises the following steps: First, two linear arrays with a common angle of α degrees and a total of M elements are mounted on the manta ray-inspired vehicle, and data is collected during platform sliding. The collected data is then low-pass filtered to reduce the sampling frequency, and the data from the M elements is divided into blocks. Next, the cross-spectrums of adjacent elements of the two linear arrays at the same time are averaged and normalized. The frequency domain weights for line spectrum enhancement are updated. The block data from the multiple elements is combined and input into the line spectrum enhancer to obtain the multi-channel line spectrum enhancement result. Finally, the above process is repeated, continuously updating the weight vector to obtain the line spectrum enhancement results for all channels. This invention overcomes the problems of slow convergence and difficulty in detecting weak target line spectra in traditional methods by using a multi-channel input method to effectively detect target line spectra.

Description

Technical Field

[0001] This invention belongs to the field of underwater acoustic target line spectrum detection technology for biomimetic vehicles, specifically relating to a multi-channel frequency domain line spectrum enhancement method and system for manta ray-inspired vehicles. Background Technology

[0002] In passive sonar detection technology, people often focus on the low-frequency line spectrum components of the received signal, which have higher power and stability than the broadband components of the received signal. However, the line spectrum of low-noise underwater targets is easily submerged in the continuous spectrum components of the received signal due to the influence of environmental noise. Therefore, it is necessary to filter out broadband interference and background noise in the signal. Effective target line spectrum detection is of great significance for improving the passive detection range and detecting low-noise underwater targets.

[0003] The manta ray-inspired underwater vehicle (UVV) is modeled after the manta ray in the ocean. It uses a single-drive slide-link mechanism to drive its flapping wings, achieving a sinusoidal flapping motion that closely resembles the movement of a real marine animal. Due to its excellent design in terms of size, structural layout, drive mechanism, and flexible wing distribution, this biomimetic vehicle generates very low self-noise during gliding, making it suitable for target identification using passive sonar technology. Since the UVV operates without human intervention, it is highly adaptable to autonomous target detection environments. Therefore, researching how to passively detect low-noise underwater targets using line spectral signals under complex conditions such as non-uniform noise backgrounds and strong interference during the movement of the UVV is of significant practical importance.

[0004] Traditional bionic robots, due to size limitations, mainly rely on a single hydrophone as their primary detection device. However, the data collected by a single hydrophone is limited, and when performing line spectrum enhancement, the algorithm converges slowly, making it difficult to effectively detect the target line spectrum. Summary of the Invention

[0005] The technical problem to be solved:

[0006] To overcome the shortcomings of existing technologies, this invention provides a multi-channel frequency domain line spectrum enhancement method and system for manta ray mimicry vehicles. The method involves deploying a passive sonar array on a manta ray mimicry vehicle and conducting research on multi-channel line spectrum enhancement technology. Data is acquired using the onboard multi-channel linear array, fully utilizing the spatial sparsity of the line spectrum signal. The cross-spectral results of the array data are used as influencing factors for frequency domain weight updates. The use of multi-channel data improves the convergence speed of the line spectrum enhancement algorithm, thereby increasing the target detection speed. To effectively detect target line spectra, this invention overcomes the problems of slow convergence and difficulty in detecting weak target line spectra in traditional methods by using a multi-channel input approach.

[0007] The technical solution of this invention is: a multi-channel frequency domain line spectrum enhancement method for manta ray-inspired aircraft, the specific steps of which are as follows:

[0008] Step 1: Mount two linear arrays with an included angle of α degrees on the two wings of the manta ray-inspired vehicle platform, with a total of M hydrophone array elements. Collect data through the hydrophone arrays during the sliding process of the vehicle.

[0009] Step 2: After low-pass filtering the collected data, reduce the sampling frequency and divide the data into blocks;

[0010] Step 3: Calculate the average of the cross spectra of the i-th block data of two adjacent linear arrays, and normalize it as prior knowledge for line spectrum enhancement. This normalization result is updated once after the line spectrum enhancement of the block data of M array elements.

[0011] Step 4: Update the frequency domain weights of the line spectrum enhancement module;

[0012] Step 5: Using the updated weights obtained in Steps 3 and 4, combined with the block data of each array element, input the data into the line spectrum enhancement module to obtain the result of multi-channel line spectrum enhancement;

[0013] Step 6: Repeat steps 3 to 5 above until all block data of M array elements are enhanced, thus obtaining the line spectrum enhancement results of all channels at each time step.

[0014] A further technical solution of the present invention is: in step 1, the included angle α between the two linear arrays is 110-130°; the collected data of the m-th array element is denoted as...

[0015] A further technical solution of the present invention is: in step 2, the collected data is low-pass filtered to below 1kHz, and then the sampling frequency is reduced to 2000.

[0016] A further technical solution of the present invention is: in step 2, the length of the block data is greater than one times the sampling frequency and less than two times the sampling frequency.

[0017] A further technical solution of the present invention is: In step 3, the received signal of the M array elements uniform linear array is first divided into blocks, each block including L points, then the signal of the i-th block of array element m is:

[0018]

[0019] Assumption This represents the cross-spectral vector between two adjacent elements of the received signal from the i-th M-element uniform linear array, i.e.

[0020]

[0021] cross-spectral vectors Find their average value, denoted as . That is, as shown in the following formula:

[0022]

[0023] The expression is normalized using the min-max method and used as the prior frequency domain weight vector. Right now

[0024]

[0025] Where k = 1, 2, ..., L-1, min x and max x Represent The maximum and minimum values ​​of a vector; the above formula This is the prior frequency domain weight vector obtained through multiple channels. This vector is updated once every M array elements after the i-th block of data enhancement is completed.

[0026] A further technical solution of the present invention is: the frequency domain weight update method of the line spectrum enhancement module in step 4 is as follows: S1. The i-th block signal of the m-th array element is represented as... The delayed signal is represented as:

[0027]

[0028] Two cascaded input signals are used as input signals for line spectrum enhancement, as follows:

[0029]

[0030] The delayed signal after FFT is represented as:

[0031]

[0032] Will Take the conjugate of the vector to participate in the frequency domain weight vector update;

[0033] S2. The i-th signal of array element m As a reference signal, it is compared with the line spectrum-enhanced output signal obtained from the previous array element. The difference is taken, and the resulting error signal is represented as follows:

[0034]

[0035] The error signal is padded with zeros to make its data length equal to the length of two data blocks, as follows:

[0036]

[0037] Error signal The error signal obtained after FFT in the frequency domain is represented as follows:

[0038]

[0039] Compare the error signal with the delay The subsequent signal multiplication is represented as:

[0040]

[0041] S3. After performing IFFT on equation (11), replace the data of the last L points with 0 and then perform FFT, which is represented as:

[0042]

[0043] After multiplying the gradient constraint and the convergence factor, and combining this with the frequency domain weight vector of the previous array element, we obtain the updated frequency domain weight vector, specifically expressed as:

[0044]

[0045] A further technical solution of the present invention is: in step 5, the prior frequency domain weight vector obtained in step 3 is used. The delay data of the i-th block and the (i-1)-th block of array element m obtained in step 4 are as follows: Frequency domain update weights of the i-th block of data in array element m Multiplying the three signals together yields the enhanced line spectrum signal in the frequency domain; then performing an IFFT gives the enhanced line spectrum signal in the time domain, represented as:

[0046]

[0047] The obtained By discarding the data from the last L points, we can obtain the enhanced signal of the line spectrum of the i-th block of array element m.

[0048] An implementation system for a multi-channel frequency domain line spectrum enhancement method for a manta ray mimicry vehicle includes a manta ray mimicry vehicle, a receiving transducer, and a central control system. The receiving transducer consists of two linear arrays mounted on the wings of the manta ray mimicry vehicle, which together contain M hydrophone array elements.

[0049] The central control system is used to receive data collected by the hydrophone array during the vehicle's sliding process. Its control system takes the collected data as input, calculates the signal after time-domain line spectrum enhancement using a multi-channel frequency domain line spectrum enhancement method, and outputs the signal.

[0050] A control system includes a processor, a memory, and an application program, wherein the application program is stored in the memory and configured to be executed by the processor, the application program being configured to perform the multi-channel frequency domain line spectrum enhancement method for a manta ray-inspired vehicle.

[0051] Beneficial effects

[0052] The beneficial effects of this invention are as follows:

[0053] 1. This invention utilizes two multi-element linear arrays mounted on a manta ray-inspired vehicle platform to increase the amount of data at any given time, providing more data for the research of multi-channel line spectrum enhancement algorithms.

[0054] 2. Compared to other line spectrum enhancement methods, this invention makes two optimizations:

[0055] (1) Multi-channel data was used to increase the amount of sample data, accelerate the convergence speed of the line spectrum enhancement algorithm, and improve the detection speed of the target line spectrum.

[0056] (2) By taking advantage of the spatial sparsity of the target line spectrum signal, the cross spectrum results of adjacent array elements are used as prior knowledge for frequency domain weight update and input into the enhancer, which greatly improves the effect of line spectrum enhancement.

[0057] like Figure 1 and Figure 2 The results show that multi-channel line spectrum enhancement was performed using simulated data with a target radiation spectrum of 90Hz, and noise was added to the simulated data, such as... Figure 1 The image shows the LoFar spectrum of the simulated signal with added noise; the 90Hz line spectrum information of the target radiation is almost completely lost. The result after line spectrum enhancement using a multi-channel line spectrum enhancement method is shown below. Figure 2 As shown, the 90Hz line spectrum information of the target radiation is fully displayed. Figure 2 It can quickly obtain the line spectrum information of the target. Attached Figure Description

[0058] Figure 1 LoFar spectrum without multi-channel line spectrum enhancement;

[0059] Figure 2 LoFar spectrum during multi-channel line spectrum enhancement;

[0060] Figure 3 Schematic diagram of an underwater unmanned platform equipped with a linear array;

[0061] Figure 4 Flowchart of the multi-channel line spectrum enhancement method according to an embodiment of the present invention;

[0062] Figure 5 Flowchart of prior frequency domain weight update. Detailed Implementation

[0063] The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the invention, and should not be construed as limiting the invention.

[0064] This invention provides a method for multi-channel line spectrum enhancement of low-frequency sound sources on a manta ray-inspired vehicle. The conformal linear array mounted on the manta ray-inspired vehicle fully utilizes the spatial sparsity of the line spectrum signal. The weight values ​​of the line spectrum enhancer are initialized using the cross power spectrum of adjacent array elements. Under the premise of keeping the number of samples unchanged, multi-channel data is used to accelerate the convergence of the line spectrum enhancement algorithm.

[0065] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0066] A method for enhancing the multi-channel frequency domain sparse line spectrum of a manta ray-inspired vehicle includes the following steps:

[0067] Step 1: Two M-element linear arrays are mounted on a small-scale underwater manta ray-inspired unmanned underwater vehicle platform. Data is collected via a hydrophone array during the descent of the manta ray-inspired unmanned underwater vehicle, such as... Figure 1 As shown, the collected data of the m-th array element is denoted as...

[0068] Step 2: After low-pass filtering the collected data, the sampling frequency is reduced. In actual use, the cutoff frequency of the low-pass filter and the required sampling rate can be adjusted according to the sampling rate of the device. The data of the M array elements are divided into blocks. The length of the block data is generally greater than one sampling frequency and less than two sampling frequencies.

[0069] Step 3: Calculate the average of the cross spectra of the i-th block data of two adjacent linear arrays and normalize it as prior knowledge for line spectrum enhancement. This normalization result is updated once after the line spectrum enhancement of the block data of M array elements.

[0070] The specific operating procedures are as follows: Figure 3 As shown:

[0071] First, divide the received signal of the M-element uniform linear array into blocks, each block including L points. Then, the signal of the i-th block of array element m is:

[0072]

[0073] Assumption This represents the cross-spectral vector between two adjacent elements of the received signal from the i-th M-element uniform linear array, i.e.

[0074]

[0075] cross-spectral vectors Find their average value, denoted as . That is, as shown in equation XX:

[0076]

[0077] The expression is normalized using the min-max method and used as the prior frequency domain weight vector. Right now

[0078]

[0079] Where k = 1, 2, ..., L-1, min x and max x Represent The maximum and minimum values ​​of a vector. (The above formula...) This is the prior frequency domain weight vector obtained through multiple channels. This vector is updated once every M array elements after the i-th block of data enhancement is completed.

[0080] Step 4: Frequency domain weight update for line spectrum enhancement.

[0081] The frequency domain weight update for line spectrum enhancement is a key part of this invention, and the specific operation is as follows:

[0082] S1. The i-th block signal of the m-th array element can be obtained from step 3. This means that the delayed signal can be represented as:

[0083]

[0084] The cascaded input signals, used as input signals for line spectrum enhancement, can be expressed as:

[0085]

[0086] The delayed signal can be represented by FFT as follows:

[0087]

[0088] Will Take the conjugate of the vector to participate in the frequency domain weight vector update.

[0089] S2. The i-th signal of array element m The reference signal and the line spectrum enhancement output signal obtained from the previous element are used as the reference signal. The error signal obtained by subtraction can be represented as:

[0090]

[0091] Padding the error signal with zeros to make its data length equal to the length of two data blocks can be represented as:

[0092]

[0093] Error signal The frequency domain error signal obtained after FFT can be expressed as:

[0094]

[0095] Multiplying the error signal by the signal after a delay of τ can be expressed as:

[0096]

[0097] S3. After performing IFFT on equation (16), replace the data of the last L points with 0 and then perform FFT, which can be expressed as:

[0098]

[0099] After multiplying the gradient constraint and the convergence factor, and combining this with the frequency domain weight vector of the previous array element, we obtain the updated frequency domain weight vector, which can be specifically expressed as:

[0100]

[0101] Step 5: Using the results obtained in Steps 3 and 4, combined with the block data of each array element, input the data into the line spectrum enhancer to obtain the result of multi-channel line spectrum enhancement.

[0102] In step 3, the prior frequency domain weight vector has already been obtained. The delay data of the i-th block and the (i-1)-th block of the array element m are In step 4, the frequency domain update weights of the i blocks of data for the m-element array are obtained. Multiplying the three signals yields the frequency-domain line spectrum-enhanced signal. Then, performing an IFFT yields the time-domain line spectrum-enhanced signal, which can be represented as:

[0103]

[0104] The obtained By discarding the data from the last L points, we can obtain the enhanced signal of the i-th block of the m-element array.

[0105] Step 6: By following the above 5 steps, the line spectrum enhancement signal of the i-th block of the m array elements can be obtained. Repeat steps 3 to 5 to obtain the line spectrum enhancement results of all array elements.

[0106] This invention is described using a practical simulation example. Multi-channel line spectrum enhancement is performed using simulation data with a target radiation spectrum of 90Hz. To simulate a real-world application environment, noise is added to the simulation data (approximately -25dB noise) to increase background noise. For example... Figure 1 The image shows the LoFar spectrum of the simulated signal with added noise in one channel; the 90Hz line spectrum information of the target radiation is almost completely lost. To improve target identification, the result of line spectrum enhancement using the multi-channel line spectrum enhancement method described in this example is shown below. Figure 2As shown, the 90Hz line spectrum information of the target radiation is fully displayed. Figure 2 It can quickly obtain the spectral information of the target, thereby achieving the effect of target identification.

[0107] In this embodiment, an implementation system for a multi-channel frequency domain line spectrum enhancement method for a manta ray mimicry vehicle includes a manta ray mimicry vehicle, a receiving transducer, and a central control system. The receiving transducer consists of two linear arrays mounted on the wings of the manta ray mimicry vehicle, containing a total of M hydrophone array elements. The central control system is used to receive data collected by the hydrophone array during the vehicle's sliding process. Its control system takes the collected data as input, calculates the signal using the multi-channel frequency domain line spectrum enhancement method, and outputs the time-domain line spectrum enhanced signal.

[0108] In this embodiment, a control system includes a processor, a memory, and an application program, wherein the application program is stored in the memory and configured to be executed by the processor, and the application program is configured to execute the multi-channel frequency domain line spectrum enhancement method for a manta ray-inspired vehicle.

[0109] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention without departing from the principles and spirit of the present invention.

Claims

1. A multi-channel frequency-domain line-spectrum enhancement method for manta ray vehicle, characterized in that The specific steps are as follows: Step 1: Two linear arrays with an included angle of a degrees are carried on the two wings of the manta ray vehicle platform, and a total of M hydrophone array elements are used to collect data during the sliding process of the vehicle; Step 2: After low-pass filtering the collected data, the sampling frequency is reduced, and the data is blocked; Step 3: the first The average of the two-way spectrum of the block data is calculated, and the result is normalized as prior knowledge for line spectrum enhancement. The normalized result is updated every M element block data after line spectrum enhancement. Step 4: Frequency domain weight update of the line spectrum enhancement module; Step 5: Using the updated weights obtained in steps 3 and 4, combined with the blocked data of each array element, input into the line spectrum enhancement module to obtain the result of multi-channel line spectrum enhancement; Step 6: Repeat the above steps 3 to 5 until all the blocked data of the M array elements are enhanced, i.e. obtain the line spectrum enhancement result of all channels at each time.

2. The method for multi-channel frequency-domain line-spectrum enhancement for manta ray vehicle according to claim 1, wherein: In the step 1, the included angle α of the two linear arrays is 110-130°; the collected data of the mth array element is denoted as .

3. The method for multi-channel frequency domain line spectrum enhancement for manta ray vehicle according to claim 1, wherein: In step 2, the collected data is low-pass filtered to below 1KHZ, and then the sampling frequency is reduced to 2000.

4. The method for multi-channel frequency-domain line-spectrum enhancement of a manta ray vehicle according to claim 3, wherein: In step 2, the length of the blocked data is greater than one sampling frequency and less than two sampling frequencies.

5. The method for multi-channel frequency-domain line spectrum enhancement of manta ray spacecraft according to any one of claims 1-4, characterized in that: In step 3, first, the received signal of the uniform linear array is divided into blocks, each block including points, and then the block signal is given by ​​​ (1) Assume representing the block The cross-spectrum vector of two adjacent elements of the received signal of the uniformly linear array is (2) Cross-spectral vectors averaging them, denoted by i.e. as follows: (3) The equation is normalized using the min-max method as a priori frequency domain weight vector i.e. (4) in, , and Represent The maximum and minimum values ​​of a vector; the above formula This is the prior frequency domain weight vector obtained through multiple channels. This vector is updated once every M array elements after the i-th block of data enhancement is completed.

6. The method for multi-channel frequency-domain line-spectrum enhancement of a manta ray vehicle according to claim 5, wherein: In step 4, the frequency domain weight update method of the line spectrum enhancement module is, S1. The first number of elements of the first block signal is represented as , and the delayed signal is represented as: (5) Cascade two input signals as the input signal of line spectrum enhancement, denoted as: (6) After delaying the signal and performing FFT, it is denoted as: (7) Will Take the conjugate of the vector to participate in the frequency domain weight vector update; S2. The i-th block signal of the array element m is As a reference signal, the output signal of the line spectrum enhancement obtained by the previous array element Subtracting, the error signal is represented as: (8) The error signal is zero-padded to have a data length of two data blocks, denoted as: (9) The error signal The error signal in the frequency domain after the FFT is represented as (10) multiplying the error signal with the delayed The multiplication of the error signal with the delayed (11) After IFFT of equation (11), replace the last L points of data with 0 and perform FFT, denoted as: (12) After adding the gradient constraint and multiplying the convergence factor, the updated frequency domain weight vector is obtained combined with the frequency domain weight vector of the previous array element, specifically denoted as: (13)。 7. The method for multi-channel frequency-domain line-spectrum enhancement of a manta ray vehicle according to claim 6, wherein: In step 5, the prior frequency domain weight vector obtained in step 3 is used , the i-th block and the i-1-th block delay data of the array element m obtained in step 4 are , the updated frequency domain weight vector of the i-th block data of the array element m ; after multiplication of the above three signals, the frequency domain line spectrum enhancement signal is obtained; and after IFFT, the time domain line spectrum enhanced signal is obtained, which is represented as: (14) The resulting The resulting signal of the i-th block of the m-th element is obtained by discarding the latter part of the L points of the resulting 8. An implementation system of a multi-channel frequency domain line spectrum enhancement method for a manta ray vehicle, for performing the method of any one of claims 1-7; characterized in that: It comprises a manta ray vehicle, a receiving transducer and a control system, the receiving transducer is two linear arrays carried on the two wings of the manta ray vehicle, and a total of M hydrophone array elements are included; The control system is used for receiving the data collected by the hydrophone array during the sliding process of the vehicle, and the control system takes the collected data as input, and outputs the time domain line spectrum enhanced signal after calculation by the multi-channel frequency domain line spectrum enhancement method.

9. A control system characterized by: It comprises a processor, a memory and an application program, the application program is stored in the memory and configured to be executed by the processor, and the application program is configured to execute the multi-channel frequency domain line spectrum enhancement method for the manta ray vehicle of claim 7.

Images