Method for enhancing ship radiation noise DEMON spectrum based on block sparsity and frequency multiplication detection method

Abstract

The invention provides a method for enhancing a ship radiation noise DEMON spectrum based on block sparsity and a frequency multiplication detection method. According to the method, frequency domain sparsity and frequency multiplication characteristics of a ship radiation noise low-frequency line spectrum are utilized, a block sparse algorithm is utilized to estimate a frequency multiplication line spectrum on a frequency domain, an envelope waveform of a time domain is reconstructed, and a DEMON spectrum is enhanced. Compared with a traditional 1 (1 / 2) spectrum, the method has a good enhancement effect, and achieves a larger signal-to-noise ratio gain. According to the enhancement method, a disordered and irregular noise part in an original DEMON spectrum is changed into frequency multiplication interference with small energy, a foundation is laid for the frequency multiplication detection method provided by the invention, the frequency multiplication detection method extracts possible frequency multiplication components of each group, removes redundant features, reduces the number of the features, and sums and sorts frequency multiplication energy of each group, so that the frequency multiplication detection efficiency is improved. The fundamental frequency of the target is estimated; and the error between the detected target fundamental frequency and the actual fundamental frequency is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of signal detection, and particularly to a method for enhancing the DEMON spectrum of ship radiated noise based on block sparsity and a frequency doubling detection method. Background Art

[0002] With the development of various marine technologies, it inevitably radiates noise into the ocean. Ship information can be obtained by extracting the characteristics of ship radiated noise. In the frequency domain, it mainly consists of three categories: modulation spectrum, continuous spectrum, and line spectrum. Among them, the modulation spectrum has been widely studied because it contains the shaft frequency and blade frequency information of the ship.

[0003] Signal detection refers to the process of identifying, extracting, and analyzing target signals from the background of noise or interference. Its core task is to judge whether a signal exists and estimate its characteristics. In terms of feature analysis, it mainly covers three dimensions: time-domain waveform analysis, frequency-domain spectrum analysis, and combined time-frequency domain analysis. For different signals, different analysis methods need to be adopted. The present invention will enhance the DEMON spectrum in the frequency domain, compare it with traditional enhancement methods under different signal-to-noise ratios, and propose a method for detecting targets and analyze the detection performance.

[0004] The DEMON spectrum is a processing method for the modulation spectrum. The general steps are as follows: 1. Band-pass filter the noise signal. 2. Square detection to extract envelope information. 3. Low-pass filter. 4. Perform a fast Fourier transform (FFT) on the low-pass signal to obtain line spectrum information. The shaft frequency and blade frequency information of the ship can be obtained after processing the ship radiated noise with the DEMON spectrum. However, in the background of low signal-to-noise ratio, the DEMON spectrum is not ideal.

[0005] Sparse signal representation provides a way - according to limited prior knowledge, construct a general signal analysis framework. Assume that an observed signal is a linear superposition of a group of unknown signals, then a group of over-complete dictionaries can be defined. The weighted sum of this group of dictionaries can represent the observed signal, and in the weight vector, only a part of the elements are non-zero, that is, the weight vector is sparse. How to seek the optimal combination to best recover the original signal is the content studied by sparse signal representation. The block sparse algorithm is a special form of the sparse algorithm. The sparse algorithm treats each sparse signal component as independent. While block sparsity utilizes the periodic characteristics of the signal in a certain dimension. Different from the former, the composition form of the block sparse dictionary is in blocks, and each block contains possible periodic elements.

[0006] In summary, based on the processing of the ship radiated noise DEMON spectrum, the present invention uses the block sparse algorithm to enhance the DEMON spectrum in the frequency domain and proposes a frequency domain detection method for the enhanced signal. Summary of the Invention

[0007] The object of the present invention is to solve the problem that the line spectrum features are not obvious and there is too much noise information after extracting the DEMON spectrum from the radiated noise of ships. A method based on block sparsity enhancement for the DEMON spectrum of ship radiated noise and a frequency doubling detection method are proposed. Compared with the traditional 1(1 / 2) spectrum method, the present invention has a greater signal-to-noise ratio gain and a smaller detection error of target information.

[0008] The present invention is realized through the following technical solutions. The present invention proposes a method based on block sparsity enhancement for the DEMON spectrum of ship radiated noise, and the method includes the following steps:

[0009] Step 1: Use DEMON to process the radiated noise of the ship to obtain the low-frequency rate spectrum information of the radiated noise of the ship. After low-pass filtering the low-frequency rate spectrum information, perform square detection, and the signal after square detection is denoted as the observed signal;

[0010] Step 2: Determine the range of the fundamental frequency from the line spectrum position of the fundamental frequency of the DEMON spectrum, construct a block sparse dictionary matrix, and each column element in the dictionary matrix matches the signal after low-pass filtering of the DEMON spectrum in the frequency domain feature, and solve the weight vector;

[0011] Step 3: Use the weight vector to re-cross multiply the constructed dictionary matrix to obtain the estimated time-domain signal to achieve the purpose of time-domain reconstruction of the signal;

[0012] Step 4: Low-pass filter the reconstructed signal and perform FFT to obtain the enhanced DEMON spectrum.

[0013] Further, in Step 1, the specific process of square detection is as follows:

[0014] When the carrier is a single-frequency signal, the modulation signal can be written as:

[0015] y(t) = (1 + mcosΩt)cosωt

[0016] where m is the modulation degree, Ω is the modulation frequency, ω is the carrier frequency, and square process the modulation signal:

[0017]

[0018] Denote the signal after square detection as Y, which is the observed signal to be solved subsequently.

[0019] Further, in the construction of the dictionary matrix, use the harmonic signals in the time domain as the elements of the matrix, and these harmonic signals have different fundamental frequency information; assume that the line spectrum at the 4-fold frequency point needs to be estimated subsequently, then the elements in the dictionary matrix can be defined as:

[0020] y n= cos(2πf n ) + cos(2π2f n ) + cos(2π3f n ) + cos(2π4f n )

[0021] where f n is different target fundamental frequencies, and the range of f n is determined by the setting range of the low-pass filtering frequency band in the DEMON spectrum.

[0022] Furthermore, the observed signal Y in the process of solving the weight vector is expressed as:

[0023] Y = Za + e

[0024] where Z is the dictionary matrix, a is the weight vector, assuming e is additive white Gaussian noise. At this time, the solution problem can be understood as a linear least squares problem approximated by convex relaxation;

[0025]

[0026] subject to ||Y - Za||2 ≤ δ

[0027] In the formula, ||·||1 is the vector 1-norm, which refers to the sum of the absolute values of each element in the vector; ||·||2 is the vector 2-norm, which is the square root of the sum of the squares of each element in the matrix; where δ ≥ ||e||2; this problem can be equivalent to that there exists λ > 0, and there is

[0028]

[0029] The above problem is the LASSO algorithm. Based on the LASSO algorithm, the weight vector a is solved in the noise, and adjusting λ can control the sparsity of the solution.

[0030] The present invention also proposes a method for detecting double frequencies based on the enhanced DEMON spectrum of ship radiated noise, and the method includes the following steps:

[0031] Step 1: Group the enhanced double-frequency line spectrum information;

[0032] Step 2: Set a threshold for the enhanced line spectrum, and screen out the line spectrum components with low energy;

[0033] Step 3: Retrieve the frequency components of the line spectrum after threshold screening: If there are corresponding double-frequency components subsequently, extract this group of double frequencies and their corresponding energy components, which are recorded as a group of double frequencies; if there are no corresponding double-frequency components subsequently, discard this frequency;

[0034] Step 4: Sum the energies of each harmonic group, sort the total energy, and consider the fundamental frequency corresponding to the harmonic group with the maximum energy as the target.

[0035] Further, in Step 1, the energy of each frequency point of the enhanced DEMON spectrum is expressed as:

[0036] E[f] = |X[f]| 2

[0037] where E[f] is the energy of each frequency point, and the energy is normalized; the reconstructed frequency-domain signal is denoted as X[f], and the enhanced line-spectrum energy is an array of a certain length, denoted as A E

[0038]

[0039] α n is the enhanced normalized line-spectrum energy, and n is the total length of the data points; the corresponding frequency array is denoted as A f

[0040]

[0041] The two are combined into a new matrix denoted as M0

[0042] M0 = [A E A f n×2 .

[0043] Further, set a threshold, find the numbers in the vector A E that are less than the threshold, and discard the row numbers where these values are located; the threshold size is set according to requirements; the new two-dimensional matrix is M1, assuming there are still m row elements left:

[0044] M1 = [A E A f m×2

[0045] Traverse the remaining frequency points of A in M1 f .

[0046] Further, in Step 4, for the classified harmonic groups, sum the energies and denote it as A En , and store the summation result of each group in the array {A E1 A E2 ... A En}, arrange the elements in the array, draw a line chart of the energy-fundamental frequency relationship, and the fundamental frequency corresponding to the highest energy point is the target fundamental frequency.

[0047] The present invention also provides an electronic device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of the method for enhancing the DEMON spectrum of ship radiated noise based on block sparsity and the frequency doubling detection method are implemented.

[0048] The present invention also provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the steps of the method for enhancing the DEMON spectrum of ship radiated noise based on block sparsity and the frequency doubling detection method are implemented.

[0049] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0050] Generally, the block sparse algorithm is often applied to echo estimation in the time domain. The present invention utilizes the frequency domain sparsity and frequency doubling characteristics of the low-frequency line spectrum of ship radiated noise, uses the block sparse algorithm to estimate the frequency doubling line spectrum in the frequency domain, reconstructs the envelope waveform in the time domain, and enhances the DEMON spectrum. Compared with the traditional 1(1 / 2) spectrum, it has a good enhancement effect and achieves a greater signal-to-noise ratio gain. This enhancement method also changes the chaotic and irregular noise part in the original DEMON spectrum into frequency doubling interference with less energy, which also lays a foundation for the frequency doubling detection method proposed by the present invention. The frequency doubling detection method extracts each group of possible frequency doubling components, removes redundant features, reduces the number of features, sums and sorts the energy of each group of frequency doublings to estimate the fundamental frequency of the target, and reduces the error between the detected fundamental frequency of the target and the actual fundamental frequency. Description of the Drawings

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0052] Figure 1 It is a flowchart specifically executed for enhancing the DEMON spectrum based on the block sparse algorithm.

[0053] Figure 2 It is a flowchart specifically executed for the frequency doubling detection method. Detailed Embodiments

[0054] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0055] The object of the present invention is achieved as follows: First, use DEMON to process the radiated noise of the ship to obtain the low-frequency rate spectrum information of the radiated noise of the ship. The present invention applies the block sparse algorithm in the frequency domain, and uses the signal after low-pass filtering and detection as the observation signal. Construct a block sparse dictionary matrix, and the elements of each column of the dictionary matrix match the frequency domain characteristics of the signal after low-pass filtering of the DEMON spectrum. Solve the weight vector, and multiply the constructed dictionary matrix again to achieve the purpose of reconstructing the signal in the time domain. Then perform FFT to enhance the DEMON spectrum. And classify and group the octave information, perform energy summation, and identify the octave group with the largest energy as the target information.

[0056] Specifically, in combination with Figure 1 - Figure 2 , the present invention proposes a method for enhancing the DEMON spectrum of ship radiated noise based on block sparsity. DEMON reflects the low-frequency line spectrum components of ship radiated noise, and these line spectra show an octave relationship and can be considered to "appear periodically" in the frequency domain. Therefore, the block sparse algorithm is introduced. Since the observation signal has octave characteristics, only the fundamental frequency range needs to be controlled to construct the element components of the dictionary matrix. The method includes the following steps:

[0057] Step 1: Use DEMON to process the radiated noise of the ship to obtain the low-frequency rate spectrum information of the radiated noise of the ship. After low-pass filtering the low-frequency rate spectrum information, perform square detection, and record the signal after square detection as the observation signal;

[0058] In Step 1, the specific process of square detection is as follows:

[0059] When the carrier is a single-frequency signal, the modulation signal can be written as:

[0060] y(t) = (1 + mcosΩt)cosωt

[0061] where m is the modulation degree, Ω is the modulation frequency, and ω is the carrier frequency. Square-process the modulation signal:

[0062]

[0063] From the above analysis, after squaring the signal y(t), y 2(t) contains a DC component, a modulation frequency and its second harmonic component, as well as other components related to the carrier frequency and the modulation frequency. By low-pass filtering y 2 (t) and removing the DC component, the modulation frequency component can be extracted.

[0064] Denote the signal after square detection as Y, which is the observation signal to be solved subsequently.

[0065] Step 2: Determine the range of the fundamental frequency from the line spectrum position of the DEMON spectrum, construct a block sparse dictionary matrix, each column of the matrix has time-domain harmonic signals with different multiple frequency information, the elements of each column in the dictionary matrix match the frequency-domain characteristics of the signal after low-pass filtering of the DEMON spectrum, and solve for the weight vector;

[0066] Considering that the line spectrum has a certain "periodicity" in the frequency domain (appearing in the form of multiple frequencies), as shown in the formula:

[0067]

[0068] In the formula, δ0(αf) is the multiple frequency line spectrum component, α is the multiple, f is the fundamental frequency information of the target, N is a positive integer, sampling is performed at t n , n = 1, 2,..., N moments respectively, and the multiple frequency line spectrum can be understood as the Fourier transform of the harmonic signal. This represents that the fundamental frequency information is an important estimation object in the subsequent solution process.

[0069] In the construction of the dictionary matrix, time-domain harmonic signals are used as the elements of the matrix, and these harmonic signals have different fundamental frequency information; assuming that the line spectrum at the 4-fold frequency point needs to be estimated subsequently, then the elements in the dictionary matrix can be defined as:

[0070] y n = cos(2πf n ) + cos(2π2f n ) + cos(2π3f n ) + cos(2π4f n )

[0071] where, f n is different target fundamental frequencies, and the range of f n is determined by the set range of the low-pass filter band in the DEMON spectrum. Generally, it only needs to cover the true fundamental frequency of the target. Too small a range may lead to inaccurate fundamental frequency solution results, and too large a range may lead to too slow solution speed.

[0072] In the process of solving the weight vector, the observation signal Y is expressed as:

[0073] Y = Za + e

[0074] Among them, Z is the dictionary matrix, a is the weight vector, which will be affected by noise in actual observations. Assuming e is additive white Gaussian noise, the problem to be solved can be understood as a linear least squares problem approximated by convex relaxation;

[0075]

[0076] subject to ||Y - Za||2 ≤ δ

[0077] In the formula, ||·||1 is the vector 1-norm, which refers to the sum of the absolute values of each element in the vector. The 1-norm of the vector is often used to induce the sparsity of the model; ||·||2 is the vector 2-norm, which is the square root of the sum of the squares of each element in the matrix; where δ ≥ ||e||2; this problem can be equivalent to the existence of λ > 0, and there is

[0078]

[0079] The above problem is the LASSO algorithm. Based on the LASSO algorithm, the weight vector a is solved in the noise, and adjusting λ can control the sparsity of the solution. The value of λ is relatively fixed under various problems to be solved.

[0080] Step 3: Use the weight vector to re-cross multiply the constructed dictionary matrix to obtain the estimated time-domain signal to achieve the purpose of time-domain reconstruction of the signal;

[0081] In the process of solving, the observed signal achieves an optimal match with a certain element in the dictionary, and this element can accurately represent the characteristic components of the signal. Since the noise does not have the spectral characteristics of the target signal, the corresponding sparse representation coefficients can be obtained through time-domain solution. Although these time-domain coefficients themselves lack clear physical meanings, after cross multiplying them with the dictionary matrix, a signal with specific frequency-domain characteristics can be reconstructed. The reconstructed signal has the same spectral characteristics as the original observed signal in the frequency domain, thus realizing the reconstruction of the signal.

[0082] Step 4: Low-pass filter the reconstructed signal and perform FFT to obtain the enhanced DEMON spectrum.

[0083] After low-pass filtering the reconstructed time-domain signal and performing FFT, an enhanced DEMON spectrum is obtained. The reconstructed frequency-domain signal is denoted as X[f].

[0084] At low signal-to-noise ratios, the enhanced DEMON spectrum of this method has certain rules. Because the elements of each column of the dictionary are all harmonic signals and they appear in the form of harmonics in the frequency domain, the interference forms also basically appear in the form of harmonics. Only by finding the correct harmonic components can the target be detected. Therefore, the present invention also proposes a harmonic detection method based on the enhanced DEMON spectrum of ship radiated noise described above. The method includes the following steps:

[0085] Step 1: Group the enhanced frequency-doubled line spectrum information;

[0086] In Step 1, the energy of each frequency point of the enhanced DEMON spectrum is expressed as:

[0087] E[f] = |X[f]| 2

[0088] where E[f] is the energy of each frequency point, and the energy is normalized; the reconstructed frequency-domain signal is denoted as X[f],

[0089] The enhanced line spectrum energy is an array of a certain length, denoted as A E

[0090]

[0091] α n is the enhanced normalized line spectrum energy, and n is the total length of the data points; the corresponding frequency array is denoted as A f

[0092]

[0093] The two are combined into a new matrix denoted as M0

[0094] M0 = [A E A f n×2 .

[0095] The purpose of synthesizing matrix M0 is to correctly correspond the energy of each frequency point after subsequent threshold setting and screening, which is convenient for subsequent data splitting.

[0096] Set a threshold to find the numbers in vector A E that are less than the threshold, and discard the row numbers where these values are located; the threshold size is set according to requirements. If the threshold is too large, it will lead to a reduction in characteristic information. If the threshold is too small, it will lead to an increase in calculation time; the new two-dimensional matrix is M1. Assume there are still m row elements left:

[0097] M1 = [A E A f m×2

[0098] Traverse the remaining frequency points of A in M1, e.g., when traversing to frequency point f0, check whether there are 2f0, 3f0, 4f0 components in the array, and record them as a frequency-doubled group and extract and classify them. f

[0099] Step 2: Set a threshold for the enhanced line spectrum to screen out the line spectrum components with low energy; ​​

[0100] Step 3: Frequency components of the line spectrum after retrieving the threshold: If there are corresponding harmonic components subsequently, extract this group of harmonics and their corresponding energy components, which is recorded as a group of harmonics; if there are no corresponding harmonic components subsequently, discard this frequency;

[0101] Step 4: Sum the energies of each harmonic group, sort the total energies, and consider the fundamental frequency corresponding to the harmonic group with the maximum energy as the target.

[0102] In Step 4, sum the energies of the classified harmonic groups and denote it as A En , and store the summation result of each group into an array {A E1 A E2 ……A En}, arrange the elements in the array, draw a line graph of the energy-fundamental frequency relationship, and the fundamental frequency corresponding to the highest energy point is the target fundamental frequency.

[0103] The present invention proposes a method based on block sparsity to enhance the DEMON spectrum of ship radiated noise and a harmonic detection method, belonging to the field of signal detection. The method first uses DEMON to process the ship radiated noise to obtain the low-frequency rate spectrum information of the ship radiated noise. The present invention applies the block sparse algorithm in the frequency domain, and uses the signal after band-pass filtering and detection as the observation signal. Construct a block sparse dictionary matrix, and the elements of each column of the dictionary matrix match the frequency domain characteristics of the signal after low-pass filtering of the DEMON spectrum. Solve the weight vector, and re-cross multiply the constructed dictionary matrix to achieve the purpose of reconstructing the signal in the time domain. Then perform FFT to realize the enhancement of the DEMON spectrum. And classify and group the harmonic information, sum the energies, and identify the harmonic group with the largest energy as the target information. Compared with the traditional 1(1 / 2) spectrum, it has a good enhancement effect. On the basis of block sparse enhancement, a new harmonic detection method is proposed, extracting each group of possible harmonic components, removing redundant features, reducing the number of features, summing and sorting the energies of each group of harmonics to estimate the target fundamental frequency; reducing the error between the detected target fundamental frequency and the actual fundamental frequency.

[0104] Here, the signal-to-noise ratio is defined as:

[0105] SNR = 10log 10 (P s / P n )

[0106] where P s is the power of the signal, and P n is the power of the noise. In the present invention, the magnitude of the signal-to-noise ratio reflects the "clarity" of the signal spectrum. The larger the signal-to-noise ratio, the clearer the spectrum structure and the more obvious the line spectrum characteristics.

[0107] Define the signal-to-noise ratio gain as:

[0108] ΔSNR = SNR1 - SNR2

[0109] In the present invention, first calculate the signal-to-noise ratio of the original DEMON spectrum, and then subtract the signal-to-noise ratios of the 1(1 / 2) spectrum and the block sparse from the signal-to-noise ratio of the original DEMON spectrum to obtain the signal-to-noise ratio gain.

[0110] Here, the effect is illustrated by the result of one-time data processing. The data is a segment of ship radiated noise signal. After the DEMON spectrum processing, the signal-to-noise ratio of the original DEMON spectrum is 18.07 dB. After the 1(1 / 2) spectrum processing, the signal-to-noise ratio is 23.82 dB, and the signal-to-noise ratio gain is 5.75 dB. After the block sparse processing, the signal-to-noise ratio reaches 28.95 dB, and the signal-to-noise ratio gain is 10.88 dB, which is 5.13 dB higher than that of the traditional method. The specific data is shown in Table 1.

[0111] Table 1 shows the processing results of multiple groups of data

[0112]

[0113] As can be seen from Table 1, in the case of large noise influence: 1. The block sparse still has a large signal-to-noise ratio, and the anti-noise effect is very good. 2. Compared with the 1(1 / 2) spectrum, the enhancement effect of the block sparse DEMON spectrum is more significant.

[0114] Under the octave detection method of the present invention, control a certain signal-to-noise ratio and estimate the target fundamental frequency multiple times. The fundamental frequency detection error of the traditional 1(1 / 2) spectrum is 11.28 Hz. Under the same conditions, the fundamental frequency detection error of the block sparse is 4.90 Hz, and the error is reduced by 6.38 Hz.

[0115] The present invention also provides an electronic device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of the method for enhancing the ship radiated noise DEMON spectrum based on block sparse and the octave detection method are implemented.

[0116] The present invention also provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the steps of the method for enhancing the ship radiated noise DEMON spectrum based on block sparse and the octave detection method are implemented.

[0117] The memory in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read only memory (ROM), a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct rambus RAM (DRRAM). It should be noted that the memory of the method described in the present invention is intended to include but not limited to these and any other suitable types of memory.

[0118] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wirelessly (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that the computer can access or a data storage device such as a server or data center that includes one or more integrated available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a high-density digital video disc (DVD)), or a semiconductor medium (such as a solid state disc (SSD)), etc.

[0119] In the implementation process, the steps of the above method can be completed by the integrated logic circuit of the hardware in the processor or the instructions in the form of software. The steps of the method disclosed in the embodiments of the present application can be directly embodied as being executed by the hardware processor or completed by a combination of the hardware and software modules in the processor. The software module can be located in a mature storage medium in the art such as a random access memory, flash memory, read-only memory, programmable read-only memory, or electrically erasable programmable memory, register, etc. This storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method. To avoid repetition, it will not be described in detail here.

[0120] It should be noted that the processor in the embodiments of the present application can be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method embodiments can be completed by the integrated logic circuit in the hardware of the processor or instructions in software form. The above processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as being executed and completed by the hardware decoding processor, or executed and completed by a combination of the hardware and software modules in the decoding processor. The software module can be located in a mature storage medium in the art such as random access memory, flash memory, read-only memory, programmable read-only memory, or electrically erasable programmable memory, registers, etc. This storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method.

[0121] The above has introduced in detail a method based on block sparse enhancement of the DEMON spectrum of ship radiated noise and a frequency doubling detection method proposed by the present invention. Specific examples are used in this article to elaborate on the principle and implementation manner of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for enhancing the DEMON spectrum of ship radiated noise based on block sparsity, characterized in that The method includes the following steps: Step 1: Use DEMON to process the radiated noise of the ship to obtain the low-frequency rate spectrum information of the radiated noise of the ship. After low-pass filtering the low-frequency rate spectrum information, perform square detection. The signal after square detection is denoted as the observation signal; Step 2: Determine the range of the fundamental frequency from the line spectrum position of the DEMON spectrum fundamental frequency, construct a block sparse dictionary matrix, where each column element in the dictionary matrix matches the frequency domain characteristics of the signal after low-pass filtering of the DEMON spectrum, and solve for the weight vector; Step 3: Use the weight vector to re-cross multiply the constructed dictionary matrix to obtain the estimated time-domain signal to achieve the purpose of time-domain reconstruction of the signal; Step 4: Perform low-pass filtering on the reconstructed signal and perform FFT to obtain the enhanced DEMON spectrum.

2. The method according to claim 1, wherein In Step 1, the specific process of square detection is as follows: When the carrier is a single-frequency signal, the modulation signal can be written as: y(t) = (1 + mcosΩt)cosωt where m is the modulation degree, Ω is the modulation frequency, ω is the carrier frequency, and square process the modulation signal: Denote the signal after square detection as Y, which is the observation signal to be solved subsequently.

3. The method according to claim 2, wherein In the construction of the dictionary matrix, use the harmonic signals in the time domain as the elements of the matrix, and these harmonic signals have different fundamental frequency information; assume that the line spectrum at 4 times the frequency point needs to be estimated subsequently, then the elements in the dictionary matrix can be defined as: y n = cos(2πf n ) + cos(2π2f n ) + cos(2π3f n ) + cos(2π4f n ) where, f n is different target fundamental frequencies, and f n is determined by the setting range of the low-pass filtering frequency band in the DEMON spectrum.

4. The method according to claim 3, wherein In the process of solving the weight vector, the observation signal Y is expressed as: Y = Za + e where Z is the dictionary matrix, a is the weight vector, assume e is additive white Gaussian noise, and at this time, the solution problem can be understood as a linear least squares problem approximated by convex relaxation; subject to ||Y - Za||2 ≤ δ In the formula, ||·||1 is the vector 1-norm, which refers to the sum of the absolute values of each element in the vector; ||·||2 is the vector 2-norm, which is the square root of the sum of the squares of each element in the matrix; where δ ≥ ||e||2; this problem can be equivalent to the existence of λ > 0, and there is The above problem is the LASSO algorithm. Based on the LASSO algorithm, solve for the weight vector a in the noise, and adjusting λ can control the sparsity of the solution.

5. A method for octave detection to enhance the DEMON spectrum of the radiated noise of a ship according to claim 1, characterized in that, The method includes the following steps: Step 1: Group the enhanced multiple-frequency line spectrum information; Step 2: Set a threshold for the enhanced line spectrum, and screen out the line spectrum components with low energy; Step 3: Retrieve the frequency components of the line spectrum after threshold screening: If there are corresponding multiple-frequency components subsequently, extract this group of multiple frequencies and their corresponding energy components, and denote them as a group of multiple frequencies; if there are no corresponding multiple-frequency components subsequently, discard this frequency; Step 4: Sum the energy of each multiple-frequency group, sort the total energy, and consider the fundamental frequency corresponding to the multiple-frequency group with the largest energy as the target.

6. The method according to claim 5, wherein In Step 1, the energy of each frequency point of the enhanced DEMON spectrum is expressed as: E[f] = |X[f]| 2 Where E[f] is the energy of each frequency point, and the energy is represented in a normalized form; the reconstructed frequency-domain signal is denoted as X[f], and the enhanced line-spectrum energy is an array of a certain length, denoted as A E α n is the enhanced normalized line spectrum energy, and n is the total length of data points; the corresponding frequency array is denoted as A f The two are combined into a new matrix denoted as M0 M0 = [A E A f n×2 。​ 7. The method according to claim 6, wherein Set a threshold value to find vector A E For the numbers in it that are less than this threshold value, discard the row numbers where these values are located; the size of the threshold value is set according to requirements; the new two-dimensional matrix is M1, assuming there are still m rows of elements left: M1 = [A E A f m×2 ​ Traverse the remaining frequency points of A in M1 f ​ 8. The method according to claim 7, characterized in that In step four, the energy summation of the classified multiple-frequency groups is denoted as A En , and the summation result of each group is stored in the array {A E1 A E2 … … A En}. For the elements in the array, they are arranged, and a line chart of the energy-fundamental frequency relationship is drawn. The fundamental frequency corresponding to the highest energy point is the target fundamental frequency.

9. An electronic device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1-8.

10. A computer-readable storage medium for storing computer instructions, characterized in that, When the computer instruction is executed by the processor, it implements the steps of the method according to any one of claims 1-8.

Images