A method and system for identifying radiation noise of underwater targets

Abstract

The invention discloses a method and system for identifying radiation noise of an underwater target. The method comprises the following steps: obtaining radiation noise of the same underwater target in different working conditions and different underwater targets in the same working condition, performing compressed sensing processing on the radiation noise signal, performing empirical mode decomposition on the processed radiation noise signal to obtain a plurality of eigenmode components; calculating the mutual information entropy between each of the eigenmode components, and obtaining the real number k of eigenmode components by using the local minimum value of the mutual information entropy; performing variational mode decomposition on the processed radiation noise signal based on the number k of eigenmode components, and obtaining the central frequency of the maximum energy eigenmode component; and constructing a feature set of radiation noise of an underwater target by using the central frequency for identification; effectively avoiding mode aliasing and boundary effect of empirical mode decomposition, improving the disadvantage of variational mode decomposition that relies on prior knowledge, and improving the adaptability of the underwater target identification method.

Description

Technical Field

[0001] The invention belongs to the field of underwater acoustic signal processing, and relates to a method for identifying radiation noise of an underwater target, and in particular to a method for identifying radiation noise of an underwater target based on compressed sensing and center frequency. Background Art

[0002] In the field of underwater acoustic signal processing, the radiation noise of underwater targets such as ships contains rich ship information, which is of great significance for the identification of underwater targets. However, due to the complex generation mechanism, diverse components, and accompanied by strong environmental noise and obvious amplitude modulation components, it is very difficult to process the radiation noise of underwater targets. Traditional methods rely too much on operators and prior knowledge, which seriously reduces the recognition efficiency and accuracy of radiation noise of underwater targets. It is of great significance to develop an adaptive feature extraction method for radiation noise of underwater targets to improve the recognition of radiation noise of underwater targets. Summary of the invention

[0003] In order to solve the problems existing in the prior art, the present invention provides a method for identifying underwater target radiated noise based on compressed sensing and center frequency, which effectively avoids the mode aliasing and boundary effect of empirical mode decomposition, improves the disadvantage of variational mode decomposition relying on prior knowledge, and improves the adaptability of the underwater target recognition method.

[0004] In order to achieve the above object, the technical solution adopted by the present invention is: a method for identifying radiated noise of underwater targets, comprising the following steps:

[0005] Acquire radiation noise of the same underwater target under different working conditions and of different underwater targets under the same working conditions, perform compressed sensing processing on the radiation noise signal, and enhance the line spectrum energy of the radiation noise signal;

[0006] Perform empirical mode decomposition on the radiation noise signal after compressed sensing processing to obtain several eigenmode components;

[0007] Calculating the mutual information entropy between each of the eigenmode components, and obtaining the true number k of eigenmode components using the local minimum value of the mutual information entropy;

[0008] The radiated noise signal after the compressed sensing processing is subjected to variational mode decomposition based on the number of eigenmode components k, and the center frequency of the eigenmode component with the maximum energy is obtained;

[0009] The central frequency of the maximum energy eigenmode component is used to construct the characteristic set of radiated noise of underwater targets for identification.

[0010] For the original water target radiation noise signal x∈R N×1 Perform compressed sensing processing and use the observation matrix Get the observation vector y∈RM×1 , so that The following steps are involved:

[0011] S11, suppose a set of orthogonal basis (Ψi is an N-dimensional column vector), the original signal x∈R N×1 Expanding on the reorganized orthogonal basis, we obtain in is the expansion coefficient, we can get x = Ψθ;

[0012] S12, using the observation matrix For the original signal x∈R N×1 During the sampling process, we can get y = ΦΨθ = A cs θ, where A cs is the compressed sensing information operator;

[0013] S13, since θ is sparse, reconstruct θ from y, and then recover the original signal x∈R according to S11 N×1 .

[0014] The radiated noise of underwater targets after compressed sensing is processed by empirical mode decomposition, and the radiated noise of underwater targets after compressed sensing is decomposed to obtain N intrinsic mode components IMF. N ;

[0015] The empirical mode decomposition first finds all the maximum points of the target radiation noise in the water, and uses the cubic spline interpolation function to fit the envelope of all the maximum values ​​of the original data ml MAX Then find all the minimum points of the target radiation noise in the water, and use the cubic spline interpolation function to fit the envelope of all the minimum values ​​of the original data ml min , the envelope of the maximum value ml MAX and the minimum envelope ml min Calculate the mean ml, calculate the difference between ml and the original data, and obtain the new data sequence IMF; the algorithm ends when there are no local maximum and local minimum.

[0016] For each adjacent eigenmode component IMF N and IMF N+1 Calculate its mutual information entropy, where k is the eigenmode component index of the first local minimum of mutual information entropy.

[0017] The k-based variational mode decomposition is performed on the radiated noise of underwater targets after compressed sensing processing to obtain k eigenmode components. Among the k eigenmode components, the eigenmode component with the largest energy is selected, and its center frequency is used as the feature.

[0018] Using variational constraint models The narrowband IMF is obtained by adaptively segmenting the signal band in the frequency domain based on the signal itself, where {u k}:={u1,u2,...u K}——K IMF components; {ω k}:={ω1,ω2,...,ω K}——center frequency of each IMF; represents the sum of all modes.

[0019] Variational mode decomposition adopts the alternating direction multiplier method to continuously update each mode and its center frequency, gradually demodulate each mode to the corresponding basic frequency band, and finally extract the center frequency of each mode. The center frequency of the maximum energy eigenmode component after variational mode decomposition is used to classify and identify underwater targets.

[0020] The present invention also provides an underwater target radiation noise identification system, comprising a compressed sensing module, an empirical mode decomposition module, a mutual information entropy calculation module and a variational mode decomposition module;

[0021] The compressed sensing module is used to detect the original target radiation noise signal x∈R in water. N×1 Perform compressed sensing processing to enhance its line spectrum energy;

[0022] The empirical mode decomposition module is used to process the underwater target radiation noise signal processed by the compressed sensing module to obtain several intrinsic mode components IMF N :

[0023] The mutual information entropy calculation module is used to calculate the intrinsic mode component IMF after the empirical mode decomposition module is processed N Calculate the mutual information entropy of two adjacent eigenmode components, and take the first local minimum value as the true eigenmode component number k;

[0024] The variational mode decomposition module is used to perform k-based variational mode decomposition on the underwater target radiation noise signal processed by the compressed sensing module to obtain k eigenmode components, and extract the center frequency of the maximum energy eigenmode component as the eigenvalue to construct a feature set for underwater target recognition.

[0025] A computer device comprises a processor and a memory, wherein the memory is used to store a computer executable program, the processor reads the computer executable program from the memory and executes it, and the method for identifying radiated noise of an underwater target according to the present invention can be realized when the processor executes the computer executable program.

[0026] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for identifying radiated noise of an underwater target described in the present invention can be implemented.

[0027] Compared with the prior art, the present invention has at least the following beneficial effects:

[0028] The compressed sensing and center frequency underwater target radiation noise recognition method of the present invention first performs compressed sensing processing on the original underwater target radiation noise and enhances the signal line spectrum. Different from the noise reduction processing of the existing method, the information contained in the signal is effectively retained; the number of real eigenmode components is obtained by calculating the empirical mode decomposition and mutual information entropy, and finally the feature set is constructed by using variational mode decomposition and the center frequency of the maximum energy eigenmode component as the eigenvalue, which effectively avoids the mode aliasing and boundary effect of the empirical mode decomposition, improves the disadvantage of the variational mode decomposition relying on prior knowledge, and improves the adaptability of the underwater target recognition method. The present invention has broad application prospects in the recognition of underwater targets such as ships. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a flow chart of the method for identifying radiation noise of underwater targets of the present invention;

[0030] Figure 2 It is the result after decomposition of the empirical mode of the radiated noise of the underwater target in the present invention;

[0031] Figure 3 It is the result after the variational mode decomposition of the radiation noise of the underwater target in the present invention;

[0032] Figure 4a It is the center frequency result of the maximum energy eigenmode component of the target radiated noise in untreated water;

[0033] Figure 4b It is the result of the center frequency of the maximum energy eigenmode component of the target radiation noise in water processed by the method of the present invention. DETAILED DESCRIPTION

[0034] The present invention is further described with reference to the accompanying drawings and specific embodiments:

[0035] like Figure 1 As shown, a method for identifying radiated noise of underwater targets based on compressed sensing and center frequency includes the following steps:

[0036] Step 1 obtains the radiation noise of the same underwater target under different working conditions and the radiation noise of different underwater targets under the same working conditions, and performs compressed sensing processing on the corresponding radiation noise signals to enhance their line spectrum energy.

[0037] Step 2: Perform empirical mode decomposition on the radiated noise signal after compressed sensing processing to obtain several eigenmode components.

[0038] Step 3 calculates the mutual information entropy between each eigenmode component, and uses the local minimum value of the mutual information entropy to obtain the actual number K of eigenmode components.

[0039] Step 4 performs variational mode decomposition on the radiated noise signal after the compressed sensing processing based on K, and obtains the center frequency of its maximum energy eigenmode component.

[0040] Step 5 uses the center frequency of the maximum energy eigenmode component to construct a feature set of radiated noise of underwater targets for identification.

[0041] The radiated noise of the same underwater target under different working conditions and different underwater targets under the same working conditions are obtained through the hydrophone, where the parameter information of the underwater target is shown in Table 1.

[0042] Original target radiation noise signal in water x∈R N×1 Perform compressed sensing processing and use the observation matrix Get the observation vector y∈R M×1 , so that The following steps are involved:

[0043] Step 1) Set a set of orthogonal bases (Ψi is an N-dimensional column vector), the original signal x∈R N×1 Expanding on this set of orthogonal bases, we obtain in As the expansion coefficient, we can get x=Ψθ.

[0044] Step 2) Using the observation matrix For the original signal x∈R N×1 During the sampling process, we can get y = ΦΨθ = A cs θ, where A cs It is a compressed sensing information operator.

[0045] Step 3) Since θ is sparse, θ can be reconstructed from y. After obtaining θ, the original signal x∈R can be finally restored according to step 1). N×1 .

[0046] like Figure 2 As shown in the figure, empirical mode decomposition is used to process the radiated noise of the underwater target after compressed sensing to obtain N intrinsic mode components IMF N ;

[0047] The empirical mode decomposition first finds all the maximum points of the target radiation noise in the water, and uses the cubic spline interpolation function to fit the envelope of all the maximum values ​​of the original data ml MAX Then find all the minimum points of the target radiation noise in the water, and use the cubic spline interpolation function to fit the envelope of all the minimum values ​​of the original data ml min, the envelope of the maximum value ml MAX and the minimum envelope ml min Calculate the mean ml, calculate the difference between ml and the original data, and obtain the new data sequence IMF; until there are no local maximum and local minimum, obtain N intrinsic mode component IMF N .

[0048] As shown in Table 2, for each adjacent eigenmode component IMF N and IMF N+1 Calculate its mutual information entropy, where k is the eigenmode component index of the first local minimum of mutual information entropy.

[0049] like Figure 3 As shown, k-based variational mode decomposition is performed on the underwater target radiation noise after compressed sensing processing to obtain k eigenmode components. The eigenmode component with the largest energy is selected from the k eigenmode components, and its center frequency is used as the feature.

[0050] like Figure 4a and Figure 4b As shown, the center frequency of the maximum energy eigenmode component after variational mode decomposition is used to identify underwater targets.

[0051] Using variational constraint models The narrowband IMF is obtained by adaptively segmenting the signal band in the frequency domain based on the signal itself, where {u k}:={u1,u2,...u K}——K IMF components; {ω k}:={ω1,ω2,...,ω K}——center frequency of each IMF; represents the sum of all modes.

[0052] Variational mode decomposition adopts the alternating direction multiplier method to continuously update each mode and its center frequency, gradually demodulate each mode to the corresponding basic frequency band, and finally extract the center frequency of each mode. The center frequency of the maximum energy eigenmode component after variational mode decomposition can be used to classify and identify underwater targets.

[0053] The present invention also provides an underwater target radiation noise identification system based on compressed sensing and center frequency, comprising a compressed sensing module, an empirical mode decomposition module, a mutual information entropy calculation module and a variational mode decomposition module;

[0054] The compressed sensing module is used to calculate the original water target radiation noise signal x∈R N×1 Compressed sensing processing is performed to enhance its line spectrum energy.

[0055] The empirical mode decomposition module processes the underwater target radiation noise signal processed by the compressed sensing module to obtain several intrinsic mode components IMF N .

[0056] The mutual information entropy calculation module calculates the intrinsic mode component IMF after the empirical mode decomposition module processes it N The mutual information entropy of two adjacent eigenmode components is calculated, and the first local minimum value is taken as the true eigenmode component number k.

[0057] The variational mode decomposition module performs k-based variational mode decomposition on the underwater target radiation noise signal processed by the compressed sensing module to obtain k eigenmode components, and extracts the center frequency of the maximum energy eigenmode component as the eigenvalue to construct a feature set for underwater target recognition.

[0058] In addition, the present invention can also provide a computer device, including a processor and a memory, the memory is used to store a computer executable program, the processor reads part or all of the computer executable program from the memory and executes it, and when the processor executes part or all of the computer executable program, it can implement the underwater target radiation noise identification method based on compressed sensing and center frequency described in the present invention.

[0059] On the other hand, the present invention provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the method for identifying underwater target radiation noise based on compressed sensing and center frequency described in the present invention can be implemented.

[0060] The computer device may be a laptop computer, a desktop computer or a workstation.

[0061] The processor may be a central processing unit (CPU), a digital signal processor (DSP), an application specific integrated circuit (ASIC), or an off-the-shelf field programmable gate array (FPGA).

[0062] The memory of the present invention may be an internal storage unit of a laptop computer, a desktop computer or a workstation, such as a memory or a hard disk; or an external storage unit, such as a mobile hard disk or a flash memory card.

[0063] Computer-readable storage media may include computer storage media and communication media. Computer storage media include volatile and non-volatile, removable and non-removable media implemented by any method or technology for storing information such as computer-readable instructions, data structures, program modules or other data. Computer-readable storage media may include: read-only memory (ROM), random access memory (RAM), solid-state drive (SSD) or optical disk, etc. Among them, random access memory may include resistance random access memory (ReRAM) and dynamic random access memory (DRAM).

Claims

1. A method for identifying radiation noise of underwater targets, characterized in that: The following steps are involved: Acquire radiation noise of the same underwater target under different working conditions and of different underwater targets under the same working conditions, perform compressed sensing processing on the radiation noise signal, and enhance the line spectrum energy of the radiation noise signal; Performing empirical mode decomposition on the radiation noise signal after compressed sensing processing to obtain several eigenmode components; Calculating the mutual information entropy between each of the eigenmode components, and obtaining the true number k of eigenmode components using the local minimum value of the mutual information entropy; The radiated noise signal after the compressed sensing process is subjected to variational mode decomposition based on the number of eigenmode components k, and the center frequency of the eigenmode component with the maximum energy is obtained; The central frequency of the maximum energy eigenmode component is used to construct the characteristic set of radiation noise of underwater targets for identification. The radiated noise of underwater targets after compressed sensing is processed by empirical mode decomposition, and the radiated noise of underwater targets after compressed sensing is decomposed to obtain N intrinsic mode components IMF. N ; The empirical mode decomposition first finds all the maximum points of the target radiation noise in the water, and uses the cubic spline interpolation function to fit the envelope of all the maximum values ​​of the original data ml MAX Then find all the minimum points of the target radiation noise in the water, and use the cubic spline interpolation function to fit the envelope of all the minimum values ​​of the original data ml min , the envelope of the maximum value ml MAX and the minimum envelope ml min Calculate the mean ml, calculate the difference between ml and the original data, and obtain the new data sequence IMF; the algorithm ends when there are no local maximum and local minimum values; for each adjacent intrinsic mode component IMF N and IMF N+1 Calculate its mutual information entropy, where k is the eigenmode component index of the first local minimum of mutual information entropy.

2. The method for identifying radiated noise of underwater targets according to claim 1, characterized in that: For the original water target radiation noise signal x∈R N×1 Perform compressed sensing processing and use the observation matrix Get the observation vector y∈R M×1 , so that The following steps are involved: S11, suppose a set of orthogonal basis (Ψi is an N-dimensional column vector), the original signal x∈R N×1 Expanding on the reorganized orthogonal basis, we obtain in is the expansion coefficient, we can get x = Ψθ; S12, using the observation matrix For the original signal x∈R N×1 During the sampling process, we can get y = ΦΨθ = A cs θ, where A cs is the compressed sensing information operator; S13, since θ is sparse, reconstruct θ from y, and then recover the original signal x∈R according to S11 N×1 .

3. The method for identifying radiated noise of underwater targets according to claim 1, characterized in that: The k-based variational mode decomposition is performed on the underwater target radiation noise after compressed sensing processing to obtain k eigenmode components. The eigenmode component with the largest energy is selected among the k eigenmode components, and its center frequency is used as the feature.

4. The method for identifying radiated noise of underwater targets according to claim 1, characterized in that: Using variational constraint models The narrowband IMF is obtained by adaptively segmenting the signal band in the frequency domain based on the signal itself, where {u k }:={u1,u2,...u K }——K IMF components; {ω k }:={ω1,ω2,...,ω K }——center frequency of each IMF; represents the sum of all modes.

5. The method for identifying radiated noise of underwater targets according to claim 1, characterized in that: Variational mode decomposition adopts the alternating direction multiplier method to continuously update each mode and its center frequency, gradually demodulate each mode to the corresponding basic frequency band, and finally extract the center frequency of each mode. The center frequency of the maximum energy eigenmode component after variational mode decomposition is used to classify and identify underwater targets.

6. A system for identifying radiated noise of underwater targets, characterized in that: It includes a compressed sensing module, an empirical mode decomposition module, a mutual information entropy calculation module and a variational mode decomposition module; The compressed sensing module is used to detect the original target radiation noise signal x∈R in water. N×1 Perform compressed sensing processing to enhance its line spectrum energy; The empirical mode decomposition module is used to process the underwater target radiation noise signal processed by the compressed sensing module to obtain several intrinsic mode components IMF N : The mutual information entropy calculation module is used to calculate the intrinsic mode component IMF after the empirical mode decomposition module is processed N Calculate the mutual information entropy of two adjacent eigenmode components, and take the first local minimum value as the true eigenmode component number k; The variational mode decomposition module is used to perform k-based variational mode decomposition on the underwater target radiation noise signal processed by the compressed sensing module to obtain k intrinsic mode components, and extract the center frequency of the maximum energy intrinsic mode component as the eigenvalue to construct a feature set for underwater target recognition; the underwater target radiation noise after compressed sensing processing is processed by empirical mode decomposition, and the underwater target radiation noise after compressed sensing processing is decomposed to obtain N intrinsic mode components IMF N ; The empirical mode decomposition first finds all the maximum points of the target radiation noise in the water, and uses the cubic spline interpolation function to fit the envelope of all the maximum values ​​of the original data ml MAX Then find all the minimum points of the target radiation noise in the water, and use the cubic spline interpolation function to fit the envelope of all the minimum values ​​of the original data ml min , the envelope of the maximum value ml MAX and the minimum envelope ml min Calculate the mean ml, calculate the difference between ml and the original data, and obtain the new data sequence IMF; the algorithm ends when there are no local maximum and local minimum values; for each adjacent intrinsic mode component IMF N and IMF N+1 Calculate its mutual information entropy, where k is the eigenmode component index of the first local minimum of mutual information entropy.

7. A computer device, characterized in that: The invention comprises a processor and a memory, wherein the memory is used to store a computer executable program, the processor reads the computer executable program from the memory and executes it, and when the processor executes the computer executable program, the method for identifying radiated noise of an underwater target as claimed in any one of claims 1 to 5 can be implemented.

8. A computer-readable storage medium, characterized in that: A computer program is stored in the computer-readable storage medium. When the computer program is executed by a processor, the method for identifying radiated noise of an underwater target according to any one of claims 1 to 5 can be implemented.

Images