A method and system for identifying coherent structures in vertical matrix three-dimensional space

Abstract

The present invention belongs to the technical field of marine turbulence microstructure identification, and discloses a method and system for identifying coherent structures in a vertical matrix three-dimensional space. The present invention analyzes and mines the coherent structural feature information contained in four groups of shear data obtained by a matrix observation instrument, and realizes the identification of coherent structures in a vertical matrix three-dimensional space driven by data. Specifically, for each group of shear data, coherent components are screened to screen out effective structural information in the nonlinear shear data, and a total of four groups of coherent characteristic fluctuation signals are obtained. The four groups of coherent characteristic fluctuation signals contain the main coherent structural information, which can provide data support for coherent core energy extraction and coherent structure identification. The cross-wavelet spectrum is used to analyze the coherent core scales corresponding to the four groups of coherent characteristic fluctuation signals, and the coherent core energy corresponding to the scale is extracted. The construction of a two-dimensional energy space fluctuation plane is completed based on the four groups of coherent core energy fluctuations, and the identification of coherent structures is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of ocean turbulence microstructure recognition, and in particular relates to a method and system for recognizing coherent structures in a vertical matrix three-dimensional space. Background Art

[0002] The ocean turbulence signal is a nonlinear and unstable signal that changes all the time. The turbulence signal is composed of the superposition of multi-scale coherent structure signals, which contains rich physical ocean information. At the same time, the multi-scale coherent structure has an important driving effect on the evolution of turbulence. The large-scale coherent structure dominates the transport of energy, momentum and matter in the turbulent water body, and the small-scale coherent structure dominates the mixing of turbulent water bodies. Therefore, the effective identification and analysis of coherent structures can promote the control and prediction of turbulent water bodies, and provide theoretical support for human beings to improve their understanding of the mechanism of physical ocean evolution. At present, the identification and mining of coherent structures is mainly concentrated in the laboratory and numerical simulation. The coherent structure is the "skeleton" of the evolution of water bodies. It exists in the water body in a three-dimensional form. At the same time, the coherent structure is in an unstable state in the water body, and it promotes the evolution of the water body in a continuous stretching and deformation manner. Traditional coherent structure research mainly relies on two-dimensional particle image velocimetry (PIV), which cuts three-dimensional coherent structures into pieces for identification and records the development direction and speed of particles in the water at high resolution. This technology provides strong technical support for the traditional identification of coherent structures, but it limits the mining of full-dimensional physical information of coherent structures. In addition, numerical simulation can provide huge data support for the identification of coherent structures, but it lacks the real evolution characteristics of the marine environment to a certain extent. In addition, some researchers have tried to use flexible observation instruments to mine coherent structures in the ocean, such as satellites. However, satellite data can only record and collect coherent structure data on the ocean surface, and track the horizontal evolution of coherent structures on the ocean surface. It is impossible to obtain coherent structure information in the deep ocean, and thus cannot provide comprehensive dimensional support for the mining of coherent structures. Summary of the invention

[0003] The purpose of the present invention is to propose a method for identifying coherent structures in a vertical matrix three-dimensional space, which is based on CEEMD multi-scale decomposition, coherent component extraction and energy evolution plane support to achieve the identification of coherent structures in the ocean vertical matrix water space. In order to achieve the above purpose, the present invention adopts the following technical solutions:

[0004] A method for identifying coherent structures in a vertical matrix three-dimensional space comprises the following steps:

[0005] Step 1. Obtain the shear signals of four sets of matrix vertical sections and perform denoising preprocessing;

[0006] Among them, the shear signals of the four sets of matrix vertical profiles are obtained during the free-falling observation process by the four sets of matrix-arranged shear probes carried by the matrix profile observation instrument for ocean turbulence observation;

[0007] Step 2. Relying on CEEMD multi-scale decomposition, combined with wavenumber spectrum analysis and multivariate parameter threshold analysis methods, the coherent structural components in the four groups of shear signals after denoising are screened to obtain effective characteristic signal components, and all effective characteristic signal components in each group of shear signals are reconstructed to obtain a total of four groups of coherent characteristic fluctuation signals;

[0008] Step 3. Pair the four groups of coherent characteristic fluctuation signals obtained in step 2, calculate the correlation between each pair of coherent characteristic fluctuation signals based on the cross wavelet spectrum, and obtain the coherent core scale;

[0009] The wavelet energy fluctuations corresponding to the four groups of coherent characteristic fluctuation signals at the coherent core scale are extracted respectively;

[0010] Step 4. Based on four sets of wavelet energy fluctuations, a two-dimensional energy fluctuation plane is established to identify coherent structures.

[0011] In addition, based on the above-mentioned vertical matrix three-dimensional space coherent structure recognition method, the present invention also proposes a corresponding vertical matrix three-dimensional space coherent structure recognition system, which adopts the following technical solutions:

[0012] A coherent structure recognition system in a vertical matrix three-dimensional space, comprising:

[0013] A preprocessing module is used to obtain the shear signals of four sets of matrix vertical sections and perform denoising preprocessing;

[0014] Among them, the shear signals of the four sets of matrix vertical profiles are obtained during the free-falling observation process by the four sets of matrix-arranged shear probes carried by the matrix profile observation instrument for ocean turbulence observation;

[0015] The coherent component screening module is used to screen the coherent structural components in the four groups of shear signals after denoising based on CEEMD multi-scale decomposition, combined with wave number spectrum analysis and multivariate parameter threshold analysis methods, to obtain effective characteristic signal components, and to reconstruct all effective characteristic signal components in each group of shear signals, to obtain a total of four groups of coherent characteristic fluctuation signals;

[0016] The coherent core energy extraction module is used to pair the four groups of coherent characteristic fluctuation signals obtained, calculate the correlation between each pair of coherent characteristic fluctuation signals based on the cross wavelet spectrum, and obtain the coherent core scale;

[0017] The wavelet energy fluctuations corresponding to the four groups of coherent characteristic fluctuation signals at the coherent core scale are extracted respectively;

[0018] And a coherent structure recognition module is used to establish a two-dimensional space energy fluctuation plane based on the four groups of wavelet energy fluctuations extracted by the coherent core energy extraction module to identify the coherent structure.

[0019] In addition, based on the above-mentioned method for identifying coherent structures in a vertical matrix stereoscopic space, the present invention also proposes a computer device for implementing the above-mentioned method for identifying coherent structures in a vertical matrix stereoscopic space.

[0020] The computer device comprises a memory and a processor. The memory stores executable codes. When the processor executes the executable codes, the computer device is used to implement the steps of the above-mentioned method for identifying coherent structures in a vertical matrix stereoscopic space.

[0021] In addition, based on the above-mentioned method for identifying coherent structures in a vertical matrix stereoscopic space, the present invention also proposes a computer-readable storage medium for implementing the above-mentioned method for identifying coherent structures in a vertical matrix stereoscopic space.

[0022] The computer-readable storage medium stores a program thereon, and when the program is executed by a processor, it is used to implement the steps of the above-mentioned method for identifying coherent structures in a vertical matrix three-dimensional space.

[0023] The present invention has the following advantages:

[0024] As described above, the present invention describes a method for identifying coherent structures in a vertical matrix three-dimensional space. The method is based on a coherent structure identification method based on multi-scale decomposition, coherent component extraction and energy evolution plane support, and realizes the identification of coherent structures in the water space of the ocean vertical matrix, extracts its spatial distribution information, and provides a structural model support for tracking the evolution mechanism of the coherent structure in the vertical space, further promoting the exploration and understanding of the field of ocean turbulence interaction and evolution mechanism. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 The figure is a recognition flow chart of a method for identifying coherent structures in a vertical matrix three-dimensional space in an embodiment of the present invention.

[0026] Figure 2 This is the first multi-scale decomposition eigenmode function component analysis screening diagram in an embodiment of the present invention.

[0027] Figure 3 This is a multi-parameter threshold analysis screening diagram of the second multi-scale decomposition of the intrinsic mode function component in an embodiment of the present invention. Figure 3 (a) in the figure corresponds to the sample entropy screening diagram corresponding to the eigenmode function. Figure 3(b) in the figure corresponds to the variance contribution rate screening diagram corresponding to the eigenmode function. Figure 3 (c) in the middle corresponds to the correlation coefficient screening diagram corresponding to the eigenmode function.

[0028] Figure 4 It is a schematic diagram of the expansion of the two-dimensional energy space wave plane in an embodiment of the present invention.

[0029] Figure 5 This is a diagram of the coherent structure recognition result of the present invention. DETAILED DESCRIPTION

[0030] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0031] Example 1

[0032] The embodiment of the present invention describes a method for identifying coherent structures in a vertical matrix three-dimensional space. The method for identifying coherent structures in a vertical matrix three-dimensional space relies on the gradual innovative development of ocean observation instruments, is supported by a multi-dimensional, synchronous, planar matrix turbulence observer, obtains multi-point matrix three-dimensional turbulence data, and uses multi-scale decomposition, wavenumber spectrum analysis, and multivariate parameter threshold analysis to achieve screening of coherent components contained in turbulence data and extraction of coherent core energy, construct a two-dimensional energy space fluctuation plane, and achieve identification of coherent structures in the vertical space of the matrix, thereby promoting the study of the evolution mechanism of coherent structures, which is of great value to the speculation of global ocean circulation and the improvement of transparent ocean cognition.

[0033] like Figure 1 The flow chart of coherent structure recognition using the method of the present invention is shown, and the general process is as follows:

[0034] The coherent structural feature information contained in the four groups of shear data obtained by the matrix observation instrument is analyzed and mined, and the identification of coherent structures in the vertical matrix three-dimensional space is realized by data-driven. Specifically, the coherent components of each group of shear data are screened to screen out the effective structural information in the nonlinear shear data, and a total of four groups of coherent characteristic fluctuation signals are obtained. The four groups of coherent characteristic fluctuation signals contain the main coherent structural information and can provide strong data support for the subsequent coherent core energy extraction and coherent structure identification. The cross wavelet spectrum is used to analyze the coherent core scales corresponding to the four groups of coherent characteristic fluctuation signals, and the coherent core energy corresponding to the coherent core scale is extracted. Based on the four groups of coherent core energy fluctuations, the construction of the two-dimensional energy space fluctuation plane is completed to achieve the identification and mining goals of the coherent structure.

[0035] The following is a detailed description of the steps of the method for identifying coherent structures in a vertical matrix stereoscopic space of the present invention. Figure 1As shown, the method for identifying coherent structures in a vertical matrix stereoscopic space in this embodiment includes the following steps:

[0036] Step 1. Observe ocean turbulence using a matrix profile observation instrument, which is equipped with four shear probes arranged in a matrix, named Bullet 1, Bullet 2, Bullet 3, and Bullet 4.

[0037] Relying on the free lowering observation process of four sets of shear probes, the shear signals of four sets of matrix vertical profiles are obtained, namely , and perform denoising preprocessing on the sheared data.

[0038] Step 1.1. Use the turbulence matrix profile observation instrument to be vertically lowered to obtain four sets of shear data. The four sets of shear data are distributed at the matrix edge positions and provide three-dimensional data in a matrix multi-point synchronous vertical detection mode.

[0039] Step 1.2. In the actual ocean observation experiment, the shear signal is collected by a high-sensitivity, high-precision shear probe. The collection process of the high-sensitivity turbulence shear probe is easily interfered by multiple noise sources in the ocean water, such as platform vibration, wave impact, plankton, etc.

[0040] In order to ensure the data quality of turbulent shear signals, the present invention combines the instrument attitude data and vibration acceleration profile data obtained synchronously during the lowering process of the observation instrument, and uses the Goodman algorithm to remove the noise disturbance signal in the shear data to obtain the denoised shear signal. , as the basis for coherent structure identification.

[0041] Step 2. Relying on CEEMD multi-scale decomposition, combined with wavenumber spectrum analysis and multivariate parameter threshold analysis methods, the coherent structural components in the four groups of shear signals after denoising are screened to obtain effective characteristic signal components, and all effective characteristic signal components in each group of shear signals are reconstructed to obtain a total of four groups of coherent characteristic fluctuation signals.

[0042] Step 2.1. Use CEEMD multi-scale decomposition to decompose the denoised clipped signal Perform the first multi-scale decomposition calculation and decompose it into Eigenmode function components:

[0043] .

[0044] in, Indicates eigenmode function components, represents the residual trend term.

[0045] Step 2.2. The turbulence signal is a multi-scale signal. The signal scale domain spans the large-scale energetic domain, the medium-scale inertial subdomain, and the small-scale dissipative domain. The effective coherent structure components mainly exist in the inertial subdomain and the energetic domain.

[0046] The present invention is to cut the signal The wave number spectrum analysis is performed with the eigenmode function components to analyze the central wave number information corresponding to each eigenmode function component, and the maximum cutoff wave number is defined as .

[0047] If the central wave number corresponding to the eigenmode function component is less than the maximum cutoff wave number , then the eigenmode function component is located in the range of the energy domain and the inertial subdomain, and its corresponding structure has a larger spatial scale in the actual space, and there is a prominent and dominant coherent structure, and the eigenmode function component is retained.

[0048] If the central wave number corresponding to the eigenmode function component is greater than or equal to the maximum cutoff wave number , then the intrinsic mode function component is located in the dissipative domain, and its corresponding structure has an extremely small spatial scale in the actual space, and there is no prominent, dominant coherent structure, and the intrinsic mode function component is eliminated.

[0049] Step 2.3. Based on the judgment of the maximum cutoff wave number, extract the waves less than the maximum cutoff wave number. All the eigenmode function components of , and reconstruct them based on this, and reconstruct the shear signal .

[0050] Step 2.4. In the first multiscale decomposition process, the signal components in the dissipation domain occupy an absolute proportion of the energy in the signal, which interferes with the decomposition result of the original signal to a certain extent. Therefore, CEEMD multiscale decomposition is used to reconstruct the sheared signal after the first multiscale decomposition. Perform a second multi-scale decomposition to refine the screening process, provide an effective signal component basis for multivariate parameter threshold analysis, and cut the signal After the second multi-scale decomposition, m intrinsic mode function components are obtained.

[0051] ;

[0052] in, Indicates eigenmode function components, is the residual trend term.

[0053] Step 2.5. Analyze the eigenmode function components The corresponding sample entropy , variance contribution rate , correlation coefficient Information, if the eigenmode function components simultaneously satisfy:

[0054] (1) Sample entropy , Represents the shear signal after reconstruction The corresponding sample entropy;

[0055] (2) The variance contribution rate satisfies V A F i ⊂ [ t 1 , 1 ] , Indicates the variance contribution rate threshold;

[0056] (3) The correlation coefficient satisfies c o r i ⊂ [ t 2 , 1 ] , Represents the correlation coefficient threshold.

[0057] Then the eigenmode function components If the signal is consistent with the coherent structure characteristics, it will be retained as a valid characteristic signal component; otherwise, if any of the above three parameters does not meet the valid threshold range, the intrinsic mode function component Those that do not conform to the coherent structure signal characteristics are discarded as invalid components.

[0058] Sample entropy is a complex measurement calculation method based on time series, which is used to characterize the degree of disorder of a piece of data. The larger the value, the more disordered the signal is. Conversely, the smaller the value, the more ordered the signal is.

[0059] Calculate separately The corresponding sample entropy SampEn and The corresponding sample entropy .

[0060] Reconstructed shear signal The corresponding sample entropy As the sample entropy judgment threshold.

[0061] If the eigenmode function components The corresponding sample entropy , it is considered that the eigenmode function component It is relatively orderly, conforms to the characteristics of coherent structure signals, and is retained.

[0062] If the eigenmode function components The corresponding sample entropy SampEn i ≥SampEn, it is considered that the eigenmode function component It is extremely disordered and does not conform to the characteristics of coherent structure signals, so it is eliminated.

[0063] The variance contribution rate is a parameter used to quantify the contribution of a component signal to the initial signal. The larger the value, the more important the component signal is in the initial signal. Conversely, the smaller the value, the less important the component signal is in the initial signal.

[0064] In this embodiment, the variance contribution rate threshold , for example, designed to be 0.2.

[0065] Calculate the eigenmode function components The corresponding variance contribution rate .

[0066] If the eigenmode function components The corresponding variance contribution rate lie in [ 0 . 2 , 1 ] Range, indicating the eigenmode function component Reconstructed shear signal The contribution degree is high, the component is more important, conforms to the characteristics of the coherent structure signal, and is retained as an effective characteristic component.

[0067] If the eigenmode function components The corresponding variance contribution rate lie in Range, indicating the eigenmode function component After reconstruction, the shear data The contribution is extremely low, the component is not important, and does not conform to the characteristics of the coherent structure signal. To be eliminated.

[0068] The correlation coefficient is a parameter used to characterize the degree of correlation between two sets of signal sequences. If the two sequences are strongly correlated, the correlation coefficient between them is high; conversely, if the two sequences are weakly correlated, the correlation coefficient between them is low.

[0069] According to the prior experimental results, the correlation coefficient threshold is set in this embodiment is 0.3.

[0070] Calculate the eigenmode function components With the reconstructed shear signal The corresponding correlation coefficient.

[0071] If the eigenmode function components With the reconstructed shear signal Correlation coefficient In [ t 2 , 1 ] Range, then the eigenmode function component With the reconstructed shear signal There is a high correlation between them, which conforms to the characteristics of coherent structure signals and is retained as an effective characteristic signal component.

[0072] If the eigenmode function components With the reconstructed shear signal Correlation coefficient In Range, then the eigenmode function component With the reconstructed shear signal The correlation between them is poor, which does not conform to the characteristics of coherent structure signals and is eliminated as invalid components.

[0073] Step 2.6. Filter all effective characteristic signal components for signal reconstruction, and reconstruct to obtain coherent characteristic fluctuation signals. Through this step, a total of four groups of coherent characteristic fluctuation signals are obtained, namely .

[0074] like Figure 2 The first multi-scale decomposition eigenmode function component analysis screening diagram of the present invention is shown. The present invention relies on CEEMD multi-scale decomposition to achieve multi-scale decomposition of the denoised shear signal. Take the shear data obtained by Bullet 1 in the 300-305 meter depth segment as an example: after the first multi-scale decomposition, 10 eigenmode function components are obtained. Compare the central wavenumber and the cutoff wavenumber corresponding to each eigenmode function component. If it is greater than or equal to the cutoff wavenumber, it means that the eigenmode function component is located in the dissipative domain, and the spatial scale corresponding to the structure in the actual space is extremely small. There is no prominent, dominant coherent structure, and the eigenmode function component needs to be eliminated; and if it is less than the cutoff wavenumber, it means that the eigenmode function component is located in the energy domain and the inertial subdomain, and the spatial scale corresponding to the structure in the actual space is relatively large. There is a prominent, dominant coherent structure, and the eigenmode function component needs to be retained. Among the 10 eigenmode function components obtained by decomposition, The corresponding central wave number is greater than the cutoff wave number and is located in the dissipative domain. These three eigenmode function components need to be eliminated. The corresponding central wave number is smaller than the cutoff wave number and is located in the energetic domain and the inertial subdomain. These seven eigenmode function components need to be retained and the shear data reconstructed. .

[0075] like Figure 3 The second multi-scale decomposition of the intrinsic mode function component multi-parameter threshold analysis screening diagram is shown. Among them, Figure 3 (a) in the figure corresponds to the sample entropy screening diagram corresponding to the eigenmode function. Figure 3 (b) in the figure corresponds to the variance contribution rate screening diagram corresponding to the eigenmode function. Figure 3(c) is the correlation coefficient screening diagram corresponding to the eigenmode function. Relying on the CEEMD algorithm, the shear data reconstructed after the first multi-scale decomposition and screening is subjected to a second multi-scale decomposition and screening to complete the extraction of the coherent characteristic fluctuation signal. Similarly, the reconstructed shear data corresponding to bullet 1 in the 300-305m depth range is taken. For example, in the second multi-scale decomposition process, a total of 10 eigenmode function components are obtained . Calculate the sample entropy, variance contribution rate and correlation coefficient of each of the 10 intrinsic mode function components, and compare them with their corresponding thresholds. In the sample entropy calculation results, the 10 intrinsic mode function components The sample entropy is smaller than the shear data after the first multi-scale decomposition screening and reconstruction. The corresponding sample entropy, so the 10 eigenmode function components All of them are consistent with the sample entropy threshold analysis. In the variance contribution rate calculation results, only the first three eigenmode function components The corresponding variance contribution rate is in the effective threshold range of [0.2,1], indicating that it is the shear data after the first multi-scale decomposition screening and reconstruction. The contribution is relatively high and it is a more important signal component; the remaining 7 eigenmode function components The corresponding variance contribution is in the range of [0,0.2), indicating that it contributes to the shear data after the first multi-scale decomposition screening and reconstruction. The contribution is very low and is an unimportant signal component. In the correlation coefficient calculation results, only the first three eigenmode function components The corresponding correlation coefficient is in the range of [0.3,1], indicating that it is consistent with the shear data reconstructed after the first multi-scale decomposition screening. There is a high degree of correlation between them, which is consistent with the characteristics of coherent structure signals; while the remaining 7 eigenmode function components The corresponding correlation coefficient is in the range of [0,0.3), indicating that it is consistent with the shear data reconstructed after the first multi-scale decomposition screening. The correlation between them is extremely poor, which does not conform to the characteristics of coherent structure signals. Threshold analysis is performed on the three parameters of sample entropy, variance contribution rate and correlation coefficient. Only the first three eigenmode function components Satisfying all threshold conditions, extracting the three components as valid coherent structural components, and reconstructing the coherent characteristic fluctuation signal of bullet 1 within the range of 300-305 meters .

[0076] Step 3. Pair the four groups of coherent characteristic fluctuation signals obtained in step 2, calculate the correlation between each pair of coherent characteristic fluctuation signals based on the cross wavelet spectrum, obtain the coherent core scale, and extract the wavelet energy fluctuations corresponding to the four groups of coherent characteristic fluctuation signals at the coherent core scale.

[0077] Step 3.1. Define four sets of coherent characteristic fluctuation signals as .

[0078] The coherent characteristic fluctuation signal is subjected to cross wavelet spectrum analysis, and the calculation formula is as follows:

[0079] .

[0080] in represents the wavelet cross spectrum, It is coherent characteristic fluctuation signal The corresponding wavelet coefficients are It is coherent characteristic fluctuation signal The corresponding wavelet coefficients are For conjugate calculations, is the integration length.

[0081] A total of six pairs of coherent characteristic fluctuation signals are obtained by pairing them two by two, which are: , , , , , .

[0082] For each pair of coherent characteristic fluctuation signals By performing cross wavelet spectrum analysis, a total of six cross wavelet spectra are calculated for the four sets of coherent characteristic fluctuation signals; among them, , .

[0083] Then, six cross wavelet spectra are calculated from the four groups of coherent characteristic fluctuation signals obtained in step 2.

[0084] Step 3.2. Based on each pair of coherent characteristic fluctuation signals The results of cross wavelet spectrum analysis are used to calculate the correlation mean corresponding to each scale in the cross wavelet spectrum and extract the correlation peak scale. .

[0085] Six cross wavelet spectrum analysis results were calculated for four sets of coherent characteristic fluctuation signals, and the correlation peak scale set was obtained. , which contains 6 correlation peak scales.

[0086] Step 3.3. Use K-means clustering method to scale the correlation peak obtained in step 3.2 The cluster is clustered and the center value of the cluster with the largest number of samples is used as the coherent core scale.

[0087] The scale of each correlation peak in the set In the clustering process, it is called a sample. The clustering process is as follows:

[0088] First, randomly select K samples as the initial cluster centers, calculate the distance between each sample and each cluster center, and assign each sample to the cluster center closest to the sample; where K < 3.

[0089] Each cluster center and the samples assigned to the corresponding cluster center represent a cluster.

[0090] If all samples are assigned, the cluster center of each cluster will be iteratively re-selected and the sample-cluster center distance will be calculated based on the existing samples in the cluster; if the cutoff condition that the cluster center no longer changes is met, the iterative process ends.

[0091] Correlation peak scale set Cluster into C clusters.

[0092] The cluster center value with the largest number of samples is used as the coherent core scale .

[0093] Step 3.4. Extract the wavelet energy fluctuations corresponding to the four groups of coherent characteristic fluctuation signals at the coherent core scale respectively , represents the wavelet energy fluctuation corresponding to the h-th group of coherent characteristic fluctuation signals at the coherent core scale, .

[0094] Step 4. Based on four sets of wavelet energy fluctuations, a two-dimensional energy fluctuation plane is established to identify coherent structures.

[0095] The wavelet energy fluctuations of the four sets of matrices arranged in three dimensions are used as the benchmark to unfold the cubic surface covered by the matrix space into a two-dimensional plane structure, with the depth data as the vertical coordinate and the bullet arrangement order as the horizontal coordinate, and coherent structure identification is performed based on the fluctuations in the two-dimensional energy space.

[0096] If there is an obvious high-energy continuous domain in the two-dimensional energy space fluctuation plane, then there is a coherent structure in this space, that is, the continuous domain with energy in the range of [1, ∞) is in the same coherent structure space range, and there is a coherent structure at this position.

[0097] like Figure 4 Figure 2 shows a schematic diagram of the expansion of the two-dimensional energy space wave plane. , , , They represent bullet 1, bullet 2, bullet 3 and bullet 4 respectively. Figure 4The left side shows the matrix arrangement order and the space domain covered by the matrix observation instrument during the free lowering process. Based on the method of the present invention, a two-dimensional energy fluctuation plane needs to be built. The vertical path of the falling bullet 1 is used as a benchmark to unfold the surface of the space cube to form a two-dimensional plane, such as Figure 4 As shown on the right, the distribution order of the expanded two-dimensional plane is Bullet 1 sequence, Bullet 2 sequence, Bullet 4 sequence, Bullet 3 sequence and repeated Bullet 1 sequence. The repeated Bullet 1 sequence is supplemented to form a closed domain on the two-dimensional plane, providing support for subsequent coherent structure identification.

[0098] Figure 5 The result diagram of the coherent structure identification of the present invention is shown. Relying on multiple coherent component screening of four groups of sheared data and extraction of coherent core energy, a two-dimensional energy space fluctuation plane is constructed. The method of the present invention can rely on the two-dimensional energy space fluctuation plane to effectively identify the coherent structure in the actual space. The coherent structure has the characteristics of synchronous evolution. Therefore, the two-dimensional energy space fluctuation plane constructed after screening is screened. If there is a high-energy continuous domain in the space, the continuous domain is identified as the same coherent structure, that is, the continuous domain with energy higher than 1 is in the same coherent structure space range.

[0099] Figure 5 There is an obvious high-energy continuous domain within the depth range of 303.8-304.3 meters. This continuous domain covers the entire spatial range of bullets 1-4, indicating that when the matrix observation instrument is lowered to this depth range, it is in an entire coherent structure, and the coherent structure can be identified in this depth domain. Figure 5 There is an obvious high-energy continuous domain within the depth range of 302.5-302.8 meters. This continuous domain only covers the spatial range of bullets 2-4, indicating that when the matrix observation instrument was lowered to this depth range, bullets 2-4 were in the same coherent structure, while bullet 1 was outside the coherent structure. Coherent structures can also be identified within this depth range.

[0100] According to the coherent structure identification method proposed in the present invention, the differentiated spatial distribution characteristics of coherent structures in different depth spaces can be effectively identified, which can provide powerful methodological support for the study of the physical evolution properties of coherent structures. At the same time, it is beneficial to promote the research progress of the interactive evolution of vortex-excited turbulence, thereby enhancing the overall understanding and control of the "transparent ocean".

[0101] The coherent structure identification method in the vertical matrix three-dimensional space proposed in the present invention can mine the coherent structure in the vertical ocean three-dimensional space. While ensuring the observation mobility, it realizes the dimensional expansion of the coherent structure cognition, improves the comprehensiveness of the coherent structure cognition, and provides strong result support for the subsequent research on the evolution mechanism of the coherent structure.

[0102] Example 2

[0103] This embodiment 2 describes a system for identifying coherent structures in a vertical matrix three-dimensional space. The system and the method for identifying coherent structures in a vertical matrix three-dimensional space described in the above embodiment 1 are based on the same inventive concept.

[0104] Specifically, the coherent structure recognition system in the vertical matrix three-dimensional space includes:

[0105] A preprocessing module is used to obtain the shear signals of four sets of matrix vertical sections and perform denoising preprocessing;

[0106] Among them, the shear signals of the four sets of matrix vertical profiles are obtained during the free-falling observation process by the four sets of matrix-arranged shear probes carried by the matrix profile observation instrument for ocean turbulence observation;

[0107] The coherent component screening module is used to screen the coherent structural components in the four groups of shear signals after denoising based on CEEMD multi-scale decomposition, combined with wave number spectrum analysis and multivariate parameter threshold analysis methods, to obtain effective characteristic signal components, and to reconstruct all effective characteristic signal components in each group of shear signals, to obtain a total of four groups of coherent characteristic fluctuation signals;

[0108] The coherent core energy extraction module is used to pair the four groups of coherent characteristic fluctuation signals obtained, calculate the correlation between each pair of coherent characteristic fluctuation signals based on the cross wavelet spectrum, and obtain the coherent core scale;

[0109] The wavelet energy fluctuations corresponding to the four groups of coherent characteristic fluctuation signals at the coherent core scale are extracted respectively;

[0110] And a coherent structure recognition module is used to establish a two-dimensional space energy fluctuation plane based on the four groups of wavelet energy fluctuations extracted by the coherent core energy extraction module to identify the coherent structure.

[0111] It should be noted that, in the vertical matrix three-dimensional space coherent structure recognition system, the implementation process of the functions and effects of each functional module is specifically described in the implementation process of the corresponding steps in the method in the above embodiment 1, and will not be repeated here.

[0112] Example 3

[0113] This embodiment 3 describes a computer device, which is used to implement the method for identifying coherent structures in a vertical matrix three-dimensional space described in the above embodiment 1.

[0114] Specifically, the computer device includes a memory and one or more processors. The memory stores executable codes, and when the processor executes the executable codes, the steps of the above-mentioned method for identifying coherent structures in a vertical matrix stereoscopic space are implemented.

[0115] In this embodiment, the computer device is any device or apparatus with data processing capability, which will not be described in detail here.

[0116] Example 4

[0117] This embodiment 4 describes a computer-readable storage medium, which is used to implement the method for identifying coherent structures in a vertical matrix three-dimensional space described in the above embodiment 1.

[0118] Specifically, the computer-readable storage medium in this embodiment 4 stores a program thereon, and when the program is executed by a processor, it is used to implement the steps of the above-mentioned method for identifying coherent structures in a vertical matrix three-dimensional space.

[0119] The computer-readable storage medium may be an internal storage unit of any device or apparatus with data processing capabilities, such as a hard disk or memory, or an external storage device of any device with data processing capabilities, such as a plug-in hard disk, a smart media card (SMC), an SD card, a flash card, etc., equipped on the device.

[0120] Of course, the above description is only a preferred embodiment of the present invention, and the present invention is not limited to the above embodiments. It should be noted that all equivalent substitutions and obvious deformation forms made by any technician familiar with the field under the guidance of this specification fall within the essential scope of this specification and should be protected by the present invention.

Claims

1. A method for identifying coherent structures in a vertical matrix three-dimensional space, characterized in that: The steps include: Step 1. Obtain the shear signals of four sets of matrix vertical sections and perform denoising preprocessing; Among them, the shear signals of the four sets of matrix vertical profiles are obtained during the free-falling observation process by the four sets of matrix-arranged shear probes carried by the matrix profile observation instrument for ocean turbulence observation; Step 2. Relying on CEEMD multi-scale decomposition, combined with wave number spectrum analysis and multivariate parameter threshold analysis methods, the coherent structural components in the four groups of shear signals after denoising are screened to obtain effective characteristic signal components, and all effective characteristic signal components in each group of shear signals are reconstructed to obtain a total of four groups of coherent characteristic fluctuation signals; The step 2 is specifically as follows: Step 2.

1. Using CEEMD multi-scale decomposition, perform the first multi-scale decomposition calculation on the denoised shear signal s(x) and decompose it into n intrinsic mode function components: Among them, IMF i represents the i-th eigenmode function component, and re represents the residual trend term; Step 2.

2. Compare the denoised shear signal s(x) with the intrinsic mode function component IMF i Perform wavenumber spectrum analysis to analyze the central wavenumber information corresponding to each eigenmode function component, and define the maximum cutoff wavenumber as k max ; If the central wave number corresponding to the eigenmode function component is less than the maximum cutoff wave number k max , then the eigenmode function component is located in the energy domain and the inertial subdomain, and the eigenmode function component is retained; If the central wave number corresponding to the eigenmode function component is greater than or equal to the maximum cutoff wave number k max , then the intrinsic mode function component is located in the dissipation domain, and the intrinsic mode function component is eliminated; Step 2.

3. Based on the judgment of the maximum cutoff wave number, extract the max All eigenmode function components of , and reconstruct based on this, and reconstruct the shear signal s'(x); Step 2.

4. Perform a second multiscale decomposition on the shear signal s'(x) reconstructed by the first multiscale decomposition using CEEMD multiscale decomposition, and then s'(x) obtains m intrinsic mode function components after the second multiscale decomposition; Among them, IMF i ' represents the i-th eigenmode function component, re' is the residual trend term; Step 2.

5. Analyze the IMF components i 'The corresponding sample entropy SampEn i , Variance Contribution Rate VAF i , correlation coefficient cor i Information, if the intrinsic mode function component IMF i 'Meet the following requirements at the same time: (1) Sample entropy SampEn i <SampEn, SampEn represents the sample entropy corresponding to the reconstructed shear signal s'(x); (2) The variance contribution rate satisfies t1 represents the variance contribution rate threshold; (3) The correlation coefficient satisfies t2 represents the correlation coefficient threshold; Then the intrinsic mode function component IMF i ' meets the coherent structure signal characteristics and is retained as an effective characteristic signal component; otherwise, the intrinsic mode function component IMF i 'Does not meet the characteristics of coherent structure signal and is eliminated as invalid component; Step 2.

6. Screen all effective characteristic signal components for signal reconstruction, and reconstruct to obtain coherent characteristic fluctuation signals; Step 3. Pair the four groups of coherent characteristic fluctuation signals obtained in step 2, calculate the correlation between each pair of coherent characteristic fluctuation signals based on the cross wavelet spectrum, and obtain the coherent core scale; The wavelet energy fluctuations corresponding to the four groups of coherent characteristic fluctuation signals at the coherent core scale are extracted respectively; The step 3 is specifically as follows: Step 3.

1. Define four sets of coherent characteristic fluctuation signals as A total of six pairs of coherent characteristic fluctuation signals are obtained by pairing them two by two, which are: For each pair of coherent characteristic fluctuation signals By performing cross wavelet spectrum analysis, a total of six cross wavelet spectra are calculated for the four groups of coherent characteristic fluctuation signals; where i = 1, 2, 3, 4, j = 1, 2, 3, 4; Step 3.

2. Relying on each pair of coherent characteristic fluctuation signals The results of cross wavelet spectrum analysis are used to calculate the correlation mean corresponding to each scale in the cross wavelet spectrum and extract the correlation peak scale scal ij ; Six cross wavelet spectrum analysis results were calculated for the four sets of coherent characteristic fluctuation signals, and the correlation peak scale set {scal 12 ,scal 13 ,scal 14 ,scal 23 ,scal 24 ,scal 34 }, which contains 6 correlation peak scales; Step 3.

3. Use K-means clustering method to cluster the correlation peak scale obtained in step 3.2 ij The cluster is clustered and the center value of the cluster with the largest number of samples is used as the coherent core scale; Step 3.

4. Extract the wavelet energy fluctuations e corresponding to the four groups of coherent characteristic fluctuation signals at the coherent core scale respectively h (x), e h (x) represents the wavelet energy fluctuation corresponding to the h-th group of coherent characteristic fluctuation signals at the coherent core scale, h = 1, 2, 3, 4; Step 4. Relying on four groups of wavelet energy fluctuations, a two-dimensional space energy fluctuation plane is established to identify the coherent structure; The step 4 is specifically as follows: The four groups of shear probes in a matrix three-dimensional arrangement carried by the matrix profile observation instrument are defined as bullet 1, bullet 2, bullet 3 and bullet 4; the wavelet energy fluctuations of the four groups of matrix three-dimensional arrangement are used as the benchmark to expand the cube surface covered by the matrix space into a two-dimensional plane structure, with the depth data as the ordinate and the bullet arrangement order as the abscissa, and the coherent structure is identified based on the fluctuation of the two-dimensional energy space; If there is an obvious high-energy continuous domain in the two-dimensional energy space fluctuation plane, then there is a coherent structure in this space, that is, if the continuous domain with energy in the range of [1,∞) is in the same coherent structure space range, then there is a coherent structure.

2. The method for identifying coherent structures in a vertical matrix three-dimensional space according to claim 1, characterized in that: In step 2.5, the sample entropy SampEn and IMF corresponding to s'(x) are calculated respectively. i 'The corresponding sample entropy SampEn i ; The sample entropy SampEn corresponding to the reconstructed shear signal s'(x) is used as the sample entropy judgment threshold; If the eigenmode function component IMF i 'The corresponding sample entropy SampEn i <SampEn, it is considered that the intrinsic mode function component IMF i 'It meets the characteristics of coherent structure signal and is retained; If the eigenmode function component IMF i 'The corresponding sample entropy SampEn i ≥SampEn, it is considered that the intrinsic mode function component IMF i ' does not meet the coherent structure signal characteristics and is eliminated.

3. The method for identifying coherent structures in a vertical matrix three-dimensional space according to claim 1, characterized in that: In step 2.5, the intrinsic mode function component IMF is calculated i 'Corresponding variance contribution rate VAF i ; If the eigenmode function component IMF i 'Corresponding variance contribution rate VAF i Located in the range [t1,1], indicating that the intrinsic mode function component IMF i 'It complies with the characteristics of the coherent structure signal and is retained as an effective characteristic component; If the eigenmode function component IMF i 'Corresponding variance contribution rate VAF i Located in the range of [0, t1), indicating that the intrinsic mode function component IMF i ' does not conform to the coherent structure signal characteristics, the intrinsic mode function component IMF i ' to be eliminated.

4. The method for identifying coherent structures in a vertical matrix three-dimensional space according to claim 1, characterized in that: In step 2.5, the intrinsic mode function component IMF is calculated i The correlation coefficient between ' and s'(x); If the intrinsic mode function component IMF i The correlation coefficient between ' and s'(x) is cor i In the range [t2,1], the intrinsic mode function component IMF i 'It complies with the coherent structure signal characteristics and is retained as an effective characteristic signal component; If the intrinsic mode function component IMF i The correlation coefficient between ' and s'(x) is cor i In the range [0, t2), the intrinsic mode function component IMF i ' does not conform to the coherent structure signal characteristics and is eliminated as an invalid component.

5. The method for identifying coherent structures in a vertical matrix three-dimensional space according to claim 1, characterized in that: The step 3.3 is specifically as follows: The scale of each correlation peak in the set is scal ij In the clustering process, it is called a sample. The clustering process is as follows: First, randomly select K samples as the initial cluster centers, calculate the distance between each sample and each cluster center, and assign each sample to the cluster center closest to the sample; K < 3; Each cluster center and the samples assigned to the corresponding cluster center represent a cluster; If all samples have been assigned, the cluster center of each cluster will be iteratively re-selected and the sample-cluster center distance will be calculated based on the existing samples in the cluster; if the cutoff condition that the cluster center no longer changes is met, the iterative process ends; Correlation peak scale set {scal 12 ,scal 13 ,scal 14 ,scal 23 ,scal 24 ,scal 34 }Clustering into C clusters; The cluster center value with the largest number of samples is used as the coherent core scale 6. A vertical matrix three-dimensional space coherent structure recognition system for implementing the vertical matrix three-dimensional space coherent structure recognition method as claimed in claim 1, characterized in that: The vertical matrix three-dimensional space coherent structure recognition system comprises: A preprocessing module is used to obtain the shear signals of four sets of matrix vertical sections and perform denoising preprocessing; Among them, the shear signals of the four sets of matrix vertical profiles are obtained during the free-falling observation process by the four sets of matrix-arranged shear probes carried by the matrix profile observation instrument for ocean turbulence observation; The coherent component screening module is used to screen the coherent structural components in the four groups of shear signals after denoising based on CEEMD multi-scale decomposition, combined with wave number spectrum analysis and multivariate parameter threshold analysis methods, to obtain effective characteristic signal components, and to reconstruct all effective characteristic signal components in each group of shear signals, to obtain a total of four groups of coherent characteristic fluctuation signals; The coherent core energy extraction module is used to pair the four groups of coherent characteristic fluctuation signals obtained, calculate the correlation between each pair of coherent characteristic fluctuation signals based on the cross wavelet spectrum, and obtain the coherent core scale; The wavelet energy fluctuations corresponding to the four groups of coherent characteristic fluctuation signals at the coherent core scale are extracted respectively; And the coherent structure recognition module is used to establish a two-dimensional space energy fluctuation plane based on the four groups of wavelet energy fluctuations extracted by the coherent core energy extraction module to identify the coherent structure.

Images