A method and device for identifying structural characteristics of hydrate-containing sediment samples

Abstract

The present invention discloses a method and device for identifying the structural characteristics of hydrate-containing sediment samples, the method comprising: constructing a finite element data model of a hydrate-containing porous medium based on a piezoelectric chip and a water-saturated porous medium; based on the finite element data model, establishing a preset number of numerical samples using a control variable method; based on the numerical samples, performing numerical simulations through a multi-physical field coupling method according to preset component parameters, and establishing an incoherent wave characteristic fingerprint library based on the results of the numerical simulation; using an acoustic wave analysis method to obtain the incoherent wave portion of the hydrate-containing porous medium sample to be analyzed; based on the incoherent wave portion, matching a target numerical sample from the incoherent wave characteristic fingerprint library through a preset similarity analysis method; using the structural characteristics of the target numerical sample as a structural characteristic identification result of the hydrate-containing porous medium sample. The present invention can accurately realize the structural characteristic identification of hydrate-containing sediment samples, and can be widely used in the field of data processing technology.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method and a device for identifying structural features of a hydrate-containing sediment sample. Background Art

[0002] Natural gas hydrate is an ice-like solid formed under high pressure and low temperature. It usually exists in seabed sediments. It has the advantages of large reserves and high energy density, but the exploitation and utilization of hydrate resources face many technical difficulties. Differences in the sedimentary environmental conditions of natural gas hydrate reservoirs and changes in the effective pressure of the formation during hydrate exploitation will cause differences and dynamic changes in topological structural characteristics such as the shape and arrangement of sediment skeleton particles and the microscopic occurrence mode of hydrates (pore filling, fracture filling). Acoustic logging is one of the important technical means for the identification and quantitative evaluation of hydrate reservoirs.

[0003] Changes in the structural characteristics of hydrate-bearing sediments will interfere with the propagation process of sound waves in the sediments, affecting the energy dissipation and attenuation characteristics of sound waves. For hydrate-bearing sediment samples with different microstructures, when the frequency, vibration direction and phase of the incident wave are the same, the frequency, vibration direction or phase of the outgoing wave may change due to different microstructures leading to different sound wave propagation characteristics, energy dissipation characteristics and attenuation characteristics, thus forming coherent and incoherent parts in the outgoing wave. The coherence of sound waves refers to the characteristics that the frequency, vibration direction and phase of two waves are similar. When sound waves propagate in inhomogeneous media, microscale units (such as mineral particles, pores, hydrate particles, etc.) with different refractive indices, densities, bulk moduli and shear moduli may cause changes in the frequency, vibration direction and phase of sound waves, thereby forming incoherent wave parts. For the outgoing waves passing through hydrate-bearing sediments, the incoherent wave part can be regarded as the fingerprint of the shape and arrangement of the skeleton particles and the hydrate distribution pattern. Therefore, the incoherent wave part can be used to identify and evaluate the microscopic topological structural characteristics of hydrate-bearing sediment samples. However, in existing research on the characteristics and applications of incoherent waves, single-component stacked solid particles (such as quartz sand particles) are used as the medium for sound wave propagation, resulting in large errors in the identification results. Summary of the invention

[0004] The present invention aims to solve the problem of related technical limitations at least to a certain extent. To this end, the present invention proposes a method and device for identifying structural characteristics of hydrate-containing sediment samples, which can accurately identify the structural characteristics of hydrate-containing sediment samples.

[0005] On the one hand, an embodiment of the present invention provides a method for identifying structural features of a hydrate-containing sediment sample, comprising:

[0006] A finite element model of porous media containing hydrates is constructed based on piezoelectric wafers and water-saturated porous media;

[0007] Based on the finite element data model, a preset number of numerical samples are established using the control variable method;

[0008] Based on the numerical samples, numerical simulation is performed through a multi-physical field coupling method according to preset component parameters, and an incoherent wave characteristic fingerprint library is established based on the results of the numerical simulation; wherein the incoherent wave characteristic fingerprint library includes each numerical sample and its corresponding incoherent wave waveform data;

[0009] The incoherent wave part of the hydrate-containing porous medium sample to be analyzed is obtained by using an acoustic wave analysis method; based on the incoherent wave part, a target numerical sample is matched from an incoherent wave feature fingerprint library by a preset similarity analysis method; and the structural characteristics of the target numerical sample are used as structural characteristic recognition results of the hydrate-containing porous medium sample.

[0010] Optionally, a water-saturated porous medium is used to characterize the simulated sediment; a predetermined number of numerical samples are established using a control variable method based on a finite element data model, including:

[0011] Based on the finite element data model, the sizes of the piezoelectric chip and the water-saturated porous medium are kept constant, at least one structural feature of the simulated sediment is changed by using a control variable method, and a preset number of numerical samples are established;

[0012] Among them, the structural characteristics include the shape and arrangement of skeleton particles and the microscopic occurrence pattern of hydrates.

[0013] Optionally, the component parameters include bulk modulus, shear modulus, density and longitudinal wave velocity; based on the numerical sample, numerical simulation is performed by a multi-physics field coupling method according to preset component parameters, including:

[0014] Based on the material composition of the numerical sample, the data volume of the component parameters corresponding to the numerical sample is determined according to a preset component parameter and a numerical mapping table of different materials;

[0015] Based on the amount of data, the piezoelectric wafer in the numerical sample is numerically simulated using electrostatic physics and solid mechanics physics;

[0016] Based on the amount of data, the pressure acoustics physics field is used to numerically simulate the water-saturated porous media in the numerical sample.

[0017] Optionally, the method further comprises:

[0018] The model geometry results of the numerical samples were meshed using free tetrahedral mesh;

[0019] Use the time-dependent solver to solve differential equations in electrostatics, solid mechanics, and pressure acoustics physics using parallel sparse direct solution methods.

[0020] Optionally, based on the incoherent wave part, matching the target numerical sample from the incoherent wave feature fingerprint library by a preset similarity analysis method includes:

[0021] The incoherent wave part is taken as the first wave train, and the first incoherent wave waveform data in the incoherent wave feature fingerprint library is taken as the second wave train;

[0022] The similarity coefficient of each group of wavelengths corresponding to the first wave train and the second wave train is obtained by similarity coefficient analysis method;

[0023] Based on the similarity coefficient of each group of wavelengths, a similarity coefficient curve between the first wave train and the second wave train is constructed;

[0024] Perform mean processing on the similarity curve to obtain the average similarity between the first wave train and the second wave train;

[0025] The next incoherent wave waveform data in the incoherent wave feature fingerprint library is used as the second wave train; then the similarity coefficient of each group of wavelengths corresponding to the first wave train and the second wave train is obtained by the similarity coefficient analysis method; until the average similarity between the incoherent wave and each incoherent wave waveform data in the incoherent wave feature fingerprint library is obtained;

[0026] The numerical sample corresponding to the incoherent wave waveform data with the largest similarity average value is taken as the target numerical sample.

[0027] Optionally, based on the incoherent wave part, matching the target numerical sample from the incoherent wave feature fingerprint library by a preset similarity analysis method includes:

[0028] The incoherent wave part is taken as the first wave train, and the first incoherent wave waveform data in the incoherent wave feature fingerprint library is taken as the second wave train;

[0029] The first wave train is normalized to obtain a first result; the second wave train is normalized to obtain a second result;

[0030] Performing wavelet decomposition on the first result to obtain a first wavelet transform coefficient; performing wavelet decomposition on the second result to obtain a second wavelet transform coefficient; wherein the wavelet decomposition includes discrete wavelet transform or continuous wavelet transform;

[0031] Performing distance measurement on the first wavelet transform coefficient and the second wavelet transform coefficient to obtain the Euclidean distance; performing regulation processing on the Euclidean distance based on a preset range;

[0032] Perform similarity measurement on the first wavelet transform coefficient and the second wavelet transform coefficient to obtain cosine similarity;

[0033] The Euclidean distance and cosine similarity after the reduction process are averaged to obtain the similarity evaluation result between the first wave train and the second wave train;

[0034] The next incoherent wave waveform data in the incoherent wave feature fingerprint library is used as the second wave train; then the second wave train is normalized to obtain the second result; until the similarity evaluation result between the incoherent wave and each incoherent wave waveform data in the incoherent wave feature fingerprint library is obtained;

[0035] The numerical sample corresponding to the incoherent wave waveform data with the largest similarity evaluation result is used as the target numerical sample.

[0036] Optionally, based on the incoherent wave part, matching the target numerical sample from the incoherent wave feature fingerprint library by a preset similarity analysis method includes:

[0037] Extracting features from each incoherent wave waveform data in the incoherent wave feature fingerprint library to obtain feature parameters of each incoherent wave waveform data;

[0038] The characteristic parameters are used as model input, and the labels of the incoherent wave waveform data corresponding to the characteristic parameters are used as model output to obtain a sample set;

[0039] Iteratively train the pre-built neural network using the sample set until the preset training conditions are met to obtain a neural network model;

[0040] The incoherent wave part is identified by using a neural network model to obtain an identification number; the numerical sample corresponding to the incoherent wave waveform data marked with the identification number is used as the target numerical sample.

[0041] On the other hand, an embodiment of the present invention provides a device for identifying structural features of a hydrate-containing sediment sample, comprising:

[0042] The first module is used to construct a finite element model of hydrate-containing porous media based on piezoelectric wafers and water-saturated porous media;

[0043] The second module is used to establish a preset number of numerical samples using a control variable method based on a finite element data model;

[0044] The third module is used to perform numerical simulation based on the numerical sample and the preset component parameters through the multi-physical field coupling method, and establish an incoherent wave feature fingerprint library based on the results of the numerical simulation; wherein the incoherent wave feature fingerprint library includes each numerical sample and its corresponding incoherent wave waveform data;

[0045] The fourth module is used to obtain the incoherent wave part of the hydrate-containing porous medium sample to be analyzed by using the acoustic wave analysis method; based on the incoherent wave part, the target numerical sample is matched from the incoherent wave feature fingerprint library through a preset similarity analysis method; and the structural characteristics of the target numerical sample are used as the structural characteristic identification result of the hydrate-containing porous medium sample.

[0046] Optionally, the device further comprises:

[0047] The fifth module is used to mesh the model geometry results of numerical samples using free tetrahedral meshes;

[0048] The sixth module is used to solve differential equations in electrostatics, solid mechanics, and pressure acoustics physics using a time-dependent solver via parallel sparse direct solution methods.

[0049] On the other hand, an embodiment of the present invention provides an electronic device, including: a processor and a memory; the memory is used to store programs; the processor executes the program to implement the above-mentioned method for identifying structural characteristics of hydrate-containing sediment samples.

[0050] On the other hand, an embodiment of the present invention provides a computer storage medium, in which a program executable by a processor is stored. When the program executable by the processor is executed by the processor, it is used to implement the above-mentioned method for identifying the structural characteristics of hydrate-containing sediment samples.

[0051] The embodiment of the present invention constructs a finite element model of a hydrate-containing porous medium based on a piezoelectric chip and a water-saturated porous medium; based on the finite element model, a preset number of numerical samples are established by using a control variable method; based on the numerical samples, numerical simulation is performed by a multi-physical field coupling method according to preset component parameters, and an incoherent wave feature fingerprint library is established based on the results of the numerical simulation; wherein the incoherent wave feature fingerprint library includes each numerical sample and its corresponding incoherent wave waveform data; an acoustic wave analysis method is used to obtain the incoherent wave part of the hydrate-containing porous medium sample to be analyzed; based on the incoherent wave part, a target numerical sample is matched from the incoherent wave feature fingerprint library by a preset similarity analysis method; and the structural characteristics of the target numerical sample are used as the structural characteristic identification result of the hydrate-containing porous medium sample. The embodiment of the present invention first uses a control variable method to establish a plurality of different numerical samples to conduct a detailed analysis of hydrate-containing sediment samples with different structural characteristics, and then a numerical simulation is performed by a multi-physical field coupling method to accurately and comprehensively analyze the characteristics of the incoherent waves of different numerical samples. The embodiment of the present invention can accurately realize the structural characteristic identification of hydrate-containing sediment samples. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The accompanying drawings are used to provide a further understanding of the technical solution of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the technical solution of the present invention and do not constitute a limitation on the technical solution of the present invention.

[0053] Figure 1 This is a schematic diagram of an implementation environment for identifying structural characteristics of hydrate-containing sediment samples provided by an embodiment of the present invention;

[0054] Figure 2 It is a schematic flow chart of a method for identifying structural characteristics of a hydrate-containing sediment sample provided by an embodiment of the present invention;

[0055] Figure 3 A schematic diagram of a geometric model of a saturated water sediment provided by an embodiment of the present invention;

[0056] Figure 4 A schematic diagram of a numerical model of a diamond-shaped arrangement of spherical particles in a pore-filling hydrate sediment provided in an embodiment of the present invention;

[0057] Figure 5 A schematic diagram of a numerical model of cubic arrangement of ellipsoidal particles in pore-filling hydrate sediments provided in an embodiment of the present invention;

[0058] Figure 6 A schematic diagram of a numerical model of cubic arrangement of spherical particles in pore-filling hydrate sediments provided in an embodiment of the present invention;

[0059] Figure 7 A schematic diagram of a numerical model of a spherical cubic arrangement of crack-filling hydrate sediments provided in an embodiment of the present invention;

[0060] Figure 8 Schematic diagram of curves of received sound wave signals of four models provided in an embodiment of the present invention;

[0061] Fig. 9 The overall process diagram of the method for identifying the structural characteristics of hydrate-containing sediment samples provided in an embodiment of the present invention is shown;

[0062] Fig.10 A schematic diagram of a first geometric model of a hydrate-containing sediment provided in an embodiment of the present invention;

[0063] Fig.11 A schematic diagram of the comparison curve of incoherent waves between the first geometric model and other models provided in an embodiment of the present invention;

[0064] Fig.12 A schematic diagram of a similarity coefficient curve between the first geometric model provided in an embodiment of the present invention and the incoherent waves of other models;

[0065] Fig.13A schematic diagram of a Euclidean distance curve for calculating wavelet transform coefficients of the first geometric model and other models provided in an embodiment of the present invention;

[0066] Fig.14 A schematic diagram of a cosine similarity curve calculated for wavelet transform coefficients of the first geometric model and other models provided in an embodiment of the present invention;

[0067] Fig.15 A schematic diagram of an average similarity coefficient curve calculated from wavelet transform coefficients of the first geometric model and other models provided in an embodiment of the present invention;

[0068] Fig.16 A schematic diagram of a second geometric model of a hydrate-containing sediment provided in an embodiment of the present invention;

[0069] Fig.17 A schematic diagram of the comparison curve of the incoherent waves of the second geometric model provided by the embodiment of the present invention and other models;

[0070] Fig.18 A schematic diagram of a similarity coefficient curve between the second geometric model provided in an embodiment of the present invention and the incoherent waves of other models;

[0071] Fig.19 A schematic diagram of a Euclidean distance curve for calculating wavelet transform coefficients of the second geometric model provided in an embodiment of the present invention and other models;

[0072] Fig. 20 A schematic diagram of a cosine similarity curve calculated for wavelet transform coefficients of the second geometric model provided in an embodiment of the present invention and other models;

[0073] Fig.21 A schematic diagram of an average similarity coefficient curve for calculating wavelet transform coefficients of the second geometric model provided in an embodiment of the present invention and other models. DETAILED DESCRIPTION

[0074] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0075] It should be noted that, although the functional modules are divided in the system schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the system or the order in the flowchart. The terms "first / S100", "second / S200", etc. in the specification, claims and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0076] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0077] It is understandable that the method for identifying the structural characteristics of hydrate-containing sediment samples provided in the embodiment of the present invention can be applied to any computer device with data processing and computing capabilities, and this computer device can be various terminals or servers. When the computer device in the embodiment is a server, the server is an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. Optionally, the terminal is a smart phone, a tablet computer, a laptop computer, a desktop computer, etc., but is not limited to this.

[0078] To facilitate the understanding of the technical solution of the present invention, the technical feature terms that may appear in the embodiments of the present invention are first explained:

[0079] like Figure 1 FIG. 1 is a schematic diagram of an implementation environment provided by an embodiment of the invention. Figure 1 The implementation environment includes at least one terminal 102 and a server 101. The terminal 102 and the server 101 can be connected to a network wirelessly or wired to complete data transmission and exchange.

[0080] Server 101 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), as well as big data and artificial intelligence platforms.

[0081] In addition, the server 101 can also be a node server in the blockchain network. Blockchain is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, and encryption algorithm.

[0082] The terminal 102 may be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, etc., but is not limited thereto. The terminal 102 and the server 101 may be directly or indirectly connected via wired or wireless communication, which is not limited in the embodiment of the present invention.

[0083] Based on the example Figure 1 In the implementation environment shown, an embodiment of the present invention provides a method for identifying structural features of hydrate-containing sediment samples. The following is explained using the example of the method for identifying structural features of hydrate-containing sediment samples being applied in the server 101. It can be understood that the method for identifying structural features of hydrate-containing sediment samples can also be applied in the terminal 102.

[0084] Reference Figure 2 , Figure 2 The flowchart of the method for identifying the structural characteristics of hydrate-containing sediment samples applied to a server provided in an embodiment of the present invention, the executor of the method for identifying the structural characteristics of hydrate-containing sediment samples can be any of the aforementioned computer devices (including a server or a terminal). Figure 2 , the method comprises the following steps:

[0085] S100, constructing a finite element model of hydrate-containing porous media based on piezoelectric wafers and water-saturated porous media;

[0086] It should be noted that the water-saturated porous medium is used to characterize the simulated sediments; in some embodiments, step S100 may include: based on the finite element data model, controlling the size of the piezoelectric chip and the water-saturated porous medium to remain unchanged, using the control variable method to change at least one structural feature of the simulated sediments, and establishing a preset number of numerical samples; wherein the structural features include the shape and arrangement of the skeleton particles and the microscopic occurrence pattern of the hydrate.

[0087] For example, in some specific implementations, the geometric structure of the model is first constructed, which is specifically implemented as follows:

[0088] Construct a finite element numerical model of porous media containing hydrates. The finite element numerical model of porous media containing hydrates was established using the multi-physics field coupling computing platform (COMSOL Multiphysics). The geometric structure of the numerical model consists of two parts: a piezoelectric chip and a porous medium. In the model, two piezoelectric chips are placed on both sides of the porous medium, respectively, to simulate the emission and reception of sound waves (respectively, the transmitting end piezoelectric chip and the receiving end piezoelectric chip). The two piezoelectric chips have the same geometric dimensions and materials, with a chip thickness of 1 mm and a side length of 1.5 mm. The water-saturated porous medium (used to simulate sediments) is a cube with a side length of 1.5 mm. The geometric structure contains 64 spherical quartz sand particles of the same size with a radius of 0.25 mm. The quartz sand particles are regularly arranged tangent to each other, such as Figure 3 shown.

[0089] S200, based on the finite element data model, using the control variable method to establish a preset number of numerical samples;

[0090] It should be noted that the water-saturated porous medium is used to characterize the simulated sediments; in some embodiments, step S100 may include: based on the finite element data model, controlling the size of the piezoelectric chip and the water-saturated porous medium to remain unchanged, using the control variable method to change at least one structural feature of the simulated sediments, and establishing a preset number of numerical samples; wherein the structural features include the shape and arrangement of the skeleton particles and the microscopic occurrence pattern of the hydrate.

[0091] For example, in some specific embodiments, on the basis of a saturated water simulated sediment model (i.e., a finite element numerical model), the sizes of the piezoelectric chip and the porous medium are controlled to remain unchanged, and the shape and arrangement of the skeleton particles in the simulated sediment, the microscopic occurrence mode of hydrates (suspension, contact, and cementation pore filling type: on a certain cross section of a porous medium containing hydrates in the suspension mode, the circular surfaces representing hydrates are uniformly distributed in the pore space, and the hydrates are not in contact with the quartz sand particles; on a cross section of a porous medium containing hydrates in the contact ... The circular surface is in contact with the quartz sand particles; on the cross section of the porous medium containing cementation mode hydrates, the circular rings representing the hydrates wrap around the quartz sand particles, and the hydrate circular rings on the surfaces of adjacent quartz sand particles are in contact with each other, cementing the quartz sand particles together. Fracture filling type of horizontal and vertical cracks: on the cross section of the porous medium containing horizontal cracks, the quartz sand particles are evenly arranged on both sides of the horizontal cracks, and the horizontal cracks are filled with hydrates; on the cross section of the porous medium containing vertical cracks, the quartz sand particles are evenly arranged on both sides of the vertical cracks, and the vertical cracks are filled with hydrates. ) and other structural features, a numerical model is established such as Figure 4 , Figure 5 , Figure 6 , Figure 7 As shown, m numerical samples are formed. Among them, Figure 4 , Figure 5 and Figure 6 The box diagram corresponding to the lower right corner of (a) the logo represents the numerical model of diamond arrangement of spherical particles of hydrate sediments containing pores in suspension state, (b) the box diagram corresponding to the lower right corner of the logo represents the numerical model of diamond arrangement of spherical particles of hydrate sediments containing pores in contact state, and (c) the box diagram corresponding to the lower right corner of the logo represents the numerical model of diamond arrangement of spherical particles of hydrate sediments containing pores in cementation state. Figure 7 The box diagram corresponding to the lower right corner of (a) represents the numerical model of spherical cubic arrangement of hydrate sediments containing fracture filling under the state of horizontal fractures, and the box diagram corresponding to the lower right corner of (b) represents the numerical model of spherical cubic arrangement of hydrate sediments containing fracture filling under the state of vertical fractures.

[0092] S300, based on the numerical sample, numerical simulation is performed by a multi-physical field coupling method according to preset component parameters, and an incoherent wave characteristic fingerprint library is established based on the results of the numerical simulation;

[0093] The incoherent wave feature fingerprint library includes each numerical sample and its corresponding incoherent wave waveform data;

[0094] It should be noted that in some embodiments, based on the numerical sample, numerical simulation is performed through a multi-physical field coupling method according to preset component parameters, which may include: based on the material composition of the numerical sample, determining the data volume of the component parameters corresponding to the numerical sample according to a numerical mapping table of preset component parameters and different materials; based on the data volume, using electrostatic physics fields and solid mechanics physics fields to perform numerical simulation of the piezoelectric chip in the numerical sample; based on the data volume, using pressure acoustic physics fields to perform numerical simulation of the water-saturated porous medium in the numerical sample.

[0095] In some embodiments, the method may further include: using a free tetrahedral grid to mesh the model geometry results of the numerical sample; using a transient solver to solve the differential equations in the electrostatic physics field, solid mechanics physics field and pressure acoustics physics field through a parallel sparse direct solution method.

[0096] For example, in some specific implementations:

[0097] First, numerical simulation is performed using the multi-physics coupling method of electrostatics, solid mechanics, and pressure acoustics. For piezoelectric chips, electrostatics and solid mechanics are used; for porous media, pressure acoustics is used.

[0098] The porous medium part of the model geometry is composed of particles and pores. Due to the complex pore structure at the microscopic scale, a free tetrahedral mesh is used to mesh the geometry (for the piezoelectric chip, the grid level is set to 5724 units, that is, the maximum grid unit size is 200μm, and the minimum grid unit size is 40μm; for the porous medium, the grid level is set to 588993 units, that is, the maximum grid unit size is 60μm, and the minimum grid unit size is 20μm.); use a transient solver (generalized alpha method: for applications such as structural mechanics and acoustics that are sensitive to excessive dissipation, a transient solver based on the generalized alpha method is preferred. Although the backward difference method is more stable and more general than the generalized alpha method, it will introduce more damping at the same time, while the generalized alpha method has smaller damping and is more suitable for this model.) to solve the differential equations in the above electrostatics, solid mechanics, and pressure acoustics physics fields, and select a parallel sparse direct solution method.

[0099] The settings of component parameters are as follows:

[0100] The porous media geometry contains skeleton particles, pore water and hydrate particles, which are defined as quartz sand, pore water and hydrate materials in the model, respectively. The parameters of bulk modulus, shear modulus, density and P-wave velocity of each component in the sediment are set according to Table 1.

[0101] Table 1

[0102] Material Bulk modulus(GPa) Shear modulus(GPa) Density (kg / m^3) P-wave velocity (m / s) Quartz sand 36.6 45.0 2650 6040 water 2.25 0 1000 1500 Hydrate 8.41 3.54 925 3770

[0103] The effective wavelength of the incident sound wave (wavelength = sound wave velocity / frequency) must be much larger than the diameter of the skeleton particles in the porous medium to ensure that the sound wave can penetrate the hydrate-containing porous medium. Therefore, a suitable wavelength needs to be selected.

[0104] Among them, the analysis of the acoustic wave data for the numerical samples is as follows:

[0105] The acoustic wave signals obtained by the acoustic wave receiving end in the numerical model are recorded and analyzed, such as Figure 8 As shown. If the waveform phase of the received sound wave signal is relatively consistent at j wavelengths and before, but inconsistent at j+1 wavelength and after, the received sound wave signal contains coherent waves of j wavelengths (such as j=2) and incoherent waves thereafter (i.e., waveforms at j+1 wavelength and after). A fingerprint library of microstructural characteristics of hydrate-bearing sediment samples (i.e., incoherent wave characteristic fingerprint library) is established based on incoherent wave waveform data of m samples (such as m=11, 100, 1000, etc.). The fingerprint library refers to: m numerical samples and corresponding "incoherent wave waveform data".

[0106] S400, using an acoustic wave analysis method to obtain an incoherent wave portion of the hydrate-containing porous medium sample to be analyzed; based on the incoherent wave portion, matching a target numerical sample from an incoherent wave feature fingerprint library using a preset similarity analysis method; and using the structural features of the target numerical sample as a structural feature recognition result of the hydrate-containing porous medium sample.

[0107] Among them, the present invention provides three methods for identifying structural characteristics of hydrate-containing sediments based on the incoherent wave feature fingerprint library and similarity analysis, namely: similarity coefficient analysis method, wavelet transform analysis method, and neural network pattern recognition method.

[0108] It should be noted that, in some embodiments, based on the incoherent wave part, matching the target numerical sample from the incoherent wave feature fingerprint library by a preset similarity analysis method can include: taking the incoherent wave part as the first wave train, and taking the first incoherent wave waveform data in the incoherent wave feature fingerprint library as the second wave train; obtaining the similarity coefficient of each group of wavelengths corresponding to the first wave train and the second wave train by a similarity coefficient analysis method; constructing a similarity coefficient curve between the first wave train and the second wave train based on the similarity coefficient of each group of wavelengths; performing mean processing on the similarity curve to obtain the average similarity value between the first wave train and the second wave train; taking the next incoherent wave waveform data in the incoherent wave feature fingerprint library as the second wave train; then returning to the step of obtaining the similarity coefficient of each group of wavelengths corresponding to the first wave train and the second wave train by a similarity coefficient analysis method; until the average similarity value of each incoherent wave waveform data in the incoherent wave feature fingerprint library is obtained; and taking the numerical sample corresponding to the incoherent wave waveform data with the largest similarity average value as the target numerical sample.

[0109] For example, in some specific implementations, the similarity coefficient analysis method is specifically implemented as follows:

[0110] The similarity coefficient is an indicator used to measure the similarity between two objects or data sets (such as two columns of waveform data). The present invention calculates the similarity coefficient between two incoherent waves to measure the similarity between them. The closer the similarity coefficient is to 1, the higher the linear correlation between the two, that is, the more similar the waveforms of the two columns of incoherent waves are.

[0111] For a hydrate-containing porous medium sample to be analyzed (not belonging to the above fingerprint library, but with the same size as the sample in the fingerprint library), the same incident wave as that used to establish the above fingerprint library is used to test the sample, obtain the corresponding outgoing wave, and cut out the incoherent wave part, which is recorded as wave train S A (i.e. the first wave train), a certain incoherent wave train in the above fingerprint library is denoted as S B (i.e. the second wave).

[0112] For the above two wave trains, namely wave train SA and wave train S B , and divide them into q groups of data (one wavelength is one group of data). Calculate the wave train S in sequence A The pth group (p≤q) of data and wave train S B The similarity coefficient of the pth group of data, until p = q. By this method, the similarity coefficients of the q groups of data divided in the two wave trains can be accurately calculated and compared.

[0113] Calculate the wave train S separately A The pth group of data and wave train S B Similarity coefficient of group p data:

[0114]

[0115] in:

[0116]

[0117] Where: T1 is the start time of the pth group of wave train data, T2 is the end time of the pth group of wave train data, S Ap is wave train S A The pth group of data, S Bp is wave train S B The pth group of data, t is the integral variable (i.e. time), R(A p ,B p ) has a value range of [-1,1], R(A p ,B p ) is closer to 1, indicating that the two waves are more similar.

[0118] After completing the above calculations, the following similarity coefficient analysis steps are used: A and wave train S B The calculated similarity coefficients of the q groups are plotted as a curve, which is the similarity coefficient curve of the two wave trains. The other incoherent wave trains in the fingerprint library are denoted as S B , the similarity coefficient is calculated by the above method. Since there are m groups of wave train data in the fingerprint library, a total of m similarity coefficient curves are obtained. The average values ​​of the m similarity coefficient curves are calculated in sequence to obtain m average values. It is considered that the fingerprint library sample corresponding to the largest average value is most similar to the structural characteristics of the hydrate-containing porous medium sample to be analyzed (such as the wave train S of the sample to be analyzed). A If the average value of the wave train similarity coefficient curve with a sample in the fingerprint library is the largest, it is considered that the sample to be analyzed has the most similar structural features to the sample in the fingerprint library, which is taken as the final result of identification).

[0119] It should be noted that, in some embodiments, matching the target numerical sample from the incoherent wave feature fingerprint library based on the incoherent wave part by a preset similarity analysis method may include: taking the incoherent wave part as the first wave train, and taking the first incoherent wave waveform data in the incoherent wave feature fingerprint library as the second wave train; normalizing the first wave train to obtain a first result; normalizing the second wave train to obtain a second result; performing wavelet decomposition on the first result to obtain a first wavelet transform coefficient; performing wavelet decomposition on the second result to obtain a second wavelet transform coefficient; wherein the wavelet decomposition includes discrete wavelet transform or continuous wavelet transform; performing distance degree analysis on the first wavelet transform coefficient and the second wavelet transform coefficient. to obtain the Euclidean distance; to reduce the Euclidean distance based on a preset range; to measure the similarity of the first wavelet transform coefficient and the second wavelet transform coefficient to obtain the cosine similarity; to average the reduced Euclidean distance and cosine similarity to obtain the similarity evaluation result of the first wave train and the second wave train; to use the next incoherent wave waveform data in the incoherent wave feature fingerprint library as the second wave train; then to return to the step of normalizing the second wave train to obtain the second result; until the similarity evaluation result of the incoherent wave and each incoherent wave waveform data in the incoherent wave feature fingerprint library is obtained; and the numerical sample corresponding to the incoherent wave waveform data with the largest similarity evaluation result is used as the target numerical sample.

[0120] For example, in some specific implementations, the wavelet transform analysis method is specifically implemented as follows:

[0121] Wavelet transform gradually refines the signal at multiple scales through scaling and translation operations to reveal the local characteristics of the signal. Expand the function x(t) under the wavelet basis function, and the continuous wavelet transform of any function x(t) is obtained as:

[0122]

[0123] in:

[0124]

[0125] Where: R1 represents the real number domain, t is the integral variable (i.e. time), W x (a,b) are wavelet transform coefficients, a and b are defined as scale parameters and translation parameters respectively. is the basic wavelet, and It is through The waveform obtained by translation and scaling. If a and b are transformed, a set of waveform functions can be obtained It is denoted as wavelet basis function.

[0126] Data preprocessing: Select the incoherent wave part of the hydrate-containing porous medium sample to be analyzed, which is recorded as wave train S A, a certain series of incoherent waves in the above fingerprint library is recorded as wave train S B , normalize the above two wave trains respectively, and get the processed result y * .

[0127]

[0128] Where: y is the selected wave train data, is the average value of the selected wave train data, σ y is the standard deviation of the selected wave train data.

[0129] Wavelet decomposition steps: Perform wavelet decomposition on the processed two wavelet data, which can be done by using discrete wavelet transform (DWT) or continuous wavelet transform (CWT). According to the characteristics of the signal and the required analysis results, select the appropriate wavelet basis function and decomposition layer number. Commonly used wavelet basis functions include: Daubechies wavelet, Haar wavelet, Morlet wavelet, etc., and decompose the processed two wavelet data into wavelet transform coefficients corresponding to different scales.

[0130] Steps to calculate the Euclidean distance: Use the wavelet transform coefficients after wavelet decomposition of the two wave trains (select a certain scale first) to calculate the Euclidean distance. The Euclidean distance is a common distance metric that can be used to measure the similarity between two vectors. The calculation formula of the Euclidean distance is:

[0131]

[0132] Where n is the total number of wavelet transform coefficients under a selected scale condition, i is the i-th index corresponding to the wavelet transform coefficient, W Ai is wave train S A The i-th wavelet transform coefficient after wavelet decomposition (at this scale), W Bi is wave train S B The i-th wavelet transform coefficient after wavelet decomposition (at this scale), W A is wave train S A The wavelet transform coefficients (of this scale) after wavelet decomposition, W B is wave train S B Wavelet transform coefficients after wavelet decomposition (at this scale).

[0133] The Euclidean distance calculated from the wavelet transform coefficients after wavelet decomposition (at a certain scale) of the wave train is a number greater than 0. In order to more effectively describe the similarity between curves (wave trains), the following formula is used to further reduce it to between (0,1] (i.e., reduction processing).

[0134]

[0135] Steps to calculate cosine similarity: Calculate cosine similarity using wavelet transform coefficients (at a certain scale) after wavelet decomposition of two wave trains. Cosine similarity is another commonly used similarity measure. It does not consider the peak value of the waveform. It measures the cosine value of the angle between two vectors, ranging from -1 to 1. The formula for calculating cosine similarity is:

[0136]

[0137] Where: θ is the wavelet transform coefficient W after wavelet decomposition of the two wave trains (at a certain scale) A and W B The angle between the two vectors.

[0138] Similarity evaluation steps: average the Euclidean distance and cosine similarity calculated from the wavelet transform coefficients of the two wave trains after wavelet decomposition (at a certain scale), and use the average value (expressed as r) as the similarity evaluation result of the two wave train data at a certain scale. The closer the average value is to 1, the higher the similarity of the two incoherent waves at this scale.

[0139] The above directly processes the wavelet transform coefficients, but the wavelet transform coefficients can also be reconstructed first, and then the above processing is performed on the reconstructed signal. Specifically, the wavelet transform coefficients of each scale are reconstructed, and the reconstructed signal is processed as above, that is, calculating the Euclidean distance, normalizing, calculating the cosine similarity, averaging the Euclidean distance and the cosine similarity), and finally using the average value to evaluate the similarity.

[0140] The above-mentioned evaluation methods of calculating Euclidean distance, cosine similarity, and taking the average value can also be directly applied to the processing of wave train data of original incoherent waves. In addition, the above-mentioned similarity coefficient analysis method can also be used for a series of wavelet transform coefficients at different scales after wavelet decomposition of wave train data or signal components of different scales obtained after reconstruction based on wavelet transform coefficients.

[0141] It should be noted that, in some embodiments, based on the incoherent wave part, matching the target numerical sample from the incoherent wave feature fingerprint library by a preset similarity analysis method may include: extracting features from each incoherent wave waveform data in the incoherent wave feature fingerprint library to obtain the characteristic parameters of each incoherent wave waveform data; using the characteristic parameters as model input and the label identification of the incoherent wave waveform data corresponding to the characteristic parameters as model output to obtain a sample set; using the sample set to iteratively train the pre-constructed neural network until the preset training conditions are met to obtain a neural network model; using the neural network model to identify the incoherent wave part to obtain an identification label; using the numerical sample corresponding to the incoherent wave waveform data marked with the identification label as the target numerical sample. Among them, the model input and / or model output of the sample set can also be normalized to ensure that all input and output data receive equal attention during the training process and to accelerate the network convergence speed.

[0142] For example, in some specific implementations, the neural network pattern recognition method is specifically implemented as follows:

[0143] The neural network is trained using sample data, and the neural network learns the relationship between input and output, and can be generalized to unknown data; it has strong parallel processing capabilities and fast computing speed; it also has distributed storage and good fault tolerance, and shows broad application prospects in the fields of target recognition and pattern recognition. The present invention uses hydrate-containing sediment samples with different structural characteristics as objects to be identified, and regards the structural characteristics of the samples as "patterns", giving full play to the pattern recognition ability of neural networks.

[0144] The first step is to extract characteristic parameters. Select the original data of m incoherent waves in the fingerprint library and extract the maximum amplitude, phase and energy of the incoherent wave signal as the characteristic parameters of the waveform.

[0145] The second step is to form a sample set, which contains m samples. Each sample includes input and output. The input is a three-dimensional vector (i.e., feature vector), including maximum amplitude, phase, and energy; the output is a number between 1 and m. In order to obtain a reasonable neural network model, the sample set needs to be divided into training samples and test samples. When the error of the test sample is as small as or slightly larger than the error of the training sample, it indicates that the established model can effectively approximate the rules contained in the training sample.

[0146] In the third step, all input and output quantities in the sample set are normalized to ensure that all input and output data receive equal attention during training and to speed up network convergence.

[0147] The fourth step is to build a neural network. Select a neural network model suitable for processing input signals, such as a three-layer BP neural network, a convolutional neural network (CNN), or a recurrent neural network (RNN). If a three-layer BP neural network is used, three neurons are set in the input layer, one neuron is set in the output layer, and the number of neurons in the middle layer (i.e., the hidden layer) is calculated using an empirical formula.

[0148] The fifth step is to train the neural network model. The sample set is input into the neural network, and the target accuracy of the training is set to a small value, so that the network cannot stop training because the target accuracy is reached, but stops training when the mean square error of the test sample continues to increase within a certain number of training times, so that the network can fully learn the training samples during the training process and have good generalization ability.

[0149] The sixth step is to use the neural network model to perform pattern recognition on the incoherent wave to be analyzed, and output the serial number of the incoherent wave with the highest similarity in the fingerprint library.

[0150] The input of the above neural network is the feature vector obtained by processing incoherent wave data. It should be noted that the above wavelet transform coefficients or signal components of different scales obtained after reconstruction based on the wavelet transform coefficients can also be used as processing objects to obtain feature vectors, which are then used as the input of the neural network to achieve the purpose of pattern recognition based on the neural network.

[0151] In order to explain the principle of the technical solution of the present invention in detail, the overall process of the present invention is described below in combination with some specific embodiments. It is easy to understand that the following is an explanation of the technical principle of the present invention and cannot be regarded as a limitation of the present invention.

[0152] First of all, it should be noted that in the existing research on the characteristics and application of incoherent waves, the single-component stacked solid particles (such as quartz sand particles) are used as the medium for sound wave propagation, and the porous medium containing natural gas hydrates, the particle shape, arrangement, and hydrate occurrence mode, etc., are not involved. The influence of factors such as the incoherent wave part in the outgoing wave is not involved, thus limiting people's comprehensive understanding of the acoustic characteristics of hydrate-containing sediments (such as sound wave frequency, amplitude, phase, sound velocity, and sound attenuation). At present, there is no method to accurately characterize the structural characteristics of hydrate-containing sediments (such as the shape and arrangement of skeleton particles and the microscopic occurrence mode of hydrates), making it difficult to establish and select applicable calculation models for evaluating hydrate saturation and permeability of hydrate-containing sediments.

[0153] The present invention proposes a method for identifying the structural characteristics of hydrate-containing sediment samples based on incoherent wave analysis for hydrate-containing granular porous media and hydrate-containing porous media samples, thereby providing a technical prerequisite for the next step of establishing and selecting a suitable model for evaluating hydrate saturation and permeability of hydrate-containing sediments.

[0154] The present invention proposes a method for identifying structural features of hydrate-containing sediments based on incoherent wave analysis, the flow chart of which is as follows: Fig. 9 As shown, the method mainly includes two parts: establishing an incoherent wave characteristic fingerprint library of typical structures of porous media containing hydrate sediments, and identifying the structural characteristics of porous media containing hydrate sediments based on the incoherent wave characteristic fingerprint library and similarity analysis.

[0155] The specific implementation process is as follows:

[0156] 1. Establish an incoherent wave characteristic fingerprint library of hydrate-containing sediment structure:

[0157] (1) Establishment of acoustic field model of hydrate-containing sediments:

[0158] 1) Constructing the geometric structure:

[0159] A finite element numerical model of hydrate-containing porous media was constructed. The geometric structure of the numerical model consists of two parts: a piezoelectric chip and a porous medium. In the model, two piezoelectric chips are placed on both sides of the porous medium to simulate the emission and reception of sound waves (the transmitting end piezoelectric chip and the receiving end piezoelectric chip, respectively). Figure 3 By controlling the variables, the shape and arrangement of the skeleton particles in the sediment, the hydrate occurrence mode (pore filling type of suspension, contact and cementation; fracture filling type of horizontal and vertical fractures) and other structural characteristics are changed in turn, and a numerical model is established as shown in FIG. Figure 4 , Figure 5 , Figure 6 , Figure 7 As shown, m numerical samples are formed.

[0160] 2) Physical field selection, meshing and solver:

[0161] The numerical simulation is realized by coupling the multi-physics field of electrostatics, solid mechanics, and pressure acoustics. For the piezoelectric chip, the physics fields of electrostatics and solid mechanics are used; for the porous medium, the physics field of pressure acoustics is used.

[0162] The geometric structure of the model consists of a particle part and a pore part. Due to the complexity of the pore structure at the microscopic scale, a free tetrahedral mesh is used to mesh the geometric structure. The differential equations in the above-mentioned electrostatics, solid mechanics, and pressure acoustics physics fields are solved using a transient solver (generalized α method), and a parallel sparse direct solution method is selected.

[0163] 3) Parameter setting:

[0164] The porous media geometry includes skeleton particles, pore water and hydrate particles, which are defined as quartz sand, pore water and hydrate materials in the model, respectively. The parameters of bulk modulus, shear modulus, density and P-wave velocity of each component in the sediment are set according to Table 1.

[0165] The effective wavelength of the incident sound wave (wavelength = sound wave velocity / frequency) must be much larger than the diameter of the skeleton particles in the porous medium to ensure that the sound wave can penetrate the hydrate-containing porous medium. Therefore, a suitable wavelength needs to be selected.

[0166] (2) Analysis of acoustic wave data:

[0167] The acoustic wave signals obtained by the acoustic wave receiving end in the numerical model are recorded and analyzed, such as Figure 8 As shown, the received sound wave signal includes coherent waves of j wavelengths (such as j=2) and subsequent incoherent waves (i.e., j+1 wavelengths and subsequent waveforms). Based on the incoherent wave waveform data of m samples (such as m=11, 100, 1000, etc.), a fingerprint library of microstructural characteristics of hydrate-containing sediment samples (i.e., an incoherent wave characteristic fingerprint library) is established.

[0168] 2. Identification of structural characteristics of hydrate-bearing sediments based on incoherent wave feature fingerprint library and similarity analysis:

[0169] The present invention provides three methods, namely: similarity coefficient analysis method, wavelet transform analysis method, neural network pattern recognition method. The logical principles of each method refer to the description of the above specific implementation methods, which will not be repeated here.

[0170] To further illustrate the principle and effect of the technical solution of the present invention, the method for identifying the structural characteristics of hydrate-containing sediment samples based on incoherent waves provided by the present invention is described in detail below in combination with specific data cases. The following is an explanation of the technical principle of the present invention and cannot be regarded as a limitation of the present invention. Specific embodiment one:

[0172] The method for identifying the structural characteristics of hydrate-containing sediment samples based on incoherent waves is as follows.

[0173] (1) Establishing the first geometric model of hydrate-containing sediments:

[0174] The first geometric model of hydrate-containing sediments was established by COMSOL. The skeleton of the model was spherical particles, arranged in a diamond shape, and the hydrate was in contact mode. The piezoelectric chips at the transmitting and receiving ends were set, such as Fig.10 As shown, the effective wavelength of the sound wave generated by the piezoelectric chip in the sample is λ=v / f≈3.25mm, which is much larger than the particle diameter used in the model. The sound wave can penetrate the particle sample well.

[0175] (2) Obtain the incoherent wave curve of the model:

[0176] The incoherent wave curve of the model is obtained through the receiving end, and the incoherent wave is compared with the incoherent wave in the fingerprint library, such as Fig.11 As shown;

[0177] (3) Conduct structural feature identification and analysis:

[0178] 1) Similarity coefficient method:

[0179] Calculate the similarity coefficient between the incoherent wave of the model and the incoherent wave in the fingerprint library,

[0180]

[0181] The similarity coefficient curve is obtained, such as Fig.12 As shown in the figure, the closer the similarity coefficient is to 1, the closer the structural characteristics of the two models are, so the shape and arrangement of the model skeleton particles and the hydrate occurrence pattern can be analyzed.

[0182] The obtained similarity coefficients are processed and the average value of each group of similarity coefficients is calculated to obtain Table 2.

[0183] Table 2

[0184] Similarity coefficient R(i, spherical diamond suspension) R(i, spherical diamond contact) R(i, spherical rhombus cementation) average value 0.8694 0.8923 0.8697 Similarity coefficient R(i, ellipsoidal cubic suspension) R(i, ellipsoidal cubic contact) R(i, ellipsoidal cubic cementation) average value 0.7399 0.7391 0.6519 Similarity coefficient R(i, spherical cubic suspension) R(i, spherical cubic contact) R(i, spherical cubic cementation) average value -0.0143 -0.0138 -0.0904 Similarity coefficient R(i, horizontal crack) R(i, vertical crack) average value 0.7798 0.7992

[0185] Comparative analysis shows that the similarity coefficient between the incoherent waves of this model and the incoherent waves of the spherical particles diamond arrangement model under the contact mode is closest to 1, that is, the incoherent waves of the two models are most similar, indicating that the structural characteristics of the two models are relatively close. Therefore, the fitting formula of the spherical particles diamond arrangement model under the contact mode can be used to estimate the hydrate content in the sediment.

[0186] 2) Wavelet transform method:

[0187] The first step is waveform data preprocessing:

[0188] The incoherent waves of the first geometric model established above and the m columns of incoherent wave curve data in the fingerprint library are normalized.

[0189] The second step is wavelet decomposition:

[0190] The processed incoherent wave data of the geometric model and the m columns of incoherent wave curve data of the fingerprint library are subjected to wavelet decomposition. Continuous wavelet transform is adopted, and Daubechies (db6) wavelet basis function and decomposition layer number 5 are selected to decompose the signal into wavelet transform coefficients of different scales.

[0191] The third step is to calculate the Euclidean distance:

[0192] The wavelet transform coefficients are used to calculate the Euclidean distance between the incoherent wave of the model and the incoherent wave in the fingerprint library.

[0193]

[0194] And constrain the Euclidean distance to between (0,1],

[0195]

[0196] Get the Euclidean distance curve, such as Fig.13 As shown in Figure 2, the closer the similarity coefficient is to 1, the closer the structural characteristics of the two models are.

[0197] The fourth step is to calculate the cosine similarity:

[0198] The wavelet transform coefficients are used to calculate the cosine similarity between the incoherent waves of the model and the incoherent waves in the fingerprint library.

[0199]

[0200] The cosine similarity curve is obtained, such as Fig.14 As shown in Figure 2, the closer the similarity coefficient is to 1, the closer the structural characteristics of the two models are.

[0201] Step 5: Result processing and analysis:

[0202] Fig.15 The calculated Euclidean distance and cosine similarity are averaged to obtain the average similarity between the incoherent wave of the model and the incoherent wave in the fingerprint library. Fig.15 It can be seen that the average similarity coefficient of the incoherent waves of this model and the incoherent waves of the spherical particle diamond arrangement model under the contact mode is closest to the y=1 curve, that is, the incoherent waves of the two models are most similar, indicating that the structural characteristics of the two models are relatively close.

[0203] 3) Neural Network Method:

[0204] In the first step, the original data of m incoherent waves in the above fingerprint library are selected, and the maximum amplitude, phase and energy of the incoherent wave signal are extracted as the characteristic parameters of the waveform.

[0205] In the second step, the m groups of three-dimensional signals of the above characteristic parameters are used as network input to form a total sample set. The output of each sample is the sequence number of the sample, i.e., 1, 2, ..., m. Before the sample set is input into the network, the sample set is divided into h groups of training samples and k groups of test samples (m = h + k).

[0206] In the third step, all input and output quantities in the sample set are normalized to ensure that all input and output data receive equal attention during training and to speed up network convergence.

[0207] The fourth step is to use a three-layer BP neural network, with 3 neurons in the input layer, 1 neuron in the output layer, and 7 neurons in the middle layer (i.e., hidden layer) (optimized after referring to the results calculated by empirical formulas and multiple tests).

[0208] The fifth step is to use the three-layer BP neural network model to train the sample set. When the mean square error of the test sample continues to increase within a certain number of training times, the training is stopped and the training is completed.

[0209] The sixth step is to extract the incoherent wave characteristic parameters of the first geometric model and input them into the trained three-layer BP neural network model to realize the recognition of the structural characteristics of the first geometric model. If the output is 2, it means that the geometric structure of this model is most similar to that of the model with serial number 2. Specific embodiment 2:

[0211] The method for identifying the structural characteristics of hydrate-containing sediment samples based on incoherent waves is as follows.

[0212] (1) Establishing the second geometric model of hydrate-containing sediments:

[0213] The second geometric model of hydrate-containing sediments was established by COMSOL. The skeleton particles of the model were arranged in ellipsoids and cubes, and hydrates coexisted in suspension and cementation modes. The transmitting and receiving piezoelectric chips were set, such as Fig.16 The effective wavelength of the sound wave generated by the piezoelectric chip in the sample is λ=v / f≈3.16mm. The wavelength of the sound wave is much larger than the particle diameter, and the sound wave can penetrate the particle sample well.

[0214] (2) Obtain the incoherent wave curve of the model:

[0215] The incoherent wave curve of the model is obtained through the receiving end, and the incoherent wave is compared with the incoherent wave of the fingerprint library, such as Fig.17 As shown;

[0216] (3) Conduct structural feature identification and analysis:

[0217] 1) Similarity coefficient method:

[0218] Calculate the similarity coefficient between the incoherent wave of the model and the incoherent wave in the fingerprint library,

[0219]

[0220] The similarity coefficient curve is obtained, such as Fig.18 As shown in the figure, the closer the similarity coefficient is to 1, the closer the structural characteristics of the two models are, so the shape and arrangement of the model skeleton particles and the hydrate occurrence pattern can be analyzed.

[0221] The obtained similarity coefficients were processed and the average value of each group of similarity coefficients was calculated to obtain Table 3.

[0222] Table 3

[0223]

[0224] Comparative analysis shows that the similarity coefficient of the incoherent waves of this model is closest to 1 with the incoherent waves of the cubic arrangement model of ellipsoidal particles in the suspension and cementation modes, that is, the incoherent waves of the three models are most similar, indicating that the second geometric model has both the structural characteristics of the cubic arrangement of ellipsoidal particles and the acoustic characteristics of hydrate-containing sediments with both suspension and cementation modes. Therefore, it is necessary to combine the structural characteristics of the two models in the fingerprint library, so that the mixed model of cubic arrangement of ellipsoidal particles in the suspension and cementation modes can be selected to estimate the structural characteristics of the sediments. In addition, the similarity coefficient of the suspension mode is greater than that of the cementation mode, so the proportion of hydrates in the suspension mode in the sediment is higher.

[0225] 2) Wavelet transform method:

[0226] The first step is waveform data preprocessing:

[0227] The incoherent waves of the second geometric model established above and the m columns of incoherent wave curve data in the fingerprint library are normalized.

[0228] The second step is wavelet decomposition:

[0229] The processed incoherent wave data of the geometric model and the m columns of incoherent wave curve data of the fingerprint library are subjected to wavelet decomposition. Continuous wavelet transform is adopted, and Daubechies (db7) wavelet basis function and decomposition layer number 5 are selected to decompose the signal into wavelet transform coefficients of different scales and frequencies.

[0230] The third step is to calculate the Euclidean distance:

[0231] The wavelet transform coefficients are used to calculate the Euclidean distance between the incoherent wave of the model and the incoherent wave in the fingerprint library.

[0232]

[0233] And constrain the Euclidean distance to between (0,1],

[0234]

[0235] Get the Euclidean distance curve, such as Fig.19 As shown in Figure 2, the closer the similarity coefficient is to 1, the closer the structural characteristics of the two models are.

[0236] The fourth step is to calculate the cosine similarity:

[0237] The wavelet transform coefficients are used to calculate the cosine similarity between the incoherent waves of the model and the incoherent waves in the fingerprint library.

[0238]

[0239] The cosine similarity curve is obtained, such as Fig. 20 As shown in Figure 2, the closer the similarity coefficient is to 1, the closer the structural characteristics of the two models are.

[0240] Step 5: Result processing and analysis:

[0241] The calculated Euclidean distance and cosine similarity are averaged to obtain the average similarity of the two curves, such as Fig.21 As shown in the figure, the average similarity coefficient of the incoherent wave of this model and the incoherent wave of the cubic arrangement model of ellipsoidal particles in the suspension and cementation modes is closest to the y=1 curve, indicating that this model may contain hydrates in both suspension and cementation modes. In addition, the average similarity curve of the suspension mode is closer to the y=1 curve, so the proportion of hydrates in the suspension mode in the sediment is higher.

[0242] 3) Neural Network Method:

[0243] In the first step, the original data of m incoherent waves in the above fingerprint library are selected, and the maximum amplitude, phase and energy of the incoherent wave signal are extracted as the characteristic parameters of the waveform.

[0244] In the second step, the m groups of three-dimensional signals of the above characteristic parameters are used as network input to form a total sample set. The output of each sample is the sequence number of the sample, i.e., 1, 2, ..., m. Before the sample set is input into the network, the sample set is divided into h groups of training samples and k groups of test samples (m = h + k).

[0245] In the third step, all input and output quantities in the sample set are normalized to ensure that all input and output data receive equal attention during training and to speed up network convergence.

[0246] In the fourth step, a three-layer BP neural network is used, with 3 neurons in the input layer, 1 neuron in the output layer, and 7 neurons in the middle layer (i.e., hidden layer).

[0247] The fifth step is to use the three-layer BP neural network model to train the sample set. When the mean square error of the test sample continues to increase within a certain number of training times, the training is stopped and the training is completed.

[0248] The sixth step is to extract the incoherent wave characteristic parameters of the second geometric model and use a three-layer BP neural network model to identify the structural features of the second geometric model.

[0249] In summary, in order to establish and select a suitable model for evaluating hydrate saturation and permeability of hydrate-containing sediments, it is necessary to identify the structural characteristics of hydrate-containing sediments. In view of the problems existing in the prior art, the present invention proposes a method for identifying the structural characteristics of hydrate-containing sediments based on incoherent wave analysis. By detecting, analyzing and processing incoherent waves, several important structural characteristics of hydrate-containing sediments are identified, such as the shape and arrangement of skeleton particles in the sediments, the hydrate occurrence pattern, etc. The use of this acoustic wave identification method of the present invention is helpful to construct and select a more suitable calculation model for natural gas hydrate saturation and permeability of hydrate-containing sediments.

[0250] Compared with the prior art, the method of the present invention has at least the following beneficial effects:

[0251] (1) The present invention provides a method for identifying the structural characteristics of hydrate-containing sediments in acoustic logging by using incoherent wave signals. By using incoherent waves as contrast signals, the resolution of acoustic detection can be improved, so that the microscopic characteristics and details of hydrate-containing sediments can be more accurately evaluated, so as to accurately identify the structural characteristics of the sediments, thereby selecting a more suitable calculation model for the natural gas hydrate saturation and the permeability of hydrate-containing sediments.

[0252] (2) The present invention provides an effective detection method for the fields of geological exploration, sediment research and hydrate resource development by accurately evaluating the structural characteristics of hydrate-containing sediments, provides more accurate guidance and decision-making basis for geological exploration and resource development, reduces the risks and costs in the exploration and development process, and has important scientific research and engineering application value.

[0253] On the other hand, an embodiment of the present invention provides a device for identifying structural features of hydrate-containing sediment samples, which may include: a first module, used to construct a finite element data model of a hydrate-containing porous medium based on a piezoelectric chip and a water-saturated porous medium; a second module, used to establish a preset number of numerical samples based on the finite element data model using a control variable method; a third module, used to perform numerical simulation based on the numerical samples through a multi-physical field coupling method according to preset component parameters, and establish an incoherent wave feature fingerprint library based on the results of the numerical simulation; wherein the incoherent wave feature fingerprint library includes each numerical sample and its corresponding incoherent wave waveform data; a fourth module, used to obtain the incoherent wave part of the hydrate-containing porous medium sample to be analyzed using an acoustic wave analysis method; based on the incoherent wave part, matching the target numerical sample from the incoherent wave feature fingerprint library through a preset similarity analysis method; and using the structural features of the target numerical sample as the structural feature identification result of the hydrate-containing porous medium sample.

[0254] It should be noted that, in some embodiments, the device may also include: a fifth module, used to use a free tetrahedral grid to mesh the model geometry results of the numerical sample; a sixth module, used to use a transient solver to solve the differential equations in the electrostatic physics field, solid mechanics physics field and pressure acoustics physics field through a parallel sparse direct solution method.

[0255] The contents of the method embodiments of the present invention are all applicable to the device embodiments. The functions specifically implemented by the device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0256] On the other hand, an embodiment of the present invention further provides an electronic device, the electronic device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the above-mentioned method for identifying the structural characteristics of hydrate-containing sediment samples when executing the computer program. The electronic device can be any intelligent terminal including a tablet computer, a car computer, etc.

[0257] It can be understood that the contents of the above method embodiments are all applicable to the present device embodiments, the functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0258] Another aspect of an embodiment of the present invention further provides a computer-readable storage medium, wherein the storage medium stores a program, and the program is executed by a processor to implement the above method.

[0259] The contents of the method embodiments of the present invention are all applicable to the computer-readable storage medium embodiments. The functions specifically implemented by the computer-readable storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0260] The embodiment of the present invention also discloses a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. A processor of a computer device can read the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the above method.

[0261] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

[0262] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the embodiments. Those skilled in the art can make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of the present invention.

Claims

1. A method for identifying structural characteristics of hydrate-containing sediment samples, characterized in that: include: A finite element model of porous media containing hydrates is constructed based on piezoelectric wafers and water-saturated porous media; Based on the finite element data model, a preset number of numerical samples are established using a control variable method; The water-saturated porous medium is used to characterize simulated sediments; the method of establishing a preset number of numerical samples based on the finite element data model using a control variable method includes: Based on the finite element data model, the sizes of the piezoelectric wafer and the water-saturated porous medium are controlled to remain unchanged, at least one structural feature of the simulated sediment is changed by using a control variable method, and a preset number of numerical samples are established; The structural characteristics include the shape and arrangement of the skeleton particles and the microscopic occurrence pattern of hydrates; Based on the numerical sample, numerical simulation is performed by a multi-physical field coupling method according to preset component parameters, and an incoherent wave characteristic fingerprint library is established based on the results of the numerical simulation; wherein the incoherent wave characteristic fingerprint library includes each of the numerical samples and its corresponding incoherent wave waveform data; Wherein, the component parameters include bulk modulus, shear modulus, density and longitudinal wave velocity; the numerical simulation based on the numerical sample and the preset component parameters is performed by a multi-physical field coupling method, including: Based on the material composition of the numerical sample, determining the data amount of the component parameter corresponding to the numerical sample according to a preset mapping table of the component parameters and the values ​​of different materials; Based on the data volume, numerically simulating the piezoelectric wafer in the numerical sample by using electrostatic physics field and solid mechanics physics field; Based on the data volume, numerically simulating the water-saturated porous medium in the numerical sample using a pressure acoustic physics field; The incoherent wave part of the hydrate-containing porous medium sample to be analyzed is obtained by using an acoustic wave analysis method; based on the incoherent wave part, a target numerical sample is matched from the incoherent wave feature fingerprint library by a preset similarity analysis method; and the structural feature of the target numerical sample is used as a structural feature identification result of the hydrate-containing porous medium sample; Wherein, the method further comprises: Meshing the model geometry results of the numerical sample using a free tetrahedral mesh; The differential equations in the electrostatic physics field, the solid mechanics physics field and the pressure acoustics physics field are solved using a transient solver through a parallel sparse direct solution method.

2. The method for identifying structural characteristics of hydrate-containing sediment samples according to claim 1, characterized in that: The matching of the target numerical sample from the incoherent wave feature fingerprint library based on the incoherent wave part by a preset similarity analysis method includes: The incoherent wave part is used as the first wave train, and the first incoherent wave waveform data in the incoherent wave feature fingerprint library is used as the second wave train; Obtaining the similarity coefficient of each group of wavelengths corresponding to the first wave train and the second wave train by the similarity coefficient analysis method; Constructing a similarity coefficient curve between the first wave train and the second wave train based on the similarity coefficient of each group of wavelengths; Performing mean processing on the similarity curve to obtain an average value of similarities between the first wave train and the second wave train; The next incoherent wave waveform data in the incoherent wave feature fingerprint library is used as the second wave train; then the process returns to the step of obtaining the similarity coefficient of each group of wavelengths corresponding to the first wave train and the second wave train by the similarity coefficient analysis method; until the average similarity between the incoherent wave and each incoherent wave waveform data in the incoherent wave feature fingerprint library is obtained; The numerical sample corresponding to the incoherent wave waveform data with the largest similarity average value is used as the target numerical sample.

3. The method for identifying structural characteristics of hydrate-containing sediment samples according to claim 1, characterized in that: The matching of the target numerical sample from the incoherent wave feature fingerprint library based on the incoherent wave part by a preset similarity analysis method includes: The incoherent wave part is used as the first wave train, and the first incoherent wave waveform data in the incoherent wave feature fingerprint library is used as the second wave train; Normalizing the first wave train to obtain a first result; Normalizing the second wave train to obtain a second result; Performing wavelet decomposition on the first result to obtain a first wavelet transform coefficient; performing wavelet decomposition on the second result to obtain a second wavelet transform coefficient; wherein the wavelet decomposition includes discrete wavelet transform or continuous wavelet transform; Performing distance measurement on the first wavelet transform coefficient and the second wavelet transform coefficient to obtain a Euclidean distance; performing regulation processing on the Euclidean distance based on a preset range; Performing similarity measurement on the first wavelet transform coefficient and the second wavelet transform coefficient to obtain cosine similarity; Averaging the reduced Euclidean distance and the cosine similarity to obtain a similarity evaluation result between the first wave train and the second wave train; The next incoherent wave waveform data in the incoherent wave feature fingerprint library is used as the second wave train; then the step of normalizing the second wave train to obtain a second result is returned to; until the similarity evaluation result between the incoherent wave and each incoherent wave waveform data in the incoherent wave feature fingerprint library is obtained; The numerical sample corresponding to the incoherent wave waveform data having the largest similarity evaluation result is used as the target numerical sample.

4. The method for identifying structural characteristics of hydrate-containing sediment samples according to claim 1, characterized in that: The matching of the target numerical sample from the incoherent wave feature fingerprint library based on the incoherent wave part by a preset similarity analysis method includes: Performing feature extraction on each of the incoherent wave waveform data in the incoherent wave feature fingerprint library to obtain feature parameters of each of the incoherent wave waveform data; The characteristic parameters are used as model inputs, and the labels of the incoherent wave waveform data corresponding to the characteristic parameters are used as model outputs to obtain a sample set; Iteratively training the pre-constructed neural network using the sample set until a preset training condition is reached to obtain a neural network model; The incoherent wave part is identified by using the neural network model to obtain an identification number; and the numerical sample corresponding to the incoherent wave waveform data marked with the identification number is used as a target numerical sample.

5. A device for identifying structural characteristics of hydrate-containing sediment samples, characterized in that: include: The first module is used to construct a finite element model of hydrate-containing porous media based on piezoelectric wafers and water-saturated porous media; The second module is used to establish a preset number of numerical samples using a control variable method based on the finite element data model; The water-saturated porous medium is used to characterize simulated sediments; the method of establishing a preset number of numerical samples based on the finite element data model using a control variable method includes: Based on the finite element data model, the sizes of the piezoelectric wafer and the water-saturated porous medium are controlled to remain unchanged, at least one structural feature of the simulated sediment is changed by using a control variable method, and a preset number of numerical samples are established; The structural characteristics include the shape and arrangement of the skeleton particles and the microscopic occurrence pattern of hydrates; The third module is used to perform numerical simulation based on the numerical sample and preset component parameters through a multi-physical field coupling method, and establish an incoherent wave feature fingerprint library based on the results of the numerical simulation; wherein the incoherent wave feature fingerprint library includes each of the numerical samples and its corresponding incoherent wave waveform data; Wherein, the component parameters include bulk modulus, shear modulus, density and longitudinal wave velocity; the numerical simulation based on the numerical sample and the preset component parameters is performed by a multi-physical field coupling method, including: Based on the material composition of the numerical sample, determining the data amount of the component parameter corresponding to the numerical sample according to a preset mapping table of the component parameters and the values ​​of different materials; Based on the data volume, numerically simulating the piezoelectric wafer in the numerical sample by using electrostatic physics field and solid mechanics physics field; Based on the data volume, numerically simulating the water-saturated porous medium in the numerical sample using a pressure acoustic physics field; The fourth module is used to obtain the incoherent wave part of the hydrate-containing porous medium sample to be analyzed by using the acoustic wave analysis method; based on the incoherent wave part, match the target numerical sample from the incoherent wave feature fingerprint library by a preset similarity analysis method; and use the structural feature of the target numerical sample as the structural feature identification result of the hydrate-containing porous medium sample; The device further comprises: The fifth module is used to mesh the model geometry results of numerical samples using free tetrahedral meshes; The sixth module is used to solve differential equations in electrostatics, solid mechanics, and pressure acoustics physics using a time-dependent solver via parallel sparse direct solution methods.

6. An electronic device, characterized in that: including a processor and a memory; The memory is used to store programs; The processor executes the program to implement the method according to any one of claims 1 to 4.

7. A computer storage medium storing a program executable by a processor, characterized in that: The program executable by the processor is used to implement the method according to any one of claims 1 to 4 when executed by the processor.

Images