CN111950192A - Method and device for modeling pore network model based on convolutional neural network - Google Patents
Method and device for modeling pore network model based on convolutional neural network Download PDFInfo
- Publication number
- CN111950192A CN111950192A CN202010680503.3A CN202010680503A CN111950192A CN 111950192 A CN111950192 A CN 111950192A CN 202010680503 A CN202010680503 A CN 202010680503A CN 111950192 A CN111950192 A CN 111950192A
- Authority
- CN
- China
- Prior art keywords
- dimensional tensor
- pore
- radius
- tensor data
- dimensional
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 239000011148 porous material Substances 0.000 title claims abstract description 162
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 24
- 238000009826 distribution Methods 0.000 claims abstract description 99
- 239000011435 rock Substances 0.000 claims abstract description 51
- 238000001228 spectrum Methods 0.000 claims abstract description 41
- 238000005481 NMR spectroscopy Methods 0.000 claims abstract description 23
- 238000013421 nuclear magnetic resonance imaging Methods 0.000 claims abstract description 7
- 230000006870 function Effects 0.000 claims description 36
- 238000010603 microCT Methods 0.000 claims description 12
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 241000023320 Luma <angiosperm> Species 0.000 claims description 4
- OSWPMRLSEDHDFF-UHFFFAOYSA-N methyl salicylate Chemical compound COC(=O)C1=CC=CC=C1O OSWPMRLSEDHDFF-UHFFFAOYSA-N 0.000 claims description 4
- BVKZGUZCCUSVTD-UHFFFAOYSA-L Carbonate Chemical compound [O-]C([O-])=O BVKZGUZCCUSVTD-UHFFFAOYSA-L 0.000 description 7
- 238000002474 experimental method Methods 0.000 description 6
- 239000012530 fluid Substances 0.000 description 4
- 238000002591 computed tomography Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000009977 dual effect Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000002247 constant time method Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- QSHDDOUJBYECFT-UHFFFAOYSA-N mercury Chemical compound [Hg] QSHDDOUJBYECFT-UHFFFAOYSA-N 0.000 description 2
- 229910052753 mercury Inorganic materials 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000005406 washing Methods 0.000 description 2
- FAPWRFPIFSIZLT-UHFFFAOYSA-M Sodium chloride Chemical compound [Na+].[Cl-] FAPWRFPIFSIZLT-UHFFFAOYSA-M 0.000 description 1
- RTAQQCXQSZGOHL-UHFFFAOYSA-N Titanium Chemical compound [Ti] RTAQQCXQSZGOHL-UHFFFAOYSA-N 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000000739 chaotic effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000001035 drying Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000035515 penetration Effects 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 230000000704 physical effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 150000003839 salts Chemical class 0.000 description 1
- 229920006395 saturated elastomer Polymers 0.000 description 1
- 239000011780 sodium chloride Substances 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000003325 tomography Methods 0.000 description 1
- 239000011800 void material Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
Abstract
The invention discloses a method and a device for modeling a pore network model based on a convolutional neural network, wherein the method comprises the following steps: acquiring a nuclear magnetic image of a rock core obtained through nuclear magnetic resonance imaging; acquiring correlation length according to the nuclear magnetic image, and calculating a three-dimensional tensor convolution kernel according to the correlation length; acquiring a T2 spectrum of a rock core obtained through nuclear magnetic resonance, and acquiring the radius frequency distribution of a pore throat according to the T2 spectrum; forming an initial three-dimensional tensor data volume according to the pore throat radius frequency distribution; generating a three-dimensional tensor data body which accords with the frequency distribution of the radius of the pore throat of the rock core by adopting a convolutional neural network forward propagation algorithm according to the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data body; constructing an unordered space structure of the pore network model according to the initial three-dimensional tensor data body; and establishing a pore network model according to the three-dimensional tensor data body which accords with the frequency distribution of the pore throat radius of the rock core and the disordered space structure.
Description
Technical Field
The disclosure relates to the field of petroleum logging interpretation, in particular to a method and a device for modeling a pore network model based on a convolutional neural network.
Background
Oil gas is an energy source with wide application and extremely important position in the world, and with the continuous development of oil gas resources in various countries, a plurality of conventional oil reservoirs are developed in a large scale and enter the middle and later stages of production. In consideration of the utilization of future oil and gas resources, research on unconventional oil and gas reservoirs with complex and various conditions needs to be deepened. Therefore, it is very important to develop research on development scale of pore space inside the porous medium, influence of spatial distribution on fluid seepage, distribution rule of fluid therein, interaction mechanism and other essential problems determining the macroscopic phenomenon of fluid flowing in the porous medium by establishing a core scale model.
Disclosure of Invention
The embodiment of the disclosure provides a method for modeling a pore network model based on a convolutional neural network, which comprises the following steps:
acquiring a nuclear magnetic image of a rock core obtained through nuclear magnetic resonance imaging;
acquiring correlation length according to the nuclear magnetic image, and calculating a three-dimensional tensor convolution kernel according to the correlation length; wherein the correlation length represents a radius average of a preselected luma in the nuclear magnetic image;
acquiring a T2 spectrum of a rock core obtained through nuclear magnetic resonance, and acquiring the radius frequency distribution of a pore throat according to the T2 spectrum; forming an initial three-dimensional tensor data volume according to the pore throat radius frequency distribution;
generating a three-dimensional tensor data body which accords with the frequency distribution of the radius of the pore throat of the rock core by adopting a convolutional neural network forward propagation algorithm according to the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data body;
constructing an unordered space structure of the pore network model according to the initial three-dimensional tensor data body;
and establishing a pore network model according to the three-dimensional tensor data body which accords with the frequency distribution of the pore throat radius of the rock core and the disordered space structure.
In an exemplary embodiment, the method further includes:
obtaining a rock core image obtained through micro-CT scanning;
acquiring the number and the direction of cracks according to the core image;
and establishing a pore network model with crack distribution according to the pore network model, the number of cracks and the crack direction.
In an exemplary embodiment, the method further comprises the following features:
the calculating a three-dimensional tensor convolution kernel according to the correlation length includes:
the three-dimensional tensor convolution kernel e (h) is calculated as follows:
E(h)＝exp(-2h/Lc)
wherein h represents (L) in a three-dimensional coordinate systemx，Ly，Lz) Is a center, the distance from the spherical surface with the radius less than or equal to Lc to the center,
Lcrepresenting the correlation length.
In an exemplary embodiment, the method further comprises the following features:
the method for obtaining the T2 spectrum of the core obtained by nuclear magnetic resonance logging and obtaining the pore throat radius distribution according to the T2 spectrum comprises the following steps:
obtaining a T2 spectrum of the core obtained by nuclear magnetic resonance; the amplitude values of the T2 spectrum are converted to a throat radius frequency distribution by a predetermined quantitative relationship.
In an exemplary embodiment, the method further comprises the following features:
the preset quantitative relation is rm＝cT2m；
Wherein r ismIs the mth pore throat radius, T2mIs the m-th amplitude value of the T2 spectrum, c is a preset conversion coefficient, and m is a positive integer.
In an exemplary embodiment, the method further comprises the following features:
forming an initial three-dimensional tensor data volume according to the pore throat radius frequency distribution, including:
establishing a three-dimensional stable random field through a random function according to the pore throat radius frequency distribution to form an initial three-dimensional tensor data body;
wherein the random function is a lognormal distribution random function as follows:
wherein the mathematical expectation μ and the standard deviation σ are obtained by fitting the frequency distribution of the throat radius, x representing the throat radius.
In an exemplary embodiment, the method further comprises the following features:
generating a three-dimensional tensor data body which accords with the frequency distribution of the radius of the pore throat of the rock core by adopting a convolutional neural network forward propagation algorithm according to the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data body, wherein the three-dimensional tensor data body comprises the following steps:
and sequentially carrying out tensor point multiplication on the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data body, and stacking point multiplication results according to the sequence in the initial three-dimensional tensor data body to generate the three-dimensional tensor data body which accords with the radius frequency distribution of the core pore throat.
In an exemplary embodiment, the method further comprises the following features:
the method for constructing the unordered space structure of the pore network model according to the initial three-dimensional tensor data body comprises the following steps:
determining the number of nodes of the disordered spatial structure according to the data quantity of the initial three-dimensional tensor data body;
constructing a three-dimensional cubic network containing X, Y, Z directions according to the number of nodes and a preset spacing distance L between the nodes;
calculating coordinates of each node in the three-dimensional cubic network;
determining whether tube bundles are communicated among all adjacent nodes in all directions, and distributing the tube bundle radius;
moving the coordinates of each node through a preset rule;
and generating an unordered space structure according to the three-dimensional cubic network, the result of determining whether the tube bundles are communicated, the distributed tube bundle radius and the moved node coordinates.
In an exemplary embodiment, the method further comprises the following features:
obtaining a rock core image obtained through micro-CT scanning;
acquiring the average throat length and coordination number according to the core image; the configuration number refers to the number of nodes which are communicated with the tube bundle and exist between one node and an adjacent node;
the preset spacing distance L between each node is the length of the average throat;
the determining whether the adjacent nodes in each direction are communicated with each other by a tube bundle comprises the following steps:
determining whether tube bundles are communicated between every two adjacent nodes in the X direction according to a preset first probability function;
determining whether tube bundles are communicated between adjacent nodes in the Y direction according to a preset second probability function;
alternatively, the first and second electrodes may be,
and determining whether the adjacent nodes in each direction are communicated with each other by the tube bundles according to the coordination number.
In an exemplary embodiment, the method further comprises the following features:
the moving the coordinates of each node through a preset rule includes:
move each node coordinate (x, y, z) as follows:
(x,y,z)＝[(i-1)L±rand( )％(0.5L),(j-1)L±rand( )％(0.5L),(k-1)L±rand( )％(0.5L)]
wherein i is a node number in the x direction, j is a node number in the y direction, k is a node number in the z direction, i, j, and k are integers greater than 0, and rand ()% (0.5L) represents an arbitrary integer randomly generated within a range of 0.5L.
In an exemplary embodiment, the method further comprises the following features: establishing a pore network model according to the three-dimensional tensor data body which accords with the frequency distribution of the pore throat radius of the rock core and the disordered space structure, wherein the pore network model comprises the following steps:
and sequentially assigning the three-dimensional tensor data body which conforms to the radius frequency distribution of the core pore throat to the nodes of the disordered space structure to establish a pore network model.
The present disclosure also provides a modeling apparatus for a pore network model based on a convolutional neural network, including: a memory and a processor;
the memory is used for storing a program for modeling a pore network model of the core;
the processor is used for reading a program for modeling the pore network model of the core, and executing the following modeling method:
acquiring a nuclear magnetic image of a rock core obtained through nuclear magnetic resonance imaging;
acquiring correlation length according to the nuclear magnetic image, and calculating a three-dimensional tensor convolution kernel according to the correlation length; wherein the correlation length represents a radius average of a preselected luma in the nuclear magnetic image;
acquiring a T2 spectrum of a rock core obtained through nuclear magnetic resonance, and acquiring the radius frequency distribution of a pore throat according to the T2 spectrum; forming an initial three-dimensional tensor data volume according to the pore throat radius frequency distribution;
generating a three-dimensional tensor data body which accords with the frequency distribution of the radius of the pore throat of the rock core by adopting a convolutional neural network forward propagation algorithm according to the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data body;
constructing an unordered space structure of the pore network model according to the initial three-dimensional tensor data body;
and establishing a pore network model according to the three-dimensional tensor data body which accords with the frequency distribution of the pore throat radius of the rock core and the disordered space structure.
In an exemplary embodiment, the processor is configured to read a program for performing the modeling of the pore network model for the core, and further perform a modeling method as follows:
obtaining a rock core image obtained through micro-CT scanning;
acquiring the number and the direction of cracks according to the core image;
and establishing a pore network model with crack distribution according to the pore network model, the number of cracks and the crack direction.
In an exemplary embodiment, said computing a three-dimensional tensor convolution kernel from said correlation length comprises:
the three-dimensional tensor convolution kernel e (h) is calculated as follows:
E(h)＝exp(-2h/Lc)
wherein h represents (L) in a three-dimensional coordinate systemx，Ly，Lz) Is the sphere center, the distance from the sphere surface with the radius less than or equal to Lc to the sphere center,
Lcrepresenting the correlation length.
In an exemplary embodiment, the obtaining a T2 spectrum of a core from a nuclear magnetic resonance log, the obtaining a pore throat radius distribution from the T2 spectrum, includes:
obtaining a T2 spectrum of the core obtained by nuclear magnetic resonance; the amplitude values of the T2 spectrum are converted to a throat radius frequency distribution by a predetermined quantitative relationship.
In an exemplary embodiment, the predetermined quantitative relationship is rm＝cT2m；
Wherein r ismIs the mth pore throat radius, T2mIs the m-th amplitude value of the T2 spectrum, c is a preset conversion coefficient, and m is a positive integer.
In an exemplary embodiment, the forming an initial three-dimensional tensor data volume from the pore throat radius frequency distribution includes:
establishing a three-dimensional stable random field through a random function according to the pore throat radius frequency distribution to form an initial three-dimensional tensor data body;
wherein the random function is a lognormal distribution random function as follows:
wherein the mathematical expectation μ and the standard deviation σ are obtained by fitting the frequency distribution of the throat radius, x representing the throat radius.
In an exemplary embodiment, the generating a three-dimensional tensor data volume conforming to a frequency distribution of a pore throat radius of a core by using a convolutional neural network forward propagation algorithm according to the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data volume includes:
and sequentially carrying out tensor point multiplication on the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data body, and stacking point multiplication results according to the sequence in the initial three-dimensional tensor data body to generate the three-dimensional tensor data body which accords with the radius frequency distribution of the core pore throat.
In an exemplary embodiment, the constructing an unordered spatial structure of a pore network model from the initial three-dimensional tensor data volume includes:
determining the number of nodes of the disordered spatial structure according to the data quantity of the initial three-dimensional tensor data body;
constructing a three-dimensional cubic network containing X, Y, Z directions according to the number of nodes and a preset spacing distance L between the nodes;
calculating coordinates of each node in the three-dimensional cubic network;
determining whether tube bundles are communicated among all adjacent nodes in all directions, and distributing the tube bundle radius;
moving the coordinates of each node through a preset rule;
and generating an unordered space structure according to the three-dimensional cubic network, the result of determining whether the tube bundles are communicated, the distributed tube bundle radius and the moved node coordinates.
In an exemplary embodiment, the processor is configured to read a program for performing the modeling of the pore network model for the core, and further perform a modeling method as follows: obtaining a rock core image obtained through micro-CT scanning;
acquiring the average throat length and coordination number according to the core image; the configuration number refers to the number of nodes which are communicated with the tube bundle and exist between one node and an adjacent node;
the preset spacing distance L between each node is the length of the average throat;
the determining whether the adjacent nodes in each direction are communicated with each other by a tube bundle comprises the following steps:
determining whether tube bundles are communicated between every two adjacent nodes in the X direction according to a preset first probability function;
determining whether tube bundles are communicated between adjacent nodes in the Y direction according to a preset second probability function;
alternatively, the first and second electrodes may be,
and determining whether the adjacent nodes in each direction are communicated with each other by the tube bundles according to the coordination number.
In an exemplary embodiment, the moving the coordinates of each node according to the preset rule includes:
move each node coordinate (x, y, z) as follows:
(x,y,z)＝[(i-1)L±rand( )％(0.5L),(j-1)L±rand( )％(0.5L),(k-1)L±rand( )％(0.5L)]
wherein i is a node number in the x direction, j is a node number in the y direction, k is a node number in the z direction, i, j, and k are integers greater than 0, and rand ()% (0.5L) represents an arbitrary integer randomly generated within a range of 0.5L.
In an exemplary embodiment, establishing a pore network model according to the three-dimensional tensor data volume conforming to the frequency distribution of the pore throat radius of the core and the disordered spatial structure includes:
and sequentially assigning the three-dimensional tensor data body which conforms to the radius frequency distribution of the core pore throat to the nodes of the disordered space structure to establish a pore network model.
Drawings
Fig. 1 is a flowchart of a method for modeling a convolutional neural network-based pore network model according to an embodiment of the present disclosure.
FIG. 2 is an example mercury intrusion pore throat distribution and relaxation distribution of an embodiment of the disclosure.
FIG. 3 is an example of a three-dimensional convolution kernel of an embodiment of the present disclosure.
Fig. 4 is a schematic diagram of generating a three-dimensional tensor data volume conforming to a pore throat radius frequency distribution according to an embodiment of the disclosure.
Fig. 5 is an example of an aperture image reconstructed from a CT scan according to an embodiment of the disclosure.
Fig. 6 is an example of a fracture image reconstructed from a CT scan according to an embodiment of the disclosure.
FIG. 7 is an example ordered spatial structure of an embodiment of the present disclosure.
Fig. 8 is an example of an unordered spatial structure of an embodiment of the present disclosure.
Fig. 9 is an example of a micropore-micropore dual media model image of an embodiment of the disclosure.
Fig. 10 is an example of a micropore-nanopore-fracture triple media model image of an embodiment of the present disclosure.
Fig. 11 is a schematic diagram of a modeling apparatus for a convolutional neural network-based pore network model according to an embodiment of the present disclosure.
Detailed Description
To make the objects, technical solutions and advantages of the present disclosure more apparent, embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. It should be noted that the embodiments and features of the embodiments in the present application may be arbitrarily combined with each other without conflict.
Fig. 1 is a flowchart of a modeling method of a pore network model based on a convolutional neural network according to an embodiment of the present disclosure, and as shown in fig. 1, the modeling method of the embodiment includes:
and S11, acquiring a nuclear magnetic image of the rock core obtained through nuclear magnetic resonance imaging.
S12, obtaining correlation length according to the nuclear magnetic image, and calculating a three-dimensional tensor convolution kernel according to the correlation length;
s13, obtaining a T2 spectrum of the rock core obtained through nuclear magnetic resonance, and obtaining the radius frequency distribution of the pore throat according to the T2 spectrum; forming an initial three-dimensional tensor data volume according to the pore throat radius frequency distribution;
s14, generating a three-dimensional tensor data body which accords with the frequency distribution of the pore throat radius of the rock core by adopting a convolutional neural network forward propagation algorithm according to the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data body;
s15, constructing an unordered space structure of the pore network model according to the initial three-dimensional tensor data body;
and S16, establishing a pore network model according to the three-dimensional tensor data body which accords with the frequency distribution of the pore throat radius of the rock core and the disordered space structure.
Among the above steps, S13 may be performed before or after S11, S12, and may be performed in parallel with S11, S12; the sequence of S14 and S15 is not limited and may be performed in parallel.
In an exemplary embodiment, the method further comprises:
obtaining a rock core image obtained through micro-CT scanning;
acquiring the number and the direction of cracks according to the core image;
and establishing a pore network model with crack distribution according to the pore network model, the number of cracks and the crack direction.
The method is not limited to the micro-CT scanning mode for obtaining the core image so as to determine the number and the direction of the cracks. Other ways of determining the number and direction of the cracks are also possible. For example, the number and direction of cracks may be set manually or determined by a random function.
In an exemplary embodiment, said computing a three-dimensional tensor convolution kernel from said correlation length comprises:
the three-dimensional tensor convolution kernel e (h) is calculated as follows:
E(h)＝exp(-2h/Lc)
wherein h represents (L) in a three-dimensional coordinate systemx，Ly，Lz) Is the sphere center, the distance from the sphere surface with the radius less than or equal to Lc to the sphere center,
Lcrepresenting the correlation length.
The three-dimensional tensor convolution kernel may be computed using the above equation or a modified equation of the equation.
In an exemplary embodiment, the obtaining a T2 spectrum of a core from a nuclear magnetic resonance log, the obtaining a pore throat radius distribution from the T2 spectrum, includes:
obtaining a T2 spectrum of the core obtained by nuclear magnetic resonance; the amplitude values of the T2 spectrum are converted to a throat radius frequency distribution by a predetermined quantitative relationship.
The preset quantitative relation may be a corresponding relation between a preset amplitude value of the T2 spectrum and the frequency distribution of the pore throat radius, or a preset calculation formula for calculating the pore throat radius through the amplitude value.
In an exemplary embodiment, the method includes:
the preset quantitative relation is rm＝cT2m；
Wherein r ismIs the mth pore throat radius, T2mIs the second of the T2 spectrumm amplitude values, c is a preset conversion coefficient, and m is a positive integer.
In an exemplary embodiment, the forming an initial three-dimensional tensor data volume from the pore throat radius frequency distribution includes:
establishing a three-dimensional stable random field through a random function according to the pore throat radius frequency distribution to form an initial three-dimensional tensor data body;
wherein the random function is a lognormal distribution random function as follows:
wherein the mathematical expectation μ and the standard deviation σ are obtained by fitting the frequency distribution of the throat radius, x representing the throat radius.
The random function is not limited to the above formula, and a logarithmic uniform random function may be used.
In an exemplary embodiment, the generating a three-dimensional tensor data volume conforming to a frequency distribution of a pore throat radius of a core by using a convolutional neural network forward propagation algorithm according to the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data volume includes:
and sequentially carrying out tensor point multiplication on the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data body, and stacking point multiplication results according to the sequence in the initial three-dimensional tensor data body to generate the three-dimensional tensor data body which accords with the radius frequency distribution of the core pore throat.
In an exemplary embodiment, the constructing an unordered spatial structure of a pore network model from the initial three-dimensional tensor data volume includes:
determining the number of nodes of the disordered spatial structure according to the data quantity of the initial three-dimensional tensor data body;
constructing a three-dimensional cubic network containing X, Y, Z directions according to the number of nodes and a preset spacing distance L between the nodes;
calculating coordinates of each node in the three-dimensional cubic network;
determining whether tube bundles are communicated among all adjacent nodes in all directions, and distributing the tube bundle radius;
moving the coordinates of each node through a preset rule;
and generating an unordered space structure according to the three-dimensional cubic network, the result of determining whether the tube bundles are communicated, the distributed tube bundle radius and the moved node coordinates.
In an exemplary embodiment, the method further comprises,
obtaining a rock core image obtained through micro-CT scanning;
acquiring the average throat length and coordination number according to the core image; the configuration number refers to the number of nodes which are communicated with the tube bundle and exist between one node and an adjacent node;
the preset spacing distance L between each node is the length of the average throat;
the determining whether the adjacent nodes in each direction are communicated with each other by a tube bundle comprises the following steps:
determining whether tube bundles are communicated between every two adjacent nodes in the X direction according to a preset first probability function;
determining whether tube bundles are communicated between adjacent nodes in the Y direction according to a preset second probability function;
alternatively, the first and second electrodes may be,
and determining whether the adjacent nodes in each direction are communicated with each other by the tube bundles according to the coordination number.
In an exemplary embodiment, the moving the coordinates of each node according to the preset rule includes:
move each node coordinate (x, y, z) as follows:
(x,y,z)＝[(i-1)L±rand( )％(0.5L),(j-1)L±rand( )％(0.5L),(k-1)L±rand( )％(0.5L)]
wherein i is a node number in the x direction, j is a node number in the y direction, k is a node number in the z direction, i, j, and k are integers greater than 0, and rand ()% (0.5L) represents an arbitrary integer randomly generated within a range of 0.5L.
In an exemplary embodiment, establishing a pore network model according to the three-dimensional tensor data volume conforming to the frequency distribution of the pore throat radius of the core and the disordered spatial structure includes:
and sequentially assigning the three-dimensional tensor data body which conforms to the radius frequency distribution of the core pore throat to the nodes of the disordered space structure to establish a pore network model.
In one exemplary embodiment, the nuclear magnetic resonance T2 spectrum and the nuclear magnetic resonance image of the core may be obtained as follows.
And measuring a target rock core, performing oil washing and salt washing on the carbonate rock core collected from the stratum for the target rock core, fully drying the target rock core at the temperature of 80 ℃ until the weight of the target rock core is unchanged, and performing a nuclear magnetic resonance measurement experiment after the carbonate rock core is saturated for 48 hours by using a vacuum pressurization saturator and KCl2 saline as a medium to obtain a rock core nuclear magnetic image.
Then, combining the nuclear magnetic T2 spectrum (obtained by nuclear magnetic resonance experiment) with the core pore size distribution (obtained by conventional rock mercury intrusion experiment), a conversion coefficient c (the value has regional experience) value can be obtained, and the abscissa of the T2 spectrum is multiplied by c to obtain the pore size distribution, as shown in fig. 2.
Images of the cross section, the coronal plane and the sagittal plane of the rock sample can be obtained by utilizing the nuclear magnetic resonance imaging. For dual medium rock, the distribution of single-phase fluid in the core space can be represented using image signals. The brighter the image pixel, the larger the pore throat radius, and the location in the image where the bright color is concentrated is indicated as a crater. Conversely, the darker the image, the more resolution-unrecognizable regions are represented, and the smaller the pore throat radius. Therefore, the distribution characteristics, the relative length, of the solution pores or solution pore development zones inside the core can be observed from the two-dimensional image.
In an exemplary embodiment, the convolution kernel may be generated as follows.
Fig. 3 is an example of a convolution kernel, which is obtained by obtaining features of a nuclear magnetic resonance image according to an empirical relationship proposed by prior geostatistices:
E(h)＝exp(-2h/Lc)
lc-correlation length, obtained from nuclear magnetic resonance image analysis; h represents (L) in a three-dimensional coordinate systemx，Ly，Lz) Is the sphere center, the distance from the sphere surface with the radius less than or equal to Lc to the sphere center,
for example, the three-dimensional convolution kernel can be implemented by the following procedure.
In an exemplary embodiment, the three-dimensional tensor data volume may be formed as follows.
According to the pore throat radius distribution characteristics extracted by a nuclear magnetic resonance T2 spectrum, a three-dimensional stable random field is established through a random function (such as logarithmic uniformity or logarithmic normality) to form a three-dimensional tensor data body. For example, the random function may be a lognormal distribution random function as follows:
wherein the mathematical expectation μ and the standard deviation σ are obtained by fitting the frequency distribution of the throat radius, x representing the throat radius.
Three-dimensional convolution is performed with the three-dimensional tensor data volume as input using a Tensorflow platform. As shown in fig. 4, a convolution kernel is used to slide each element of the input three-dimensional stable random field tensor data volume, the result is formed into a new matrix, finally, the result matrices of all channels are stacked in the original sequence to form a third-order tensor, and the third-order tensor is used as an output to obtain the pore throat radius three-dimensional tensor data volume, wherein the data volume contains pore throat radius distribution space information of (micro) pore-pore double media.
In one exemplary embodiment, a carbonate CT image with a microporosity-solvus-fracture triple medium can be generated as follows.
As shown in fig. 5 and 6, a sample is scanned based on a computer high-resolution tomography (i.e., MicroCT) imaging technology, a pore network is constructed by an equivalent sphere method and a maximum sphere method respectively for 3D reconstruction of a digital core, and structural characteristics of a reservoir are analyzed. Wherein the throat length can be calculated by:
L＝D-R1-R2
wherein, R1 and R2 are respectively the radius of two pores connected by the throat and have the unit of mum; d is the actual coordinate distance of the central points of the two pores and has the unit of mum.
The results of the micro-CT experiment treatment were as follows: resolution was 8 μm, size was 710 × 710 μm, the colour concentrate was porous, the remainder was porous, analytical porosity was 1.12%, together with a volume percentage of 71.3%, an average pore radius of 18.97 μm, an average throat radius of 17.8 μm, an average throat length of 131.8 μm, a coordination number of 1.13. The red ball in fig. 5 is a void; the white bar is the throat; the individual connected pore volumes of fig. 6 are shown in size order, and it can be readily seen that large pores are distributed less and small pores predominate in the range where resolution is visible. Meanwhile, the distribution of the dissolved pores and cracks of the carbonate rock can be obviously observed, and the number, the direction, the number, the volume ratio and the pore throat radius distribution of the dissolved pores are counted on the basis of the distribution of the dissolved pores and the cracks.
In an exemplary embodiment, the chaotic space structure may be constructed as follows.
The modeling method of the unordered network model (as shown in fig. 8) is described by a network model of a square grid. The ordered network model (as shown in fig. 7) is characterized by regular shape of the model and fixed positions of the nodes. According to the characteristic, a modeling method of the unordered network model is provided.
(1) The number of nodes of the model (the number of nodes corresponds to the number of data of the three-dimensional tensor data volume) is specified, and an X multiplied by Y multiplied by Z three-dimensional simple cubic grid is constructed. Each node represents a pore, and the nodes are connected by a throat. Six throats are connected around each node representing pore in the network established by the method; similarly, six apertures are connected around each throat. The spacing distance between nodes in each direction (namely the x direction, the y direction and the z direction) is set to be L (the average roar length can be determined to be 131.8 mu m as the node distance according to the result of the micro CT experiment), the number of nodes is set to be d, and the side length of the model is (d-1) multiplied by L.
(2) And calculating the coordinates of each node in the network model. The calculation formula is as follows: (x, y, z) [ (i-1) l, (j-1) l, (k-1) l ], where i, j, and k are node numbers in the x, y, and z directions, respectively, and take values of 1, 2, and 3 …, respectively.
(3) And setting a probability function with the probability p in a program, and determining whether the pipe bundles are communicated between every two adjacent nodes in the x direction through a (pseudo) random number generator. A function is used to generate random numbers and thus random probabilities. In the C/C + + programming language, random numbers can be generated using the rand () function, thereby generating random probabilities. The specific C/C + + code is:
if(rand( )％100＜p×100)
wherein rand ()% 100-computer randomly generates any integer in the range of 0-99.
When the penetration probability p is 50%, the probability (probability) of 50% among the randomly generated integers of the rand () function is less than 50, and the probability (probability) of the other 50% is more than 50. Therefore, the expression can realize bundle connection with a probability p of 50%, that is, when a number smaller than 50 is generated, the expression is true, and the task of allocating the bundle radius (allocating the bundle radius r) is performed; otherwise, false, no operation is performed.
(4) And establishing a connection probability function through a (pseudo) random number generator to determine whether tube bundle connection exists between every two adjacent nodes in the y direction and the z direction. The method is the same as the x-direction tube bundle distribution process.
The tube bundle communication of the node can be determined by the coordination number obtained by the micro CT experiment result to replace the (3) and (4).
(5) And moving each node coordinate, and generating a disordered space structure according to the three-dimensional cube network, the result of determining whether the tube bundles are communicated, the distributed tube bundle radius and the moved node coordinate.
The node coordinates (x, y, z) may be moved as follows:
(x,y,z)＝[(i-1)L±rand( )％(0.5L),(j-1)L±rand( )％(0.5L),(k-1)L±rand( )％(0.5L)]
wherein i is a node number in the x direction, j is a node number in the y direction, k is a node number in the z direction, i, j, and k are integers greater than 0, and rand ()% (0.5L) represents an arbitrary integer randomly generated within a range of 0.5L.
In an exemplary embodiment, a pore network model of a (micro) pore-pore dual media feature may be generated as follows.
And sequentially assigning the data in the three-dimensional tensor data volume to the pore throat radius grid points of the disordered structure network model to construct a pore network model with (micro) pore-pore dual-dissolved-pore dual-medium characteristics (as shown in fig. 9, the model dimension is length multiplied by width multiplied by height multiplied by 3cm multiplied by 1 cm).
In an exemplary embodiment, a pore network model of the (micro) pore-fracture triple media characteristics may be generated as follows.
And (micro) pore-crack triple media are constructed by inserting a plane equation into the constructed pore network model nodes with the characteristics of the (micro) pore-pore double media to generate a two-dimensional plane crack. Let n be the plane normal vector:
m (x, y, z), N (x0, y0, z0) are two arbitrary points on the plane, then:
the point-normal equation for the plane is thus:
A(x-x0)+B(y-y0)+C(z-z0)＝0
the spatial circular face crack can be generated by the following method:
taking a random point on a normal vector as a circle center o (a, B, C), wherein a is a random number from 0 to A, B is a random number from 0 to B, and C is a random number from 0 to C, firstly calculating a spherical equation:
(x-a)2+(y-b)2+(z-c)2＝r2
wherein the radius r is the radius of the sphere; so that the equation of the plane where the circle is located satisfies
From both, the equation of the circular plane (circular plane fracture) can be derived. The number of the cracks is determined by the number of the plane equations, and the number i of the cracks in the rock core with the diameter of 2.5cm and the length of 3cm of a plurality of samples is counted through CT experimental results, wherein the cracks approximately trend; and controlling the angle and the distribution position of the cracks by taking the ni normal vector as a reference (generating random point coordinates), and controlling the number of generated cracks by using the cycle number.
The GPU is utilized to calculate and call a workstation provided with two TITAN display cards in parallel, a surface equation is inserted into the micropore-pore dissolving model, and a crack medium is generated as shown in figure 11 (n is a normal vector, and M, N is any two points on a plane). In a CT experimental sample with the diameter of 2.5cm, the length of 4cm and the resolution of 8 mu m, the number of cracks (containing microcracks) is counted to be 4, the distribution is random, the trend of the cracks is random, and a random surface equation is inserted in a mode of randomly generating a normal vector and a central point position to obtain a final carbonate rock micropore-pore-solution-crack model (as shown in figure 10).
The carbonate rock core model established in the embodiment combines an experimental method and a computer algorithm, a nuclear magnetic imaging graph characteristic value and a correlation length are processed through a convolutional neural network to generate a model, and meanwhile, reservoir physical property parameters are added into the model through a micro-nano CT scanning experiment, so that the problem that a geological model established by the existing modeling technology lacks certain physical significance is solved to a certain extent. By combining with CT experimental analysis, a micropore-dissolved pore-crack network model is generated in a mode of adding a random surface equation into the model, and the problem that the carbonate rock micropore-dissolved pore-crack triple medium is difficult to model is solved to a certain extent.
Fig. 11 is a schematic diagram of a modeling apparatus for a convolutional neural network-based pore network model according to an embodiment of the present disclosure. The modeling apparatus includes: a memory and a processor;
the memory is used for storing a program for modeling a pore network model of the core;
the processor is used for reading a program for modeling the pore network model of the core, and executing the following modeling method:
acquiring a nuclear magnetic image of a rock core obtained through nuclear magnetic resonance imaging;
acquiring correlation length according to the nuclear magnetic image, and calculating a three-dimensional tensor convolution kernel according to the correlation length; wherein the correlation length represents a radius average of a preselected luma in the nuclear magnetic image;
acquiring a T2 spectrum of a rock core obtained through nuclear magnetic resonance, and acquiring the radius frequency distribution of a pore throat according to the T2 spectrum; forming an initial three-dimensional tensor data volume according to the pore throat radius frequency distribution;
generating a three-dimensional tensor data body which accords with the frequency distribution of the radius of the pore throat of the rock core by adopting a convolutional neural network forward propagation algorithm according to the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data body;
constructing an unordered space structure of the pore network model according to the initial three-dimensional tensor data body;
and establishing a pore network model according to the three-dimensional tensor data body which accords with the frequency distribution of the pore throat radius of the rock core and the disordered space structure.
In an exemplary embodiment, the processor is configured to read a program for performing the modeling of the pore network model for the core, and further perform a modeling method as follows:
obtaining a rock core image obtained through micro-CT scanning;
acquiring the number and the direction of cracks according to the core image;
and establishing a pore network model with crack distribution according to the pore network model, the number of cracks and the crack direction.
In an exemplary embodiment, said computing a three-dimensional tensor convolution kernel from said correlation length comprises:
the three-dimensional tensor convolution kernel e (h) is calculated as follows:
E(h)＝exp(-2h/Lc)
wherein h represents (L) in a three-dimensional coordinate systemx，Ly，Lz) Is the sphere center, the distance from the sphere surface with the radius less than or equal to Lc to the sphere center,
Lcrepresenting the correlation length.
In an exemplary embodiment, the obtaining a T2 spectrum of a core from a nuclear magnetic resonance log, the obtaining a pore throat radius distribution from the T2 spectrum, includes:
obtaining a T2 spectrum of the core obtained by nuclear magnetic resonance; the amplitude values of the T2 spectrum are converted to a throat radius frequency distribution by a predetermined quantitative relationship.
In an exemplary embodiment, the predetermined quantitative relationship is rm＝cT2m；
Wherein r ismIs the mth pore throat radius, T2mIs the m-th amplitude value of the T2 spectrum, c is a preset conversion coefficient, and m is a positive integer.
In an exemplary embodiment, the forming an initial three-dimensional tensor data volume from the pore throat radius frequency distribution includes:
establishing a three-dimensional stable random field through a random function according to the pore throat radius frequency distribution to form an initial three-dimensional tensor data body;
wherein the random function is a lognormal distribution random function as follows:
wherein the mathematical expectation μ and the standard deviation σ are obtained by fitting the frequency distribution of the throat radius, x representing the throat radius.
In an exemplary embodiment, the generating a three-dimensional tensor data volume conforming to a frequency distribution of a pore throat radius of a core by using a convolutional neural network forward propagation algorithm according to the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data volume includes:
and sequentially carrying out tensor point multiplication on the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data body, and stacking point multiplication results according to the sequence in the initial three-dimensional tensor data body to generate the three-dimensional tensor data body which accords with the radius frequency distribution of the core pore throat.
In an exemplary embodiment, the constructing an unordered spatial structure of a pore network model from the initial three-dimensional tensor data volume includes:
determining the number of nodes of the disordered spatial structure according to the data quantity of the initial three-dimensional tensor data body;
constructing a three-dimensional cubic network containing X, Y, Z directions according to the number of nodes and a preset spacing distance L between the nodes;
calculating coordinates of each node in the three-dimensional cubic network;
determining whether tube bundles are communicated among all adjacent nodes in all directions, and distributing the tube bundle radius;
moving the coordinates of each node through a preset rule;
and generating an unordered space structure according to the three-dimensional cubic network, the result of determining whether the tube bundles are communicated, the distributed tube bundle radius and the moved node coordinates.
In an exemplary embodiment, the processor is configured to read a program for performing the modeling of the pore network model for the core, and further perform a modeling method as follows: obtaining a rock core image obtained through micro-CT scanning;
acquiring the average throat length and coordination number according to the core image; the configuration number refers to the number of nodes which are communicated with the tube bundle and exist between one node and an adjacent node;
the preset spacing distance L between each node is the length of the average throat;
the determining whether the adjacent nodes in each direction are communicated with each other by a tube bundle comprises the following steps:
determining whether tube bundles are communicated between every two adjacent nodes in the X direction according to a preset first probability function;
determining whether tube bundles are communicated between adjacent nodes in the Y direction according to a preset second probability function;
alternatively, the first and second electrodes may be,
and determining whether the adjacent nodes in each direction are communicated with each other by the tube bundles according to the coordination number.
In an exemplary embodiment, the moving the coordinates of each node according to the preset rule includes:
move each node coordinate (x, y, z) as follows:
(x,y,z)＝[(i-1)L±rand( )％(0.5L),(j-1)L±rand( )％(0.5L),(k-1)L±rand( )％(0.5L)]
wherein i is a node number in the x direction, j is a node number in the y direction, k is a node number in the z direction, i, j, and k are integers greater than 0, and rand ()% (0.5L) represents an arbitrary integer randomly generated within a range of 0.5L.
In an exemplary embodiment, establishing a pore network model according to the three-dimensional tensor data volume conforming to the frequency distribution of the pore throat radius of the core and the disordered spatial structure includes:
and sequentially assigning the three-dimensional tensor data body which conforms to the radius frequency distribution of the core pore throat to the nodes of the disordered space structure to establish a pore network model.
It will be understood by those skilled in the art that all or part of the steps of the above methods may be implemented by instructing the relevant hardware through a program, and the program may be stored in a computer readable storage medium, such as a read-only memory, a magnetic or optical disk, and the like. Alternatively, all or part of the steps of the above embodiments may be implemented using one or more integrated circuits. Accordingly, each module/unit in the above embodiments may be implemented in the form of hardware, and may also be implemented in the form of a software functional module. The present disclosure is not limited to any specific form of combination of hardware and software.
The foregoing is only a preferred embodiment of the present disclosure, and there are certainly many other embodiments of the present disclosure, which will become apparent to those skilled in the art from this disclosure and it is therefore intended that various changes and modifications can be made herein without departing from the spirit and scope of the disclosure as defined in the appended claims.
Claims (12)
1. A method for modeling a pore network model based on a convolutional neural network comprises the following steps:
acquiring a nuclear magnetic image of a rock core obtained through nuclear magnetic resonance imaging;
acquiring correlation length according to the nuclear magnetic image, and calculating a three-dimensional tensor convolution kernel according to the correlation length; wherein the correlation length represents a radius average of a preselected luma in the nuclear magnetic image;
acquiring a T2 spectrum of a rock core obtained through nuclear magnetic resonance, and acquiring the radius frequency distribution of a pore throat according to the T2 spectrum; forming an initial three-dimensional tensor data volume according to the pore throat radius frequency distribution;
generating a three-dimensional tensor data body which accords with the frequency distribution of the radius of the pore throat of the rock core by adopting a convolutional neural network forward propagation algorithm according to the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data body;
constructing an unordered space structure of the pore network model according to the initial three-dimensional tensor data body;
and establishing a pore network model according to the three-dimensional tensor data body which accords with the frequency distribution of the pore throat radius of the rock core and the disordered space structure.
2. The method of claim 1, further comprising:
obtaining a rock core image obtained through micro-CT scanning;
acquiring the number and the direction of cracks according to the core image;
and establishing a pore network model with crack distribution according to the pore network model, the number of cracks and the crack direction.
3. The method of claim 1, said computing a three-dimensional tensor convolution kernel from the correlation length, comprising:
the three-dimensional tensor convolution kernel e (h) is calculated as follows:
E(h)＝exp(-2h/Lc)
wherein h represents (L) in a three-dimensional coordinate systemx，Ly，Lz) Is a center, the distance from the spherical surface with the radius less than or equal to Lc to the center,
Lcrepresenting the correlation length.
4. The method of claim 1, wherein the obtaining a T2 spectrum of a core from a nuclear magnetic resonance log, and the obtaining a pore throat radius distribution from the T2 spectrum, comprises:
obtaining a T2 spectrum of the core obtained by nuclear magnetic resonance; the amplitude values of the T2 spectrum are converted to a throat radius frequency distribution by a predetermined quantitative relationship.
5. The method of claim 4, comprising:
the preset quantitative relation is rm＝cT2m；
Wherein r ismIs the mth pore throat radius, T2mIs the m-th amplitude value of the T2 spectrum, c is a preset conversion coefficient, and m is a positive integer.
6. The method of claim 4, said forming an initial three-dimensional tensor data volume from the pore throat radius frequency distribution, comprising:
establishing a three-dimensional stable random field through a random function according to the pore throat radius frequency distribution to form an initial three-dimensional tensor data body;
wherein the random function is a lognormal distribution random function as follows:
wherein the mathematical expectation μ and the standard deviation σ are obtained by fitting the frequency distribution of the throat radius, x representing the throat radius.
7. The method of claim 1, wherein generating a three-dimensional tensor data volume conforming to a core pore throat radius frequency distribution using a convolutional neural network forward propagation algorithm from the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data volume comprises:
and sequentially carrying out tensor point multiplication on the three-dimensional tensor convolution kernel and the initial three-dimensional tensor data body, and stacking point multiplication results according to the sequence in the initial three-dimensional tensor data body to generate the three-dimensional tensor data body which accords with the radius frequency distribution of the core pore throat.
8. The method of claim 1, said constructing an unordered spatial structure of a pore network model from said initial three-dimensional tensor data volume, comprising:
determining the number of nodes of the disordered spatial structure according to the data quantity of the initial three-dimensional tensor data body;
constructing a three-dimensional cubic network containing X, Y, Z directions according to the number of nodes and a preset spacing distance L between the nodes;
calculating coordinates of each node in the three-dimensional cubic network;
determining whether tube bundles are communicated among all adjacent nodes in all directions, and distributing the tube bundle radius;
moving the coordinates of each node through a preset rule;
and generating an unordered space structure according to the three-dimensional cubic network, the result of determining whether the tube bundles are communicated, the distributed tube bundle radius and the moved node coordinates.
9. The method of claim 8, comprising:
obtaining a rock core image obtained through micro-CT scanning;
acquiring the average throat length and coordination number according to the core image; the configuration number refers to the number of nodes which are communicated with the tube bundle and exist between one node and an adjacent node;
the preset spacing distance L between each node is the length of the average throat;
the determining whether the adjacent nodes in each direction are communicated with each other by a tube bundle comprises the following steps:
determining whether tube bundles are communicated between every two adjacent nodes in the X direction according to a preset first probability function;
determining whether tube bundles are communicated between adjacent nodes in the Y direction according to a preset second probability function;
alternatively, the first and second electrodes may be,
and determining whether the adjacent nodes in each direction are communicated with each other by the tube bundles according to the coordination number.
10. The method according to claim 8 or 9, wherein the moving the coordinates of each node according to the preset rule comprises:
move each node coordinate (x, y, z) as follows:
(x,y,z)＝[(i-1)L±rand()％(0.5L),(j-1)L±rand()％(0.5L),(k-1)L±rand()％(0.5L)]
wherein i is a node number in the x direction, j is a node number in the y direction, k is a node number in the z direction, i, j, and k are integers greater than 0, and rand ()% (0.5L) represents an arbitrary integer randomly generated within a range of 0.5L.
11. The method as claimed in claim 1, establishing a pore network model according to the three-dimensional tensor data volume conforming to the frequency distribution of the pore throat radius of the core and the disordered spatial structure, comprising:
and sequentially assigning the three-dimensional tensor data body which conforms to the radius frequency distribution of the core pore throat to the nodes of the disordered space structure to establish a pore network model.
12. A convolutional neural network-based modeling apparatus for a pore network model, comprising: a memory and a processor; the method is characterized in that:
the memory is used for storing a program for modeling a pore network model of the core;
the processor is configured to read a program for performing the modeling of the pore network model for the core, and perform the modeling method according to any one of claims 1 to 11.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010680503.3A CN111950192B (en) | 2020-07-15 | 2020-07-15 | Modeling method and device for pore network model based on convolutional neural network |
| PCT/CN2020/126213 WO2022011894A1 (en) | 2020-07-15 | 2020-11-03 | Convolutional neural network-based modeling method and device for pore network model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010680503.3A CN111950192B (en) | 2020-07-15 | 2020-07-15 | Modeling method and device for pore network model based on convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111950192A true CN111950192A (en) | 2020-11-17 |
| CN111950192B CN111950192B (en) | 2023-10-24 |
Family
ID=73341738
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010680503.3A Active CN111950192B (en) | 2020-07-15 | 2020-07-15 | Modeling method and device for pore network model based on convolutional neural network |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN111950192B (en) |
| WO (1) | WO2022011894A1 (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114283254A (en) * | 2021-12-31 | 2022-04-05 | 西南石油大学 | Core digital pore network model construction method based on nuclear magnetic resonance data |
| CN117077499A (en) * | 2023-10-16 | 2023-11-17 | 西安交通大学 | Numerical modeling method for pore structure of thermal barrier coating of heavy gas turbine |
| CN117115370A (en) * | 2023-08-15 | 2023-11-24 | 西南石油大学 | High-precision digital core model construction method |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114169182B (en) * | 2022-02-14 | 2022-05-31 | 煤炭科学研究总院有限公司 | Method and equipment for reconstructing ellipsoid model of rock mass surface crack |
| CN115254213B (en) * | 2022-06-24 | 2024-05-03 | 中国计量大学 | Microfluidic chip device based on true soil pore network |
| CN115374599B (en) * | 2022-07-13 | 2023-06-16 | 河海大学 | 3D fibrous pore ice numerical modeling method based on random distribution |
| CN115688590B (en) | 2022-11-09 | 2023-05-16 | 中国石油大学(华东) | Multi-mineral-phase core permeability prediction method based on deep learning |
| CN115993376B (en) * | 2022-12-06 | 2023-09-15 | 东北石油大学 | Shale matrix digital core reconstruction method based on random growth method |
| CN116342815B (en) * | 2023-05-12 | 2023-08-01 | 中国石油大学(华东) | Multi-scale pore network model construction method for shale pore space |
| CN116858630B (en) * | 2023-06-14 | 2024-04-09 | 中国石油大学(华东) | Method and system for constructing real shale kerogen pore model |
| CN116519731B (en) * | 2023-07-03 | 2023-08-25 | 中国石油大学(华东) | Shale oil movable limit determination method based on molecular dynamics simulation |
| CN117152373B (en) * | 2023-11-01 | 2024-02-02 | 中国石油大学(华东) | Core-level pore network model construction method considering cracks |
| CN117313589B (en) * | 2023-11-30 | 2024-02-06 | 山东科技大学 | Method for constructing, evaluating and simulating seepage of three-dimensional coarse discrete fracture network |
| CN117745979B (en) * | 2024-02-21 | 2024-05-03 | 山东科技大学 | Three-dimensional fracture-pore coupling network simulation generation method and system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104331579A (en) * | 2014-11-19 | 2015-02-04 | 中国石油大学(华东) | Simulation method of low-permeability reservoir crude oil boundary layer |
| US20180321127A1 (en) * | 2015-10-02 | 2018-11-08 | Repsol, S.A. | Method for Providing a Numerical Model of a Sample of Rock |
| CN108876901A (en) * | 2018-05-14 | 2018-11-23 | 中国石油大学(华东) | A kind of digital cores method for reconstructing based on two dimensional image and multi-point statistic |
| CN110618082A (en) * | 2019-10-29 | 2019-12-27 | 中国石油大学(北京) | Reservoir micro-pore structure evaluation method and device based on neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106443802B (en) * | 2016-09-20 | 2018-11-13 | 中石化石油工程技术服务有限公司 | A kind of quantitatively characterizing method and system of the Electrical imaging reservoir fracture hole body based on morphologic filtering |
-
2020
- 2020-07-15 CN CN202010680503.3A patent/CN111950192B/en active Active
- 2020-11-03 WO PCT/CN2020/126213 patent/WO2022011894A1/en active Application Filing
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104331579A (en) * | 2014-11-19 | 2015-02-04 | 中国石油大学(华东) | Simulation method of low-permeability reservoir crude oil boundary layer |
| US20180321127A1 (en) * | 2015-10-02 | 2018-11-08 | Repsol, S.A. | Method for Providing a Numerical Model of a Sample of Rock |
| CN108876901A (en) * | 2018-05-14 | 2018-11-23 | 中国石油大学(华东) | A kind of digital cores method for reconstructing based on two dimensional image and multi-point statistic |
| CN110618082A (en) * | 2019-10-29 | 2019-12-27 | 中国石油大学(北京) | Reservoir micro-pore structure evaluation method and device based on neural network |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114283254A (en) * | 2021-12-31 | 2022-04-05 | 西南石油大学 | Core digital pore network model construction method based on nuclear magnetic resonance data |
| CN117115370A (en) * | 2023-08-15 | 2023-11-24 | 西南石油大学 | High-precision digital core model construction method |
| CN117115370B (en) * | 2023-08-15 | 2024-03-19 | 西南石油大学 | High-precision digital core model construction method |
| CN117077499A (en) * | 2023-10-16 | 2023-11-17 | 西安交通大学 | Numerical modeling method for pore structure of thermal barrier coating of heavy gas turbine |
| CN117077499B (en) * | 2023-10-16 | 2024-01-12 | 西安交通大学 | Numerical modeling method for pore structure of thermal barrier coating of heavy gas turbine |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111950192B (en) | 2023-10-24 |
| WO2022011894A1 (en) | 2022-01-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111950192B (en) | Modeling method and device for pore network model based on convolutional neural network | |
| Kempe et al. | Entangled games are hard to approximate | |
| Reuter et al. | Fractal space-times under the microscope: A Renormalization Group view on Monte Carlo data | |
| CN105279790B (en) | Fracture network 3-dimensional digital rock core modeling method | |
| Gleiser et al. | Configurational entropy as a lifetime predictor and pattern discriminator for oscillons | |
| WO2022011893A1 (en) | Reservoir-based modeling method and device for pore network model | |
| Moroni et al. | Three‐dimensional particle tracking velocimetry studies of the transition from pore dispersion to Fickian Dispersion for homogeneous porous media | |
| CN112098293A (en) | Unsteady gas-water two-phase seepage simulation method based on pore fracture dual-medium gas reservoir | |
| CN111462020A (en) | Method, system, storage medium and device for correcting motion artifact of heart image | |
| US9041721B2 (en) | System, method, and computer program product for evaluating an integral utilizing a low discrepancy sequence and a block size | |
| Holodovsky et al. | In-situ multi-view multi-scattering stochastic tomography | |
| CN112462438B (en) | Method, device and equipment for quantitatively evaluating formation fluid based on nuclear magnetic resonance logging | |
| Li et al. | Reconstructing the 3D digital core with a fully convolutional neural network | |
| CN110335196A (en) | A kind of super-resolution image reconstruction method and system based on fractal decoding | |
| Kang et al. | Efficient randomized hierarchy construction for interactive visualization of large scale point clouds | |
| CN109377552A (en) | Image occlusion test method, apparatus calculates equipment and storage medium | |
| RU2812143C1 (en) | Method and device for measuring characteristics of rock column to create model of pore system | |
| CN109934906B (en) | Crude oil saturation acquisition method and device, computer equipment and storage medium | |
| Ouksel et al. | The interpolation-based bintree and encoding of binary images | |
| Campagnolo et al. | An experimental study on volumetric visualization of black oil reservoir models | |
| Zhang et al. | 3D reconstruction of porous media using a batch normalized variational auto-encoder | |
| Franc et al. | Image-based effective medium approximation for fast permeability evaluation of porous media core samples | |
| CN106657966B (en) | A kind of integration imaging 3D film source rapid generations based on CPU multithreadings | |
| CN112669427A (en) | Distance sampling method, device, equipment and storage medium | |
| Al-Oraiqat et al. | Specialized Computer systems for environment visualization |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |