CN109191423A - A kind of porous media Permeability Prediction method based on machine image intelligence learning - Google Patents
A kind of porous media Permeability Prediction method based on machine image intelligence learning Download PDFInfo
- Publication number
- CN109191423A CN109191423A CN201810788898.1A CN201810788898A CN109191423A CN 109191423 A CN109191423 A CN 109191423A CN 201810788898 A CN201810788898 A CN 201810788898A CN 109191423 A CN109191423 A CN 109191423A
- Authority
- CN
- China
- Prior art keywords
- image
- permeability
- porous media
- sem
- gray level
- 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
- 230000035699 permeability Effects 0.000 title claims abstract description 78
- 238000000034 method Methods 0.000 title claims abstract description 36
- 239000000463 material Substances 0.000 claims abstract description 60
- 238000001878 scanning electron micrograph Methods 0.000 claims abstract description 58
- 238000012360 testing method Methods 0.000 claims abstract description 21
- 238000003062 neural network model Methods 0.000 claims abstract description 10
- 238000013528 artificial neural network Methods 0.000 claims description 15
- 230000006870 function Effects 0.000 claims description 13
- 230000008569 process Effects 0.000 claims description 11
- 239000011159 matrix material Substances 0.000 claims description 10
- 238000005259 measurement Methods 0.000 claims description 9
- 238000001514 detection method Methods 0.000 claims description 6
- 238000013135 deep learning Methods 0.000 claims description 4
- 238000002347 injection Methods 0.000 claims description 4
- 239000007924 injection Substances 0.000 claims description 4
- QSHDDOUJBYECFT-UHFFFAOYSA-N mercury Chemical compound [Hg] QSHDDOUJBYECFT-UHFFFAOYSA-N 0.000 claims description 4
- 229910052753 mercury Inorganic materials 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 238000012549 training Methods 0.000 claims description 4
- 230000001052 transient effect Effects 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 3
- 238000012512 characterization method Methods 0.000 claims description 3
- 239000006185 dispersion Substances 0.000 claims description 3
- 238000009826 distribution Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 238000007477 logistic regression Methods 0.000 claims description 3
- 241000894007 species Species 0.000 claims 4
- 241000208340 Araliaceae Species 0.000 claims 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims 1
- 235000003140 Panax quinquefolius Nutrition 0.000 claims 1
- 235000008434 ginseng Nutrition 0.000 claims 1
- 229910000278 bentonite Inorganic materials 0.000 description 10
- 239000000440 bentonite Substances 0.000 description 10
- SVPXDRXYRYOSEX-UHFFFAOYSA-N bentoquatam Chemical compound O.O=[Si]=O.O=[Al]O[Al]=O SVPXDRXYRYOSEX-UHFFFAOYSA-N 0.000 description 10
- 239000002689 soil Substances 0.000 description 4
- 239000011148 porous material Substances 0.000 description 3
- 239000000243 solution Substances 0.000 description 3
- 230000004888 barrier function Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 210000004218 nerve net Anatomy 0.000 description 2
- 238000010998 test method Methods 0.000 description 2
- 230000003139 buffering effect Effects 0.000 description 1
- 239000003245 coal Substances 0.000 description 1
- 239000003989 dielectric material Substances 0.000 description 1
- 238000009375 geological disposal Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 230000003746 surface roughness Effects 0.000 description 1
- 239000011800 void material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N15/00—Investigating characteristics of particles; Investigating permeability, pore-volume or surface-area of porous materials
- G01N15/08—Investigating permeability, pore-volume, or surface area of porous materials
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration using histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10056—Microscopic image
- G06T2207/10061—Microscopic image from scanning electron microscope
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Pathology (AREA)
- Data Mining & Analysis (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Immunology (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Dispersion Chemistry (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Analysing Materials By The Use Of Radiation (AREA)
Abstract
The porous media Permeability Prediction method based on machine image intelligence learning that the invention discloses a kind of chooses the same porous media material of multiple groups difference dry density, and determines the true permeability of each group porous media material；Its SEM image is shown using SEM electron-microscope scanning to each group porous media material, gray average, gray variance, image energy, Image entropy and the fractal dimension of each SEM image is then calculated；Study is trained to five image features of each SEM image and its corresponding true permeability using extreme learning machine neural network model, determines the variation relation between five image features and true permeability；The SEM image parameter of the porous media material of unknown permeability is inputted when prediction, extreme learning machine neural network model can predict the permeability of the porous media material.Operation of the present invention is simple, and test period is short, and the accuracy of the permeability of prediction is high, and its testing cost is cheap.
Description
Technical field
The porous media Permeability Prediction method based on machine image intelligence learning that the present invention relates to a kind of.
Background technique
Permeability is all critical technical indicator, such as coal bed gas and shale gas exploitation in many engineering fields, and core is useless
Expect barrier system gas migration problem in disposal process, CO2The fields such as deep geological storage.Currently, porous media material permeability
Test method mainly have: mercury injection method, gas steady state method and gas Transient Method etc..These more mature macro-test methods its
Although test result is very accurate, test process is cumbersome, the not easy to get started operation of lay operators；In addition above-mentioned each side
The test period of method is with the difference of porous media material at several days to some months etc.；And equipment manufacturing cost required for testing is high
It is expensive.
Summary of the invention
In view of the above existing problems in the prior art, the present invention provides a kind of porous Jie based on machine image intelligence learning
Matter Permeability Prediction method, easy to operate, test period is short, and the accuracy of the permeability of prediction is high, and its testing cost
It is cheap.
To achieve the goals above, the technical solution adopted by the present invention is that: it is a kind of based on the more of machine image intelligence learning
Hole medium permeability prediction technique, specific steps are as follows:
A, porous media material permeability data library is established:
A, by 30 groups of selection of measurement dry density or more, (the test group number of selection is more in same porous media material
Better, group number is more, as a result more accurate) (porous media material is mainly by ground for the porous media material of different dry density
Grain is mutually cementing to be formed, and including soil skeleton and hole, and the permeability of porous media material is mainly exactly by these soil bodys
Between interconnected pore determine.Therefore the ratio of the total volume of the quality of ground solid particle and soil, i.e. dry density, reflect
The void ratio of ground.Dry density can be used to judge the size of porous media material permeability, and dry density is bigger, and permeability is smaller；
Dry density is smaller, and permeability is bigger), then each group porous media material is carried out using existing permeability macroscopic view detection method
Test, obtains the true permeability value of each group porous media material；
B, each group porous media material is subjected to SEM Electronic Speculum (i.e. scanning electron microscope) scanning, obtains porous Jie of each group
The SEM image of material, by the true permeability value phase of the SEM image of every group of porous media material and this group of porous media material
It is corresponding；
The sequence of step a and b are interchangeable；
B, SEM image feature information extraction and analysis:
Since SEM image is gray level image, gray value knowledge is carried out to each pixel in SEM image using MATLAB software
Not, to form the gray scale value matrix of each SEM image, each SEM is then calculated according to the gray scale value matrix of SEM image
Image grayscale mean value, image grayscale variance, image energy, Image entropy and the fractal dimension of image, using this five image spies
Levy the parameter characterization SEM image；
C, depth is carried out using five image features of the computer neural network to each SEM image obtained
It practises:
Using known extreme learning machine neural network model to five image features of each SEM image and its institute
Corresponding true permeability is trained and learns, which, with after study, determines five image spies by training
The logic variation relation between parameter and true permeability is levied, logistic regression fitting is carried out；Due to extreme learning machine neural network
The threshold value that weight and hidden layer are generated when model running, it is trained with learning process in data, it is determined after capable of being adjusted automatically
Only optimal solution；
D, Permeability Prediction is carried out to the same porous media material of unknown permeability:
1. the same porous media material of unknown permeability is used SEM electron-microscope scanning, the porous media material is obtained
SEM image, then by the way that the image grayscale mean value, image grayscale variance, image energy, image of the SEM image is calculated
Entropy and fractal dimension；
2. five image features of 1. SEM image that step is obtained input the computer nerve net of deep learning
In network, computer neural network is according to the SEM image of input and has determined that between five image features and true permeability
Variation relation analyzed after, computer neural network predicts permeability corresponding to the SEM image, and pass through display
Equipment is shown.
Further, the existing permeability macroscopic view detection method includes mercury injection method, gas steady state method and gas Transient Method.
Further, in the step B five image features specific calculating process are as follows:
I, grey level histogram will be first obtained after gray value matrix normalization, grey level histogram is the discrete function of gray level,
It can reflect that the number of the pixel in image with the gray level accounts for the percentage of the total pixel of image, i.e., have gray level in image
The frequency that the pixel of i occurs, such as following formula:
Wherein, i indicates gray level, and N indicates total number of image pixels, niIndicate the summation of the pixel of gray level i in image, L
Indicate the species number of gray level；
II, image grayscale mean value: the gray average of image is the measurement for reflecting entire gray level image texture average brightness, more
Hole dielectric material permeability is higher, and the hole in SEM image is more, whole darker, the gray average of image that can also seem of image
It also can be lower；Grey level histogram will be obtained after gray value matrix normalization, image grayscale mean value such as following formula:
In formula, m is image grayscale mean value, and i is gray level, and p (i) is the discrete function of gray level, and L indicates the kind of gray level
Class number；
III, image grayscale variance: image grayscale variance is to reflect the gray value dispersion degree and image line of gray level image
The measurement of average contrast is managed, image grayscale variance such as following formula:
In formula, σ2For image grayscale variance, i is gray level, and m is image grayscale mean value, and p (i) is the discrete letter of gray level
Number, L indicate the species number of gray level；
IV, image energy is the uniformity coefficient for reflecting gray level image gray value, under normal conditions the grey value profile of image
The energy of more homogeneous image is bigger, conversely, the energy of image will be smaller.Image energy is in porous media material SEM image
It is able to reflect the degree that is evenly distributed of ground hole and soil skeleton, image energy such as following formula:
In formula, U is image energy, and p (i) is the discrete function of gray level, and L indicates the species number of gray level；
V, Image entropy: Image entropy is the uniformity for reflecting grey level histogram distribution, Image entropy shows more greatly figure
The randomness of picture is also bigger, conversely, the randomness of image is smaller, Image entropy such as following formula:
In formula, e is Image entropy, and p (i) is the discrete function of gray level, and L indicates the species number of gray level；
VI, fractal dimension: fractal dimension is the variation relation for reflecting porous media material aperture structure and hole surface, with hole
Complexity, heterogeneity, pore surface roughness degree, the systematicness of gap structure are related；Fractal dimension is higher, and hole surface is not advised
Then, pore structure heterogeneity is stronger；Fractal dimension is defined as follows: setting A as RnAny non-empty bounded subset in space, for appointing
R > 0 of meaning, it is N that the n that side length needed for covering A is r, which ties up the minimal amount of cube,r(A).If making there are d as r → 0:
Nr(A)∝1/rd
So d is referred to as the box counting dimension of A.One positive number k of existence anduniquess at this time, so that:
Logarithm is taken to above formula the right and left, can be obtained:
In calculating process, box number N required for covering A difference when counting different r values according to the actual situationr
(A), using lgr as abscissa, with lgNr(A) for ordinate logarithmic coordinates system in drawFinally lead to
The fitting line slope for crossing these points seeks absolute value to get the fractal dimension of set A is arrived.
Compared with prior art, the present invention combines mode using image recognition and neural network, by first to multiple groups not
Same porous media material with dry density carries out permeability test, obtains the true permeability of each group porous media material,
Then SEM electron-microscope scanning is used to each group porous media material, obtains the SEM image of each group, then passes through each SEM image
MATLAB software obtains gray average, the ash of SEM image to each pixel progress gray value identification in SEM image and after calculating
Spend variance, image energy, Image entropy and fractal dimension；Using extreme learning machine neural network model to each SEM image
Five image features and its corresponding true permeability are trained and learn, and determine five image features and true
Variation relation between real permeability, it is when being predicted, the SEM image of the same porous media material of unknown permeability is defeated
After entering, and calculate image grayscale mean value, image grayscale variance, image energy, Image entropy and the fractal dimension of image, the limit
Learning machine neural network model can predict the permeability of the porous media material.Operation of the present invention is simple, and test period is short,
The accuracy of the permeability of prediction is high, and its testing cost is cheap.
Detailed description of the invention
Fig. 1 is the equipment connection schematic diagram used in the present invention；
Fig. 2 is the structure chart of extreme learning machine neural network in the present invention；
Fig. 3 is the SEM image of porous media material in the present invention；
Fig. 4 is the comparison diagram that permeability value and true permeability value are predicted during present invention test proves.
Specific embodiment
The invention will be further described below.
As shown, a kind of porous media Permeability Prediction method based on machine image intelligence learning, specific steps are as follows:
A, porous media material permeability data library is established:
A, by 30 groups of selection of measurement dry density or more, (the test group number of selection is more in same porous media material
Better, group number is more, as a result more accurate) porous media material of different dry density, then examined using existing permeability macroscopic view
Survey method tests each group porous media material, obtains the true permeability value of each group porous media material；
B, each group porous media material is subjected to SEM electron-microscope scanning, obtains the SEM image of each group porous media material, it will
The SEM image of every group of porous media material is corresponding with the true permeability value of this group of porous media material；
B, SEM image feature information extraction and analysis:
Since SEM image is gray level image, gray value knowledge is carried out to each pixel in SEM image using MATLAB software
Not, to form the gray scale value matrix of each SEM image, each SEM is then calculated according to the gray scale value matrix of SEM image
Image grayscale mean value, image grayscale variance, image energy, Image entropy and the fractal dimension of image, using this five image spies
Levy the parameter characterization SEM image；
C, depth is carried out using five image features of the computer neural network to each SEM image obtained
It practises:
Using known extreme learning machine neural network model to five image features of each SEM image and its institute
Corresponding true permeability is trained and learns, and extreme learning machine is a kind of New Algorithm for feedforward neural network, should
Neural network model is closed by training with the logic variation between five image features and true permeability after study, is determined
System carries out logistic regression fitting, and the threshold value of weight and hidden layer is generated when due to stochastic neural net model running, instructs in data
In experienced and learning process, only optimal solution is determined after capable of being adjusted automatically；It can be obtained by Fig. 2, if setting input layer and hidden layer
Connection weight is w, is shown below:
If the connection weight of hidden layer and output layer is β, it is shown below:
Can proper input sample integrate the output of extreme learning machine neural network when as Q as T, be shown below:
D, Permeability Prediction is carried out to the same porous media material of unknown permeability:
1. the same porous media material of unknown permeability is used SEM electron-microscope scanning, the porous media material is obtained
SEM image, then by the way that the image grayscale mean value, image grayscale variance, image energy, image of the SEM image is calculated
Entropy and fractal dimension；
2. five image features of 1. SEM image that step is obtained input the computer nerve net of deep learning
In network, computer neural network is according to the SEM image of input and has determined that between five image features and true permeability
Variation relation analyzed after, computer neural network predicts permeability corresponding to the SEM image, and is set by display
It is standby to be shown.
Further, which is characterized in that the existing permeability macroscopic view detection method include mercury injection method, gas steady state method and
Gas Transient Method.
Further, which is characterized in that the specific calculating process of five image features in the step B are as follows:
I, grey level histogram will be first obtained after gray value matrix normalization, grey level histogram is the discrete function of gray level,
Such as following formula:
Wherein, i indicates gray level, and N indicates total number of image pixels, niIndicate the summation of the pixel of gray level i in image, L
Indicate the species number of gray level；
II, image grayscale mean value: the gray average of image is the measurement for reflecting entire gray level image texture average brightness, figure
As gray average such as following formula:
In formula, m is image grayscale mean value, and i is gray level, and p (i) is the discrete function of gray level, and L indicates the kind of gray level
Class number；
III, image grayscale variance: image grayscale variance is to reflect the gray value dispersion degree and image line of gray level image
The measurement of average contrast is managed, image grayscale variance such as following formula:
In formula, σ2For image grayscale variance, i is gray level, and m is image grayscale mean value, and p (i) is the discrete letter of gray level
Number, L indicate the species number of gray level；
IV, image energy be reflect gray level image gray value uniformity coefficient, image energy such as following formula:
In formula, U is image energy, and p (i) is the discrete function of gray level, and L indicates the species number of gray level；
V, Image entropy: Image entropy be reflect grey level histogram distribution uniformity, Image entropy such as following formula:
In formula, e is Image entropy, and p (i) is the discrete function of gray level, and L indicates the species number of gray level；
VI, fractal dimension: fractal dimension is the variation relation for reflecting porous media material aperture structure and hole surface, divides shape
Dimension is defined as follows: setting A as RnAny non-empty bounded subset in space, for arbitrary r > 0, side length needed for covering A is the n of r
The minimal amount for tieing up cube is Nr(A).If there is d, make as r → 0:
Nr(A)∝1/rd
So d is referred to as the box counting dimension of A.One positive number k of existence anduniquess at this time, so that:
Logarithm is taken to above formula the right and left, can be obtained:
In calculating process, box number N required for covering A difference when counting different r values according to the actual situationr
(A), using lgr as abscissa, with lgNr(A) for ordinate logarithmic coordinates system in drawFinally lead to
The fitting line slope for crossing these points seeks absolute value to get the fractal dimension of set A is arrived.
Test proves: choosing the bentonite (abbreviation GMZ bentonite) from Inner Mongol Gao Miao, which can
As one of the buffering barrier material in high radiation nuke rubbish depth geological disposal library, permeability index is related to entire disposition library
Airtightness；Different 34 groups of GMZ bentonite in lump of dry density are chosen, by 30 groups of bentonite in lump step A according to the invention therein
To C, the training learning process of extreme learning machine neural network model is completed, then using other 4 groups of bentonite in lump as required pre-
The porous media material for surveying permeability, respectively obtains its SEM image by 4 groups of bentonite in lump of SEM electron-microscope scanning, then according to this
The step D of invention, extreme learning machine neural network model export the pre- of 4 groups of bentonite in lump by result output display unit respectively
Survey permeability；Then its true permeability is tested out respectively using permeability macroscopic view detection method to this 4 groups of bentonite in lump, finally
4 groups of bentonite in lump are carried out permeability drafting pattern Fig. 4 of its prediction result and test after serial numbers；
As shown in Figure 4, through actual test, after the permeability and actual test of the 4 groups of bentonite in lump predicted through the invention
The error between true permeability obtained illustrates that the present invention is quasi- to the Permeability Prediction of porous media material less than 5%
Exactness is higher.
Claims (3)
1. a kind of porous media Permeability Prediction method based on machine image intelligence learning, which is characterized in that specific steps are as follows:
A, porous media material permeability data library is established:
A, the porous media material of 30 groups or more different dry densities is chosen by measurement dry density in same porous media material
Then material tests each group porous media material using known permeability macroscopic view detection method, obtains porous Jie of each group
The true permeability value of material；
B, each group porous media material is subjected to SEM electron-microscope scanning, the SEM image of each group porous media material is obtained, by every group
The SEM image of porous media material is corresponding with the true permeability value of this group of porous media material；
B, SEM image feature information extraction and analysis:
Since SEM image is gray level image, gray value identification is carried out to each pixel in SEM image using MATLAB software,
To form the gray scale value matrix of each SEM image, each SEM is then calculated according to the gray scale value matrix of SEM image and is schemed
Image grayscale mean value, image grayscale variance, image energy, Image entropy and the fractal dimension of picture, using this five characteristics of image
The parameter characterization SEM image；
C, deep learning is carried out using five image features of the computer neural network to each SEM image obtained:
Using known extreme learning machine neural network model to five image features of each SEM image and its corresponding
True permeability be trained and learn, which, with after study, determines five characteristics of image ginseng by training
Logic variation relation between several and true permeability, carries out logistic regression fitting；
D, Permeability Prediction is carried out to the same porous media material of unknown permeability:
1. the same porous media material of unknown permeability is used SEM electron-microscope scanning, the SEM of the porous media material is obtained
Image, then by be calculated the image grayscale mean value of the SEM image, image grayscale variance, image energy, Image entropy and
Fractal dimension；
2. five image features of 1. SEM image that step is obtained input the computer neural network of deep learning
Interior, computer neural network is according to the SEM image of input and has determined that between five image features and true permeability
After variation relation is analyzed, computer neural network predicts permeability corresponding to the SEM image, and passes through display equipment
It is shown.
2. a kind of porous media Permeability Prediction method based on machine image intelligence learning according to claim 1,
It is characterized in that, the existing permeability macroscopic view detection method includes mercury injection method, gas steady state method and gas Transient Method.
3. a kind of porous media Permeability Prediction method based on machine image intelligence learning according to claim 1,
It is characterized in that, the specific calculating process of five image features in the step B are as follows:
I, grey level histogram will be first obtained after gray value matrix normalization, grey level histogram is the discrete function of gray level, as follows
Formula:
Wherein, i indicates gray level, and N indicates total number of image pixels, niIndicate the summation of the pixel of gray level i in image, L indicates ash
Spend the species number of grade；
II, image grayscale mean value: the gray average of image is the measurement for reflecting entire gray level image texture average brightness, image ash
Spend mean value such as following formula:
In formula, m is image grayscale mean value, and i is gray level, and p (i) is the discrete function of gray level, and L indicates the type of gray level
Number；
III, image grayscale variance: image grayscale variance is to reflect that the gray value dispersion degree of gray level image and image texture are flat
The measurement of equal contrast, image grayscale variance such as following formula:
In formula, σ2For image grayscale variance, i is gray level, and m is image grayscale mean value, and p (i) is the discrete function of gray level, L table
Show the species number of gray level；
IV, image energy be reflect gray level image gray value uniformity coefficient, image energy such as following formula:
In formula, U is image energy, and p (i) is the discrete function of gray level, and L indicates the species number of gray level；
V, Image entropy: Image entropy be reflect grey level histogram distribution uniformity, Image entropy such as following formula:
In formula, e is Image entropy, and p (i) is the discrete function of gray level, and L indicates the species number of gray level；
VI, fractal dimension: fractal dimension is the variation relation for reflecting porous media material aperture structure and hole surface, fractal dimension
It is defined as follows: setting A as RnAny non-empty bounded subset in space, for arbitrary r > 0, the n dimension that side length needed for covering A is r is vertical
The minimal amount of cube is Nr(A).If there is d, make as r → 0:
Nr(A)∝1/rd
So d is referred to as the box counting dimension of A.One positive number k of existence anduniquess at this time, so that:
Logarithm is taken to above formula the right and left, can be obtained:
In calculating process, box number N required for covering A difference when counting different r values according to the actual situationr(A), exist
Using lgr as abscissa, with lgNr(A) for ordinate logarithmic coordinates system in drawFinally by these
The fitting line slope of point seeks absolute value to get the fractal dimension of set A is arrived.
Priority Applications (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810788898.1A CN109191423B (en) | 2018-07-18 | 2018-07-18 | A kind of porous media Permeability Prediction method based on machine image intelligence learning |
| CA3062555A CA3062555C (en) | 2018-07-18 | 2018-09-07 | Porous medium permeability prediction method based on machine image intelligent learning |
| AU2018424207A AU2018424207B2 (en) | 2018-07-18 | 2018-09-07 | Porous medium permeability prediction method based on machine image intelligent learning |
| PCT/CN2018/104640 WO2020015086A1 (en) | 2018-07-18 | 2018-09-07 | Porous medium permeability prediction method based on intelligent machine image learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810788898.1A CN109191423B (en) | 2018-07-18 | 2018-07-18 | A kind of porous media Permeability Prediction method based on machine image intelligence learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109191423A true CN109191423A (en) | 2019-01-11 |
| CN109191423B CN109191423B (en) | 2019-07-02 |
Family
ID=64936163
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810788898.1A Active CN109191423B (en) | 2018-07-18 | 2018-07-18 | A kind of porous media Permeability Prediction method based on machine image intelligence learning |
Country Status (3)
| Country | Link |
|---|---|
| CN (1) | CN109191423B (en) |
| AU (1) | AU2018424207B2 (en) |
| WO (1) | WO2020015086A1 (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110458816A (en) * | 2019-08-06 | 2019-11-15 | 北京工商大学 | A kind of fibrous material analysis of porosity method returned based on threshold value |
| CN111141653A (en) * | 2019-12-30 | 2020-05-12 | 上海地铁维护保障有限公司 | Tunnel leakage rate prediction method based on neural network |
| CN112881637A (en) * | 2021-02-08 | 2021-06-01 | 东华大学 | Visual osmotic membrane treatment wastewater detection device |
| CN112949213A (en) * | 2021-04-14 | 2021-06-11 | 南方科技大学 | Round filling type isotropic porous medium generation method based on machine learning |
| CN112991277A (en) * | 2021-02-25 | 2021-06-18 | 广东工业大学 | Porous medium permeability prediction method based on three-dimensional convolutional neural network |
| CN116342541A (en) * | 2023-03-29 | 2023-06-27 | 中国矿业大学 | Rock-soil body permeability calculation method based on adjacent image pore fusion reconstruction |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113192119B (en) * | 2021-05-27 | 2023-01-06 | 宜宾学院 | Quantitative statistical method for porosity of multi-scale pore surface |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2010113141A1 (en) * | 2009-04-03 | 2010-10-07 | Schlumberger Canada Limited | Use of calibration well injections with microseismic monitoring |
| CN104335042A (en) * | 2012-03-30 | 2015-02-04 | 领英股份有限公司 | An efficient method for selecting representative elementary volume in digital representations of porous media |
| CN104866706A (en) * | 2015-04-13 | 2015-08-26 | 中国石油大学(北京) | Method and device for determining permeability of carbonate rocks |
| CN105021506A (en) * | 2015-07-09 | 2015-11-04 | 中国石油大学(华东) | Pore network model-based calculation method of three-phase relative permeability |
| CN105510204A (en) * | 2015-12-03 | 2016-04-20 | 大连理工大学 | Permeability predication method based on CT (computed tomography) images |
| CN106716114A (en) * | 2015-09-11 | 2017-05-24 | 数岩科技（厦门）股份有限公司 | Porous medium analysis system and method |
| CN106869990A (en) * | 2017-03-02 | 2017-06-20 | 新疆大学 | Coal gas Permeability Prediction method based on LVQ CPSO BP algorithms |
| CN107240132A (en) * | 2017-05-09 | 2017-10-10 | 中国人民解放军装甲兵工程学院 | A kind of method that utilization gray level co-occurrence matrixes calculate fractal dimension |
| CN107917865A (en) * | 2016-10-11 | 2018-04-17 | 中国石油化工股份有限公司 | A kind of tight sandstone reservoir multi-parameter Permeability Prediction method |
| CN108133474A (en) * | 2017-12-22 | 2018-06-08 | 西安石油大学 | Permeability Prediction method based on core sample two dimension aperture image |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103426156A (en) * | 2012-05-15 | 2013-12-04 | 中国科学院声学研究所 | SAS image segmentation method and system based on SVM classifier |
| CN106709515A (en) * | 2016-12-16 | 2017-05-24 | 北京华航无线电测量研究所 | Downward-looking scene matching area selection criteria intervention method |
| CN106651856A (en) * | 2016-12-31 | 2017-05-10 | 湖南文理学院 | Detection method for foamed nickel surface defects |
-
2018
- 2018-07-18 CN CN201810788898.1A patent/CN109191423B/en active Active
- 2018-09-07 AU AU2018424207A patent/AU2018424207B2/en active Active
- 2018-09-07 WO PCT/CN2018/104640 patent/WO2020015086A1/en active Application Filing
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2010113141A1 (en) * | 2009-04-03 | 2010-10-07 | Schlumberger Canada Limited | Use of calibration well injections with microseismic monitoring |
| CN104335042A (en) * | 2012-03-30 | 2015-02-04 | 领英股份有限公司 | An efficient method for selecting representative elementary volume in digital representations of porous media |
| CN104866706A (en) * | 2015-04-13 | 2015-08-26 | 中国石油大学(北京) | Method and device for determining permeability of carbonate rocks |
| CN105021506A (en) * | 2015-07-09 | 2015-11-04 | 中国石油大学(华东) | Pore network model-based calculation method of three-phase relative permeability |
| CN106716114A (en) * | 2015-09-11 | 2017-05-24 | 数岩科技（厦门）股份有限公司 | Porous medium analysis system and method |
| CN105510204A (en) * | 2015-12-03 | 2016-04-20 | 大连理工大学 | Permeability predication method based on CT (computed tomography) images |
| CN107917865A (en) * | 2016-10-11 | 2018-04-17 | 中国石油化工股份有限公司 | A kind of tight sandstone reservoir multi-parameter Permeability Prediction method |
| CN106869990A (en) * | 2017-03-02 | 2017-06-20 | 新疆大学 | Coal gas Permeability Prediction method based on LVQ CPSO BP algorithms |
| CN107240132A (en) * | 2017-05-09 | 2017-10-10 | 中国人民解放军装甲兵工程学院 | A kind of method that utilization gray level co-occurrence matrixes calculate fractal dimension |
| CN108133474A (en) * | 2017-12-22 | 2018-06-08 | 西安石油大学 | Permeability Prediction method based on core sample two dimension aperture image |
Non-Patent Citations (2)
| Title |
|---|
| 杨伟斌、朱庆勇等: ""基于SEM图像纹理增强的岩石多孔介质渗透率计算"", 《中国力学大会-2017暨庆祝中国力学学会成功60周年大会论文集（B）》 * |
| 赵玲: ""基于图像的多孔材料特征分析与物性测量研究"", 《中国优秀硕士学位论文数据库 信息科技辑》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110458816A (en) * | 2019-08-06 | 2019-11-15 | 北京工商大学 | A kind of fibrous material analysis of porosity method returned based on threshold value |
| CN111141653A (en) * | 2019-12-30 | 2020-05-12 | 上海地铁维护保障有限公司 | Tunnel leakage rate prediction method based on neural network |
| CN111141653B (en) * | 2019-12-30 | 2022-08-09 | 上海地铁维护保障有限公司 | Tunnel leakage rate prediction method based on neural network |
| CN112881637A (en) * | 2021-02-08 | 2021-06-01 | 东华大学 | Visual osmotic membrane treatment wastewater detection device |
| CN112991277A (en) * | 2021-02-25 | 2021-06-18 | 广东工业大学 | Porous medium permeability prediction method based on three-dimensional convolutional neural network |
| CN112949213A (en) * | 2021-04-14 | 2021-06-11 | 南方科技大学 | Round filling type isotropic porous medium generation method based on machine learning |
| CN112949213B (en) * | 2021-04-14 | 2023-07-07 | 南方科技大学 | Round filling type isotropic porous medium generation method based on machine learning |
| CN116342541A (en) * | 2023-03-29 | 2023-06-27 | 中国矿业大学 | Rock-soil body permeability calculation method based on adjacent image pore fusion reconstruction |
| CN116342541B (en) * | 2023-03-29 | 2024-03-22 | 中国矿业大学 | Rock-soil body permeability calculation method based on adjacent image pore fusion reconstruction |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2020015086A1 (en) | 2020-01-23 |
| AU2018424207B2 (en) | 2020-10-29 |
| CN109191423B (en) | 2019-07-02 |
| AU2018424207A1 (en) | 2020-02-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109191423B (en) | A kind of porous media Permeability Prediction method based on machine image intelligence learning | |
| US10991078B2 (en) | Inferring petrophysical properties of hydrocarbon reservoirs using a neural network | |
| Cuddy | Litho-facies and permeability prediction from electrical logs using fuzzy logic | |
| CN110276827B (en) | Effectiveness evaluation method based on shale reservoir | |
| US8081796B2 (en) | Method for determining properties of fractured rock formations using computer tomograpic images thereof | |
| Dvorkin et al. | The future of rock physics: computational methods vs. lab testing | |
| CN103026202A (en) | Method for obtaining consistent and integrated physical properties of porous media | |
| CN113553780B (en) | Stratum pore pressure prediction method based on machine learning | |
| CN112396130A (en) | Intelligent identification method and system for rock stratum in static sounding test, computer equipment and medium | |
| KR101625660B1 (en) | Method for making secondary data using observed data in geostatistics | |
| US8452539B2 (en) | Method and apparatus utilizing nuclear magnetic resonance measurements for estimating residual carbon dioxide saturation in aquifiers | |
| Vega et al. | 2D multifractal analysis and porosity scaling estimation in Lower Cretaceous carbonates | |
| CA3062555C (en) | Porous medium permeability prediction method based on machine image intelligent learning | |
| CN117251802B (en) | Heterogeneous reservoir parameter prediction method and system based on transfer learning | |
| Cuddy et al. | The application of fuzzy logic and genetic algorithms to reservoir characterization and modeling | |
| Huang et al. | An innovative approach to permeability estimation of the fractured-vuggy carbonate reservoirs based on 2D images | |
| Datta et al. | Determination of porosity of rock samples from photomicrographs using image analysis | |
| Hakiki et al. | Formulation of rock type prediction in cored well using fuzzy substractive clustering algorithm | |
| CN112836163A (en) | Diagenetic logging quantitative division method, device and equipment | |
| Zhang et al. | A statistical information reconstruction method of images based on multiple-point geostatistics integrating soft data with hard data | |
| Lin et al. | Machine learning for multiple petrophysical properties regression based on core images and well logs in a heterogenous reservoir | |
| Zhang et al. | Multifractal Analysis of Pore Structure and Evaluation of Deep‐Buried Cambrian Dolomite Reservoir with Image Processing: A Case from Tarim Basin, NW China | |
| CN108898286B (en) | Evaluation method and device for reservoir fracture development degree | |
| Ucan et al. | A New Approach to Gravity Anomaly Separation: Differential Markov Random Field (DMRF) | |
| CN111612049B (en) | Classification and naming analysis method for fine-grained soil engineering on the basis of artificial neural network |
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 |