CN108288092A - A method of obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form - Google Patents
A method of obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form Download PDFInfo
- Publication number
- CN108288092A CN108288092A CN201810017910.9A CN201810017910A CN108288092A CN 108288092 A CN108288092 A CN 108288092A CN 201810017910 A CN201810017910 A CN 201810017910A CN 108288092 A CN108288092 A CN 108288092A
- Authority
- CN
- China
- Prior art keywords
- magnetic resonance
- permeability
- nuclear magnetic
- spectrum
- neural network
- 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.)
- Pending
Links
- 230000035699 permeability Effects 0.000 title claims abstract description 94
- 238000001228 spectrum Methods 0.000 title claims abstract description 67
- 238000005481 NMR spectroscopy Methods 0.000 title claims abstract description 66
- 238000000034 method Methods 0.000 title claims abstract description 41
- 239000004576 sand Substances 0.000 title claims abstract description 27
- 238000004458 analytical method Methods 0.000 claims abstract description 42
- 238000013507 mapping Methods 0.000 claims abstract description 33
- 238000013528 artificial neural network Methods 0.000 claims description 31
- 210000002569 neuron Anatomy 0.000 claims description 23
- 230000005311 nuclear magnetism Effects 0.000 claims description 16
- 230000008569 process Effects 0.000 claims description 16
- 238000003062 neural network model Methods 0.000 claims description 15
- 238000012549 training Methods 0.000 claims description 12
- 238000010606 normalization Methods 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 5
- 239000011435 rock Substances 0.000 claims description 3
- 238000013499 data model Methods 0.000 claims 1
- 238000012795 verification Methods 0.000 claims 1
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 description 8
- 239000003345 natural gas Substances 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 239000011148 porous material Substances 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000012530 fluid Substances 0.000 description 2
- 239000007789 gas Substances 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 229910052799 carbon Inorganic materials 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005553 drilling Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000008595 infiltration Effects 0.000 description 1
- 238000001764 infiltration Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Classifications
-
- 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
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N3/061—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using biological neurons, e.g. biological neurons connected to an integrated circuit
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21B—EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- E21B49/00—Testing the nature of borehole walls; Formation testing; Methods or apparatus for obtaining samples of soil or well fluids, specially adapted to earth drilling or wells
-
- 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
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Geology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Mining & Mineral Resources (AREA)
- Neurology (AREA)
- Chemical & Material Sciences (AREA)
- Evolutionary Computation (AREA)
- Dispersion Chemistry (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Environmental & Geological Engineering (AREA)
- Fluid Mechanics (AREA)
- Analytical Chemistry (AREA)
- Microelectronics & Electronic Packaging (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geochemistry & Mineralogy (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A method of tight sand permeability being obtained using nuclear magnetic resonance T 2 spectrum form, is included the following steps:1) the core analysis permeability and nuclear magnetic resonance T 2 spectrum log data of multigroup well section are collected, sample data is used as after depth match;2) the core analysis permeability of collection is normalized with nuclear magnetic resonance T 2 spectrum log data so that sample data range is between [0,1];3) the Nonlinear Mapping relationship between nuclear magnetic resonance T 2 spectrum and core analysis permeability is established;4) the tight sand permeability value of the Nonlinear Mapping Relation acquisition well section to be calculated in step 3) between trained nuclear magnetic resonance T 2 spectrum and core analysis permeability is utilized.
Description
Technical field
The present invention relates to a kind of methods obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form, belong to oil and open
Adopt field.
Background technology
As China's environmental problem is increasingly serious, under the overall background of green low-carbon energy transition, natural gas conduct is most clear
Clean fossil energy, the importance in energy strategy have reached unprecedented height, and the demand of natural gas continues to increase,
In unconventional tight sand natural gas be the following important source.
In the exploration and development of tight sand natural gas, the accurate of reservoir permeability is estimated for evaluating production capacity with important
Effect.The method for obtaining reservoir permeability mainly has core analysis method and well logging calculating method, wherein core analysis permeability is the most
Accurately, but drilling and coring delivery is of high cost, and the reservoir permeability of acquisition is not comprehensive enough nor continuous;It is estimated using well-log information
Permeability is a kind of economic and feasible method.Therefore, research and utilization well-log information estimation permeability has highly important meaning
Justice.Influence of the complex pore structure of tight sand to permeability is big, and nuclear magnetic resonance log T2 spectrums can be with describing reservoir porosity, hole
A variety of formation informations such as structure can play a significant role in the evaluation of tight sand permeability.Currently, being surveyed using nuclear magnetic resonance
The interpretation model that well calculates reservoir permeability mainly has:Coates models and SDR models, wherein Coates models utilize nuclear-magnetism T2
It composes obtained movable fluid porosity, constraint fluid porosity, nuclear-magnetism total porosity parameter and establishes relationship with permeability;SDR
The relationship of nuclear-magnetism total porosity, T2 geometric mean parameters foundation and permeability that model is composed using nuclear-magnetism T2.But these
Not satisfactory using utilization effect of the model of nuclear magnetic resonance T 2 spectrum calculating permeability in tight sandstone reservoir, main cause is
Reservoir permeability is calculated using the above method to need to obtain nuclear-magnetism T2 spectrum cutoff values, since tight sandstone reservoir anisotropism is tight
Weight, therefore, it is difficult to accurately obtain continually changing nuclear-magnetism T2 to compose cutoff value, computing permeability resultant error is larger;In addition, above-mentioned
Method fails the whole rock pore structure information for making full use of nuclear magnetic resonance T 2 spectrum to be reflected.
Invention content
The problem of for background technology, utilizing nuclear magnetic resonance T 2 spectrum form the purpose of the present invention is to provide a kind of
The method for obtaining tight sand permeability, can Accurate Prediction tight sand permeability.
To achieve the above object, the present invention uses following technical scheme:A kind of obtained using nuclear magnetic resonance T 2 spectrum form is caused
The method of close Permeability of Sandstone, includes the following steps:
1) the core analysis permeability and nuclear magnetic resonance T 2 spectrum log data of multigroup well section are collected, sample is used as after depth match
Notebook data;
2) the core analysis permeability of collection is normalized with nuclear magnetic resonance T 2 spectrum log data so that sample data
Range is between [0,1];
3) the Nonlinear Mapping relationship between nuclear magnetic resonance T 2 spectrum and core analysis permeability is established;
4) Nonlinear Mapping in step 3) between trained nuclear magnetic resonance T 2 spectrum and core analysis permeability is utilized to close
System obtains the tight sand permeability value of well section to be calculated.
In the step 2), the core analysis permeability of collection is normalized with nuclear magnetic resonance T 2 spectrum log data,
So that process of the sample data range between [0,1] is as follows:
Nuclear magnetic resonance T 2 spectrum sample data is normalized using formula (1)
T2i,jFor i-th of component of j-th of depth point nuclear-magnetism T2 spectrums;T2_normali,jFor T2i,jNormalization result;
T2minFor all T2i,jIn minimum value;T2maxFor all T2i,jIn maximum value;
Core analysis permeability sample data is normalized using formula (2)
KjFor j-th of depth point core analysis permeability;K_normaljFor KjNormalization result;KminFor all KjIn
Minimum value;KmaxFor all KjIn maximum value.
In the step 3), the mistake of the Nonlinear Mapping relationship between nuclear magnetic resonance T 2 spectrum and core analysis permeability is established
Journey is as follows:
1. choosing three layers of BP neural network containing one layer of hidden layer;2. determining BP neural network input, output layer variable
Number;3. determining the hidden layer neuron number of BP neural network;4. obtaining BP neural network model;5. by being obtained in step 2)
Normalized nuclear magnetic resonance T 2 spectrum data T2_normali,j(i=1 ..., m, j=1 ..., n) be used as input sample data,
The core analysis permeability K_normal obtained in step 2)j(j=1 ..., n) utilizes BP nerves as output sample data
Network model carries out network training and establishes the Nonlinear Mapping relationship between nuclear magnetic resonance T 2 spectrum and core analysis permeability.
The step 2. in, determine BP neural network input, the process of output layer variable number it is as follows:
The nuclear magnetic resonance T 2 spectrum log data of use is made of 64 components, therefore input layer variable number is 64;Output
Layer variable number is the permeability K exported, therefore output layer variable number is 1;
The step 3. in, determine that the process of the hidden layer neuron number of BP neural network is as follows:
Hidden neuron number N is determined using such as following formula (3)2
Wherein, N1For input layer variable layer number;N3For output layer variable number;Constants of the α between [1,10];
It is the integer between 9 to 18 by hidden layer neuron number known to calculating, respectively with 9 to 18 neurons
It is attempted, it is found that effect is best when 18 neurons, is determined as 18 by hidden layer neuron number.
The step 4. in, obtain BP neural network model process it is as follows:
BP neural network is first subjected to netinit, then by the number of plies of the aforementioned BP neural network set, input
The parameters such as variable layer number, output variable layer number and hidden layer neuron number, which input, carries out network instruction in BP neural network
Practice.
The step 5. in, the normalized nuclear magnetic resonance T 2 spectrum data T2_normal that will obtain in step 2)i,j(i=
1 ..., m, j=1 ..., n) it is used as input sample data, the core analysis permeability K_normal obtained in step 2)j(j=
1 ..., n) as output sample data, using BP neural network model carry out network training establish nuclear magnetic resonance T 2 spectrum and rock core
The process for analyzing the Nonlinear Mapping relationship between permeability is as follows:
A. it constantly randomly selects 70% sample data using existing iterative algorithm to be calculated, each iteration completes it
Afterwards, assessing for 50% pair of Nonlinear Mapping relationship is randomly selected in the sample data being never extracted;
B. it when 70% sample data computational accuracy is difficult to improve, then is randomly selected in the sample data that is never extracted
50% sample data verified for model is calculated, and corresponding Nonlinear Mapping relationship is most when by error sum of squares minimum
Whole Nonlinear Mapping relationship;
C. the remaining sample data not being extracted in step b tests to final Nonlinear Mapping relationship.
8, a kind of method obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form as claimed in claim 7,
It is characterized in that:In the step 4), using in step 3) between trained nuclear magnetic resonance T 2 spectrum and core analysis permeability
Nonlinear Mapping Relation acquisition well section to be calculated tight sand permeability value process it is as follows:
1. the nuclear magnetic resonance T 2 spectrum log data for the well section of being calculated is normalized using formula (1) in step 2);
2. the data after step 1. middle normalization are inputted into trained BP neural network model, it is total according to the nuclear-magnetism of foundation
The T2 that shakes composes the Nonlinear Mapping relationship between core analysis permeability, obtains output result K_normal;
3. being converted to final permeability value K using formula (4) to output result K_normal:
The invention adopts the above technical scheme, which has the following advantages:It is proposed by the present invention to be based on machine learning BP
The method that neural network algorithm calculates tight sand permeability using nuclear-magnetism T2 spectrums establishes two-dimentional nuclear-magnetism T2 spectrums and one-dimensional infiltration
Nonlinear Mapping relationship between rate value takes full advantage of nuclear-magnetism T2 and composes all pore structural informations for including, effectively improves
The computational accuracy of tight sandstone reservoir permeability, greatly improves than nuclear magnetic resonance permeability evaluation model precision before, real
The Continuous plus for having showed tight sandstone reservoir permeability meets the demand of compact sandstone gas evaluating production capacity.
Description of the drawings
Fig. 1 is that proposed by the present invention composed using nuclear-magnetism T2 based on machine learning BP neural network algorithm calculates permeability method
In BP neural network network topology structure figure;
Fig. 2 is sample data of the present invention for the training of BP neural network algorithm;
Fig. 3 is that the present invention selects the BP neural network model accuracy obtained after the training of 70% sample data;
Fig. 4 is the BP neural network model accuracy that the present invention is assessed 15% test samples not being extracted;
Fig. 5 is the BP neural network model accuracy that the present invention trains 15% sample data that residue is not extracted;
Fig. 6 is the BP neural network model accuracy that all sample datas of the present invention are trained;
Fig. 7 is that proposed by the present invention composed using nuclear-magnetism T2 based on machine learning BP neural network algorithm calculates permeability method
The practical nuclear magnetic resonance log computing permeability result figure of certain well section.
Specific implementation mode
The present invention is described in detail below with reference to the accompanying drawings and embodiments.
As shown in Figure 1, the present invention proposes a kind of side obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form
Method includes the following steps:
1) as shown in Figure 1, collecting the core analysis permeability and nuclear magnetic resonance T 2 spectrum log data of multigroup well section, depth pair
Sample data is used as after neat.
2) the core analysis permeability of collection is normalized with nuclear magnetic resonance T 2 spectrum log data so that sample data
For range between [0,1], detailed process is as follows:
Nuclear magnetic resonance T 2 spectrum sample data is normalized using formula (1)
T2i,jFor i-th of component of j-th of depth point nuclear-magnetism T2 spectrums;T2_normali,jFor T2i,jNormalization result;
T2minFor all T2i,jIn minimum value;T2maxFor all T2i,jIn maximum value.
Core analysis permeability sample data is normalized using formula (2)
KjFor j-th of depth point core analysis permeability;K_normaljFor KjNormalization result;KminFor all KjIn
Minimum value;KmaxFor all KjIn maximum value.
3) the Nonlinear Mapping relationship between nuclear magnetic resonance T 2 spectrum and core analysis permeability is established, detailed process is as follows:
1. determining the number of plies of BP neural network
It chooses and only has three layers of BP neural network of one layer of hidden layer to be trained, specific reason is as follows:
One three layers of BP neural network can be approached a nonlinear function by arbitrary the required accuracy, and excessive layer instead can
The convergence rate for slowing down BP neural network, makes run time increase, and training speed accelerates that increase hidden layer node can be used
Number realize.
2. determining BP neural network input, output layer variable number
As shown in Fig. 2, the nuclear magnetic resonance T 2 spectrum log data used is made of 64 components, therefore input layer variable number
It is 64;Output layer variable number is the permeability K exported, therefore output layer variable number is 1.
3. determining the hidden layer neuron number of BP neural network
Hidden neuron number N is determined using such as following formula (3)2
Wherein, N1For input layer variable layer number;N3For output layer variable number;Constants of the α between [1,10].
It is the integer between 9 to 18 by hidden layer neuron number known to calculating, respectively with 9 to 18 neurons
It is attempted, it is found that effect is best when 18 neurons, is determined as 18 by hidden layer neuron number.
4. obtaining BP neural network model
BP neural network is first subjected to netinit, then by the number of plies of the aforementioned BP neural network set, input
The parameters such as variable layer number, output variable layer number and hidden layer neuron number, which input, carries out network instruction in BP neural network
Practice.
5. the normalized nuclear magnetic resonance T 2 spectrum data T2_normal that will be obtained in step 2)i,j(i=1 ..., m, j=
1 ..., n) it is used as input sample data, the core analysis permeability K_normal obtained in step 2)j(j=1 ..., n) makees
To export sample data, carries out network training using BP neural network model and establish nuclear magnetic resonance T 2 spectrum and core analysis permeability
Between Nonlinear Mapping relationship, detailed process is as follows:
A. it constantly randomly selects 70% sample data using existing iterative algorithm to be calculated, each iteration completes it
Afterwards, assessing for 50% pair of Nonlinear Mapping relationship is randomly selected in the sample data being never extracted;
B. it when 70% sample data computational accuracy is difficult to improve, then is randomly selected in the sample data that is never extracted
50% sample data verified for model is calculated, and corresponding Nonlinear Mapping relationship is most when by error sum of squares minimum
Whole Nonlinear Mapping relationship;
C. the remaining sample data not being extracted in step b tests to final Nonlinear Mapping relationship.
Fig. 3 is that the model time after selecting the training of 70% training sample data sentences precision R=0.96, Fig. 4 for that will not be extracted
The models of 15% test samples data return discrimination precision R=0.71, Fig. 5 is that the model of remaining 15% test samples data returns and sentences
Precision R=0.66, Fig. 6 are the precision of prediction R=0.87 of all training sample data.Network model precision of prediction has reached R=
0.87, it is greatly improved than NMR Permeability Models computational solution precision before.
4) Nonlinear Mapping in step 3) between trained nuclear magnetic resonance T 2 spectrum and core analysis permeability is utilized to close
System obtains the tight sand permeability value of well section to be calculated, and detailed process is as follows:
1. the nuclear magnetic resonance T 2 spectrum log data for the well section of being calculated is normalized using formula (1) in step 2);
2. the data after step 1. middle normalization are inputted into trained BP neural network model, it is total according to the nuclear-magnetism of foundation
The T2 that shakes composes the Nonlinear Mapping relationship between core analysis permeability, obtains output result K_normal;
3. being converted to final permeability value K using formula (4) to output result K_normal:
Fig. 7 is the real processing results of certain well section nuclear magnetic resonance log data, from the graph, it is apparent that using the party
The permeability result that method calculates is coincide very well with core analysis permeability, can meet compact sandstone gas evaluating production capacity demand.
The present invention is only illustrated with above-described embodiment, and structure, installation position and its connection of each component are all can have
Changed, based on the technical solution of the present invention, all improvement that individual part is carried out according to the principle of the invention and equivalent
Transformation, should not exclude except protection scope of the present invention.
Claims (8)
1. a kind of method obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form, includes the following steps:
1) the core analysis permeability and nuclear magnetic resonance T 2 spectrum log data of multigroup well section are collected, sample number is used as after depth match
According to;
2) the core analysis permeability of collection is normalized with nuclear magnetic resonance T 2 spectrum log data so that sample data range
Between [0,1];
3) the Nonlinear Mapping relationship between nuclear magnetic resonance T 2 spectrum and core analysis permeability is established;
4) the Nonlinear Mapping relationship in step 3) between trained nuclear magnetic resonance T 2 spectrum and core analysis permeability is utilized to obtain
Take the tight sand permeability value of well section to be calculated.
2. a kind of method obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form as described in claim 1, special
Sign is:In the step 2), the core analysis permeability of collection is normalized with nuclear magnetic resonance T 2 spectrum log data, is made
It is as follows to obtain process of the sample data range between [0,1]:
Nuclear magnetic resonance T 2 spectrum sample data is normalized using formula (1)
T2i,jFor i-th of component of j-th of depth point nuclear-magnetism T2 spectrums;T2_normali,jFor T2i,jNormalization result;T2minFor
All T2i,jIn minimum value;T2maxFor all T2i,jIn maximum value;
Core analysis permeability sample data is normalized using formula (2)
KjFor j-th of depth point core analysis permeability;K_normaljFor KjNormalization result;KminFor all KjIn minimum
Value;KmaxFor all KjIn maximum value.
3. a kind of method obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form as claimed in claim 2, special
Sign is:In the step 3), the mistake of the Nonlinear Mapping relationship between nuclear magnetic resonance T 2 spectrum and core analysis permeability is established
Journey is as follows:
1. choosing three layers of BP neural network containing one layer of hidden layer;2. determining BP neural network input, output layer variable number;
3. determining the hidden layer neuron number of BP neural network;4. obtaining BP neural network model;5. returning what is obtained in step 2)
The one nuclear magnetic resonance T 2 spectrum data T2_normal changedi,j(i=1 ..., m, j=1 ..., n) be used as input sample data, step
2) the core analysis permeability K_normal obtained inj(j=1 ..., n) utilizes BP neural network as output sample data
Model carries out network training and establishes the Nonlinear Mapping relationship between nuclear magnetic resonance T 2 spectrum and core analysis permeability.
4. a kind of method obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form as claimed in claim 3, special
Sign is:The step 2. in, determine BP neural network input, the process of output layer variable number it is as follows:
The nuclear magnetic resonance T 2 spectrum log data of use is made of 64 components, therefore input layer variable number is 64;Output layer becomes
It is the permeability K exported to measure number, therefore output layer variable number is 1.
5. a kind of method obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form as claimed in claim 4, special
Sign is:The step 3. in, determine that the process of the hidden layer neuron number of BP neural network is as follows:
Hidden neuron number N is determined using such as following formula (3)2
Wherein, N1For input layer variable layer number;N3For output layer variable number;Constants of the α between [1,10];
It is the integer between 9 to 18 by hidden layer neuron number known to calculating, is carried out respectively with 9 to 18 neurons
It attempts, it is found that effect is best when 18 neurons, is determined as 18 by hidden layer neuron number.
6. a kind of method obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form as claimed in claim 5, special
Sign is:The step 4. in, obtain BP neural network model process it is as follows:
BP neural network is first subjected to netinit, then by the number of plies, the input variable of the aforementioned BP neural network set
Network training is carried out in the parameters input BP neural network such as layer number, output variable layer number and hidden layer neuron number.
7. a kind of method obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form as claimed in claim 6, special
Sign is:The step 5. in, the normalized nuclear magnetic resonance T 2 spectrum data T2_normal that will obtain in step 2)i,j(i=
1 ..., m, j=1 ..., n) it is used as input sample data, the core analysis permeability K_normal obtained in step 2)j(j=
1 ..., n) as output sample data, using BP neural network model carry out network training establish nuclear magnetic resonance T 2 spectrum and rock core
The process for analyzing the Nonlinear Mapping relationship between permeability is as follows:
A. 70% sample data is constantly randomly selected using existing iterative algorithm to be calculated, after each iteration is completed,
Never assessing for 50% pair of Nonlinear Mapping relationship is randomly selected in the sample data being extracted;
B. when 70% sample data computational accuracy is difficult to improve, then 50% is randomly selected in the sample data that is never extracted
Sample data for model verification is calculated, and corresponding Nonlinear Mapping relationship is final when by error sum of squares minimum
Nonlinear Mapping relationship;
C. the remaining sample data not being extracted in step b tests to final Nonlinear Mapping relationship.
8. a kind of method obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form as claimed in claim 7, special
Sign is:In the step 4), using non-between trained nuclear magnetic resonance T 2 spectrum and core analysis permeability in step 3)
The process that linear mapping relation obtains the tight sand permeability value of well section to be calculated is as follows:
1. the nuclear magnetic resonance T 2 spectrum log data for the well section of being calculated is normalized using formula (1) in step 2);
2. the data after step 1. middle normalization are inputted trained BP neural network model, according to the nuclear magnetic resonance T2 of foundation
Nonlinear Mapping relationship between spectrum and core analysis permeability obtains output result K_normal;
3. being converted to final permeability value K using formula (4) to output result K_normal:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810017910.9A CN108288092A (en) | 2018-01-09 | 2018-01-09 | A method of obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810017910.9A CN108288092A (en) | 2018-01-09 | 2018-01-09 | A method of obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108288092A true CN108288092A (en) | 2018-07-17 |
Family
ID=62835234
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810017910.9A Pending CN108288092A (en) | 2018-01-09 | 2018-01-09 | A method of obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108288092A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109655394A (en) * | 2018-12-21 | 2019-04-19 | 中国海洋石油集团有限公司 | A kind of nuclear magnetic resonance T 2 spectrum calculation of permeability under pore throat character restriction on the parameters |
CN109800521A (en) * | 2019-01-28 | 2019-05-24 | 中国石油大学(华东) | A kind of oil-water relative permeability curve calculation method based on machine learning |
CN109932297A (en) * | 2019-02-28 | 2019-06-25 | 中国石油天然气集团有限公司 | A kind of calculation method of tight sandstone reservoir permeability |
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 |
CN111583148A (en) * | 2020-05-07 | 2020-08-25 | 苏州闪掣智能科技有限公司 | Rock core image reconstruction method based on generation countermeasure network |
CN112986309A (en) * | 2021-04-01 | 2021-06-18 | 中海石油(中国)有限公司 | Method for measuring porosity of coal seam by using rock debris crushed coal sample |
-
2018
- 2018-01-09 CN CN201810017910.9A patent/CN108288092A/en active Pending
Non-Patent Citations (3)
Title |
---|
余凡 等: "《主被动遥感协同反演地表土壤水分方法》", 31 July 2016, 北京:测绘出版社 * |
朱林奇 等: "《基于改进BPNN与T2全谱的致密砂岩储层渗透率预测》", 《石油物探》 * |
柳炳祥 等: "《智能优化方法及应用》", 31 August 2018, 江苏凤凰美术出版社 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109655394A (en) * | 2018-12-21 | 2019-04-19 | 中国海洋石油集团有限公司 | A kind of nuclear magnetic resonance T 2 spectrum calculation of permeability under pore throat character restriction on the parameters |
CN109655394B (en) * | 2018-12-21 | 2022-05-27 | 中国海洋石油集团有限公司 | Nuclear magnetic resonance T2 spectrum permeability calculation method under constraint of throat structure parameters |
CN109800521A (en) * | 2019-01-28 | 2019-05-24 | 中国石油大学(华东) | A kind of oil-water relative permeability curve calculation method based on machine learning |
CN109932297A (en) * | 2019-02-28 | 2019-06-25 | 中国石油天然气集团有限公司 | A kind of calculation method of tight sandstone reservoir permeability |
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 |
CN111583148A (en) * | 2020-05-07 | 2020-08-25 | 苏州闪掣智能科技有限公司 | Rock core image reconstruction method based on generation countermeasure network |
CN112986309A (en) * | 2021-04-01 | 2021-06-18 | 中海石油(中国)有限公司 | Method for measuring porosity of coal seam by using rock debris crushed coal sample |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108288092A (en) | A method of obtaining tight sand permeability using nuclear magnetic resonance T 2 spectrum form | |
CN109102180B (en) | Comprehensive parameter evaluation method for double-dessert evaluation of tight sandstone reservoir | |
CN104533400B (en) | Method for reconstructing logging curve | |
CN103590827B (en) | Based on the compact clastic rock natural gas well PRODUCTION FORECASTING METHODS of Reservoir Classification | |
CN109611087A (en) | A kind of Volcanic Reservoir reservoir parameter intelligent Forecasting and system | |
CN104318109B (en) | Shale Gas Reservoir Identification Method Based on Support Vector Machine | |
CN106372402A (en) | Parallelization method of convolutional neural networks in fuzzy region under big-data environment | |
CN105447762B (en) | A kind of calculation method of the low-permeability oil deposit water logging information of fluid replacement | |
CN106529667A (en) | Logging facies identification and analysis method based on fuzzy depth learning in big data environment | |
CN102830442B (en) | A kind of potential coefficient partition method of prediction Coalbed Methane Productivity | |
CN104047598A (en) | Heterogeneous paleo-karst carbonate reservoir productivity prediction method | |
CN109184677A (en) | Reservoir evaluation methods for heterogeneous alternating layers sand body | |
CN116305850B (en) | Stratum thermal conductivity prediction method based on random forest model | |
CN103376468A (en) | Reservoir parameter quantitative characterization method based on neural network function approximation algorithm | |
CN106446514A (en) | Fuzzy theory and neural network-based well-log facies recognition method | |
CN110988997A (en) | Hydrocarbon source rock three-dimensional space distribution quantitative prediction technology based on machine learning | |
CN114114414A (en) | Artificial intelligence prediction method for 'dessert' information of shale reservoir | |
CN109856688A (en) | Flow net model method based on the double TW Polarimetric enhancement methods of nuclear magnetic resonance log | |
CN117251802A (en) | Heterogeneous reservoir parameter prediction method and system based on transfer learning | |
CN117093922A (en) | Improved SVM-based complex fluid identification method for unbalanced sample oil reservoir | |
CN117131971A (en) | Reservoir dominant seepage channel prediction method based on Xgboost algorithm | |
CN115964667A (en) | River-lake lithofacies well logging identification method based on deep learning and resampling | |
CN114202036A (en) | Static sounding automatic soil layer dividing technology based on pattern recognition technology | |
CN111706323A (en) | Water flooded layer fine interpretation and evaluation method based on GWO-LSSVM algorithm | |
CN113344729B (en) | Residual oil submergence digging method based on small sample learning |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180717 |