CN113033648A - Method for realizing logging interpretation by using machine learning algorithm - Google Patents
Method for realizing logging interpretation by using machine learning algorithm Download PDFInfo
- Publication number
- CN113033648A CN113033648A CN202110293510.2A CN202110293510A CN113033648A CN 113033648 A CN113033648 A CN 113033648A CN 202110293510 A CN202110293510 A CN 202110293510A CN 113033648 A CN113033648 A CN 113033648A
- Authority
- CN
- China
- Prior art keywords
- logging
- interpretation
- data
- machine learning
- model
- 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
- 238000010801 machine learning Methods 0.000 title claims abstract description 25
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 8
- 238000000034 method Methods 0.000 title claims abstract description 5
- 238000012549 training Methods 0.000 claims abstract description 16
- 238000007635 classification algorithm Methods 0.000 claims abstract description 4
- 238000012360 testing method Methods 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 7
- 230000000704 physical effect Effects 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 3
- 238000012952 Resampling Methods 0.000 claims 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 abstract description 9
- 238000005553 drilling Methods 0.000 abstract description 6
- 239000004215 Carbon black (E152) Substances 0.000 abstract description 5
- 229930195733 hydrocarbon Natural products 0.000 abstract description 5
- 150000002430 hydrocarbons Chemical class 0.000 abstract description 5
- 239000011435 rock Substances 0.000 abstract description 3
- 230000035699 permeability Effects 0.000 abstract description 2
- 230000007547 defect Effects 0.000 abstract 1
- 239000003921 oil Substances 0.000 description 9
- 239000007789 gas Substances 0.000 description 8
- 239000010779 crude oil Substances 0.000 description 3
- 238000003066 decision tree Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 description 2
- 238000013508 migration Methods 0.000 description 2
- 230000005012 migration Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000009933 burial Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 238000005755 formation reaction Methods 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 239000003345 natural gas Substances 0.000 description 1
- 239000003209 petroleum derivative Substances 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V20/00—Geomodelling in general
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Medical Informatics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geophysics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A method for realizing logging interpretation by using machine learning algorithm, logging data records are generally various physical parameters such as resistivity, natural potential, sound wave speed, rock volume density and the like, which can be collectively called logging information. The well logging interpretation includes calculating porosity, permeability and water saturation of stratum and determining oil, gas and water layer with well logging data in different templates and formulas. The logging data measure the geophysical information of the stratum, the geophysical information is converted into geological information, logging experts are required to summarize various works such as formulas, template making, intersection drawing and the like, and the acquisition of the empirical plate and the formulas is related to the level of the logging experts. Machine learning collects, sorts and learns internal logics among logging information, drilling information, logging information and hydrocarbon reservoir interpretation by logging experts through a big data concept, a machine learning classification algorithm is used for training a model, and the model is directly applied to the interpretation of a target well and a target reservoir, so that the defects of horizontal factors of logging personnel and difficult hydrocarbon reservoir identification errors are avoided, and the logging interpretation speed and the interpretation precision are greatly improved.
Description
One, the technical field
Petroleum and natural gas are fossil energy and basic chemical raw materials, are important strategic resources and are commodities for nationally-planned nations, the main work of well logging interpretation is to find a stratum rich in oil and gas in a plurality of kilometers of a drilled stratum, and well logging interpreters complete lithology identification and interpretation of oil, gas and water layers by using geophysical well logging, mud logging and other data through a chart, a rendezvous graph, an empirical formula and the like. The machine learning can complete classified work by training sample characteristics, well logging data and well drilling data are used as characteristics, lithology classification and oil-gas-water interpretation schemes are used as targets, training is completed through a machine learning algorithm, and a trained model is directly applied to wells needing interpretation so as to complete machine learning automatic interpretation work.
Second, background Art
Whether the stratum has oil and gas resources is restricted by various factors, such as the abundance, thickness and maturity of crude oil-bearing rock, whether a migration channel of oil and gas from the crude oil-bearing rock to a reservoir is communicated, the development degree of a cover layer and the quality of the reservoir, and logging interpretation needs to apply a large amount of data and requires that interpreters have rich experience.
In the area which has been detected by exploratory well, the combination of crude oil, reservoir stratum and cover stratum, the trap, migration and storage types have been detected. Reservoir interpretation is an important ring of oil development, is it reservoir? Is the reservoir oil? Knowledge of hydrocarbon reservoirs is limited by a variety of factors such as reservoir framework, porosity, fluid, water saturation, etc. The logging experts need to complete the establishment of an interpretation model, calculate porosity, calculate water saturation, calculate permeability, judge lithology and oil-gas-water layer, and also need to make a series of cross charts such as resistivity-porosity cross chart, sound wave-gamma cross chart and the like, so that the work is complicated, and the experience correlation degree with the interpretation experts is high.
Machine learning is a multi-domain interdiscipline, and is a special study on how a computer simulates or realizes human learning behaviors to acquire new knowledge or skills and reorganize an existing knowledge structure to continuously improve the performance of the computer. By utilizing a machine learning algorithm and learning the internal logic relation between the data used by logging experts and the interpretation result, a suitable interpretation model is obtained by training, and the interpretation model is applied to logging interpretation, so that errors caused by insufficient interpretation experience of difficult hydrocarbon reservoirs can be avoided, and a plurality of complex template making and derivation of empirical formulas can be avoided.
By means of big data technology, the interpretation experience of logging experts is collected, information used by the logging experts for interpretation is integrated, and interpretation results are combined with used drilling data, logging data and logging data to form a machine learning database.
The characteristic of machine learning is that the richer the sample, the higher the model accuracy, so the result of each well logging interpretation is taken as a sample, and the conditions required by the well logging interpretation are taken as the machine learning characteristics, which are not limited to the spatial coordinates, sedimentary formations, burial depth, electrical properties, lithology, physical properties, logging gas logging, drilling rate, mud density, etc. of the sample.
Third, the invention
The reservoir interpretation experience of logging experts is the basis of machine learning, manual logging interpretation result samples are collected, and a logging interpretation scheme is sampled at intervals according to logging depth sampling.
Designing a machine learning characteristic table, and establishing the machine learning characteristic table according to information which can be collected such as spatial position, geological age, lithology, physical property, electrical property, oil-gas content, drilling parameters, logging parameters and the like, wherein the related parameters can be as many as possible.
The characteristics in the characteristic table (0008) are selected as X variables, and the logging interpretation conclusion is selected as y variables, wherein the logging interpretation conclusion is non-digital and is difficult to classify and calculate, and in order to enable machine learning to automatically run, digital decoding needs to be carried out according to name-number corresponding relation specified by a user.
The [ 0009 ] sample library part X, y was chosen as training data and named X _ train, y _ train, and the remaining part of the samples were chosen as test data and named X _ test, y _ test. And inputting the training data into a machine learning classification algorithm to obtain a logging interpretation model with high precision and good generalization capability.
And selecting the same characteristics as X in the training model for the interpretation well, and calculating by using the model obtained by machine learning to obtain the well logging interpretation conclusion predicted by each depth sample point.
The gradient boosting decision tree algorithm can be used for classification, each decision tree generates a weak classifier in each step, and then the weak classifiers are converted into a strong classifier through weighted accumulation:
for example, each step will produce an f (x),
F(x)＝∑fi(x) In essence, a stack of classifiers is combined into a strong classifier by weighting.
......
ft(xi) New function added in the t round
The objective function is:
meaning that the t-th better weak classifier is found, so that the error becomes smaller, omega (f)t) Is a regular term to prevent overfitting.
And y is the determined and interpreted conclusion of the calculated target interval, such as an oil layer, a water layer, a gas layer, a suspicious oil layer, a poor oil layer, an oil-water layer and the like.
Fourthly, the attached drawings of the specification
FIG. 1 is a table of machine learning features and neural network configurations.
Figure 2 character variable decoding.
FIG. 3 is a graph comparing machine learning log interpretation with manual interpretation.
Fifth, detailed description of the invention
1. And collecting the interpretation samples of the logging expert hydrocarbon reservoir, and obtaining a sample library by sampling according to the depth of the logging curve.
2. And establishing a machine learning characteristic table according to the parameters such as the space coordinate, the sedimentary stratum, the buried depth, the electrical property, the lithology, the physical property, the logging gas logging, the drilling speed, the mud and the like [ 006 ].
3. Selecting [ 0017 ] feature table as X input, and decoding the numerical value of the interpretation conclusion of each sample as y.
4. The obtained samples [ 0018 ] are divided into training data (X _ train, y _ train) and test data (X _ test, y _ test) according to a certain ratio.
5. And calling a model training function in a machine learning classification algorithm to train the model by using the training data, and verifying the test data by using the prediction function until the model has better precision and generalization capability.
6. And (4) collecting and sorting the data of the target well according to the same characteristic name in [ 0018 ], and obtaining characteristic data of the target well.
And (3) using the model obtained by training (0020), and calling a model prediction algorithm to finish logging interpretation to obtain an interpretation scheme.
Claims (1)
1. A method for realizing logging interpretation by using a machine learning algorithm is characterized by comprising the following steps:
step 1, collecting logging, perforation oil testing and gas logging information of each well in a work area;
step 2, establishing a sample library through interpolation and resampling operation according to the depth sampling interval of the logging information of the logging interpretation section;
step 3, extracting the position characteristics, physical properties characteristics, lithology characteristics, electrical characteristics and oil-gas containing characteristics of each sample in the step 2 as X of sample data; selecting a sample well logging interpretation conclusion as y of sample data;
step 4, dividing the sample obtained in the step 3 into training data (X _ train, y _ train) and testing data (X _ test, y _ test) according to a certain proportion, and carrying out numerical decoding on the interpretation conclusion of the text description;
step 5, calling a model training function in a machine learning classification algorithm, training a model by using training data, and verifying the model by using test data to ensure that the model has good accuracy and generalization capability;
step 6, collecting target well data according to the method and the characteristic type in the step 3;
and 7, using the model obtained by training in the step 5, calling a model prediction function to directly carry out target well logging interpretation, and obtaining a target well interpretation conclusion.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110293510.2A CN113033648A (en) | 2021-03-19 | 2021-03-19 | Method for realizing logging interpretation by using machine learning algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110293510.2A CN113033648A (en) | 2021-03-19 | 2021-03-19 | Method for realizing logging interpretation by using machine learning algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113033648A true CN113033648A (en) | 2021-06-25 |
Family
ID=76471657
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110293510.2A Pending CN113033648A (en) | 2021-03-19 | 2021-03-19 | Method for realizing logging interpretation by using machine learning algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113033648A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113283811A (en) * | 2021-06-30 | 2021-08-20 | 中国海洋石油集团有限公司 | Gradient lifting method well cementation quality evaluation method |
| CN113671569A (en) * | 2021-08-23 | 2021-11-19 | 中油奥博(成都)科技有限公司 | Method for predicting formation temperature by using acoustic logging information machine learning nonlinearity |
| CN115049173A (en) * | 2022-08-17 | 2022-09-13 | 中国石油大学(华东) | Deep learning and Eaton method coupling driving stratum pore pressure prediction method |
-
2021
- 2021-03-19 CN CN202110293510.2A patent/CN113033648A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113283811A (en) * | 2021-06-30 | 2021-08-20 | 中国海洋石油集团有限公司 | Gradient lifting method well cementation quality evaluation method |
| CN113671569A (en) * | 2021-08-23 | 2021-11-19 | 中油奥博(成都)科技有限公司 | Method for predicting formation temperature by using acoustic logging information machine learning nonlinearity |
| CN113671569B (en) * | 2021-08-23 | 2023-12-22 | 中油奥博(成都)科技有限公司 | Method for machine learning nonlinear prediction of formation temperature by using acoustic logging data |
| CN115049173A (en) * | 2022-08-17 | 2022-09-13 | 中国石油大学(华东) | Deep learning and Eaton method coupling driving stratum pore pressure prediction method |
| CN115049173B (en) * | 2022-08-17 | 2022-10-21 | 中国石油大学(华东) | Deep learning and Eaton method coupling driving stratum pore pressure prediction method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Pyrcz et al. | Geostatistical reservoir modeling | |
| CN113033648A (en) | Method for realizing logging interpretation by using machine learning algorithm | |
| Alnahwi et al. | Mineralogical composition and total organic carbon quantification using x-ray fluorescence data from the Upper Cretaceous Eagle Ford Group in southern Texas | |
| US20070016389A1 (en) | Method and system for accelerating and improving the history matching of a reservoir simulation model | |
| Luo et al. | Production-strategy insights using machine learning: Application for bakken shale | |
| Chehrazi et al. | Pore-facies as a tool for incorporation of small-scale dynamic information in integrated reservoir studies | |
| CN109653725A (en) | A layer water flooding degree log interpretation method is stored up based on sedimentary micro and the mixed of rock phase | |
| Amosu et al. | Effective machine learning identification of TOC-rich zones in the Eagle Ford Shale | |
| Euzen et al. | Well log cluster analysis: an innovative tool for unconventional exploration | |
| Masoudi et al. | Net pay determination by artificial neural network: Case study on Iranian offshore oil fields | |
| CN117272841B (en) | Shale gas dessert prediction method based on hybrid neural network | |
| CN116168224A (en) | Machine learning lithology automatic identification method based on imaging gravel content | |
| Larson et al. | Machine learning classification of Austin Chalk chemofacies from high-resolution x-ray fluorescence core characterization | |
| KR101893800B1 (en) | Method of sedimentary environment interpretation through electrofacies construction | |
| CN111472763B (en) | Stratum thickness prediction method and device | |
| KR101175072B1 (en) | Estimation system and method for pore fluids, including hydrocarbon and non-hydrocarbon, in oil sands reservoir using statistical analysis of well logging data | |
| CN106570524A (en) | Reservoir fluid type identifying method and device | |
| CN104834934A (en) | Nuclear body capturing method used for identifying reservoir fluid | |
| Katterbauer et al. | A Deep Learning Wag Injection Method for Co2 Recovery Optimization | |
| Amanipoor | Static modeling of the reservoir for estimate oil in place using the geostatistical method | |
| Carrasquilla et al. | Using facies, data mining and artificial intelligence concepts in the evaluation of a carbonate reservoir in Campos basin, Southeastern Brazil | |
| CN110930020A (en) | Method for determining economic recoverable resource amount of unconventional oil and gas resources | |
| Murphy | A geospatial investigation of the potential for inter-aquifer communication in Shelby County, Tennessee: A multi-scale Spatial Dependency Model | |
| Mahmoud | Automatic characterization and quantitative analysis of seismic facies in naturally fractured reservoir: Case study of Amguid Messaoud field, Algeria. | |
| Eze | Modeling the Spatial Distribution of Natural Fractures in Shale Reservoirs using Machine Learning and Geostatistical Methods |
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 |