CN106886043A - Reservoir detecting method based on geological data deep learning - Google Patents
Reservoir detecting method based on geological data deep learning Download PDFInfo
- Publication number
- CN106886043A CN106886043A CN201710115720.6A CN201710115720A CN106886043A CN 106886043 A CN106886043 A CN 106886043A CN 201710115720 A CN201710115720 A CN 201710115720A CN 106886043 A CN106886043 A CN 106886043A
- Authority
- CN
- China
- Prior art keywords
- reservoir
- deep learning
- destination layer
- layer position
- data
- 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
- 238000013135 deep learning Methods 0.000 title claims abstract description 30
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000013136 deep learning model Methods 0.000 claims abstract description 32
- 238000001514 detection method Methods 0.000 claims abstract description 19
- 229930195733 hydrocarbon Natural products 0.000 claims abstract description 4
- 150000002430 hydrocarbons Chemical class 0.000 claims abstract description 4
- 230000006870 function Effects 0.000 claims description 13
- 239000000284 extract Substances 0.000 claims description 6
- 210000002569 neuron Anatomy 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 2
- 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
- 230000001143 conditioned effect Effects 0.000 claims 1
- 235000008434 ginseng Nutrition 0.000 claims 1
- 230000004044 response Effects 0.000 abstract description 10
- 238000000605 extraction Methods 0.000 abstract description 5
- 239000004215 Carbon black (E152) Substances 0.000 abstract description 3
- 239000012530 fluid Substances 0.000 description 8
- 230000008859 change Effects 0.000 description 6
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 6
- 230000000007 visual effect Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000008555 neuronal activation Effects 0.000 description 4
- 238000006073 displacement reaction Methods 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 238000007689 inspection Methods 0.000 description 2
- 239000011435 rock Substances 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 229910002056 binary alloy Inorganic materials 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013075 data extraction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 239000011148 porous material Substances 0.000 description 1
- 230000008569 process Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
- G01V1/28—Processing seismic data, e.g. analysis, for interpretation, for correction
- G01V1/282—Application of seismic models, synthetic seismograms
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
- G01V1/28—Processing seismic data, e.g. analysis, for interpretation, for correction
- G01V1/30—Analysis
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V2210/00—Details of seismic processing or analysis
- G01V2210/60—Analysis
- G01V2210/62—Physical property of subsurface
- G01V2210/622—Velocity, density or impedance
- G01V2210/6226—Impedance
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V2210/00—Details of seismic processing or analysis
- G01V2210/60—Analysis
- G01V2210/66—Subsurface modeling
Abstract
A kind of reservoir detecting method provided in an embodiment of the present invention, belongs to geophysical prospecting for oil technical field, and methods described includes:According to the destination layer position demarcated, seismic trace near well data corresponding with destination layer position are obtained;Based on the seismic trace near well data, reservoir detection deep learning model is set up;Deep learning model and destination layer position are detected based on the reservoir, the high-level characteristic of the destination layer position is obtained;Obtain the reservoir characteristic of the specified reference position on the destination layer position;Based on the reservoir characteristic of the reference position, the reservoir characteristic with the high-level characteristic identical region of the reference position on the destination layer position is determined.This method detects the deep learning weak seismic response features of model extraction reservoir by setting up reservoir characteristic, can more simply and efficiently determine reservoir characteristic, improves the reservoir such as oil gas, hydrocarbon indication precision of seismic exploration data.
Description
Technical field
The present invention relates to geophysical prospecting for oil technical field, more particularly to a kind of reservoir detection based on deep learning
Method.
Background technology
The target of geophysical exploration increasingly deep, microminiaturization, exploration environment more become to complicating, and some are in shallow condition
The lower significant Gas potential detection method and technology of effectiveness, such as " bright spot " technology, AVO effects, high frequency shade cannot meet exploration will
Ask.The oil and gas prediction of reservoir its essence is blowhole fluid properties and saturation degree is sentenced knowledge and evaluated, reservoir pore space stream
The volume of body only accounts for a minimum part for reservoir rocks with quality, and is filled in the hole of solid-state rock matrix, ground
Ringing should be very faint.If earthquake record has response to blowhole change of fluid, it is only possible to be reflected in the thin of seismic events
In structure.The wave equation for describing seimic wave propagation is the approximate of (such as perfectly elastic media) acquisition under certain assumed condition
Equation, can well characterize " principal phase " of fluctuation, but may not necessarily reflect " microfacies " of pore-fluid response.Therefore, based on fluctuation
The oil and gas detection of equation lacks strict mathematics basis.Earthquake record is objectively responding for actual geology dielectric response, is not existed
It is any approximate.If the amplitude Observable of blowhole fluid seismic response, then it is necessarily present in earthquake record.Problem
Key be converted into how the pore-fluid " response " in appreciation earthquake record.Deep learning is development under the big data epoch
The Automatic Feature Extraction method got up, has been successfully applied to speech recognition, Face datection, target tracking, semantic parsing etc..But
Current deep learning is also seldom applied to field of seismic exploration.
The content of the invention
In order to solve the above technical problems, it is an object of the invention to provide a kind of reservoir inspection based on geological data deep learning
Survey method.Deep learning is the Automatic Feature Extraction method grown up under the big data epoch, shows as hierarchy characteristic extraction;It is low
Layer feature belongs to locality characteristic, and high-level characteristic is the nonlinear combination of low-level feature, belongs to abstract Structural Characteristics, high-rise
Feature is indicative with more distinction and classification.It is weak that the present invention innovatively introduces deep learning feature extracting method extraction reservoir
Seismic response features, can more simply and efficiently determine reservoir characteristic, improve the hydrocarbon indication precision of seismic exploration data.
The purpose of the present invention is realized by following technical scheme:
Based on geological data deep learning reservoir detecting method, comprise the following steps:
Target zone is demarcated using well logging, well logging and synthetic seismogram;
Window width extracts training data of the geological data as deep learning pre-training model when being specified along destination layer position, its
Middle individualized training data sample is that Dow Jones index timing window data cube computation is formed around per pass, when general the taking of window displacement be less than or equal to
Time window length;
Joined using limiting Boltzmann machine (RBM) or continuously limiting Boltzmann machine (CRBM) pre-training deep learning model
Number;
By experimental selection optimal models depth, every layer of neuron node number of model, neuron activation functions and sparse limit
System;
Window width extracts the training number that well lie geological data finely tunes model as deep learning when being specified along destination layer position
According to the classification for finely tuning model includes gas and water;
Deep learning model parameter is finely tuned using batch stochastic gradient descent algorithm;
Deep learning every layer of base of model is calculated, destination layer seismic response value is extracted, is determined with base correlation using the sample
Deep learning target signature, this category feature can reflect the faint change of seismic signal, strengthen oil gas seismic response features, strengthen
The difference of reservoir and non-reservoir.
Well or even well geological data were extracted by training data extracting method, well will be crossed or even well geological data is input to instruction
The deep learning model perfected obtains target signature.Determine that the earthquake depth that different lithology, fluid cause learns according to well data
The difference of feature, then the different earthquake deep learning characteristic that different lithology, fluid are caused is extrapolated to without well area, and then
Carry out lithology, hydrocarbon indication.
The calculating of deep learning high level nonlinear characteristic is applicable to two dimension and three-dimensional data, and calculation is versatile and flexible,
Isochronous surface, horizon slice etc. can according to the actual requirements be calculated.
Brief description of the drawings
In order to become apparent from illustrating the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
The accompanying drawing to be used needed for technology description is briefly described, it should be apparent that, the accompanying drawing in below describing is only the present invention
Some embodiments.
Fig. 1 is based on geological data deep learning reservoir detecting method flow chart;
Fig. 2 is that data extract schematic diagram;
Fig. 3 is the high-rise base that deep learning is obtained;
Fig. 4 is that W1 wells, W2 wells, the well of W3 wells three connect well profile;
Fig. 5 is the high-level characteristic 1 at the destination layer position obtained using the scheme of embodiments of the invention;
Fig. 6 is the high-level characteristic 1 at the lower 10ms in destination layer position obtained using the scheme of embodiments of the invention;
Fig. 7 is clustered to 10ms high-level characteristics 1 at the destination layer position obtained using the scheme of embodiments of the invention;
Fig. 8 is W1 wells, W2 wells, the section of company well high-level characteristic 1 of W3 wells obtained using the scheme of embodiments of the invention;
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, with reference to embodiment and accompanying drawing to the present invention
It is described in further detail;
As shown in figure 1, being that, based on geological data deep learning reservoir detecting method flow, methods described includes following step
Suddenly:
Step 10 is using well logging, well logging and synthetic seismogram accurate calibration target zone;
Window width extracts well lie geological data as deep learning model training data when step 20 is specified along destination layer,
Extracting mode is shown in Fig. 2;
Using certain point as origin in Fig. 2, by adjacent 9 road as an entirety, it is 20ms that per pass opens window when taking, by 2ms
Sampling interval, training sample is 90 sampled points.Dimensional slip distance for 1 interval, when window displacement typically take it is small
In equal to time window length, until area of space border and time zone boundary.Above-mentioned origin neighboring track number, sliding distance, when window
Displacement, area of space border, time zone border can be selected by actual conditions;
Step 30 is to train deep learning model using training data, is specifically divided into two parts:Deep learning model is instructed in advance
Practice, deep learning model is finely tuned.The deep learning model pre-training stage can use limitation Boltzmann machine or continuously limit Bohr
Hereby graceful machine pre-training parameter, finally stacks and constitutes depth confidence network (Deep Belief Networks);
Limitation Boltzmann machine is rooted in statistical mechanics, and energy function is to describe estimating for whole system state, and system has
Sequence or probability distribution get over concentration, and the energy of whole system is just smaller.Training RBM parameters are mainly makes following energy function most
It is small, it is assumed that RBM is input into layer unit and m Hidden unit comprising n, do not represent visual layer unit with vector v and h and hidden layer list
Unit
State so energy function of RBM be defined as follows (single sample):
θ={ wij,ai,bjBe RBM parameter.Wherein, wijIt is the connection weight between visual element i and hidden unit j, ai
Represent the biasing of visual layer unit i, bjRepresent the biasing of Hidden unit j.Sdpecific dispersion algorithm can be obtained using what Hinton was proposed
To the renewal of weights,
Wherein,
Hinton is pointed out to work as and is used training data to initialize v(0)When, it is good enough that we only need sampling k steps can be obtained by
It is approximate, that is to say, that Gibbs sampling k step after, RBM generation models will become closer to the distribution of initial data;
In order to simulate continuous data, it is σ that CRBM adds a variance in visual layers sigmoid functions2, average be 0
Gauss unit, constant σ and Nj(0,1) Gauss input component n is generated jointlyj=σ Nj(0,1), its probability distribution is
Therefore, layer state h is hiddenjBy visual layer state viIt is expressed as
Wherein, functionExpression formula it is as follows:
θLAnd θHThe respectively lower asymptote and upper asymptote of sigmoid functions, parameter ajIt is that control sigmoid functions are oblique
The variable of rate, works as ajWhen changing from small to big, unit just can be smoothed to binary system stochastic regime from muting certainty state
Cross;
If ajSigmoid functions are made to be changed into linear in noise range, then hjAverage will be obeyed isSide
Difference is σ2Gaussian Profile.Right value update formula is
The object function of sparse RBM is as follows
Deep learning is finely tuned, it is assumed that have sample set { (x(1),y(1)),...(x(m),y(m)), common m sample, x refer to input to
Amount, y is object vector, and the multilayer neural network after launching is solved using batch stochastic gradient descent.Generation on whole data set
Valency function is:
L represents the layer depth of whole neutral net, including input layer, hidden layer, output layer.NlRepresent l layers of neuron
Number, Nl+1Represent the neuron number of l+1;
After reservoir detection deep learning model is set up, the coding layer segment for only taking reservoir detection deep learning model is used to carry
Take feature.Designated area data extraction procedure:First determine layer position, then extract layer position data, target area is determined, by target
Area data pulls into the deep learning model that one-dimensional vector input is trained;
Step 40 is deep learning feature selecting, and every layer of neuronal activation value can be used as initial data in deep learning model
New feature representation, not all feature is adapted to describing reservoir feature;
Theoretical according to deep learning, low-level feature detects the local message of initial data, and high-level characteristic is low-level feature
Combination, is the abstract representation of initial data, and burst change of the high-level characteristic to initial data has stronger robustness;
In order to determine the high-rise deep learning feature for being suitable for describing reservoir feature, it is necessary to calculate every layer of deep learning model
Base, ground floor base is the weights of deep learning model ground floor, behind the base of each layer be that ground floor multiplies to this layer of the tired of weights;
Seismic target earthquakes response is extracted, deep learning target signature is determined with base correlation using the sample.Specific tasks
In, a range of coefficient correlation is typically chosen, determine final goal feature by characteristic area indexing.
Fig. 3 be target sample and with the high-rise base that the target sample coefficient correlation is 0.9007.Meanwhile, high-rise basic function
It is more smooth, there is stronger robustness to the burst change of initial data.
Step 50 is the reservoir characteristic that destination layer position is determined using preferred feature.Fig. 4 is W1 wells, W2 wells, three mouthfuls of W3 wells
Company's well profile of well, red elliptic mark is that wherein W1 wells are aqueous, and W2 well gassiness, W3 well air contents are few at destination layer position.From
Even find out at water-bearing layer and gas-bearing bed that earthquake record (oval position) shows difference on well profile small, it is virtually impossible to distinguish, because
And need scheme of the invention further to process.
Fig. 5 and Fig. 6 are respectively the corresponding high-rise neuronal activation values of Fig. 3 bases on the middle and senior level, i.e. new feature expression.In Fig. 5 and
In the embodiment of Fig. 6, abscissa is Lian Jing CDP, and ordinate is deep learning high-level characteristic value.As illustrated, water-bearing layer and containing
The deep learning feature difference of gas-bearing formation substantially, can be easier to distinguish water-bearing layer and gas-bearing bed, illustrate water-bearing layer and gas-bearing bed it
Between a shade of difference of geologic parameter can be reflected from deep learning high-level characteristic.
Fig. 7 be the corresponding high-rise neuronal activation value of Fig. 3 bases on the middle and senior level at destination layer position to the cluster of 10ms, can see
Go out W1 wells, W2 wells, W3 wells to belong to a different category, W2 is closer with W3.And category distribution is more continuous on locus.
Fig. 8 be the corresponding high-rise neuronal activation value of Fig. 3 bases on the middle and senior level W1 wells, W2 wells, W3 wells company's well profile, can be compared with
Identification water outlet is gentle well, and can weaken other irrelevant informations.
Practical application shows:The deep learning high-level characteristic used in the present invention to the susceptibility of the feature of seismic signal very
Height, can distinguish the faint change of seismic signal that different lithology, fluid etc. cause, and be appropriate for the inspection such as lithology, oil and gas reservoir
Survey.
Although disclosed herein implementation method as above, described content is only to facilitate understanding the present invention and adopting
Implementation method, is not limited to the present invention.Any those skilled in the art to which this invention pertains, are not departing from this
On the premise of the disclosed spirit and scope of invention, any modification and change can be made in the formal and details implemented,
But scope of patent protection of the invention, must be still defined by the scope of which is defined in the appended claims.
Claims (9)
1. a kind of reservoir detecting method, it is characterised in that methods described includes:
According to the destination layer position demarcated, seismic channel data corresponding with destination layer position is obtained, based on the seismic channel data,
Set up the reservoir detection unsupervised pre-training model of deep learning;
According to well logging, log data, seismic trace near well data corresponding with destination layer position are obtained, set up reservoir detection depth
Study supervision Feature Selection Model;
Deep learning model and destination layer position are detected based on the reservoir, the high level for obtaining destination layer position study is special
Levy;
Obtain the reservoir characteristic of the specified reference position on the destination layer position;
Based on the reservoir characteristic of the specified reference position, the high-level characteristic with the reference position on the destination layer position is determined
The reservoir characteristic in identical region.
2. method according to claim 1, it is characterised in that described based on the seismic trace near well data, sets up reservoir
Detection deep learning model, including:
Using destination layer position seismic channel data as training data, the configuration ginseng of pre-training reservoir detection deep learning model
Number;
Based on the seismic trace near well data and target data corresponding with the seismic trace near well data, adjustment pre-training is completed
Reservoir detection deep learning model, so as to set up reservoir detection deep learning model.
3. method according to claim 2, it is characterised in that described using destination layer position seismic channel data as training
Data, pre-training reservoir detects the configuration parameter of deep learning model, including:
Using destination layer position geological data as training data, Boltzmann using limited Boltzmann machine or is continuously limited
Machine, pre-training reservoir detects the configuration parameter of deep learning model;The configuration parameter is updated, it is limited comprising multilayer until completing
The reservoir of Boltzmann machine or continuous limited Boltzmann machine detects the depth confidence network of deep learning model, wherein, it is described
Configuration parameter includes model depth, every layer of neuron node number of model, neuron activation functions and sparse object function.
4. method according to claim 3, it is characterised in that it is described based on seismic trace near well data and with well side ground
The corresponding target data of shake track data, the reservoir detection deep learning model that adjustment pre-training is completed, so as to set up reservoir detection
Deep learning model, including:
Based on the seismic trace near well data and target data corresponding with the seismic trace near well data, calculated based on backpropagation
Method, the storage of the limited Boltzmann machine of the multilayer or continuous limited Boltzmann machine is solved using batch stochastic gradient descent algorithm
The depth confidence network of layer detection deep learning model, the reservoir detection deep learning model parameter that adjustment pre-training is completed, directly
It is pre-conditioned to meeting, so as to set up reservoir detection deep learning model.
5. method according to claim 4, it is characterised in that described that deep learning model and institute are detected based on the reservoir
Destination layer position is stated, the high-level characteristic of the destination layer position is obtained, including:
Every layer of base that the reservoir detects deep learning model is calculated, every layer of base includes high-rise base;
Coefficient correlation based on destination layer position with the high-rise base, determines the high-level characteristic of the destination layer position.
6. method according to claim 1, it is characterised in that the reservoir characteristic include aqueous, gassiness, air content it is few,
At least one of lithology and hydro carbons.
7. method according to claim 1, it is characterised in that described according to the destination layer demarcated position, obtains and the mesh
The corresponding seismic trace near well data in mark layer position, including:
Along demarcate destination layer position, with it is set in advance when window width extract hoistway side geological data.
8. method according to claim 1, it is characterised in that according to the destination layer position demarcated, obtain and the target
Before the corresponding seismic trace near well data step in layer position, methods described also includes:
The destination layer position is demarcated using well logging, well logging and synthetic seismogram.
9. seismic data unit is extracted, for according to the destination layer position demarcated, obtaining well side ground corresponding with destination layer position
Shake track data;
Reservoir detection deep learning unit is set up, for reservoir detection depth will to be set up based on the seismic trace near well data
Practise model;
High-level characteristic unit is obtained, for detecting deep learning model and destination layer position based on the reservoir, obtains described
The high-level characteristic of destination layer position;
Specified location reservoir characteristic unit is obtained, the reservoir characteristic for obtaining the specified reference position on the destination layer position;
Determine destination layer position reservoir characteristic unit, for the reservoir characteristic based on the specified reference position, determine the target
Reservoir characteristic on layer position with the high-level characteristic identical region of the specified reference position.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710115720.6A CN106886043B (en) | 2017-03-01 | 2017-03-01 | Reservoir detection method based on seismic data deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710115720.6A CN106886043B (en) | 2017-03-01 | 2017-03-01 | Reservoir detection method based on seismic data deep learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106886043A true CN106886043A (en) | 2017-06-23 |
CN106886043B CN106886043B (en) | 2021-05-14 |
Family
ID=59180245
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710115720.6A Expired - Fee Related CN106886043B (en) | 2017-03-01 | 2017-03-01 | Reservoir detection method based on seismic data deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106886043B (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107392155A (en) * | 2017-07-25 | 2017-11-24 | 西安电子科技大学 | The Manuscripted Characters Identification Method of sparse limited Boltzmann machine based on multiple-objection optimization |
CN107678059A (en) * | 2017-09-05 | 2018-02-09 | 中国石油大学(北京) | A kind of method, apparatus and system of reservoir gas-bearing identification |
CN108629072A (en) * | 2018-03-12 | 2018-10-09 | 山东科技大学 | Convolutional neural networks study towards the distribution of earthquake oil and gas reservoir and prediction technique |
CN109271898A (en) * | 2018-08-31 | 2019-01-25 | 电子科技大学 | Solution cavity body recognizer based on optimization convolutional neural networks |
CN109828304A (en) * | 2019-03-08 | 2019-05-31 | 中国海洋石油集团有限公司 | A method of lithological sequence model is predicted using seismic data based on deep learning |
CN110320557A (en) * | 2019-06-10 | 2019-10-11 | 北京有隆科技服务有限公司 | Multiple dimensioned geologic feature detection fusion method based on deep learning and evolutionary learning |
CN110927798A (en) * | 2018-09-20 | 2020-03-27 | 中国石油化工股份有限公司 | Logging curve prediction method and system based on deep learning |
CN113433589A (en) * | 2021-08-03 | 2021-09-24 | 成都理工大学 | Weathered crust reservoir bottom interface identification method based on mathematical statistics |
WO2021237327A1 (en) * | 2020-05-29 | 2021-12-02 | Faculdades Catolicas | Method for detecting gas-reservoir signatures in seismic surveys |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2012015515A1 (en) * | 2010-07-29 | 2012-02-02 | Exxonmobil Upstream Research Company | Methods and systems for machine-learning based simulation of flow |
CN106226816A (en) * | 2016-09-12 | 2016-12-14 | 电子科技大学 | A kind of pre-stack seismic signal waveform sorting technique |
CN106291701A (en) * | 2016-10-13 | 2017-01-04 | 成都理工大学 | Reservoir detecting method and device |
-
2017
- 2017-03-01 CN CN201710115720.6A patent/CN106886043B/en not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2012015515A1 (en) * | 2010-07-29 | 2012-02-02 | Exxonmobil Upstream Research Company | Methods and systems for machine-learning based simulation of flow |
CN106226816A (en) * | 2016-09-12 | 2016-12-14 | 电子科技大学 | A kind of pre-stack seismic signal waveform sorting technique |
CN106291701A (en) * | 2016-10-13 | 2017-01-04 | 成都理工大学 | Reservoir detecting method and device |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107392155A (en) * | 2017-07-25 | 2017-11-24 | 西安电子科技大学 | The Manuscripted Characters Identification Method of sparse limited Boltzmann machine based on multiple-objection optimization |
CN107678059B (en) * | 2017-09-05 | 2019-06-28 | 中国石油大学(北京) | A kind of method, apparatus and system of reservoir gas-bearing identification |
CN107678059A (en) * | 2017-09-05 | 2018-02-09 | 中国石油大学(北京) | A kind of method, apparatus and system of reservoir gas-bearing identification |
CN108629072A (en) * | 2018-03-12 | 2018-10-09 | 山东科技大学 | Convolutional neural networks study towards the distribution of earthquake oil and gas reservoir and prediction technique |
WO2019174366A1 (en) * | 2018-03-12 | 2019-09-19 | 山东科技大学 | Seismic data-based oil and gas reservoir distribution-oriented convolutional neural network learning and predicting method |
CN109271898A (en) * | 2018-08-31 | 2019-01-25 | 电子科技大学 | Solution cavity body recognizer based on optimization convolutional neural networks |
CN110927798A (en) * | 2018-09-20 | 2020-03-27 | 中国石油化工股份有限公司 | Logging curve prediction method and system based on deep learning |
CN110927798B (en) * | 2018-09-20 | 2021-12-31 | 中国石油化工股份有限公司 | Logging curve prediction method and system based on deep learning |
CN109828304A (en) * | 2019-03-08 | 2019-05-31 | 中国海洋石油集团有限公司 | A method of lithological sequence model is predicted using seismic data based on deep learning |
CN110320557A (en) * | 2019-06-10 | 2019-10-11 | 北京有隆科技服务有限公司 | Multiple dimensioned geologic feature detection fusion method based on deep learning and evolutionary learning |
CN110320557B (en) * | 2019-06-10 | 2021-08-17 | 北京有隆科技服务有限公司 | Multi-scale geological feature detection fusion method based on deep learning and evolutionary learning |
WO2021237327A1 (en) * | 2020-05-29 | 2021-12-02 | Faculdades Catolicas | Method for detecting gas-reservoir signatures in seismic surveys |
CN113433589A (en) * | 2021-08-03 | 2021-09-24 | 成都理工大学 | Weathered crust reservoir bottom interface identification method based on mathematical statistics |
Also Published As
Publication number | Publication date |
---|---|
CN106886043B (en) | 2021-05-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106886043A (en) | Reservoir detecting method based on geological data deep learning | |
CN109709603B (en) | Seismic horizon identification and tracking method and system | |
CN106291701B (en) | Reservoir detecting method and device | |
CN109799533A (en) | A kind of method for predicting reservoir based on bidirectional circulating neural network | |
US11226424B2 (en) | Method for detecting geological objects in a seismic image | |
Yanhu et al. | A method of seismic meme inversion and its application | |
CN111596978A (en) | Web page display method, module and system for lithofacies classification by artificial intelligence | |
CN105259572A (en) | Seismic facies calculation method based on non-linear automatic classification of multiple attribute parameters of earthquake | |
Ouadfeul et al. | Lithofacies classification using the multilayer perceptron and the self-organizing neural networks | |
CN109581487A (en) | A kind of lithosomic body waveform space structural analysis and body control Inverse modeling method | |
Liu et al. | Microseismic event detection and classification based on convolutional neural network | |
Birnie et al. | Bidirectional recurrent neural networks for seismic event detection | |
Bruno et al. | Groundwater characterization in arid regions using seismic and gravity attributes: Al Jaww Plain, UAE | |
Obafemi et al. | Characterization of deep water turbidite channels and submarine fan lobes using artificial intelligence; Case study of Frem field deep offshore Niger Delta | |
WO2021226614A1 (en) | Framework for integration of geo-information extraction, geo-reasoning and geologist-responsive inquiries | |
Liu* et al. | Seismic lithofacies computation method based on deep learning | |
Yang et al. | Porosity prediction from pre-stack seismic data via a data-driven approach | |
Legget et al. | 3D horizon tracking using artificial neural networks | |
Malvić et al. | Neural networks in petroleum geology as interpretation tools | |
DELL'AVERSANA | An integrated multi-physics Machine Learning approach for exploration risk mitigation. | |
Carrasquilla et al. | Using facies, data mining and artificial intelligence concepts in the evaluation of a carbonate reservoir in Campos basin, Southeastern Brazil | |
Malviæ et al. | Some benefits of the neural approach in porosity prediction (Case study from Benièanci field) | |
Arzuman | Comparison of geostatistics and artificial neural networks in reservoir property estimation | |
Chikhi et al. | Using Probabilistic Unsupervised Neural Method for Lithofacies Identification. | |
US20240052734A1 (en) | Machine learning framework for sweep efficiency quantification |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210514 |