CN106291701A - Reservoir detecting method and device - Google Patents
Reservoir detecting method and device Download PDFInfo
- Publication number
- CN106291701A CN106291701A CN201610891005.7A CN201610891005A CN106291701A CN 106291701 A CN106291701 A CN 106291701A CN 201610891005 A CN201610891005 A CN 201610891005A CN 106291701 A CN106291701 A CN 106291701A
- Authority
- CN
- China
- Prior art keywords
- reservoir
- layer position
- destination layer
- learning model
- detection degree
- 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
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
- G01V1/30—Analysis
- G01V1/301—Analysis for determining seismic cross-sections or geostructures
Abstract
A kind of reservoir detecting method of embodiment of the present invention offer and device, belong to geophysical prospecting for oil technical field, and described method includes: according to the destination layer position demarcated, obtains the seismic trace near well data corresponding with described destination layer position；Based on described seismic trace near well data, set up reservoir detection degree of depth learning model；Based on described reservoir detection degree of depth learning model and described destination layer position, it is thus achieved that the high-level characteristic of described destination layer position；Obtain the reservoir characteristic of preset reference position on described destination layer position；Reservoir characteristic based on described preset reference position, determines the reservoir characteristic in region identical with the high-level characteristic of described preset reference position on described destination layer position.This method extracts the weak seismic response features of reservoir by setting up reservoir characteristic detection degree of depth learning model, it is possible to more simply and efficiently determines reservoir characteristic, improves the reservoir such as oil gas of seismic exploration data, hydrocarbon indication precision.
Description
Technical field
The present invention relates to geophysical prospecting for oil technical field, in particular to a kind of reservoir detecting method and dress
Put.
Background technology
The target of geophysical exploration day by day deep, microminiaturization, exploration environment more becomes to complicating, and some are at shallow condition
Lower effectiveness significant Gas potential detection method and technology, wants as " bright spot " technology, AVO effect, high frequency shade etc. cannot meet exploration
Ask.Its essence of the oil and gas prediction of reservoir is that blowhole fluid properties is known with sentencing of saturation and evaluated, reservoir pore space stream
The volume of body and quality only account for a minimum part for reservoir rocks, and are to be filled in the hole of solid-state rock matrix, ground
Ringing should be the faintest.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 describing seimic wave propagation is the approximation obtained (such as perfectly elastic media etc.) under certain assumed condition
Equation, can characterize " principal phase " of fluctuation well, but may not necessarily reflect " microfacies " that pore-fluid responds.Therefore, based on fluctuation
The oil and gas detection of equation lacks strict mathematics basis, can cause bigger error.
Summary of the invention
In view of this, the purpose of the embodiment of the present invention is to provide a kind of reservoir detecting method and device, to solve multiple
Under miscellaneous exploration environment, the problem that prior art cannot detect its reservoir characteristic such as oil-gas possibility.
First aspect, embodiments provides a kind of reservoir detecting method, and described method includes:
According to the destination layer position demarcated, obtain the seismic trace near well data corresponding with described destination layer position；
Based on described seismic trace near well data, set up reservoir detection degree of deep learning model；
Based on described reservoir detection degree of deep learning model and described destination layer position, it is thus achieved that the high-rise spy of described destination layer position
Levy；
Obtain the reservoir characteristic of preset reference position on described destination layer position；
Reservoir characteristic based on described preset reference position, determine on described destination layer position with described preset reference position
The reservoir characteristic in the region that high-level characteristic is identical.
Second aspect, embodiments provides a kind of reservoir detection device, and described device includes:
Extract seismic data unit, for according to the destination layer position demarcated, extract seismic trace near well data；
Set up reservoir detection degree of depth unit, for reservoir detection will be set up deep based on described seismic trace near well data
Degree learning model；
Obtain high-level characteristic unit, for based on described reservoir detection degree of deep learning model and described destination layer position, it is thus achieved that
The high-level characteristic of described destination layer position；
Obtain predeterminated position reservoir characteristic unit, special for obtaining the reservoir of the preset reference position on described destination layer position
Levy；
Determine destination layer position reservoir characteristic unit, for reservoir characteristic based on described preset reference position, determine described
The reservoir characteristic in region identical with the high-level characteristic of described preset reference position on destination layer position；
A kind of reservoir detecting method of embodiment of the present invention offer and device, the seismic trace near well extracted by destination layer position
Data and degree of deep learning method, set up reservoir characteristic detection degree of deep learning model and extract the weak seismic response features of reservoir, it is possible to more
Simply and efficiently determine reservoir characteristic, improve the reservoir such as oil gas of seismic exploration data, hydrocarbon indication precision.
Other features and advantages of the present invention will illustrate in description subsequently, and, partly become from description
It is clear that or understand by implementing the embodiment of the present invention.The purpose of the present invention and other advantages can be by saying of being write
Structure specifically noted in bright book, claims and accompanying drawing realizes and obtains.
Accompanying drawing explanation
In order to be illustrated more clearly that the technical scheme of the embodiment of the present invention, below by embodiment required use attached
Figure is briefly described, it will be appreciated that the following drawings illustrate only certain embodiments of the present invention, and it is right to be therefore not construed as
The restriction of scope, for those of ordinary skill in the art, on the premise of not paying creative work, it is also possible to according to this
A little accompanying drawings obtain other relevant accompanying drawings.Shown in accompanying drawing, above and other purpose, feature and the advantage of the present invention will more
Clearly.The part that reference instruction identical in whole accompanying drawings is identical.The most deliberately paint by actual size equal proportion scaling
Accompanying drawing processed, it is preferred that emphasis is illustrate the purport of the present invention.
Fig. 1 is the structured flowchart of a kind of electronic equipment that can be applicable in this embodiment of the present invention；
The flow chart of the reservoir detecting method that Fig. 2 provides for first embodiment of the invention；
The seismic trace near well data that Fig. 3 provides for first embodiment of the invention extract schematic diagram；
The reservoir detection degree of deep learning model flow chart that Fig. 4 provides for first embodiment of the invention；
The schematic diagram of the high-rise basic function of the reservoir detection degree of deep learning model that Fig. 5 provides for first embodiment of the invention；
Company's well profile figure of W1 well, W2 well and W3 well that Fig. 6 provides for first embodiment of the invention；
The schematic diagram of the high-level characteristic 1 of the reservoir detecting method that Fig. 7 provides for first embodiment of the invention；
The schematic diagram of the high-level characteristic 2 of the reservoir detecting method that Fig. 8 provides for first embodiment of the invention；
The flow chart of the reservoir detecting method that Fig. 9 provides for second embodiment of the invention；
The structured flowchart of the reservoir detection device that Figure 10 provides for third embodiment of the invention；
The structured flowchart of the reservoir detection device that Figure 11 provides for fourth embodiment of the invention.
Detailed description of the invention
Below in conjunction with accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Ground describes, it is clear that described embodiment is only a part of embodiment of the present invention rather than whole embodiments.Generally exist
Can arrange and design with various different configurations with the assembly of the embodiment of the present invention that illustrates described in accompanying drawing herein.Cause
This, be not intended to limit claimed invention to the detailed description of the embodiments of the invention provided in the accompanying drawings below
Scope, but it is merely representative of the selected embodiment of the present invention.Based on embodiments of the invention, those skilled in the art are not doing
The every other embodiment obtained on the premise of going out creative work, broadly falls into the scope of protection of the invention.
It should also be noted that similar label and letter represent similar terms, therefore, the most a certain Xiang Yi in following accompanying drawing
Individual accompanying drawing is defined, then need not it be defined further and explains in accompanying drawing subsequently.Meanwhile, the present invention's
In description, term " first ", " second " etc. are only used for distinguishing and describe, and it is not intended that indicate or hint relative importance.
Refer to Fig. 1, Fig. 1 and show the structured flowchart of a kind of electronic equipment 100 that can be applicable in the embodiment of the present application.
This electronic equipment 100 can be as user terminal, computer or server, as it is shown in figure 1, electronic equipment 100 can include depositing
Reservoir 110, storage control 111, processor 112 and reservoir detection device.
Between memorizer 110, storage control 111, processor 112, the reservoir detection each element of device directly or indirectly
Electrical connection, to realize the transmission of data or mutual.Such as, one or more communication bus or letter can be passed through between these elements
Number bus realizes electrical connection.Described reservoir detecting method includes that at least one can be with software or firmware (firmware) respectively
Form is stored in the software function module in memorizer 110, software function module that the most described reservoir detection device includes or
Computer program.
Memorizer 110 can store various software program and module, the reservoir detection side provided such as the embodiment of the present application
Method and programmed instruction/module corresponding to device.Processor 112 by operation store software program in the memory 110 and
Module, thus perform the application of various function and data process, i.e. realize the reservoir detecting method in the embodiment of the present application.Storage
Device 110 can include but not limited to random access memory (Random Access Memory, RAM), read only memory (Read
Only Memory, ROM), programmable read only memory (Programmable Read-Only Memory, PROM), erasable
Read only memory (Erasable Programmable Read-Only Memory, EPROM), electricallyerasable ROM (EEROM)
(Electric Erasable Programmable Read-Only Memory, EEPROM) etc..
Processor 112 can be a kind of IC chip, has signal handling capacity.Above-mentioned processor can be general
Processor, including central processing unit (Central Processing Unit is called for short CPU), network processing unit (Network
Processor, is called for short NP) etc.；Can also is that digital signal processor (DSP), special IC (ASIC), ready-made able to programme
Gate array (FPGA) or other PLDs, discrete gate or transistor logic, discrete hardware components.It can
To realize or to perform disclosed each method, step and the logic diagram in the embodiment of the present application.General processor can be micro-
Processor or this processor can also be the processors etc. of any routine.
First embodiment
Referring to Fig. 2, embodiments provide a kind of reservoir detecting method, what the present embodiment described is aimed at
The handling process of the seismic trace near well data obtained, described method includes:
Step S210: according to the destination layer position demarcated, obtains the well corresponding with described destination layer position other ground track data；
Step S220: based on described seismic trace near well data, sets up reservoir detection degree of deep learning model；
Step S230: based on described reservoir detection degree of deep learning model and described destination layer position, it is thus achieved that described destination layer position
High-level characteristic；
Step S240: obtain the reservoir characteristic of preset reference position on described destination layer position；
Step S250: reservoir characteristic based on described preset reference position, determines and presets with described on described destination layer position
The reservoir characteristic in the region that the high-level characteristic of reference position is identical.
Further, based on step S110, described according to the destination layer position demarcated, extract geological data, including: along demarcating
Destination layer position, extract the other geological data of hoistway with window width time set in advance.Along the destination layer position demarcated, first extract three
Mouth JingW1Jing, W2 well and W3 seismic trace near well data, refer to Fig. 3, and black round dot represents together, using certain point as initial point,
Each along together to neighbouring peripheral direction, the raw 3*3=9 road of common property, it is 20ms that per pass opens window when taking, by the sampling interval of 2ms, per pass
10 sampled points.One training sample is 90 sampled points.Dimensional slip distance is 1 interval, time window displacement typically take little
In equal to time window length, until area of space border and time zone boundary.Number of channels around above-mentioned initial point, sliding distance, time window
Displacement, area of space border, time zone border can be selected by practical situation.As a kind of embodiment, the present embodiment
In set in advance time window width be 150ms, then the 150ms of per pass is classified into each hour i.e. 20ms of window width.
Step S220: based on described seismic trace near well data, sets up reservoir detection degree of deep learning model.
Further, refer to Fig. 4, set up reservoir detection degree of deep learning model, be specifically divided into two parts: pre-training
Degree of deep learning model and percentage regulation learning model.
Step S221: using described seismic trace near well data as training data, pre-training reservoir detection degree of deep learning model
Configuration parameter.
Specifically, using other for described well geological data as training data, limited Boltzmann machine or continuous limited glass are utilized
The graceful machine of Wurz, the configuration parameter of pre-training reservoir detection degree of deep learning model；Update described configuration parameter, until completing to comprise many
The degree of depth confidence network of the reservoir detection degree of deep learning model of the limited Boltzmann machine of layer or continuous limited Boltzmann machine, its
In, described configuration parameter includes model depth, every layer of model neuron node number, neuron activation functions and sparse target letter
Number.
Presently used restricted Boltzmann machine is Boltzmann machine and the combination of products of experts system, and model structure is still
Maintaining single hidden layer stochastic neural net of Boltzmann machine, model training algorithm is calculating sdpecific dispersion in products of experts system
Method.
Boltzmann machine (Boltzmann Machine, BM) is that Hinton proposed in 1986, and this network is a kind of base
Stochastic neural net in statistical mechanics.Neuron in network belongs to probabilistic neural unit, only two kinds output states (activate with
Un-activation), typically represent with Binary Zero and 1, the value of state determines according to statistical rules.Boltzmann machine is by probabilistic neural
Unit connects the Feedback Neural Network of composition entirely, and symmetry connects, and without self feed back, comprises a visual layers and a hidden layer.Bohr
The most graceful facility have powerful unsupervised learning ability, it is possible to the Structural Characteristics in study initial data, but the training time is
Considerably long, be not only difficult to accurately calculate the distribution represented by Boltzmann machine, and obtain obeying ANALOGY OF BOLTZMANN DISTRIBUTION with
Press proof is the most also the most difficult.The such issues that of in order to solve, Smolensky introduces the limited version of Boltzmann machine
(Restricted Boltzmann Machine, RBM), RBM is to have a visible layer and a hidden layer equally, and interlayer connects entirely
Connect, without connecting in layer.When given visible layer location mode, the state of activation conditional sampling of each Hidden unit, in like manner, given
During Hidden unit state, the state of activation of visual layers unit is also conditional sampling.Roux and Bengio demonstrates theoretically
As long as the number of hidden unit is abundant, RBM just can any Discrete Distribution of matching.
Limit Boltzmann machine (Restricted Boltzmann Machine, RBM) and be rooted in statistical mechanics, energy letter
Number is to describe estimating of whole system state, system order or probability distribution more to concentrate, and the energy of whole system is the least.Training
RBM parameter mainly makes following energy function minimum, it is assumed that RBM comprises n input layer unit and m Hidden unit, with to
Amount v and h does not represent visual layers unit and Hidden unit state, then the energy function of RBM is defined as follows (single sample):
θ={ wij,ai,bjIt it is the parameter of RBM.Wherein, wijIt is the connection weight between visual element i and hidden unit j, ai
Represent the biasing of visual layers unit i, bjRepresent the biasing of Hidden unit j.Utilize what Hinton proposed sdpecific dispersion algorithm can be obtained
Renewal to weights:
Wherein,
Further, when using training data to initialize v(0)Time, it is only necessary to sampling k step can be obtained by the best near
Seemingly, say, that after Gibbs sampling k step, RBM generates model will become closer to the distribution of initial data.
In order to simulate continuous data, continuous limited Boltzmann machine (Continuous Restricted Boltzmann
Machine, CRBM) add in visual layers unit sigmoid function average be 0, variance be, constant σ and Nj(0,1) altogether
Component n is inputted with creating GaussjObey σ Nj(0,1) is distributed, and its probability density is:
Therefore, hidden layer state hjBy visual layers location mode viIt is expressed as:
Wherein, functionExpression formula as follows:
θLAnd θHIt is respectively lower asymptote and upper asymptote, parameter a of sigmoid functionjIt is that control sigmoid function is oblique
The variable of rate, works as ajWhen changing from small to big, unit just can be smoothed to binary system random manner from muting definitiveness state
Cross；If ajSigmoid function is made to become linear in noise range, then hjWill obey average isVariance is σ2
Gauss distribution.Right value update formula is:
The object function expression formula of sparse RBM is as follows:
Step S222: based on described seismic trace near well data and the target data corresponding with described seismic trace near well data,
Adjust the reservoir detection degree of deep learning model that pre-training completes, thus set up reservoir detection degree of deep learning model.
Specifically, based on step S222, based on described seismic trace near well data and corresponding with described seismic trace near well data
Target data, based on back-propagation algorithm, utilize batch stochastic gradient descent algorithm to solve the limited Boltzmann of described multilamellar
The degree of depth confidence network of the reservoir detection degree of deep learning model of machine or continuous limited Boltzmann machine, adjusts the storage that pre-training completes
Layer detection degree of deep learning model parameter, until meeting pre-conditioned, thus sets up reservoir detection degree of deep learning model.
Assume there is sample set { (x(1),y(1)),...(x(m),y(m)), altogether m sample, x refer to input vector, y be target to
Amount.Use the limited Boltzmann machine of multilamellar after stochastic gradient descent algorithm solves expansion in batches or continuous limited Boltzmann machine
Reservoir detection degree of deep learning model degree of depth confidence network.Cost function on whole data set is:
L represents the degree of depth of entire depth confidence network, including input layer, hidden layer, output layer.NlRepresent the god of l layer
Through unit's number, Nl+1Represent the neuron number of l+1.
As a kind of embodiment, in the present embodiment, visual layers unit i.e. input layer is 1, and hidden layer is 5, and entire depth is put
The degree of depth L=2* hidden layer number+1 of communication network is 11, J (W, b；X, Y) it is mean square error function.
Step 230: based on described reservoir detection degree of deep learning model and described destination layer position, it is thus achieved that described destination layer position
High-level characteristic.
Specifically, based on step 230, calculate every layer of basic function of described reservoir detection degree of deep learning model, described every layer
Basic function includes high-rise basic function；Correlation coefficient based on described destination layer position with described high-rise basic function, determines described target
The high-level characteristic of layer position.
After reservoir detection degree of deep learning model is set up, only take the coding layer segment of reservoir detection degree of deep learning model for carrying
Take feature.Area data is specified to extract process: first to determine layer position, then extract a layer bit data, determine target area, by target
Area data pulls into one-dimensional vector and inputs the degree of deep learning model trained.
In degree of deep learning model, every layer of neuronal activation value can be not all as the new feature representation of initial data
Feature be adapted to describing reservoir feature；
According to degree of depth theory of learning, the local message of low-level feature detection initial data, high-level characteristic is low-level feature
Combination, is the abstract representation of initial data, and initial data is happened suddenly to change by high-level characteristic higher robustness；In order to determine
It is suitable for the high layer deep learning characteristic of describing reservoir feature, needs to calculate every layer of basic function of degree of deep learning model, ground floor base
Function is the weights of degree of deep learning model ground floor, after the basic function of each layer be that ground floor is taken advantage of to the tired of this layer of weights；Extract
Seismic target earthquakes response value, utilizes this sample and basic function correlation coefficient to determine the high-level characteristic of described destination layer position.
Refer to the high-rise base that Fig. 5, Fig. 5 are target sample and positive correlation very big with this target sample and very big negative correlation
Function, correlation coefficient is respectively 0.8779 ,-0.8766.Meanwhile, high-rise basic function is the most smooth, and the burst to initial data becomes
Change and have higher robustness.
Step S240: obtain the reservoir characteristic of preset reference position on described destination layer position.
Described reservoir characteristic include aqueous, gassiness, air content are few, in lithology and hydro carbons at least one.
Referring to Fig. 6, Fig. 6 is W1 well, W2 well and company's well profile of three mouthfuls of wells of W3 well, and oval marks is destination layer position
Place, wherein W1 well is aqueous, and W2 well gassiness, W3 well air content is few.Can be seen that aqueous from the section of the synthetic seismogram of Fig. 6
Layer and gas-bearing bed earthquake record performance difference are small, it is virtually impossible to difference, so that locate further according to the solution of the present invention
Reason.
Refer to the high-level characteristic 1 of the reservoir detecting method that Fig. 7 and Fig. 8, Fig. 7 provide for first embodiment of the invention；Fig. 8
High-level characteristic 2 for the reservoir detecting method that first embodiment of the invention provides.In figures 7 and 8, abscissa is common depth point
(CDP) high-level characteristic that number, vertical coordinate is corresponding positive correlation high level basic function respectively and negative correlation high level basic function obtains.CDP
The basic road collection form i.e. just having during seismic data process, corresponding each recorded trace with common center pip forms concentrically
Point road collection.In the figure 7, W1, W2 and W3 labelling represents that W1 well, W2 well and W3 well, W1 well, W2 well and W3 well represent destination layer respectively
Reference position in advance on position, the high-level characteristic value obvious difference that W1 well, W2 and W3 well are corresponding, the high-level characteristic value that W1 well is corresponding
For about-1.1, the high-level characteristic value that W2 well is corresponding is about-1.25, and the high-level characteristic value that W3 well is corresponding is about-1.28.?
In Fig. 8, it is pre-that W1, W2 and W3 labelling represents that W1 well, W2 well and W3 well, W1 well, W2 well and W3 well represent on destination layer position respectively
Referring initially to position, the high-level characteristic value obvious difference that W1 well, W2 and W3 well are corresponding, the high-level characteristic value that W1 well is corresponding is 1.1 left sides
The right side, the high-level characteristic value that W2 well is corresponding is about-1.18, and the high-level characteristic value that W3 well is corresponding is about 1.16.
Aqueous in conjunction with known W1 well, W2 well gassiness, the geologic information that W3 well air content is few, draw water-bearing layer W1 well and gassiness
The respective heights feature difference of layer W2 well is obvious, it is possible to be easier to water-bearing layer and gas-bearing bed at difference, W2 well gassiness and W3 well gassiness
Measure few respective heights feature very close to, this is consistent with practical situation, illustrates that the embodiment of the present invention can illustrate water-bearing layer and contain
The a shade of difference of the geologic parameter between gas-bearing formation can reflect from degree of depth study high-level characteristic.Additionally, it is true according to well data
Determine different lithology, the difference of earthquake deep learning characteristic that fluid causes, the more different ground that different lithology, fluid are caused
Focal depth degree learning characteristic is extrapolated to without well area, and then carries out oil-containing, lithology, hydrocarbon indication.
Step S250: reservoir characteristic based on described preset reference position, determines and presets with described on described destination layer position
The reservoir characteristic in the region that the high-level characteristic of reference position is identical.
Referring to Fig. 7 and Fig. 8, W1, W2 and W3 labelling represents W1 well, W2 well and W3 well, W1 well, W2 well and W3 well table respectively
Show the reference position in advance on destination layer position, aqueous in conjunction with known W1 well, W2 well gassiness, the geologic information that W3 well air content is few,
It is the most aqueous that the region identical with W1 well altitude feature there may be identical reservoir characteristic；The region identical with W2 well altitude feature
There may be identical reservoir characteristic such as gassiness；The region identical with W3 well altitude feature there may be identical reservoir characteristic such as
Air content is few.Therefore, the embodiment of the present invention can detect same or analogous reservoir characteristic by altitude feature.
A kind of reservoir detection that the embodiment of the present invention provides is according to the destination layer position demarcated, and it is right with described destination layer position to obtain
The seismic trace near well data answered；Based on described seismic trace near well data, set up reservoir detection degree of deep learning model；Based on described storage
Layer detection degree of deep learning model and described destination layer position, it is thus achieved that the high-level characteristic of described destination layer position；Obtain described destination layer position
On the reservoir characteristic of preset reference position；Reservoir characteristic based on described preset reference position, determines on described destination layer position
The reservoir characteristic in the region identical with the high-level characteristic of described preset reference position.The degree of depth study used in the present invention is high-rise special
The sensitivity levying the feature to seismic signal is the highest, it is possible to distinguish the faint change of seismic signal that different lithology, fluid etc. cause
Change, be appropriate to the detection such as lithology, oil and gas reservoir.
Second embodiment
Refer to Fig. 9, a kind of reservoir detecting method that second embodiment of the invention provides, what the present embodiment described is for
In spotting layer position and the handling process of seismic trace near well data, described method includes:
Step S310: utilize well logging, well logging and synthetic seismogram spotting layer position.
Step S320: according to the destination layer position demarcated, obtain the seismic trace near well data corresponding with described destination layer position；
Step S330: based on described seismic trace near well data, sets up reservoir detection degree of deep learning model；
Step S340: based on described reservoir detection degree of deep learning model and described destination layer position, it is thus achieved that described destination layer position
High-level characteristic；
Step S350: obtain the reservoir characteristic of preset reference position on described destination layer position；
Step S360: reservoir characteristic based on described preset reference position, determines and presets with described on described destination layer position
The reservoir characteristic in the region that the high-level characteristic of reference position is identical.
It is appreciated that the present embodiment difference topmost with the information processing method that first embodiment provides is, this reality
The reservoir detecting method that executing example provides also includes utilizing well logging, well logging and synthetic seismogram spotting layer position.Other number
Similar to the principle in first embodiment according to handling principle, similarity is referred to the content in first embodiment, this enforcement
Example repeats no more.
A kind of reservoir detecting method that the embodiment of the present invention provides, utilizes well logging, well logging and synthetic seismogram to demarcate mesh
Mark layer position；According to the destination layer position demarcated, obtain the seismic trace near well data corresponding with described destination layer position；By described well
Seismic channel data, sets up reservoir detection degree of deep learning model；Based on described reservoir detection degree of deep learning model and described destination layer
Position, it is thus achieved that the high-level characteristic of described destination layer position；Obtain the reservoir characteristic of preset reference position on described destination layer position；Based on
The reservoir characteristic of described preset reference position, determines on described destination layer position identical with the high-level characteristic of described preset reference position
The reservoir characteristic in region.The degree of depth study high-level characteristic used in the present invention is the highest to the sensitivity of the feature of seismic signal,
The faint change of seismic signal that different lithology, fluid etc. cause can be distinguished, be appropriate to the detection such as lithology, oil and gas reservoir.
3rd embodiment
Referring to Figure 10, third embodiment of the invention provides a kind of reservoir detection device 400, and described device 400 includes:
Extract seismic data unit 410, for according to the destination layer position demarcated, obtaining the well corresponding with described destination layer position
Other seismic channel data.
Set up reservoir detection degree of depth unit 420, for reservoir detection will be set up based on described seismic trace near well data
Degree of deep learning model.
Obtain high-level characteristic unit 430, for based on described reservoir detection degree of deep learning model and described destination layer position, obtaining
Obtain the high-level characteristic of described destination layer position；
Obtain predeterminated position reservoir characteristic unit 440, for obtaining the storage of the preset reference position on described destination layer position
Layer feature；
Determine destination layer position reservoir characteristic unit 450, for reservoir characteristic based on described preset reference position, determine institute
State the reservoir characteristic in region identical with the high-level characteristic of described preset reference position on destination layer position.
As a kind of embodiment, described reservoir of setting up detects degree of depth unit 420, including:
Pre-training reservoir detection degree of depth unit 421, is used for described seismic trace near well data as training data, in advance
The configuration parameter of training reservoir detection degree of deep learning model；
Adjust reservoir detection degree of depth unit 422, based on described seismic trace near well data and with described seismic trace near well
The target data that data are corresponding, adjusts the reservoir detection degree of deep learning model that pre-training completes, thus sets up the reservoir detection degree of depth
Learning model.
It should be noted that each unit in the present embodiment can be by software code realization, above each unit equally may be used
To be realized by hardware such as IC chip.
4th embodiment
Referring to Figure 11, fourth embodiment of the invention provides a kind of reservoir detection device 500, and described device 500 includes:
Spotting layer bit location 510: utilize well logging, well logging and synthetic seismogram spotting layer position.
Extract seismic data unit 520, for according to the destination layer position demarcated, obtaining the well corresponding with described destination layer position
Other seismic channel data.
Set up reservoir detection degree of depth unit 530, for reservoir detection will be set up based on described seismic trace near well data
Degree of deep learning model.
Obtain high-level characteristic unit 540, for based on described reservoir detection degree of deep learning model and described destination layer position, obtaining
Obtain the high-level characteristic of described destination layer position；
Obtain predeterminated position reservoir characteristic unit 550, for obtaining the storage of the preset reference position on described destination layer position
Layer feature；
Determine destination layer position reservoir characteristic unit 560, for reservoir characteristic based on described preset reference position, determine institute
State the reservoir characteristic in region identical with the high-level characteristic of described preset reference position on destination layer position.
As a kind of embodiment, described reservoir of setting up detects degree of depth unit 530, including:
Pre-training reservoir detection degree of depth unit 531, is used for described seismic trace near well data as training data, in advance
The configuration parameter of training reservoir detection degree of deep learning model；
Adjust reservoir detection degree of depth unit 532, based on described seismic trace near well data and with described seismic trace near well
The target data that data are corresponding, adjusts the reservoir detection degree of deep learning model that pre-training completes, thus sets up the reservoir detection degree of depth
Learning model.
It should be noted that each unit in the present embodiment can be by software code realization, above each unit equally may be used
To be realized by hardware such as IC chip.
In several embodiments provided herein, it should be understood that disclosed apparatus and method, it is also possible to pass through
Other mode realizes.Device embodiment described above is only schematically, such as, and the flow chart in accompanying drawing and block diagram
Show the device of multiple embodiments according to the present invention, the architectural framework in the cards of method and computer program product,
Function and operation.In this, each square frame in flow chart or block diagram can represent a module, program segment or the one of code
Part, a part for described module, program segment or code comprises holding of one or more logic function for realizing regulation
Row instruction.It should also be noted that at some as in the implementation replaced, the function marked in square frame can also be to be different from
The order marked in accompanying drawing occurs.Such as, two continuous print square frames can essentially perform substantially in parallel, and they are the most also
Can perform in the opposite order, this is depending on involved function.It is also noted that every in block diagram and/or flow chart
The combination of the square frame in individual square frame and block diagram and/or flow chart, can be with function or the special base of action performing regulation
System in hardware realizes, or can realize with the combination of specialized hardware with computer instruction.
It addition, each functional module in each embodiment of the present invention can integrate one independent portion of formation
Point, it is also possible to it is modules individualism, it is also possible to two or more modules are integrated to form an independent part.
If described function is using the form realization of software function module and as independent production marketing or use, permissible
It is stored in a computer read/write memory medium.Based on such understanding, technical scheme is the most in other words
The part contributing prior art or the part of this technical scheme can embody with the form of software product, this meter
Calculation machine software product is stored in a storage medium, including some instructions with so that a computer equipment (can be individual
People's computer, server, or the network equipment etc.) perform all or part of step of method described in each embodiment of the present invention.
And aforesaid storage medium includes: USB flash disk, portable hard drive, read only memory (ROM, Read-Only Memory), random access memory are deposited
The various media that can store program code such as reservoir (RAM, Random Access Memory), magnetic disc or CD.Need
Illustrate, in this article, the relational terms of such as first and second or the like be used merely to by an entity or operation with
Another entity or operating space separate, and there is any this reality between not necessarily requiring or imply these entities or operating
The relation on border or order.And, term " includes " or its any other variant is intended to comprising of nonexcludability, thus
Make to include that the process of a series of key element, method, article or equipment not only include those key elements, but also include the most clearly
Other key elements listed, or also include the key element intrinsic for this process, method, article or equipment.The most more
In the case of many restrictions, statement " including ... " key element limited, it is not excluded that including the process of described key element, side
Method, article or equipment there is also other identical element.
The foregoing is only the preferred embodiments of the present invention, be not limited to the present invention, for the skill of this area
For art personnel, the present invention can have various modifications and variations.All within the spirit and principles in the present invention, that is made any repaiies
Change, equivalent, improvement etc., should be included within the scope of the present invention.It should also be noted that similar label and letter exist
Figure below represents similar terms, therefore, the most a certain Xiang Yi accompanying drawing is defined, is then not required in accompanying drawing subsequently
It is defined further and explains.
The above, the only detailed description of the invention of the present invention, but protection scope of the present invention is not limited thereto, and any
Those familiar with the art, in the technical scope that the invention discloses, can readily occur in change or replace, should contain
Cover within protection scope of the present invention.Therefore, protection scope of the present invention should described be as the criterion with scope of the claims.
It should be noted that in this article, the relational terms of such as first and second or the like is used merely to a reality
Body or operation separate with another entity or operating space, and deposit between not necessarily requiring or imply these entities or operating
Relation or order in any this reality.And, term " includes " or its any other variant is intended to non-exclusive
Comprising of property, so that include that the process of a series of key element, method, article or equipment not only include those key elements, and
Also include other key elements being not expressly set out, or also include intrinsic for this process, method, article or equipment
Key element.In the case of there is no more restriction, statement " including ... " key element limited, it is not excluded that described in including
The process of key element, method, article or equipment there is also other identical element.
Claims (10)
1. a reservoir detecting method, it is characterised in that described method includes:
According to the destination layer position demarcated, obtain the seismic trace near well data corresponding with described destination layer position；
Based on described seismic trace near well data, set up reservoir detection degree of deep learning model；
Based on described reservoir detection degree of deep learning model and described destination layer position, it is thus achieved that the high-rise spy of described destination layer position study
Levy；
Obtain the reservoir characteristic of preset reference position on described destination layer position；
Reservoir characteristic based on described preset reference position, determines the high level with described preset reference position on described destination layer position
The reservoir characteristic in the region that feature is identical.
Method the most according to claim 1, it is characterised in that described based on described seismic trace near well data, sets up reservoir
Detection degree of deep learning model, including:
Using described seismic trace near well data as training data, the configuration parameter of pre-training reservoir detection degree of deep learning model；
Based on described seismic trace near well data and the target data corresponding with described seismic trace near well data, adjust pre-training and complete
Reservoir detection degree of deep learning model, thus set up reservoir detection degree of deep learning model.
Method the most according to claim 2, it is characterised in that described using described seismic trace near well data as training number
According to, the configuration parameter of pre-training reservoir detection degree of deep learning model, including:
Using other for described well geological data as training data, utilize limited Boltzmann machine or continuous limited Boltzmann machine, in advance
The configuration parameter of training reservoir detection degree of deep learning model；Update described configuration parameter, until completing to comprise the limited Bohr of multilamellar
The degree of depth confidence network of the reservoir detection degree of deep learning model of the most graceful machine or continuous limited Boltzmann machine, wherein, described configuration
Parameter includes model depth, every layer of model neuron node number, neuron activation functions and sparse object function.
Method the most according to claim 3, it is characterised in that described based on described seismic trace near well data and with described well
The target data that other seismic channel data is corresponding, adjusts the reservoir detection degree of deep learning model that pre-training completes, thus sets up reservoir
Detection degree of deep learning model, including:
Based on described seismic trace near well data and the target data corresponding with described seismic trace near well data, calculate based on back propagation
Method, utilizes batch stochastic gradient descent algorithm to solve the limited Boltzmann machine of described multilamellar or the storage of continuous limited Boltzmann machine
The degree of depth confidence network of layer detection degree of deep learning model, adjusts the reservoir detection degree of deep learning model parameter that pre-training completes, directly
Pre-conditioned to meeting, thus set up reservoir detection degree of deep learning model.
Method the most according to claim 4, it is characterised in that described based on described reservoir detection degree of deep learning model and institute
State destination layer position, it is thus achieved that the high-level characteristic of described destination layer position, including:
Calculating every layer of basic function of described reservoir detection degree of deep learning model, described every layer of basic function includes high-rise basic function；
Correlation coefficient based on described destination layer position with described high-rise basic function, determines the high-level characteristic of described destination layer position.
Method the most according to claim 1, it is characterised in that described reservoir characteristic include aqueous, gassiness, air content are few,
In lithology and hydro carbons at least one.
Method the most according to claim 1, it is characterised in that described according to the destination layer position demarcated, obtains and described mesh
The seismic trace near well data that mark layer position is corresponding, including:
Along the destination layer position demarcated, extract the other geological data of hoistway with window width time set in advance.
Method the most according to claim 1, it is characterised in that according to the destination layer position demarcated, obtain and described target
Before the seismic trace near well data step that layer position is corresponding, described method also includes:
Well logging, well logging and synthetic seismogram is utilized to demarcate described destination layer position.
9. a reservoir detection device, it is characterised in that described device includes:
Extract seismic data unit, for according to the destination layer position demarcated, obtain the other earthquake of the well corresponding with described destination layer position
Track data；
Set up reservoir detection degree of depth unit, for the reservoir detection degree of depth will be set up based on described seismic trace near well data
Practise model；
Obtain high-level characteristic unit, for based on described reservoir detection degree of deep learning model and described destination layer position, it is thus achieved that described
The high-level characteristic of destination layer position；
Obtain predeterminated position reservoir characteristic unit, for obtaining the reservoir characteristic of the preset reference position on described destination layer position；
Determine destination layer position reservoir characteristic unit, for reservoir characteristic based on described preset reference position, determine described target
The reservoir characteristic in region identical with the high-level characteristic of described preset reference position on layer position.
Device the most according to claim 9, it is characterised in that described reservoir detection degree of depth unit of setting up includes:
Pre-training reservoir detection degree of depth unit, for storing up described seismic trace near well data as training data, pre-training
The configuration parameter of layer detection degree of deep learning model；
Adjust reservoir detection degree of depth unit, for based on described seismic trace near well data and with described seismic trace near well data
Corresponding target data, adjusts the reservoir detection degree of deep learning model that pre-training completes, thus sets up reservoir detection degree of depth study
Model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610891005.7A CN106291701B (en) | 2016-10-13 | 2016-10-13 | Reservoir detecting method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610891005.7A CN106291701B (en) | 2016-10-13 | 2016-10-13 | Reservoir detecting method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106291701A true CN106291701A (en) | 2017-01-04 |
| CN106291701B CN106291701B (en) | 2019-03-15 |
Family
ID=57718595
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610891005.7A Expired - Fee Related CN106291701B (en) | 2016-10-13 | 2016-10-13 | Reservoir detecting method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106291701B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106886043A (en) * | 2017-03-01 | 2017-06-23 | 成都理工大学 | Reservoir detecting method based on geological data deep learning |
| 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 |
| CN109510805A (en) * | 2017-09-15 | 2019-03-22 | 中国科学院声学研究所 | A kind of network data security detection method and system based on security baseline model |
| CN110300095A (en) * | 2019-05-13 | 2019-10-01 | 江苏大学 | A kind of deep learning network inbreak detection method based on improvement learning rate |
| CN110462445A (en) * | 2017-02-09 | 2019-11-15 | 地质探索系统公司 | Geophysics deep learning |
| CN110806602A (en) * | 2018-08-06 | 2020-02-18 | 中国石油化工股份有限公司 | Intelligent seismic data random noise suppression method based on deep learning |
| CN112433248A (en) * | 2020-11-10 | 2021-03-02 | 成都理工大学 | Method for detecting hidden reservoir stratum in carbonate rock deposition environment |
| CN112507631A (en) * | 2020-11-19 | 2021-03-16 | 中国核动力研究设计院 | Narrow-channel flow instability outlet boundary gas-containing rate limit value testing method and system |
| CN113253346A (en) * | 2021-04-20 | 2021-08-13 | 中国石油大学(北京) | Prediction method for horizontal fractures of thin interbed reservoir |
| CN113325471A (en) * | 2021-05-21 | 2021-08-31 | 成都理工大学 | Seismic wave field subcomponent extraction method based on singular value decomposition |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105182431A (en) * | 2015-09-23 | 2015-12-23 | 中国石油大学(北京) | Petroleum reservoir rock constituent identification and quantitative evaluation method |
| WO2016089711A1 (en) * | 2014-12-05 | 2016-06-09 | Microsoft Technology Licensing, Llc | Quantum deep learning |
-
2016
- 2016-10-13 CN CN201610891005.7A patent/CN106291701B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016089711A1 (en) * | 2014-12-05 | 2016-06-09 | Microsoft Technology Licensing, Llc | Quantum deep learning |
| CN105182431A (en) * | 2015-09-23 | 2015-12-23 | 中国石油大学(北京) | Petroleum reservoir rock constituent identification and quantitative evaluation method |
Non-Patent Citations (5)
| Title |
|---|
| HINTON G E，SALAKHUTDINOV Ｒ Ｒ: "Ｒeducing the dimensionality of data with neural networks", 《SCIENCE》 * |
| YANN LECUN，ET AL.: "Deep learning", 《REVIEW》 * |
| 卢溜，等: "卷积神经网络在地震多属性分析中的应用", 《中国地球科学联合学术年会 2015》 * |
| 吴施楷，等: "基于连续限制玻尔兹曼机的支持向量机岩性识别方法", 《地球物理学进展》 * |
| 李建国，等: "深度学习在测井岩性识别中的应用", 《科技创新与应用》 * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110462445A (en) * | 2017-02-09 | 2019-11-15 | 地质探索系统公司 | Geophysics deep learning |
| CN106886043A (en) * | 2017-03-01 | 2017-06-23 | 成都理工大学 | Reservoir detecting method based on geological data deep learning |
| 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 |
| CN109510805A (en) * | 2017-09-15 | 2019-03-22 | 中国科学院声学研究所 | A kind of network data security detection method and system based on security baseline model |
| CN109510805B (en) * | 2017-09-15 | 2020-06-16 | 中国科学院声学研究所 | Network data security detection method and system based on security baseline model |
| 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 |
| CN110806602A (en) * | 2018-08-06 | 2020-02-18 | 中国石油化工股份有限公司 | Intelligent seismic data random noise suppression method based on deep learning |
| CN110806602B (en) * | 2018-08-06 | 2022-04-22 | 中国石油化工股份有限公司 | Intelligent seismic data random noise suppression method based on deep learning |
| CN110300095A (en) * | 2019-05-13 | 2019-10-01 | 江苏大学 | A kind of deep learning network inbreak detection method based on improvement learning rate |
| CN112433248A (en) * | 2020-11-10 | 2021-03-02 | 成都理工大学 | Method for detecting hidden reservoir stratum in carbonate rock deposition environment |
| CN112507631A (en) * | 2020-11-19 | 2021-03-16 | 中国核动力研究设计院 | Narrow-channel flow instability outlet boundary gas-containing rate limit value testing method and system |
| CN113253346A (en) * | 2021-04-20 | 2021-08-13 | 中国石油大学(北京) | Prediction method for horizontal fractures of thin interbed reservoir |
| CN113325471A (en) * | 2021-05-21 | 2021-08-31 | 成都理工大学 | Seismic wave field subcomponent extraction method based on singular value decomposition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106291701B (en) | 2019-03-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106291701A (en) | Reservoir detecting method and device | |
| US11010629B2 (en) | Method for automatically extracting image features of electrical imaging well logging, computer equipment and non-transitory computer readable medium | |
| US11320551B2 (en) | Training machine learning systems for seismic interpretation | |
| Ye et al. | A new tool for electro-facies analysis: multi-resolution graph-based clustering | |
| Deng et al. | Support vector machine as an alternative method for lithology classification of crystalline rocks | |
| Qi et al. | Neural network prediction of carbonate lithofacies from well logs, Big Bow and Sand Arroyo Creek fields, Southwest Kansas | |
| Wang et al. | Marcellus shale lithofacies prediction by multiclass neural network classification in the Appalachian Basin | |
| CN106886043B (en) | Reservoir detection method based on seismic data deep learning | |
| CN107678059B (en) | A kind of method, apparatus and system of reservoir gas-bearing identification | |
| Zhou et al. | A pattern‐search‐based inverse method | |
| CN104914467A (en) | Classification model channel extracting seismic facies clustering analysis method | |
| CN105044775B (en) | A kind of earthquake fluid inverting recognition methods and device | |
| Bhattacharya | A primer on machine learning in subsurface geosciences | |
| CN109113729A (en) | Lithology Identification Methods and device based on log | |
| CN110412662A (en) | Method for prediction of reservoirs of thin interbeded based on seismic multi-attribute deep learning | |
| Karimpouli et al. | Neuro-Bayesian facies inversion of prestack seismic data from a carbonate reservoir in Iran | |
| Ji et al. | HVSR-based site classification approach using general regression neural network (GRNN): case study for china strong motion stations | |
| Yang et al. | Deep-learning missing well-log prediction via long short-term memory network with attention-period mechanism | |
| Wu et al. | Effective parametrization for reliable reservoir performance predictions | |
| Udegbe et al. | Big Data Analytics for Seismic Fracture Identification, Using Amplitude-Based Statistics | |
| Bauer et al. | Neural network analysis of crosshole tomographic images: The seismic signature of gas hydrate bearing sediments in the Mackenzie Delta (NW Canada) | |
| Chikhi et al. | Probabilistic neural method combined with radial-bias functions applied to reservoir characterization in the Algerian Triassic province | |
| Smith et al. | Exploratory analysis of machine learning techniques in the Nevada geothermal play fairway analysis | |
| CN114707597A (en) | River facies tight sandstone reservoir complex lithofacies intelligent identification method and system | |
| Fan et al. | Quantitative Prediction of Low-Permeability Sandstone Grain Size Based on Conventional Logging Data by Deep Neural Network-Based BP Algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into 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 |
Granted publication date: 20190315 Termination date: 20211013 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |