CN109214025A - Reservoir parameter predication method and system based on Bayes's classification - Google Patents
Reservoir parameter predication method and system based on Bayes's classification Download PDFInfo
- Publication number
- CN109214025A CN109214025A CN201710548360.9A CN201710548360A CN109214025A CN 109214025 A CN109214025 A CN 109214025A CN 201710548360 A CN201710548360 A CN 201710548360A CN 109214025 A CN109214025 A CN 109214025A
- Authority
- CN
- China
- Prior art keywords
- reservoir
- parameter
- bayes
- classification
- likelihood function
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/08—Probabilistic or stochastic CAD
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Mathematical Analysis (AREA)
- Probability & Statistics with Applications (AREA)
- Algebra (AREA)
- Computer Hardware Design (AREA)
- Computational Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Geometry (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of reservoir parameter predication method and system based on Bayes's classification establish the first inversion equation of reservoir physical parameter this method comprises: being based on known reservoir elastic parameter；Based on Bayesian formula, the first inversion equation is converted to the second inversion equation of the likelihood function about reservoir elastic parameter；Based on Bayes-sequence Gauss method, the likelihood function of reservoir elastic parameter is solved；Likelihood function value based on obtained reservoir elastic parameter solves the second inversion equation and obtains reservoir physical parameter value.The present invention does not need to carry out complicated model initialization, fully consider the advantage of the stochastic behaviour of geological & geophysical characteristics, make inversion result with more practical geological Significance, while also can solve variogram not high problem of inversion result reliability in the case where well data is less, even Nei Jing is unevenly distributed in work area in geostatistics Method of Stochastic.
Description
Technical field
The invention belongs to petrochemical industry field of geophysical exploration, more particularly, to one kind based on Bayes point
The reservoir parameter predication method and system of class.
Background technique
Currently, most common two kinds in the domestic reservoir physical parameter inversion method based on conventional longitudinal wave reflection seismic data
Method is reservoir physical parameter non-linear inversion neural network based and the Method of Stochastic based on geostatistics respectively,
Both methods belongs to the scope of non-linear inversion.Reservoir physical parameter probability inversion method based on bayesian theory is not required to
Carry out complicated model initialization, but established based on reservoir model reservoir physical parameter with reservoir elastic parameter it
Between probability statistics relationship, and then combine with pre-stack seismic inversion, moreover, it also has both the random mould of geostatistics
Quasi- method can fully consider the advantage of the stochastic behaviour of geological & geophysical characteristics, and inversion result is made to anticipate with more practical geology
Justice, at the same also can solve in geostatistics Method of Stochastic variogram well data it is less, even in work area
Interior well becomes problem that is extremely unstable, and keeping inversion result reliability not high in the case where being unevenly distributed.But currently based on
Reservoir physical parameter probability inversion method likelihood function in calculating process of bayesian theory is difficult to accurately seek, and therefore, has
Necessity develops a kind of reservoir parameter predication method based on Bayes's classification, can seek stable and accurate likelihood function.
The information for being disclosed in background of invention part is merely intended to deepen the reason to general background technique of the invention
Solution, and it is known to those skilled in the art existing to be not construed as recognizing or imply that the information is constituted in any form
Technology.
Summary of the invention
The invention proposes a kind of reservoir parameter predication method and system based on Bayes's classification, can be by known
Reservoir elastic parameter establish reservoir physical parameter inversion equation, and reservoir properties are realized based on Bayes-sequential Gaussian simulation
Parametric joint inverting, this method are based on probability distribution theory, can preferably characterize the shadow of determining, uncertain error component
It rings, obtains reliable and stable reservoir physical parameter inversion result.
According to an aspect of the invention, it is proposed that a kind of reservoir parameter predication method based on Bayes's classification.The side
Method may include: 1) to establish the first inversion equation of reservoir physical parameter based on known reservoir elastic parameter；2) it is based on Bayes
First inversion equation is converted to the second inversion equation of the likelihood function about the reservoir elastic parameter by formula；3)
Based on Bayes-sequence Gauss method, the likelihood function of the reservoir elastic parameter is solved；4) it is based on the obtained reservoir
The likelihood function value of elastic parameter solves the second inversion equation and obtains the reservoir physical parameter value, wherein the reservoir parameter
Including the reservoir elastic parameter and the reservoir physical parameter.
Preferably, the reservoir physical parameter includes porosity φ, water saturation sw and shale content C；The reservoir
Elastic parameter includes velocity of longitudinal wave Vp, shear wave velocity VsAnd density p.
Preferably, reservoir physical parameter described in inverting is any in porosity φ, water saturation sw and shale content C
A kind of or any combination.
Preferably, first inversion equation are as follows:
Wherein, [φ, sw, C] is reservoir physical parameter, and φ is porosity, and sw is water saturation, and C is shale content；For reservoir elastic parameter, VpFor velocity of longitudinal wave, VsFor shear wave velocity, ρ is density；J is the classification of reservoir physical parameter
Number.
Preferably, second inversion equation are as follows:
Wherein, [φ, sw, C] is reservoir physical parameter, and φ is porosity, and sw is water saturation, and C is shale content；For reservoir elastic parameter, VpFor velocity of longitudinal wave, VsFor shear wave velocity, ρ is density；J is the classification of reservoir physical parameter
Number；P([[φ,sw,C]j) it is reservoir physical parameter prior distribution；P(Vp|[φ,sw,C]j) be velocity of longitudinal wave likelihood function, P
(Vs|[φ,sw,C]j) be shear wave velocity likelihood function, P (ρ | [φ, sw, C]j) be density likelihood function.
It is preferably based on Bayes-sequence Gauss method, the likelihood function for solving the reservoir elastic parameter includes:
3.1) three dimensional reservoir framework is created, by original well data and includes the intracorporal reservoir of column to simulation lattice unit
Mean parameter is transformed into the data of normal distribution；
3.2) random walk of the access to simulation lattice unit is specified；
3.3) from local distribution function p (xo|xs,z1,z2) in random sampling obtain xoAnalog result, wherein xoIt indicates
Value to be simulated, xsIndicate the value simulated in neighbouring unit, z1And z2It is comprising xoThe column at place it is intracorporal the first
Reservoir parameter and second of reservoir parameter；
3.4) by xoAs the value simulated, continue to simulate other grid lists according to the random walk that step 3.2) is specified
Member, until having accessed all grid cells；
3.5) normal state inverse transformation is carried out to step 3.4) analog result, obtains the likelihood function of the reservoir elastic parameter
Value.
Preferably, pass through following steps in step 3.3) from local distribution function p (xo|xs,z1,z2) in random sampling
Obtain xoAnalog result include:
A) based on Bayes' theorem to the local distribution function p (xo|xs,z1,z2) converted, it obtains:
p(xo|xs,z1,z2)∝p(xo|xs)f(z1|xs,xo)g(z2|xs,xo,z1) (3)
Wherein, xoIndicate value to be simulated, x in current unitsIndicate the value simulated in neighbouring unit, p (xo|
xs) be distributed for the condition of sequential Gaussian simulation；f(z1|xs,xo) be distributed for the condition of the first reservoir parameter；g(z2|xs,xo,z1)
For the condition distribution of second of reservoir parameter；
B) similar based on the normal state condition distribution during being simulated to the condition distribution in formula (3) with Gauss, by formula
(3) it converts are as follows:
p(x0|xs,z1,z2)∝N(mSK,σSK 2)×N(mf,σf 2)×N(mg,σg 2) (4)
Wherein, N (mSK,σSK 2) indicate x0Gaussian Profile, mSK,σSK 2Indicate x0Mean value and variance, N (mf,σf 2) indicate
The Gaussian Profile of the first reservoir parameter, mf, σf 2Indicate z1Mean value and variance, N (mg,σg 2) indicate second of reservoir parameter
Gaussian Profile, mg,σg 2Indicate z2Mean value and variance.
C) by seeking mSK,σSK 2, mf,σf 2, mg,σg 2To obtain local distribution function value.
According to another aspect of the invention, it is proposed that a kind of reservoir parameter forecast system based on Bayes's classification, described
System include memory, processor and storage on a memory and the computer program that can run on a processor, the place
1) reason device is performed the steps of when executing described program based on known reservoir elastic parameter, establish reservoir physical parameter first
Inversion equation；2) it is based on Bayesian formula, first inversion equation is converted into the likelihood about the reservoir elastic parameter
Second inversion equation of function；3) it is based on Bayes-sequence Gauss method, solves the likelihood function of the reservoir elastic parameter；
4) the likelihood function value based on the obtained reservoir elastic parameter solves the second inversion equation and obtains the reservoir properties ginseng
Numerical value, wherein the reservoir parameter includes the reservoir elastic parameter and the reservoir physical parameter.
Preferably, the reservoir physical parameter includes porosity φ, water saturation sw and shale content C；The reservoir
Elastic parameter includes velocity of longitudinal wave Vp, shear wave velocity VsAnd density p.
Preferably, reservoir physical parameter described in inverting is any in porosity φ, water saturation sw and shale content C
A kind of or any combination.
The beneficial effects of the present invention are: the reservoir physical parameter probability inversion method based on bayesian theory do not need into
The complicated model initialization of row, have both geostatistics Method of Stochastic can fully consider geological & geophysical characteristics with
The advantage of machine characteristic makes inversion result with more practical geological Significance, while also can solve the random mould of geostatistics
Variogram becomes extremely unstable in the case where well data is less, even Nei Jing is unevenly distributed in work area in quasi- method, and
The problem for keeping inversion result reliability not high.
Other features and advantages of the present invention will then part of the detailed description can be specified.
Detailed description of the invention
Exemplary embodiment of the invention is described in more detail in conjunction with the accompanying drawings, it is of the invention above-mentioned and its
Its purpose, feature and advantage will be apparent, wherein in exemplary embodiment of the invention, identical reference label
Typically represent same parts.
Fig. 1 shows the flow chart of the step of reservoir parameter predication method according to the present invention based on Bayes's classification.
Fig. 2 shows the schematic diagrames of three dimensional reservoir framework according to an embodiment of the invention.
Specific embodiment
The preferred embodiment of the present invention is described in more detail below.Although the following describe preferred implementations of the invention
Mode, however, it is to be appreciated that may be realized in various forms the present invention without that should be limited by the embodiments set forth herein.Phase
Instead, these embodiments are provided so that the present invention is more thorough and complete, and can be by the scope of the present invention completely
It is communicated to those skilled in the art.
Embodiment 1
In this embodiment, the reservoir parameter predication method according to the present invention based on Bayes's classification may include: step
Rapid 1, it is based on known reservoir elastic parameter, establishes the first inversion equation of reservoir physical parameter；Step 2, it is based on Bayesian formula,
First inversion equation is converted to the second inversion equation of the likelihood function about reservoir elastic parameter；Step 3, it is based on pattra leaves
This-sequence Gauss method, solve the likelihood function of reservoir elastic parameter；And step 4, it is based on obtained reservoir elastic parameter
Likelihood function value solve the second inversion equation obtain reservoir physical parameter value, wherein reservoir parameter includes reservoir elastic parameter
And reservoir physical parameter.
The embodiment establishes reservoir physical parameter inversion equation by known reservoir elastic parameter, and is based on Bayes-
Sequential Gaussian simulation realizes reservoir physical parameter joint inversion, and this method is based on probability distribution theory, can preferably characterize determination
, the influence of uncertain error component, obtain reliable and stable reservoir physical parameter inversion result.
Fig. 1 shows the flow chart of the step of reservoir parameter predication method according to the present invention based on Bayes's classification.
The specific steps of the reservoir parameter predication method according to the present invention based on Bayes's classification are described in detail below with reference to Fig. 1.
Step 1, it is based on known reservoir elastic parameter, establishes the first inversion equation of reservoir physical parameter.
In one example, reservoir physical parameter includes porosity φ, water saturation sw and shale content C；Reservoir bullet
Property parameter includes velocity of longitudinal wave Vp, shear wave velocity VsAnd density p.In one example, inverting reservoir physical parameter is porosity
Any one in φ, water saturation sw and shale content C or any combination.
In one example, the first inversion equation are as follows:
Wherein, [φ, sw, C] is reservoir physical parameter, and φ is porosity, and sw is water saturation, and C is shale content；For reservoir elastic parameter, VpFor velocity of longitudinal wave, VsFor shear wave velocity, ρ is density；J is the classification of reservoir physical parameter
Number.
Specifically, the reservoir parameter predication method based on Bayes's classification is established under Bayes's inverting framework, such as formula
(1) shown in, the first inversion equation is, under the conditions of known reservoir elastic parameter, the maximum a posteriori probability point of reservoir physical parameter
Cloth.
Step 2, it is based on Bayesian formula, the first inversion equation is converted into the likelihood function about reservoir elastic parameter
Second inversion equation.
In one example, the second inversion equation are as follows:
Wherein, [φ, sw, C] is reservoir physical parameter, and φ is porosity, and sw is water saturation, and C is shale content；For reservoir elastic parameter, VpFor velocity of longitudinal wave, VsFor shear wave velocity, ρ is density；J is the classification of reservoir physical parameter
Number；P([[φ,sw,C]j) it is reservoir physical parameter prior distribution；P(Vp|[φ,sw,C]j) be velocity of longitudinal wave likelihood function, P
(Vs|[φ,sw,C]j) be shear wave velocity likelihood function, P (ρ | [φ, sw, C]j) be density likelihood function.
Specifically, according to Bayesian formulaFormula (1) can be write as:
P([φ,sw,C]j) it is constant term, result is not had an impact, directly can give up to fall:
Assuming that it is mutually indepedent between reservoir elastic parameter, then available:
Bring formula (7) into (6) formula, available the second inversion equation as shown in formula (2):
[φ, sw, C]=
argMax{P([φ,sw,C]j)*P(Vp|[φ,sw,C]j)*P(Vs|[φ,sw,C]j)*P(ρ|[φ,sw,C]j)} (2)
P ([φ, sw, C] in formula (2)j) it is reservoir physical parameter prior distribution, statistical can be passed through by well-log information
Analysis obtains, likelihood function P (Vp|[φ,sw,C]j)、P(Vs|[φ,sw,C]j) and P (ρ | [φ, sw, C]j) pass through below step 3
It acquires.
Step 3, it is based on Bayes-sequence Gauss method, solves the likelihood function of reservoir elastic parameter.
In one example, it is based on Bayes-sequence Gauss method, the likelihood function for solving reservoir elastic parameter includes:
3.1) three dimensional reservoir framework is created, by original well data and includes the intracorporal reservoir of column to simulation lattice unit
Mean parameter is transformed into the data of normal distribution；
3.2) random walk of the access to simulation lattice unit is specified；
3.3) from local distribution function p (xo|xs,z1,z2) in random sampling obtain xoAnalog result, wherein xoIt indicates
Value to be simulated, xsIndicate the value simulated in neighbouring unit, z1And z2It is comprising xoThe column at place it is intracorporal the first
Reservoir parameter and second of reservoir parameter, wherein the first reservoir parameter (second of reservoir parameter) be include porosity φ, contain
Water saturation sw, shale content C, velocity of longitudinal wave Vp, shear wave velocity VsWith the one of which in density p；
3.4) by xoAs the value simulated, continue to simulate other grid lists according to the random walk that step 3.2) is specified
Member, until having accessed all grid cells；
3.5) normal state inverse transformation is carried out to step 3.4) analog result, obtains the likelihood function value of reservoir elastic parameter.
Specifically, the likelihood function for solving reservoir elastic parameter based on Bayes-sequence Gauss method initially sets up three-dimensional
Reservoir model uses xiIt indicates the reservoir parameter data at the grid cell i of three dimensional reservoir framework, uses z1,iAnd z2,iIt respectively indicates
In the mean value of the intracorporal two kinds of seismic properties of the vertical column comprising unit i, i.e.,
Wherein, nzIt is two attribute z in vertical direction in model1,iAnd z2,iBetween grid number；amAnd bmInclude unit i
Cylinder m layers of weighting coefficient, these weighting coefficients can be constant and are also possible to spatial variations；xiIt is on schedule
Support, and z1,iAnd z2,iIt indicates entire vertical cylinder, is block support.Our purpose is to generate the 3D realization of an x, this reality
Now other than the histogram of given data to be met and space covariance condition, also to meet the limitation of formula (8) and formula (9)
Condition.
As shown in Fig. 2, allowing xoValue to be simulated in current unit is indicated, such as the grid x in Fig. 20, analog result is logical
It crosses from local distribution function p (xo|xs,z1,z2) in random sampling obtain；xsIndicate the value simulated in neighbouring unit,
Including grid x1、x2、x3、x4、x5And x6；xcExpression is including xoThe intracorporal value simulated of column, including grid x2And x3；z1
And z2It is comprising xoCylinder at attribute data.
In one example, pass through following steps in step 3.3) from local distribution function p (xo|xs,z1,z2) in
Machine samples to obtain xoAnalog result include:
A) based on Bayes' theorem to local distribution function p (xo|xs,z1,z2) converted, it obtains:
p(xo|xs,z1,z2)∝p(xo|xs)f(z1|xs,xo)g(z2|xs,xo,z1) (3)
Wherein, xoIndicate value to be simulated, x in current unitsIndicate the value simulated in neighbouring unit, p (xo|
xs) be distributed for the condition of sequential Gaussian simulation；f(z1|xs,xo) be distributed for the condition of the first reservoir parameter；g(z2|xs,xo,z1)
For the condition distribution of second of reservoir parameter；
B) similar based on the normal state condition distribution during being simulated to the condition distribution in formula (3) with Gauss, by formula
(3) it converts are as follows:
p(x0|xs,z1,z2)∝N(mSK,σSK 2)×N(mf,σf 2)×N(mg,σg 2) (4)
Wherein, N (mSK,σSK 2) indicate x0Gaussian Profile, mSK,σSK 2Indicate x0Mean value and variance, N (mf,σf 2) indicate
The Gaussian Profile of the first reservoir parameter, mf, σf 2Indicate z1Mean value and variance, N (mg,σg 2) indicate second of reservoir parameter
Gaussian Profile, mg,σg 2Indicate z2Mean value and variance.
C) by seeking mSK,σSK 2, mf,σf 2, mg,σg 2To obtain local distribution function value.
Specifically, it is assumed that f (z1|xs,xo)=f (z1|xc,xo) and g (z2|xs,xo,z1)=g (z2|xc,xo,z1), that is,
It says, z1And z2Condition distribution be dependent only on the value simulated in cylinder, it is unrelated with the value simulated in neighbouring unit,
It can be obtained in conjunction with formula (3):
p(xo|xs,z1,z2)∝p(xo|xs)f(z1|xc,xo)g(z2|xc,xo,z1) (10)
The first item on the right, is in the value x simulated in formula (10)sUnder conditions of xoCondition distribution, this in Gauss
By the obtained mean value (m) of common Ke Lijin and variance (σ in simulation process2) normal state condition distribution it is similar, it may be assumed that
Wherein, SK represents common Ke Lijin.If allowing xc+oIndicate the value simulated in current cylinder and just in mould
The set of quasi- value, then the Section 2 f (z on the right of formula (10)1|xc,xo)≡f(z1|xc+o) it is known as z1Likelihood function,
It is by Gaussian Profile.If we use ∑j∈c+oIndicate the unit simulated in cylinder and the grid cell being modeled
Upper summation is usedExpression is summed on the grid cell that remaining is not accessed, then f (z1|xc+o) mean value and variance are as follows:
Wherein, ajIt is the weighting coefficient of the first attribute data in cylinder；λjIt is the weighting system of golden system in following gram
Number,
CkrIndicate the covariance between k and r point.
Section 3 g (z on the right of formula (10)2|xc,xo,z1)≡g(z2|xc+o,z1) it is known as z2Likelihood function, it meet
Normal distribution, mean value and variance are as follows:
Wherein bjIt is the weight coefficient of second of attribute data in cylinder, μjAnd μz1It is the weighting of golden system in following gram
Coefficient:
Therefore, available:
p(x0|xs,z1,z2)∝N(mSK,σSK 2)×N(mf,σf 2)×N(mg,σg 2) (4)
Here N (m, σ2) indicate mean value be m, variance σ2Gaussian Profile.The part on the right of formula (4) is rearranged,
Cancellation contains xoItem, obtain final likelihood function:
p(x0|xs,z1,z2)∝N(mo,σo 2) (18)
Wherein,
σo 2=σSK 2σf 2σg 2/d (20)
D in formula (19), (20) are as follows:
D=σf 2σg 2+(ao+λo)2σSK 2σg 2+(bo+μo)2σSK 2σf 2 (21)
From N (m, σ2) in random sampling obtain xoValue, by xoAs the value simulated, according to access specified in advance
To the random walk of simulation lattice unit, continue to simulate other grid cells, until having accessed all grid cells.
In conjunction with the P (V in formula (5)p|[φ,sw,C]j), only consider VpIn the case where φ, formula (18) can be rewritten
Are as follows:
p(vp0|vps,z1,z2)∝N(mo,σo 2) (22)
z1Indicate VpMean value, z2Indicate the mean value of φ, it follows that the likelihood function of velocity of longitudinal wave.In formula (22)
φ could alternatively be sw and C, it will be able to acquire the likelihood function of shear wave velocity and the likelihood function of density respectively.
Step 4, the likelihood function value based on obtained reservoir elastic parameter solves the second inversion equation and obtains reservoir object
Property parameter value, wherein reservoir parameter includes reservoir elastic parameter and reservoir physical parameter.
It specifically, will be after normal state changes, so being carried out first to it by each likelihood function that formula (22) obtain
Normal state anti-change, then substitute into formula (2) in, can inverting obtain reservoir physical parameter.
The embodiment does not need to carry out complicated model based on the reservoir physical parameter probability inversion method of bayesian theory
Initialization, have both geostatistics Method of Stochastic can fully consider geological & geophysical characteristics stochastic behaviour it is excellent
Gesture makes inversion result with more practical geological Significance, while also can solve and becoming in geostatistics Method of Stochastic
Difference function becomes extremely unstable in the case where well data is less, even Nei Jing is unevenly distributed in work area, and makes inversion result
The not high problem of reliability.
It will be understood by those skilled in the art that above to the purpose of the description of the embodiment of the present invention only for illustratively saying
The beneficial effect of bright the embodiment of the present invention is not intended to limit embodiments of the invention to given any example.
Embodiment 2
According to an embodiment of the invention, providing a kind of reservoir parameter forecast system based on Bayes's classification, system packet
The computer program that includes memory, processor and storage on a memory and can run on a processor, processor execute journey
It is performed the steps of when sequence based on known reservoir elastic parameter, establishes the first inversion equation of reservoir physical parameter；Based on pattra leaves
First inversion equation is converted to the second inversion equation of the likelihood function about reservoir elastic parameter by this formula；Based on pattra leaves
This-sequence Gauss method, solve the likelihood function of reservoir elastic parameter；And the likelihood based on obtained reservoir elastic parameter
Functional value solves the second inversion equation and obtains reservoir physical parameter value, wherein reservoir parameter includes reservoir elastic parameter and reservoir
Physical parameter.
The embodiment establishes reservoir physical parameter inversion equation by known reservoir elastic parameter, and is based on Bayes-
Sequential Gaussian simulation realizes reservoir physical parameter joint inversion, and this method is based on probability distribution theory, can preferably characterize determination
, the influence of uncertain error component, obtain reliable and stable reservoir physical parameter inversion result.
In one example, reservoir physical parameter includes porosity φ, water saturation sw and shale content C；Reservoir bullet
Property parameter includes velocity of longitudinal wave Vp, shear wave velocity VsAnd density p.
In one example, inverting reservoir physical parameter is appointing in porosity φ, water saturation sw and shale content C
It anticipates a kind of or any combination.
The embodiment does not need to carry out complicated model based on the reservoir physical parameter probability inversion method of bayesian theory
Initialization, have both geostatistics Method of Stochastic can fully consider geological & geophysical characteristics stochastic behaviour it is excellent
Gesture makes inversion result with more practical geological Significance, while also can solve and becoming in geostatistics Method of Stochastic
Difference function becomes extremely unstable in the case where well data is less, even Nei Jing is unevenly distributed in work area, and makes inversion result
The not high problem of reliability.
It will be understood by those skilled in the art that above to the purpose of the description of the embodiment of the present invention only for illustratively saying
The beneficial effect of bright the embodiment of the present invention is not intended to limit embodiments of the invention to given any example.
Various embodiments of the present invention are described above, above description is exemplary, and non-exclusive, and
It is not limited to disclosed each embodiment.Without departing from the scope and spirit of illustrated each embodiment, for this skill
Many modifications and changes are obvious for the those of ordinary skill in art field.
Claims (10)
1. a kind of reservoir parameter predication method based on Bayes's classification, which is characterized in that this method comprises:
1) it is based on known reservoir elastic parameter, establishes the first inversion equation of reservoir physical parameter；
2) it is based on Bayesian formula, first inversion equation is converted into the likelihood function about the reservoir elastic parameter
Second inversion equation；
3) it is based on Bayes-sequence Gauss method, solves the likelihood function of the reservoir elastic parameter；
4) the likelihood function value based on the obtained reservoir elastic parameter solves the second inversion equation and obtains the reservoir object
Property parameter value,
Wherein, the reservoir parameter includes the reservoir elastic parameter and the reservoir physical parameter.
2. the reservoir parameter predication method according to claim 1 based on Bayes's classification, wherein
The reservoir physical parameter includes porosity φ, water saturation sw and shale content C；
The reservoir elastic parameter includes velocity of longitudinal wave Vp, shear wave velocity VsAnd density p.
3. the reservoir parameter predication method according to claim 2 based on Bayes's classification, wherein reservoir object described in inverting
Property parameter is any one or any combination in porosity φ, water saturation sw and shale content C.
4. the reservoir parameter predication method according to claim 2 based on Bayes's classification, wherein the first inverting side
Journey are as follows:
Wherein, [φ, sw, C] is reservoir physical parameter, and φ is porosity, and sw is water saturation, is shale content；
For reservoir elastic parameter, VpFor velocity of longitudinal wave, VsFor shear wave velocity, ρ is density；J is the classification number of reservoir physical parameter.
5. the reservoir parameter predication method according to claim 4 based on Bayes's classification, wherein the second inverting side
Journey are as follows:
Wherein, [φ, sw, C] is reservoir physical parameter, and φ is porosity, and sw is water saturation, and C is shale content；For reservoir elastic parameter, VpFor velocity of longitudinal wave, VsFor shear wave velocity, ρ is density；J is the classification of reservoir physical parameter
Number；P([[φ,sw,C]j) it is reservoir physical parameter prior distribution；P(Vp|[φ,sw,C]j) be velocity of longitudinal wave likelihood function, P
(Vs|[φ,sw,C]j) be shear wave velocity likelihood function, P (ρ | [φ, sw, C]j) be density likelihood function.
6. the reservoir parameter predication method according to claim 1 based on Bayes's classification, wherein be based on Bayes-sequence
Gauss method is passed through, the likelihood function for solving the reservoir elastic parameter includes:
3.1) three dimensional reservoir framework is created, by original well data and includes the intracorporal reservoir parameter of column to simulation lattice unit
Mean value transformation at normal distribution data；
3.2) random walk of the access to simulation lattice unit is specified；
3.3) from local distribution function p (xo|xs,z1,z2) in random sampling obtain xoAnalog result, wherein xoIt indicates to mould
Quasi- value, xsIndicate the value simulated in neighbouring unit, z1And z2It is comprising xoThe first intracorporal reservoir of the column at place
Parameter and second of reservoir parameter；
3.4) by xoAs the value simulated, continue to simulate other grid cells according to the random walk that step 3.2) is specified, directly
To having accessed all grid cells；
3.5) normal state inverse transformation is carried out to step 3.4) analog result, obtains the likelihood function value of the reservoir elastic parameter.
7. the reservoir parameter predication method according to claim 6 based on Bayes's classification, in step 3.3) by with
Lower step is from local distribution function p (xo|xs,z1,z2) in random sampling obtain xoAnalog result include:
A) based on Bayes' theorem to the local distribution function p (xo|xs,z1,z2) converted, it obtains:
p(xo|xs,z1,z2)∝p(xo|xs)f(z1|xs,xo)g(z2|xs,xo,z1) (3)
Wherein, xoIndicate value to be simulated, x in current unitsIndicate the value simulated in neighbouring unit, p (xo|xs) it is sequence
Pass through the condition distribution of Gauss simulation；f(z1|xs,xo) be distributed for the condition of the first reservoir parameter；g(z2|xs,xo,z1) it is second
The condition distribution of kind reservoir parameter；
B) similar based on the normal state condition distribution during being simulated to the condition distribution in formula (3) with Gauss, formula (3) are become
It is changed to:
p(x0|xs,z1,z2)∝N(mSK,σSK 2)×N(mf,σf 2)×N(mg,σg 2) (4)
Wherein, N (mSK,σSK 2) indicate x0Gaussian Profile, mSK,σSK 2Indicate x0Mean value and variance, N (mf,σf 2) indicate first
The Gaussian Profile of kind reservoir parameter, mf, σf 2Indicate z1Mean value and variance, N (mg,σg 2) indicate second of reservoir parameter Gauss
Distribution, mg,σg 2Indicate z2Mean value and variance.
C) by seeking mSK,σSK 2, mf,σf 2, mg,σg 2To obtain local distribution function value.
8. a kind of reservoir parameter forecast system based on Bayes's classification, which is characterized in that the system comprises memories, processing
Device and storage on a memory and the computer program that can run on a processor, the reality when processor executes described program
Existing following steps:
1) it is based on known reservoir elastic parameter, establishes the first inversion equation of reservoir physical parameter；
2) it is based on Bayesian formula, first inversion equation is converted to second of the likelihood function about reservoir elastic parameter
Inversion equation；
3) it is based on Bayes-sequence Gauss method, solves the likelihood function of the reservoir elastic parameter；
4) the likelihood function value based on obtained reservoir elastic parameter solves the second inversion equation and obtains the reservoir properties ginseng
Numerical value,
Wherein, the reservoir parameter includes the reservoir elastic parameter and the reservoir physical parameter.
9. the reservoir parameter forecast system according to claim 8 based on Bayes's classification, wherein
The reservoir physical parameter includes porosity φ, water saturation sw and shale content C；
The reservoir elastic parameter includes velocity of longitudinal wave Vp, shear wave velocity VsAnd density p.
10. the reservoir parameter forecast system according to claim 8 based on Bayes's classification, wherein reservoir described in inverting
Physical parameter is any one or any combination in porosity φ, water saturation sw and shale content C.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710548360.9A CN109214025A (en) | 2017-07-06 | 2017-07-06 | Reservoir parameter predication method and system based on Bayes's classification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710548360.9A CN109214025A (en) | 2017-07-06 | 2017-07-06 | Reservoir parameter predication method and system based on Bayes's classification |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109214025A true CN109214025A (en) | 2019-01-15 |
Family
ID=64993040
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710548360.9A Pending CN109214025A (en) | 2017-07-06 | 2017-07-06 | Reservoir parameter predication method and system based on Bayes's classification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109214025A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109948682A (en) * | 2019-03-12 | 2019-06-28 | 湖南科技大学 | Laser radar point cloud power line classification method based on normal state random sampling distribution |
| CN110175721A (en) * | 2019-05-30 | 2019-08-27 | 长江大学 | A kind of quantitative preferred method of potentiality recess and system based on Bayesian statistics prediction |
| CN112906242A (en) * | 2021-03-18 | 2021-06-04 | 吉林大学 | Geophysical modeling method based on combination of naive Bayes method and proximity classification method |
| WO2022226856A1 (en) * | 2021-04-29 | 2022-11-03 | 浙江海洋大学 | Method and apparatus for inverting geoacoustic parameters of multilayer seabed in shallow sea, and computer device and storage medium |
| CN115879344A (en) * | 2022-12-13 | 2023-03-31 | 水利部交通运输部国家能源局南京水利科学研究院 | Soil moisture migration model parameter inversion method based on random sampling |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103760600A (en) * | 2014-01-07 | 2014-04-30 | 中国石油天然气股份有限公司 | Gas saturation inversion method |
| CN104297785A (en) * | 2014-09-29 | 2015-01-21 | 中国石油天然气股份有限公司 | Lithofacies constrained reservoir physical property parameter inversion method and device |
-
2017
- 2017-07-06 CN CN201710548360.9A patent/CN109214025A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103760600A (en) * | 2014-01-07 | 2014-04-30 | 中国石油天然气股份有限公司 | Gas saturation inversion method |
| CN104297785A (en) * | 2014-09-29 | 2015-01-21 | 中国石油天然气股份有限公司 | Lithofacies constrained reservoir physical property parameter inversion method and device |
Non-Patent Citations (2)
| Title |
|---|
| 印兴耀等: "贝叶斯-序贯高斯模拟方法", 《石油大学学报(自然科学版)》 * |
| 胡华锋: "基于叠前道集的储层参数反演方法研究", 《中国优秀硕士学位论文全文数据库基础科学辑》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109948682A (en) * | 2019-03-12 | 2019-06-28 | 湖南科技大学 | Laser radar point cloud power line classification method based on normal state random sampling distribution |
| CN110175721A (en) * | 2019-05-30 | 2019-08-27 | 长江大学 | A kind of quantitative preferred method of potentiality recess and system based on Bayesian statistics prediction |
| CN112906242A (en) * | 2021-03-18 | 2021-06-04 | 吉林大学 | Geophysical modeling method based on combination of naive Bayes method and proximity classification method |
| CN112906242B (en) * | 2021-03-18 | 2023-02-28 | 吉林大学 | Geophysical modeling method based on combination of naive Bayes method and proximity classification method |
| WO2022226856A1 (en) * | 2021-04-29 | 2022-11-03 | 浙江海洋大学 | Method and apparatus for inverting geoacoustic parameters of multilayer seabed in shallow sea, and computer device and storage medium |
| CN115879344A (en) * | 2022-12-13 | 2023-03-31 | 水利部交通运输部国家能源局南京水利科学研究院 | Soil moisture migration model parameter inversion method based on random sampling |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109214025A (en) | Reservoir parameter predication method and system based on Bayes's classification | |
| Kanevski | Advanced mapping of environmental data | |
| CN109709603B (en) | Seismic horizon identification and tracking method and system | |
| Aleardi et al. | 1D elastic full‐waveform inversion and uncertainty estimation by means of a hybrid genetic algorithm–Gibbs sampler approach | |
| Bai et al. | Hybrid geological modeling: Combining machine learning and multiple-point statistics | |
| Ailliot et al. | An autoregressive model with time‐varying coefficients for wind fields | |
| Biau et al. | Estimation of precipitation by kriging in the EOF space of thesea level pressure field | |
| CN107688201B (en) | RBM-based seismic prestack signal clustering method | |
| de Almeida | Stochastic simulation methods for characterization of lithoclasses in carbonate reservoirs | |
| CN109272029B (en) | Well control sparse representation large-scale spectral clustering seismic facies partitioning method | |
| CN108563837B (en) | Method and system for correcting model parameters of alluvial river water sand model in real time | |
| Seiler et al. | Structural surface uncertainty modeling and updating using the ensemble Kalman filter | |
| Erdogan Erten et al. | Combination of machine learning and kriging for spatial estimation of geological attributes | |
| Tzu-hao et al. | Reservoir uncertainty quantification using probabilistic history matching workflow | |
| Fossum et al. | Parameter sampling capabilities of sequential and simultaneous data assimilation: II. Statistical analysis of numerical results | |
| Shen et al. | Distributed Markov Chain Monte Carlo Method on Big-Data Platform for Large-Scale Geosteering Inversion Using Directional Electromagnetic Well Logging Measurements. | |
| CN104391325A (en) | Discontinuous heterogeneous geologic body detection method and device | |
| Martin et al. | Non-stationary geostatistical modeling: a case study comparing LVA estimation frameworks | |
| US20230205842A1 (en) | Apparatus and Methods for Improved Subsurface Data Processing Systems | |
| CN112444850B (en) | Seismic data velocity modeling method, storage medium and computing device | |
| CN113970787A (en) | Physical property parameter inversion method, physical property parameter inversion device, computer equipment and storage medium | |
| CN115639605B (en) | Automatic identification method and device for high-resolution fault based on deep learning | |
| Cornford et al. | Modelling frontal discontinuities in wind fields | |
| Le et al. | Physics-constrained deep learning for production forecast in tight reservoirs | |
| GB2584449A (en) | Apparatus method and computer-program product for calculating a measurable geological metric |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190115 |