CN103967478B - A kind of peupendicular hole meteor trail echoes method based on conducting probe - Google Patents
A kind of peupendicular hole meteor trail echoes method based on conducting probe Download PDFInfo
- Publication number
- CN103967478B CN103967478B CN201410216750.2A CN201410216750A CN103967478B CN 103967478 B CN103967478 B CN 103967478B CN 201410216750 A CN201410216750 A CN 201410216750A CN 103967478 B CN103967478 B CN 103967478B
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- munderover
- mtd
- individual
- 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.)
- Active
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The present invention relates to a kind of peupendicular hole meteor trail echoes method based on conducting probe, meteor trail echoes are carried out available for the vertical well section to water-oil phase withdrawal well.The method is characterized in that comprising the following steps:Training sample and test sample are obtained by dynamic experimental device, the output signal to conducting probe carries out multi-angle feature extraction, utilize the multiple scalar characterization construction features vector extracted;Dimensionality reduction is compressed to characteristic vector using principal component analytical method, the linear dependence between each characteristic parameter is eliminated;For a variety of flow patterns in peupendicular hole, multiclass classification of flow patterns device is set up using support vector machine classification method;The kernel function of SVMs is selected, penalty factor is optimized with kernel functional parameter using test sample.On-line identification can be carried out to the flow pattern of water-oil phase peupendicular hole, with the higher meteor trail echoes degree of accuracy, can meet underground survey demand the beneficial effects of the invention are as follows this method.
Description
Technical field
The present invention relates to a kind of peupendicular hole meteor trail echoes method based on conducting probe, available for water-oil phase withdrawal well
Vertical well section carry out meteor trail echoes.
Background technology
Production logging typically refers to the logging operation carried out after Oil/gas Well is gone into operation, including the injection well after completion and production
Go out borehole logging tool technology.In recent years, the task of production logging extends to the whole of oil well from each stage at drilling well initial stage and opened
Course is adopted, main purpose is the dynamic change for the situation and monitoring oil reservoir for evaluating Oil/gas Well itself, and foundation is provided for field management.
Different with measurement purpose according to measurement object, production logging can substantially be divided into three important components:Engineering technology well logging,
Evaluation of producing is logged well and Production development well logging.Wherein, production profile logging belongs to the category of Production development well logging, through oil gas
Overall process of the well from going into operation to scrapping.Main task is to carry out dynamic monitoring to the flow profile of withdrawal well, understands each payzone
Output situation.The well log interpretation of production profile is the basic data of reservoir description of becoming more meticulous, and is played in oil-gas field development field
Very important effect.In underground, gas phase is dissolved mainly in oil, and the meltage very little in water.It is less for gas production
Oil well, when fluid from stratum flow into test well section when, if well pressure be higher than oil phase bubble point pressure, gas then will not be from oil phase
Separate out, the fluid-mixing in such pit shaft is just considered oil-water two-phase flow.It is this kind of at home in the exploitation of onshore oil field
Oil well occupies certain proportion, therefore has very important meaning to the research of oil-water two-phase flow Tech of Production Profile Logging.
The content of the invention
Flow pattern is the important parameter of two phase flow, characterizes the distribution situation of fluid each phase medium in flow process.Flow pattern pair
The influence of oil well production characteristic and production performance observation is always the focus problem in production logging field.Flow pattern is different,
The flow behavior and heat and mass transfer performance of two-phase fluid in well are not only influenceed, the accurate survey of other diphasic stream parameters is also contributed to
Amount.Under normal circumstances, if the flow pattern in well can be predicted before well logging, it is possible to the measuring instrument that selection is more suitable for, obtain
More preferable measurement effect.Therefore, research peupendicular hole meteor trail echoes method is of great significance to engineer applied tool, wants reality
The now accurate measurement of vertical well yield and moisture content, it is necessary first to determine the flow pattern in well.In order to realize flow and moisture content
Accurate measurement, a kind of peupendicular hole meteor trail echoes method based on conducting probe of invention, it is characterised in that including following
Step:
(1) training sample and test sample are obtained by dynamic experimental device, the output signal to conducting probe uses many
The method of kind carries out feature extraction, utilizes the multiple scalar characterization construction features vector extracted;The output signal of conducting probe is
Single channel time series signal, feature extracting method includes:Statistical analysis, Symbolic time series analysis, chaos time sequence point
Analysis, wavelet transformation, Hilbert-Huang conversion, fractal theory and density-wave theory;
, all can not be directly sharp for the measurement of peupendicular hole parameter detecting, either meteor trail echoes or flow and moisture content
It is modeled with voltage signal, and feature extraction should be carried out to signal first before modeling.The feature extraction of voltage signal is real
It is the analysis to its wave characteristic and quantitative description in matter, and the quality of feature extraction can directly affect parameter measurement model
Validity and measurement accuracy.Influence the factor of signal fluctuation characteristic a lot, when pipe parameter and probe size structure determination, this
A little factors are substantially from two aspects:On the one hand it is that fluid comes from two-phase medium and conducting probe in itself, on the other hand
Interaction, wherein existing certainty factor contains some enchancement factors again.Therefore, feature extracting method should take into account this two
The information of aspect.
(2) characteristic vector is compressed using principal component analytical method linear between dimensionality reduction, each characteristic parameter of elimination
Correlation;Principal component Y1, Y2..., YpIt is expressed as primitive character parameter X1, X2..., XpLinear combination, be designated as following algebraically shape
Formula:
In formula, Yi=ui' X is i-th of principal component of primitive character parameter, ui=(ui1, ui2..., uip) ' for coefficient to
Amount, X=(X1, X2..., Xp) ' be primitive character parameter vector, XiFor i-th of primitive character parameter;Wherein, vectorial ui' it is vector
uiTransposition, the linear combination is limited to tri- constraint principles of following a, b, c:
a.ui′ui=1;
B. as i ≠ j, YiWith YjIt is mutually orthogonal;
c.Y1It is X1, X2..., XpAll linear combinations in variance the maximum, Y2For except Y1X in addition1, X2..., Xp's
Variance the maximum in all linear combinations, the rest may be inferred, YpFor except Y1, Y2..., Yp-1X in addition1, X2..., XpIt is one tangent linear
Variance the maximum in combination;
For useful information that is as complete as possible, comprehensively reflecting in signal, using a variety of methods to electricity in step (1)
Pressure fluctuation signal has carried out feature extraction to construct characteristic vector.But, the characteristic parameter extracted is not screened by careful,
The useful information of reflection flow of fluid rule is not only contained in characteristic vector, the correlation between feature is also contains and produces
Redundancy and noise.If characteristic vector to be directly used for the structure of classification of flow patterns device, supporting vector will be had a strong impact on
The nicety of grading and generalization ability of machine.In this case, general resolving ideas is the feature that these are had into dependency relation
Parameter is replaced with a few overall target.Principal component analysis is a kind of multivariate statistics and processing method, and its basic thought is
Polytomy variable with certain correlation in luv space is converted to the orthogonal principal component of a few in new space, it is right
While initial data is compressed dimensionality reduction, just less information loss is protected.
(3) a variety of flow patterns in peupendicular hole are directed to, multiclass classification of flow patterns device is set up using support vector machine classification method;Hang down
Flow pattern in straight well includes:Water-In-Oil flow pattern, oil-in-water flow pattern, transition flow pattern, scattered oil vacuole stream and scattered bubble stream;It can use
Many classification policys have one-to-one, one-to-many, directed acyclic graph and directly construction one polytypic object function;Utilize training
Sample is trained to the mathematical modeling of SVMs, and the training of supporting vector machine model is one with the excellent of Prescribed Properties
Change problem, can be described as following form:
In formula, w is weight vectors, and b is biasing, and C is penalty factor,For mapping function, ξiFor slack variable, yiFor
The output of model;Solving the optimization problem needs to be translated into following dual form:
In formula, K (xi, xj) it is the kernel function for meeting Mercer conditions, xiFor the input of model, αiFor Lagrange multiplier.
Finally the mathematical modeling of determination SVMs is:
The decision function of classification of flow patterns device is:
The interaction process of water-oil phase fluid and conducting probe is sufficiently complex and is difficult to predict, wants by having set up
It is substantially unworkable that standby theoretical model, which carries out meteor trail echoes,.In this case, a more satisfactory resolving ideas
It is an attempt to use soft-measuring technique.SVMs is a kind of emerging artificial intelligence technology, in treatment classification and regression problem
On show huge potentiality and advantage.SVMs utilizes limited sample using structural risk minimization as guideline
Information seeks optimal compromise between model training precision and Generalization Capability, is solving small sample, non-linear and high problem of dimension
When show the advantage of uniqueness, be maximally effective machine learning method so far.
(4) kernel function of SVMs is selected, penalty factor is carried out with kernel functional parameter σ using test sample excellent
Change;One kind in the following kernel function of Selection of kernel function of SVMs:Linear kernel function, multiphase formula kernel function, radial direction base core
Function and neutral net kernel function;The optimization method used is genetic algorithm;Genetic algorithm used for reference simultaneously it is Darwinian enter
Change opinion and Mendelian theory Of heredity are a kind of by simulating the global random searching that living nature natural evolution mechanism grows up
And optimization method.Genetic algorithm to object function without can lead or it is successional limitation and directly solve;It need not set
Specific search rule and the direction of search, the optimal solution of condition is met in global parallel search.Genetic algorithm be intellectual analysis with
The key technology of calculating field, is widely used in machine learning, signal transacting, intelligent control and artificial life etc. by people
Field.Optimization Steps based on genetic algorithm are as follows:
A., initial population scale, maximum evolutionary generation T, penalty factor, kernel functional parameter σ hunting zone, intersection are set
And mutation probability;
B. chromosome coding is carried out to parameter C, σ, randomly generates initial population, initialization evolutionary generation t=0;
C. fitness R individual in population is calculatedcv(C, σ);
D. according to individual adaptation degree, individual is selected into the next generation from current population by the way of roulette;
E. crossover operation is carried out to the individual in population, produces new individual and enter of future generation;
F. some genes in mutation operation, random change individual are performed to the individual in population and produce new individual;
If being unsatisfactory for end condition and t≤T, step b is jumped to, if meeting end condition, selection is adapted to
The maximum individual of degree is used as optimum results..
Can carry out on-line identification to the flow pattern of water-oil phase peupendicular hole the beneficial effects of the invention are as follows this method, with compared with
The high meteor trail echoes degree of accuracy, can meet underground survey demand.
Brief description of the drawings
Fig. 1 is that classification of flow patterns device sets up process in embodiment;
Fig. 2 is the structure chart of the conducting probe employed in embodiment, in figure:Metal shell (201), insulating barrier
(202), electrode (203);
Fig. 3 is the meteor trail echoes process of peupendicular hole in embodiment.
Embodiment
Fig. 1 is that classification of flow patterns device sets up process in embodiment;Flow pattern in peupendicular hole is with scattered stream (oil vacuole
Stream, bubble stream and transition flow) based on, therefore mainly consider detection energy of the conducting probe to discrete phase in probe structure design
Power, and the influence that sleeve pipe and instrument arm are exported to probe.Fig. 2 is the structure of the conducting probe employed in embodiment
In figure, figure:Metal shell (201), insulating barrier (202), electrode (203).Conducting probe entire outer diameter 3mm, probe electrode is exposed
3mm, during actual measurement, 3mm oil vacuole or bubble can be more than for detection diameter and is not influenceed by continuous phase.Fig. 3 is tool
The meteor trail echoes process of peupendicular hole in body embodiment, in conjunction with the embodiment of the brief description of the drawings present invention.
(1) training sample and test sample are obtained by dynamic experimental device, the output signal progress to conducting probe is more
Angle character is extracted, and utilizes the multiple scalar characterization construction features vector extracted;When the output signal of conducting probe is single channel
Between sequence signal, adoptable feature extracting method has:Statistical analysis, Symbolic time series analysis, chaos time sequence point
Analysis, wavelet transformation, Hilbert-Huang conversion, fractal theory and density-wave theory;
The collection of training sample and test sample is completed on Simulation of Multiphase Flow device.Experiment is diesel oil, density with oil
825kg/m3, viscosity 3 × 10-3Pas, surface tension 28.62 × 10-3N/m;Water is normal domestic water, density 1000kg/m3、
Viscosity 0.890 × 10-3Pas, surface tension 71.25 × 10-3N/m.Total (volume) the flow Q of profittAdjustable range for 10~
200m3/ d (cubic meter is daily), control interval is 10m3/d;Moisture content CwAdjustable range be 0.1~0.9, control interval is
0.1.Matched for every kind of different flow and moisture content, gather one group of experimental data, the acquisition time of every group of data is
15min, sample frequency is 100Hz, amounts to the data obtained under 180 groups of different parameters conditions of mixture ratios.In 180 groups of whole experiments
In data, wherein oil-in-water flow pattern is 89 groups, and transition flow pattern is 57 groups, and Water-In-Oil flow pattern is 34 groups.Sample is trained in three kinds of flow patterns
Originally the allocation proportion with test sample is respectively:Water-In-Oil flow pattern 19:15, transition flow pattern 33:24, oil-in-water flow pattern 48:41, always
The ratio of training data and total test data is 100:80.
Because conducting probe is output as voltage fluctuation signal, it is impossible to be directly used in the structure of parameter measurement model.In order to
The quantitative description of voltage signal is obtained, multi-angle feature extraction has been carried out to voltage fluctuation signal:Based on statistical analysis technique, carry
Average, variance, coefficient skewness and coefficient of kurtosis are taken;Based on symbolic analysis, the length of sequence, local water holding are extracted
Rate, average and variance;Based on wavelet packet decomposition algorithm, the energy and comentropy of each band signal are extracted;Using chaotic time
Sequence analysis method, is extracted correlation dimension, kolmogorov entropys and lyapunov indexes.Each characteristic parameter be it is effective,
And change with flow and moisture content in regular.But, there is obvious correlation, Individual features are carried between Partial Feature parameter
Take result while certain regularity is shown, also show the presence of measurement noise.
(2) characteristic vector is compressed using principal component analytical method linear between dimensionality reduction, each characteristic parameter of elimination
Correlation;Principal component Y1, Y2..., YpIt is expressed as primitive character parameter X1, X2..., XpLinear combination, be designated as following algebraically shape
Formula:
In formula, Yi=ui' X is i-th of principal component of primitive character parameter, ui=(ui1, ui2..., uip) ' for coefficient to
Amount, X=(X1, X2..., Xp) ' be primitive character parameter vector, XiFor i-th of primitive character parameter, the linear combination is limited to
Lower column constraint principle:
a.ui′ui=1;
B. as i ≠ j, YiWith YjIt is mutually orthogonal;
c.Y1It is X1, X2..., XpAll linear combinations in variance the maximum, Y2It is in Y1Variance is maximum premise
Under, it is X1, X2..., XpAll linear combinations in variance the maximum, the rest may be inferred, YpIt is in Yp-1Before variance is maximum
Put, be X1, X2..., XpAll linear combinations in variance the maximum;
Before classification of flow patterns device is set up, it is necessary first to which sample data is carried out feature extraction to construct characteristic vector.So
And, there is a certain degree of correlation between unscreened characteristic parameter, the structure for being directly used in sorter model will be serious
Influence its nicety of grading and Generalization Capability.For this reason, it may be necessary to be joined using a few mutually orthogonal principal component instead of these features
Number is modeled.Principal component analysis is carried out to characteristic vector, before several principal components just embody very well and contain primitive character
The most information of parameter, in order to be able to retain the information of former characteristic vector, selection preceding 4 as much as possible while compressed data
Individual principal component replaces characteristic vector.
(3) a variety of flow patterns in peupendicular hole are directed to, multiclass classification of flow patterns device is set up using support vector machine classification method;Hang down
Flow pattern in straight well includes:Water-In-Oil flow pattern, oil-in-water flow pattern, transition flow pattern, scattered oil vacuole stream and scattered bubble stream;It can use
Many classification policys have:One-to-one, one-to-many, directed acyclic graph and directly one polytypic object function of construction;Utilize instruction
Practice sample to be trained the mathematical modeling of SVMs, the training of supporting vector machine model is one with Prescribed Properties
Optimization problem, can be described as following form:
In formula, w is weight vectors, and b is biasing, and C is penalty factor,For mapping function, ξiFor slack variable, yiFor
The output of model;Solving the optimization problem needs to be translated into following dual form:
In formula, K (xi, xj) it is the kernel function for meeting Mercer conditions, xiFor the input of model, αiFor Lagrange multiplier,
Finally the mathematical modeling of determination SVMs is:
The decision function of classification of flow patterns device is:
Classification of flow patterns device is set up using one-to-one many classification policys, one-to-one strategy is directed to any two in n classification problems
Class sample sets up sub-classifier respectively, and n (n -1)/2 sub-classifier need to be set up altogether.Final classification results are total to by sub-classifier
Determined with " ballot ", as shown in Figure 1.The characteristics of One-against-one be the number of grader as classification number is sharply increased, training
Efficiency is low.It is by the flow pattern rough segmentation in peupendicular hole:Oil-in-water, transition flow and oil-in-water flow pattern, then be total to using One-against-one
3 sub-classifiers, respectively oil-in-water-transition flow sub-classifier, transition flow-Water-In-Oil sub-classifier and water bag need to be set up
Oil-Water-In-Oil sub-classifier.
(4) kernel function of SVMs is selected, penalty factor is carried out with kernel functional parameter σ using test sample excellent
Change;Selectable kernel function has:Linear kernel function, multiphase formula kernel function, Radial basis kernel function and neutral net kernel function;Adopted
Optimization method is genetic algorithm, and the Optimization Steps of genetic algorithm are as follows:
A., initial population scale, maximum evolutionary generation T, penalty factor, kernel functional parameter σ hunting zone, intersection are set
And mutation probability;
B. chromosome coding is carried out to parameter C, σ, randomly generates initial population, initialization evolutionary generation t=0;
C. fitness R individual in population is calculatedcv(C, σ);
D. according to individual adaptation degree, individual is selected into the next generation from current population by the way of roulette;
E. crossover operation is carried out to the individual in population, produces new individual and enter of future generation;
F. some genes in mutation operation, random change individual are performed to the individual in population and produce new individual;
If being unsatisfactory for end condition and t≤T, step b is jumped to.
Parameter optimization is carried out to each sub-classifier using the classification accuracy under ten folding cross validations as object function, such as
Really other condition all sames, and different pretreatment modes simply is used to voltage signal, different classification of flow patterns will be obtained
Device.Various classification of flow patterns devices are tested, overall classification accuracy reaches more than 90%, illustrate that characteristic parameter is sensitive to
Variations in flow patterns, and the meteor trail echoes method based on SVMs is feasible.When setting up flow pattern using whole characteristic parameters
During grader, the Generalization Capability of grader is slightly worse;And when building classification of flow patterns device using the principal component of characteristic parameter, grader
Performance it is again relevant with the quantity of principal component, but difference is not obvious on the whole.When the negligible amounts of principal component, so that it may
To obtain higher nicety of grading;When setting up model using preceding 4 principal components, classification accuracy is to reach maximum 95.00%.
In summary, a kind of peupendicular hole meteor trail echoes method based on conducting probe that this patent is proposed can effectively solve peupendicular hole oil
The meteor trail echoes problem of water two phase flow.
Above to the description of the present invention and embodiments thereof, it is not limited to which this, is only the reality of the present invention shown in accompanying drawing
Apply one of mode.Without departing from the spirit of the invention, it is similar with the technical scheme without designing with creating
Structure or embodiment, belong to the scope of the present invention.
Claims (1)
1. a kind of peupendicular hole meteor trail echoes method based on conducting probe, it is characterised in that the foundation of classification of flow patterns device in this method
Comprise the following steps:
(1) training sample and test sample are obtained by dynamic experimental device, the output signal to conducting probe uses a variety of sides
Method carries out feature extraction, utilizes the multiple scalar characterization construction features vector extracted;The output signal of conducting probe is single channel
Time series signal, feature extracting method includes:It is statistical analysis, Symbolic time series analysis, Chaotic Time Series Analysis, small
Wave conversion, Hilbert-Huang conversion, fractal theory and density-wave theory;
(2) dimensionality reduction is compressed to characteristic vector using principal component analytical method, eliminates the linear correlation between each characteristic parameter
Property;Principal component Y1, Y2..., YpIt is expressed as primitive character parameter X1, X2..., XpLinear combination, be designated as following quantic:
In formula, Yi=ui' X is i-th of principal component of primitive character parameter, ui=(ui1, ui2..., uip) ' be coefficient vector, X=
(X1, X2..., Xp) ' be primitive character parameter vector, XiFor i-th of primitive character parameter;Wherein, vectorial ui' it is vector uiTurn
Put, the linear combination is limited to following tri- constraint principles of a, b, c:
a.ui′ui=1;
B. as i ≠ j, YiWith YjIt is mutually orthogonal;
c.Y1It is X1, X2..., XpAll linear combinations in variance the maximum, Y2For except Y1X in addition1, X2..., XpAll
Variance the maximum in linear combination, the rest may be inferred, YpFor except Y1, Y2..., Yp-1X in addition1, X2..., XpAll linear combinations
In variance the maximum;
(3) a variety of flow patterns in peupendicular hole are directed to, multiclass classification of flow patterns device is set up using support vector machine classification method;Peupendicular hole
Interior flow pattern includes:Water-In-Oil flow pattern, oil-in-water flow pattern, transition flow pattern, scattered oil vacuole stream and scattered bubble stream;Many classification policys
Including:One-to-one, one-to-many, directed acyclic graph and directly one polytypic object function of construction;Using training sample to branch
The mathematical modeling for holding vector machine is trained, and the training of supporting vector machine model is an optimization problem with Prescribed Properties,
It is described as following form:
<mfenced open = "" close = "">
<mtable>
<mtr>
<mtd>
<munder>
<mrow>
<mi>m</mi>
<mi>i</mi>
<mi>n</mi>
</mrow>
<mrow>
<mi>w</mi>
<mo>,</mo>
<mi>b</mi>
<mo>,</mo>
<mi>&xi;</mi>
</mrow>
</munder>
</mtd>
<mtd>
<mrow>
<mfrac>
<mn>1</mn>
<mn>2</mn>
</mfrac>
<mo>|</mo>
<mo>|</mo>
<mi>w</mi>
<mo>|</mo>
<msup>
<mo>|</mo>
<mn>2</mn>
</msup>
<mo>+</mo>
<mi>C</mi>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>l</mi>
</munderover>
<msub>
<mi>&xi;</mi>
<mi>i</mi>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
ξi>=0 i=1,2 ..., l
In formula, w is weight vectors, and b is biasing, and C is penalty factor,For mapping function, ξiFor slack variable, yiFor model
Output;Solving the optimization problem needs to be translated into following dual form:
<mfenced open = "" close = "">
<mtable>
<mtr>
<mtd>
<munder>
<mrow>
<mi>m</mi>
<mi>a</mi>
<mi>x</mi>
</mrow>
<mi>&alpha;</mi>
</munder>
</mtd>
<mtd>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>l</mi>
</munderover>
<msub>
<mi>&alpha;</mi>
<mi>i</mi>
</msub>
<mo>-</mo>
<mfrac>
<mn>1</mn>
<mn>2</mn>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>l</mi>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>l</mi>
</munderover>
<msub>
<mi>&alpha;</mi>
<mi>i</mi>
</msub>
<msub>
<mi>&alpha;</mi>
<mi>j</mi>
</msub>
<msub>
<mi>y</mi>
<mi>i</mi>
</msub>
<msub>
<mi>y</mi>
<mi>j</mi>
</msub>
<mi>K</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>&CenterDot;</mo>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mfenced open = "" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mi>s</mi>
<mo>.</mo>
<mi>t</mi>
<mo>.</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>l</mi>
</munderover>
<msub>
<mi>&alpha;</mi>
<mi>i</mi>
</msub>
<msub>
<mi>y</mi>
<mi>i</mi>
</msub>
<mo>=</mo>
<mn>0</mn>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
0≤αi≤C
In formula, K (xi, xj) it is the kernel function for meeting Mercer conditions, xiFor the input of model, αiFor Lagrange multiplier, finally
The mathematical modeling for determining SVMs is:
The decision function of classification of flow patterns device is:
<mrow>
<mi>y</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mi>sgn</mi>
<mo>&lsqb;</mo>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>&rsqb;</mo>
<mo>=</mo>
<mi>sgn</mi>
<mo>&lsqb;</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>l</mi>
</munderover>
<msub>
<mi>&alpha;</mi>
<mi>i</mi>
</msub>
<msub>
<mi>y</mi>
<mi>i</mi>
</msub>
<mi>K</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>&CenterDot;</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>b</mi>
<mo>&rsqb;</mo>
</mrow>
(4) kernel function of SVMs is selected, penalty factor is optimized with kernel functional parameter σ using test sample;Branch
Hold one kind in the following kernel function of Selection of kernel function of vector machine:Linear kernel function, multiphase formula kernel function, Radial basis kernel function and
Neutral net kernel function;The optimization method used is genetic algorithm, and the Optimization Steps of genetic algorithm are as follows:
A., initial population scale, maximum evolutionary generation T, penalty factor, kernel functional parameter σ hunting zone, intersection and change are set
Different probability;
B. chromosome coding is carried out to parameter C, σ, randomly generates initial population, initialization evolutionary generation t=0;
C. fitness R individual in population is calculatedcv(C, σ);
D. according to individual adaptation degree, individual is selected into the next generation from current population by the way of roulette;
E. crossover operation is carried out to the individual in population, produces new individual and enter of future generation;
F. some genes in mutation operation, random change individual are performed to the individual in population and produce new individual;
If being unsatisfactory for end condition and t≤T, step b is jumped to, if meeting end condition, selection fitness is most
Big individual is used as optimum results.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410216750.2A CN103967478B (en) | 2014-05-21 | 2014-05-21 | A kind of peupendicular hole meteor trail echoes method based on conducting probe |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410216750.2A CN103967478B (en) | 2014-05-21 | 2014-05-21 | A kind of peupendicular hole meteor trail echoes method based on conducting probe |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103967478A CN103967478A (en) | 2014-08-06 |
CN103967478B true CN103967478B (en) | 2017-10-27 |
Family
ID=51237436
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410216750.2A Active CN103967478B (en) | 2014-05-21 | 2014-05-21 | A kind of peupendicular hole meteor trail echoes method based on conducting probe |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103967478B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107300577A (en) * | 2016-12-07 | 2017-10-27 | 江苏国泉泵业制造有限公司 | A kind of multiphase flow pattern detection method based on soft-measuring technique |
Families Citing this family (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105697002A (en) * | 2014-11-24 | 2016-06-22 | 中国石油化工股份有限公司 | Method for recognizing coal measure strata lithology |
CN105064993B (en) * | 2015-08-06 | 2018-01-09 | 北京航空航天大学 | A kind of peupendicular hole measurement of water ratio method based on the fusion of conducting probe array information |
CN105715253A (en) * | 2016-01-30 | 2016-06-29 | 上海大学 | Prediction method for flowing bottomhole pressure of gas well |
CN107834551A (en) * | 2017-11-20 | 2018-03-23 | 国网湖南省电力有限公司 | A kind of power distribution network low-voltage Forecasting Methodology based on SVMs |
WO2019216891A1 (en) * | 2018-05-09 | 2019-11-14 | Landmark Graphics Corporation | Learning based bayesian optimization for optimizing controllable drilling parameters |
CN109472285B (en) * | 2018-09-29 | 2020-12-22 | 北京中油瑞飞信息技术有限责任公司 | Lost circulation identification method and device and computer equipment |
CN109359692A (en) * | 2018-10-23 | 2019-02-19 | 中国石油大学(北京) | A kind of crude oil water content on-line measurement model and its construction method, application |
CN113054881B (en) * | 2021-02-22 | 2022-10-28 | 江苏大学 | Method for constructing outer rotor stator coreless bearingless permanent magnet synchronous motor controller |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1410774A (en) * | 2002-11-29 | 2003-04-16 | 清华大学 | Two phase fluid phase concentration measuring method based on main component analysis and neuron network |
US6577700B1 (en) * | 2001-06-22 | 2003-06-10 | Liang-Shih Fan | Neural network based multi-criteria optimization image reconstruction technique for imaging two- and three-phase flow systems using electrical capacitance tomography |
CN1904581A (en) * | 2006-08-01 | 2007-01-31 | 东北电力大学 | Oil gas water multiphase flow type identification method based on main component analysis and supporting vector machine |
CN101608994A (en) * | 2009-07-17 | 2009-12-23 | 清华大学 | Dense medium suspension liquid coal slurry content determination with principal component analysis (PCA) and support vector machine |
CN102567742A (en) * | 2010-12-15 | 2012-07-11 | 中国科学院电子学研究所 | Automatic classification method of support vector machine based on selection of self-adapting kernel function |
CN102590030A (en) * | 2012-01-18 | 2012-07-18 | 浙江大学 | Small-passage gas-liquid phase flow pattern identification device and method based on photovoltaic array sensor |
-
2014
- 2014-05-21 CN CN201410216750.2A patent/CN103967478B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6577700B1 (en) * | 2001-06-22 | 2003-06-10 | Liang-Shih Fan | Neural network based multi-criteria optimization image reconstruction technique for imaging two- and three-phase flow systems using electrical capacitance tomography |
CN1410774A (en) * | 2002-11-29 | 2003-04-16 | 清华大学 | Two phase fluid phase concentration measuring method based on main component analysis and neuron network |
CN1904581A (en) * | 2006-08-01 | 2007-01-31 | 东北电力大学 | Oil gas water multiphase flow type identification method based on main component analysis and supporting vector machine |
CN101608994A (en) * | 2009-07-17 | 2009-12-23 | 清华大学 | Dense medium suspension liquid coal slurry content determination with principal component analysis (PCA) and support vector machine |
CN102567742A (en) * | 2010-12-15 | 2012-07-11 | 中国科学院电子学研究所 | Automatic classification method of support vector machine based on selection of self-adapting kernel function |
CN102590030A (en) * | 2012-01-18 | 2012-07-18 | 浙江大学 | Small-passage gas-liquid phase flow pattern identification device and method based on photovoltaic array sensor |
Non-Patent Citations (4)
Title |
---|
基于LS-SVM的气液两相流参数测量研究;彭珍瑞;《中国博士学位论文全文数据库 信息科技辑》;20080215(第2期);第10-14、28-42、60-72页 * |
基于图像小波包信息熵和遗传神经网络的气-液两相流流型识别;周云龙 等;《核动力工程》;20080215;第29卷(第1期);第115-119页 * |
水平井油水两相流相含率测井资料解释方法研究;贾晓辉;《中国优秀硕士学位论文全文数据库 基础科学辑》;20111215(第S2期);全文 * |
油气水三相流流型识别方法研究;谭平;《中国优秀硕士学位论文全文数据库 基础科学辑》;20111015(第10期);第1-13、40-54页 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107300577A (en) * | 2016-12-07 | 2017-10-27 | 江苏国泉泵业制造有限公司 | A kind of multiphase flow pattern detection method based on soft-measuring technique |
Also Published As
Publication number | Publication date |
---|---|
CN103967478A (en) | 2014-08-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103967478B (en) | A kind of peupendicular hole meteor trail echoes method based on conducting probe | |
Ahmadi et al. | Developing a robust proxy model of CO2 injection: Coupling Box–Behnken design and a connectionist method | |
CN105760673A (en) | Fluvial facies reservoir earthquake sensitive parameter template analysis method | |
CN103617147A (en) | Method for identifying mine water-inrush source | |
CN105715253A (en) | Prediction method for flowing bottomhole pressure of gas well | |
Ikpeka et al. | Application of machine learning models in predicting initial gas production rate from tight gas reservoirs | |
Zhang et al. | SVM model for estimating the parameters of the probability-integral method of predicting mining subsidence | |
Kamari et al. | Modeling the permeability of heterogeneous oil reservoirs using a robust method | |
CN112302594A (en) | Method, device and equipment for determining connection structure of heterogeneous water-drive reservoir | |
CN104834934B (en) | A kind of nucleome capture method for being used to identify reservoir fluid | |
Wang et al. | Combined application of unsupervised and deep learning in absolute open flow potential prediction: a case study of the Weiyuan Shale gas reservoir | |
Huang et al. | Quantitative analysis of the main controlling factors of oil saturation variation | |
CN105003249A (en) | Horizontal well flow pattern recognition method based on total flow and conductance probe array signals | |
CN114036774A (en) | Determination method, calculation method and determination system for quasi-relative permeability of semi-filled fracture-cavity reservoir | |
CN115438823A (en) | Borehole wall instability mechanism analysis and prediction method and system | |
Luo et al. | Optimization Method of Hydraulic Fracturing Design for Horizontal Shale Gas Well Based on Artificial Neural Network and Genetic Algorithm | |
Tom et al. | Volumetric approach based DOE and ANN models for estimating reservoirs oil in place | |
Zarepakzad et al. | A comparative analysis and rapid performance prediction of polymer flooding process by coupling reservoir simulation with neural networks | |
Liu et al. | Application of Water Injection Profile Recognition Based on Machine Learning Method in F Oilfield | |
Roustazadeh et al. | Estimating oil and gas recovery factors via machine learning: Database-dependent accuracy and reliability | |
Shawkat et al. | Research Article Fluid Flow Behavior Prediction in Naturally Fractured Reservoirs Using Machine Learning Models | |
Elgibaly et al. | Optimisation of gas lift performance using artificial neural network | |
Zhao et al. | The fracture density and fractal dimension prediction based on support vector machine | |
Liu et al. | Infill Well Location Optimization Method based on Remaining Oil Recoverable Potential Evaluation | |
Songyi et al. | Uncertainty assessment on the performance of horizontal wells prediction in bottom-water reservoir based on analytic hierarchy process |
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 |