CN114575802A - High water cut oil reservoir oil well yield prediction method based on machine learning - Google Patents
High water cut oil reservoir oil well yield prediction method based on machine learning Download PDFInfo
- Publication number
- CN114575802A CN114575802A CN202011385850.XA CN202011385850A CN114575802A CN 114575802 A CN114575802 A CN 114575802A CN 202011385850 A CN202011385850 A CN 202011385850A CN 114575802 A CN114575802 A CN 114575802A
- Authority
- CN
- China
- Prior art keywords
- well
- oil
- production
- water injection
- yield
- 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
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 64
- 238000000034 method Methods 0.000 title claims abstract description 48
- 239000003129 oil well Substances 0.000 title claims abstract description 46
- 238000010801 machine learning Methods 0.000 title claims abstract description 19
- 238000004519 manufacturing process Methods 0.000 claims abstract description 81
- 238000002347 injection Methods 0.000 claims abstract description 73
- 239000007924 injection Substances 0.000 claims abstract description 73
- 238000011161 development Methods 0.000 claims abstract description 12
- 239000007788 liquid Substances 0.000 claims abstract description 8
- 230000006870 function Effects 0.000 claims description 14
- 239000011159 matrix material Substances 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 7
- 238000012360 testing method Methods 0.000 claims description 7
- 238000004422 calculation algorithm Methods 0.000 abstract description 13
- 230000008859 change Effects 0.000 abstract description 4
- 238000004364 calculation method Methods 0.000 abstract description 3
- 230000008569 process Effects 0.000 abstract description 3
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000007689 inspection Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000012843 least square support vector machine Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 239000011435 rock Substances 0.000 description 2
- 239000000243 solution Substances 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 230000000704 physical effect Effects 0.000 description 1
- 238000011158 quantitative evaluation Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000011425 standardization method Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 235000020681 well water Nutrition 0.000 description 1
- 239000002349 well water Substances 0.000 description 1
- 230000002087 whitening effect Effects 0.000 description 1
Images
Classifications
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21B—EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- E21B43/00—Methods or apparatus for obtaining oil, gas, water, soluble or meltable materials or a slurry of minerals from wells
- E21B43/16—Enhanced recovery methods for obtaining hydrocarbons
- E21B43/20—Displacing by water
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21B—EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- E21B47/00—Survey of boreholes or wells
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
- Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Mining & Mineral Resources (AREA)
- Geology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Environmental & Geological Engineering (AREA)
- Mathematical Optimization (AREA)
- Computational Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Geochemistry & Mineralogy (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Fluid Mechanics (AREA)
- Computing Systems (AREA)
- Geophysics (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Algebra (AREA)
Abstract
The invention relates to the technical field of oilfield development, in particular to a method for predicting the yield of an oil reservoir and an oil well in a high water cut period based on machine learning. According to the method, the oil yield of the oil well and the injection amount of the water injection well are jointly used as influence factors of a yield prediction process through a vector autoregressive algorithm, a time sequence model is constructed by using liquid amount change curves of the water injection well and the oil production well, the time step length of a fitting curve is determined through selection of a lag order, the oil well yield under different time step lengths in the future is calculated through iteration, an oil well yield prediction method based on a machine learning algorithm is formed, and a new thought is provided for oil well yield prediction. The method has the advantages of simple model, few considered variables, easy calculation and operation, high accuracy and contribution to popularization and application.
Description
Technical Field
The invention relates to the technical field of oilfield development, in particular to a method for predicting the yield of an oil reservoir and an oil well in a high water cut period based on machine learning.
Background
As the oil field enters a high water content development stage, a large amount of static oil reservoir data and development dynamic data provide an important data base for predicting the oil reservoir development effect through a machine learning algorithm. Currently, machine learning models are applied in different fields of oil exploration and development, such as decision tree methods, support vector machine methods, neural network methods, and the like. The machine learning algorithm determines the relationship between input and output parameters to obtain the result of the test data set. Factors influencing the accuracy of the prediction model mainly include the size and quality of a data set, the selection of data characteristics, the establishment of the model, the fusion of an algorithm and the like.
The Anifovose utilizes an integration algorithm in machine learning to research the classification characteristics of the reservoir without oil exploitation and proposes the follow-up application in the field in urgent need. Martins performs petrophysical classification on the carbonate-siliciclastic rock through well logging information by using a machine learning method and provides an accurate reservoir heterogeneity identification result. According to the method, the Anemangely predicts the transverse wave velocity in the rock physical logging through a machine learning algorithm, selects different logging parameters as the optimal characteristics for estimating the transverse wave velocity by adopting a Least Square Support Vector Machine (LSSVM) and Particle Swarm Optimization (PSO), compares a regression model with an empirical model, and shows that the error of the model is obviously smaller than that of the empirical model. Ao predicts the logging curve by using a random forest algorithm and researches the application of the logging curve in logging regression modeling.
In the aspect of carrying out oil well yield prediction research by using a machine learning method, Jinbao strong predicts the medium and short-term oil well yield by using a support vector machine method; the GouJianwei proposes an ARIMA-Kalman filter-based oil well yield prediction model, and then proposes a deep learning model utilizing a long-short term memory network to predict time sequences. However, the previously studied yield prediction model only considers the yield variation curve of an oil well as an input condition, has certain limitations, and cannot quantitatively evaluate the development effect of each oil well in a block.
The Chinese invention patent CN104951842B discloses a new oil field yield prediction method, which comprises the following steps: step 1, preliminarily screening and determining factors influencing yield; step 2, processing the basic data; step 3, carrying out independent variable whitening, and establishing a time sequence model of the independent variable; step 4, determining the decision factors influencing the yield; and step 5, establishing a yield prediction model by using the time series model, and checking the reasonability of the model. The independent variable time series model established by the method is as follows: 1) comprehensive water content: ARIMA (0,1, 0) (2, 1, 0); 2) controlling reserves by a single well: ARIMA (1, 1, 0) (1, 0, 0); 3) the extraction degree is as follows: ARIMA (0, 2, 1); 4) oil extraction speed: ARIMA (1, 1, 0); 5) the number of the new wells put into production: ARIMA (0,1,1) (1, 1, 0); 6) annual water injection amount: ARIMA (0,1, 0).
Chinese patent application CN104732303A discloses an oil field yield prediction method based on a dynamic radial basis function neural network, comprising the following steps: A. determining factors influencing yield according to the oil field condition, acquiring historical data and dividing the historical data into a training data set and a detection data set; B. normalizing the data set by using a dispersion standardization method; C. dynamically adjusting the RBF neural network structure by using a sensitivity method, and establishing a temporary RBF neural network prediction model; D. correcting the model error by using the state transition probability matrix to obtain a stable RBF neural network oil field yield prediction model; E. b, verifying the model by using the detection data set obtained in the step A, and judging whether the model meets the expectation; F. and E, predicting the oil yield by using the yield prediction model which is obtained in the step E and meets the expectation. The method avoids the problem that RBF hidden layer neurons are too much or too small, and the obtained model has a self-adaptive adjustment function; and the prediction error is secondarily corrected, so that the prediction result is more accurate and reasonable.
The chinese patent application CN110400006A discloses an oil well yield prediction method based on a deep learning algorithm, which comprises the following steps: step 1, acquiring data and performing quality inspection; step 2, processing and dividing data; step 3, establishing a learning model; step 4, training and verifying the model built in the step 3; and 5, predicting the oil well yield. According to the oil well yield prediction method based on the deep learning algorithm, the relation between factors such as reservoir physical properties, a working system and a development stage and oil production and liquid production is established through training, the advantages of a data driving algorithm are exerted, and a multi-factor oil well yield prediction model is established.
The method has numerous considered factors, large actual calculation amount and relatively complex steps. The yield change influence factors of the oil deposit wells in the high water cut period are complex, wherein the main factor is the influence of the injection amount of the water injection well on the yield of the oil wells. Therefore, it is necessary to provide a prediction method which has a simple step, is easy to operate, and has high accuracy, with the main influencing factor as the influencing factor.
Disclosure of Invention
The invention mainly aims to provide an oil well yield prediction method of a high water-cut oil reservoir based on machine learning, which is characterized in that the oil yield of an oil well and the injection amount of a water injection well are jointly used as influence factors of a yield prediction process through a vector autoregressive algorithm, a time sequence model is constructed by using liquid amount change curves of the water injection well and an oil production well, the time step length of a fitting curve is determined through selection of a lag order, the oil well yield under different time step lengths in the future is calculated through iteration, the oil well yield prediction method based on the machine learning algorithm is formed, and a new thought is provided for oil well yield prediction.
In order to achieve the purpose, the invention adopts the following technical scheme:
the invention provides a machine learning-based high water-cut oil reservoir oil well yield prediction method, which comprises the following steps of:
judging a target oil reservoir, and selecting a proper block;
respectively detecting the data stationarity of a water injection curve of a water injection well and a production curve of a production well, and if the curves have the phenomenon of unstable fluctuation, carrying out differential processing on the data until the curve data of two variables are stable;
determining the hysteresis order of the water injection well and the oil production well;
and establishing a vector autoregressive yield prediction model to predict the oil well yield.
Further, the method for judging the target oil reservoir comprises the following steps: judging the correlation between the water injection well and the oil production well of the target oil deposit, selecting the injection well with high correlation, drawing the production curve of the oil production well and the water injection curve of the water injection well, judging whether the conditions of simultaneous increase and simultaneous decrease exist between the water injection rate of the water injection well and the output of the oil production well according to the two curves, and if so, judging that the oil deposit is a proper oil deposit.
Further, a co-correlation matrix of the water injection wells and the oil production wells is drawn, wherein the brighter the color block is, the higher the correlation of the production curve between the water injection wells and the oil production wells is, and the darker the color block is, the lower the correlation is.
Further, the stability of the data is calculated by adopting an ADF (automatic document delivery) inspection method, and the ADF inspection judges whether the sequence has a unit root or not, and if so, the sequence is not stable.
Further, determining the hysteresis order of the water injection well and the oil production well through the AIC function and the BIC function:
AIC＝2k-2ln(L)
BIC＝kln(n)-2ln(L)
in the formula, k is the number of model parameters, n is the number of samples, and L is a likelihood function.
Furthermore, the values of the AIC and BIC functions are calculated between every two water injection wells and oil production wells, and the optimal hysteresis order of the whole block is obtained through iteration.
Furthermore, the hysteresis order corresponding to the minimum value of the AIC and the BIC in the model is considered preferentially.
Further, the vector autoregressive yield prediction model is:
Yt＝εt+B1Xt+B2Xt-1+…+BqEt-q+ut+A1Yt-1+…+ApYt-p+u
in the formula, Yt＝(y1t......ykt) Representing a K x 1 order random vector; a. the1To ApA parameter matrix of K multiplied by K order; xtRepresenting an M × 1 order extrinsic variable vector; b is1To BqA coefficient matrix to be estimated of K multiplied by M order; u is a white noise sequence; epsilontIs the model residual error; p and q are the hysteresis order of the two variables, respectively.
Further, for a particular block of the reservoir, its YtAnd XtThe construction vector of (a) can be expressed as:
Yt＝[y1，t，y2，t，...，yi，t]T
Xt＝[x1，t，x2，t，...，xi，t]T
in the formula, yi，tThe liquid production amount of the ith oil production well at the tth moment; x is the number ofi，tThe injection quantity of the ith water injection well at the t-th moment is shown.
And further, predicting the oil well yield of the t time point by taking the development curves of the injection well and the oil production well in the lag order as the basis, and iteratively calculating the oil well yield of the t +1 time point by taking the oil well yield of the t time point as an input parameter, thereby predicting the oil well yield in a future time step.
Compared with the prior art, the invention has the following beneficial effects:
the method comprises the steps of using the oil yield of the oil well and the injection amount of the water injection well as the influence factor of the yield prediction process to construct a time sequence model by using liquid amount change curves of the water injection well and the oil production well, determining the time step length of a fitting curve through selection of a lag order, and calculating the oil well yield under different time step lengths in the future through iteration. The method realizes quantitative evaluation on the development effect of each oil well, and has high accuracy of prediction results, and the relative error is 2.85%.
The method has the advantages of simple model, few considered variables, easy calculation and operation, high accuracy and contribution to popularization and application.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, are included to provide a further understanding of the invention, and are incorporated in and constitute a part of this specification, illustrate exemplary embodiments of the invention and together with the description serve to explain the invention and not to limit the invention.
FIG. 1 is a graph of the heat of a co-correlation matrix of production curves for a water injection well and a production well in an actual block according to an embodiment of the present invention;
FIG. 2 is a graph of production from a typical water injection and production well in accordance with an embodiment of the present invention;
FIG. 3 is a graph comparing the production curves of oil-water wells before and after treatment according to an embodiment of the present invention.
Detailed Description
It is to be understood that the following detailed description is exemplary and is intended to provide further explanation of the invention as claimed. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the invention. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of the stated features, steps, operations, and/or combinations thereof, unless the context clearly indicates otherwise.
In order to make the technical solutions of the present invention more clearly understood by those skilled in the art, the technical solutions of the present invention will be described in detail below with reference to specific embodiments.
Example 1 method for predicting oil well yield of high water-cut reservoir based on machine learning
The method comprises the following steps:
step 1, judging a target oil reservoir, and selecting a proper block:
the method for judging the target oil reservoir comprises the following steps: judging the correlation between the water injection well and the oil production well of the target oil deposit, selecting the injection well with high correlation, drawing the production curve of the oil production well and the water injection curve of the water injection well, judging whether the conditions of simultaneous increase and simultaneous decrease exist between the water injection rate of the water injection well and the output of the oil production well according to the two curves, and if so, judging that the oil deposit is a proper oil deposit.
And 2, respectively detecting the data stability of the water injection curve of the water injection well and the production curve of the oil production well, and if the fluctuation of the curves is unstable, carrying out differential processing on the data until the curve data of the two variables are stable.
Step 3, determining the hysteresis orders of the water injection well and the oil production well;
selecting an Akaichi Information Criterion (AIC) and a Bayesian Information Criterion (BIC) to determine a lag period when the statistics of the Akaichi information criterion and the Bayesian information criterion are minimum, wherein the expressions are respectively as follows:
AIC＝2k-2ln(L) (1)
BIC＝kln(n)-2ln(L) (2)
in the formula, k is the number of model parameters, n is the number of samples, and L is a likelihood function.
And 4, establishing a vector autoregressive yield prediction model to predict the oil well yield.
The vector autoregressive yield prediction model is as follows:
Yt＝εt+B1Xt+B2Xt-1+…+BqEt-q+ut+A1Yt-1+…+ApYt-p+u(3)
in the formula, Yt＝(y1t......ykt) Representing a K x 1 order random vector; a. the1To ApA parameter matrix of K multiplied by K order; xtRepresenting an M × 1 order extrinsic variable vector; b is1To BqA coefficient matrix to be estimated of K multiplied by M order; u is a white noise sequence; epsilontIs the model residual error; p and q are the hysteresis order of the two variables, respectively.
For a particular block of the reservoir, YtAnd XtThe constructed vector of (a) can be expressed as:
Yt＝[y1，t，y2，t，...，yi，t]T (4)
Xt＝[x1，t，x2，t，...，xi，t]T (5)
in the formula, yi，tThe liquid production amount of the ith oil production well at the tth moment; x is the number ofi，tThe injection quantity of the ith water injection well at the t-th moment is shown.
And predicting the oil well yield at the t-th time point by taking the development curves of the water injection well and the oil production well in the lag order as the basis, and iteratively calculating the oil well yield at the t +1 time point by taking the oil well yield at the t-th time point as an input parameter, thereby predicting the oil well yield at a future time step.
Embodiment 2 high water cut oil reservoir oil well yield prediction method based on machine learning
Taking a certain practical block of China as an example, the method for predicting the oil well yield of the oil reservoir in the high water cut period based on machine learning specifically comprises the following steps:
and (4) drawing a time series curve for the production data of the block oil-water well, wherein the production date of the block oil-water well ranges from 3 months in 1969 to 12 months in 2018. And drawing a co-correlation matrix of the water injection wells and the oil production wells of the block, wherein the brighter the color block is, the higher the correlation of the production curves among the water injection wells and the oil production wells is, and the darker the color block is, the lower the correlation is. It can be seen that better correlation exists among the injection and production wells of the block part, and a West 46-6-1 oil production well production curve and a West 43-8-1 water injection well water injection curve with better correlation are drawn. Fig. 2 shows that under the condition that the production curve of the injection and production wells with high liquid quantity correlation has 'simultaneous increase and simultaneous decrease', the increase of the water injection quantity between the injection and production wells supplements the energy of the stratum, and simultaneously, the effect of oil displacement is achieved, so that the yield of the oil well is increased. Therefore, a vector autoregressive model can be established for the whole injection and production well network system, the production curve dependency relationship among injection and production wells is captured, the oil well yield is predicted, and the oil well production effect is evaluated.
The stability of the data was calculated before model building using ADF inspection. ADF test judges whether the sequence has unit root, if it exists, the sequence is not stable. It first assumes that there is a unit root, and if the resulting significance test statistic is less than three confidences (10%, 5%, 1%), then the correspondence (90%, 95, 99%) is held down to reject the original hypothesis. ADF tests were performed using West 46-6-1 wells in the study block as an example. The results obtained by testing the raw production curve are shown in table 1. Before data differentiation is performed, the block ADF value is-0.243, ADF values well above the 1% confidence level, P value is 0.899, and is not close to 0, so the well production data is a non-stationary sequence.
TABLE 1 ADF test results for original time series production curves
The production data was subjected to moving average processing at 10 time steps, and subjected to 1-order difference, and a contrast curve was plotted, as shown in fig. 3. The ADF value of the curve is-11.883 after the difference is carried out, and is far less than the ADF value of the 1% confidence level, and the time sequence obtained after the difference processing is a stable sequence, so that the time sequence prediction can be carried out.
The lag order of the water injection well and the oil production well is determined through the AIC function and the BIC function, and the lag order is determined through the mutual influence between different endogenous variables and exogenous variables of the vector autoregressive model, so that the values of the AIC function and the BIC function are calculated between every two water injection wells and oil production wells, and the optimal lag order of the whole block is obtained through iteration.
And (3) statistically analyzing the time sequence of the injection and production wells after the difference, and respectively selecting 1 to 10 lag orders of the water injection well and the oil production well to obtain the values of the AIC function and the BIC function under different orders, which is shown in the table 2. And (3) preferably considering the hysteresis order corresponding to the minimum value of AIC and BIC in the model to obtain the optimum injection and production well hysteresis order on the actual block time sequence to be 3.
TABLE 2 statistical table of different optimal hysteresis numbers of blocks
And (3) predicting the yield of the oil wells in the block from 2018, 2 months to 2018, 12 months by using the constructed vector autoregressive model, comparing the yield with the actual oil yield, and obtaining a result shown in table 3.
TABLE 3 comparison of actual oil production in an oil field to predicted oil production using a vector autoregressive model
As can be seen from Table 3, the actual oil production in the time period is 456t, the predicted yield established by the vector autoregressive model is 453.3t, the relative error is 2.85%, and the accuracy of the predicted result is high.
The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.
Claims (10)
1. The method for predicting the oil well yield of the oil reservoir in the high water cut period based on machine learning is characterized by comprising the following steps of:
judging a target oil reservoir, and selecting a proper block;
respectively detecting the data stationarity of a water injection curve of a water injection well and a production curve of a production well, and if the curves have the phenomenon of unstable fluctuation, carrying out differential processing on the data until the curve data of two variables are stable;
determining the hysteresis order of the water injection well and the oil production well;
and establishing a vector autoregressive yield prediction model to predict the oil well yield.
2. The prediction method according to claim 1, wherein the method of determining the target reservoir is: judging the correlation between the water injection well and the oil production well of the target oil deposit, selecting the injection well with high correlation, drawing the production curve of the oil production well and the water injection curve of the water injection well, judging whether the conditions of simultaneous increase and simultaneous decrease exist between the water injection rate of the water injection well and the output of the oil production well according to the two curves, and if so, judging that the oil deposit is a proper oil deposit.
3. The prediction method of claim 2, wherein a co-correlation matrix is plotted for water injection wells and production wells, wherein the brighter the color block the higher the correlation of the production curve between injection wells, and the darker the color block the lower.
4. The prediction method of claim 1 wherein the stationarity of the data is calculated using an ADF test that determines if the sequence has a unit root and if so, the sequence is not stationary.
5. Prediction method according to claim 1, characterized in that the hysteresis order of the water injection and production wells is determined by the AIC and BIC functions:
AIC＝2k-2ln(L)
BIC＝kln(n)-2ln(L)
in the formula, k is the number of model parameters, n is the number of samples, and L is a likelihood function.
6. Prediction method according to claim 5, characterized in that the values of the AIC, BIC functions are calculated for each pair of injection and production wells, and the optimal hysteresis order for the whole block is obtained by iteration.
7. The prediction method of claim 6, wherein the hysteresis order corresponding to the minimum of AIC and BIC in the model is prioritized.
8. The prediction method of claim 1, wherein the vector autoregressive yield prediction model is:
Yt＝εt+B1Xt+B2Xt-1+…+BqEt-q+ut+A1Yt-1+…+ApYt-p+u
in the formula, Yt＝(y1t……ykt) Representing a K x 1 order random vector; a. the1To ApA parameter matrix of K x K order; xtRepresenting an M × 1 order extrinsic variable vector; b is1To BqA coefficient matrix to be estimated of K multiplied by M order; u is a white noise sequence; epsilontIs the model residual error; p and q are the hysteresis order of the two variables, respectively.
9. The prediction method of claim 8, wherein for a particular block of the reservoir, its YtAnd XtThe constructed vector of (a) can be expressed as:
Yt＝[y1,t,y2,t,…,yi,t]T
Xt＝[x1,t,x2,t,…,xi,t]T
in the formula, yi,tThe liquid production amount of the ith oil production well at the tth moment; x is the number ofi,tThe injection quantity of the ith water injection well at the t-th moment is shown.
10. The prediction method according to claim 8, wherein the well production at the t-th time point is predicted based on the development curves of the injection well and the production well in the lag order, and the well production at the t +1 time point is iteratively calculated using the well production at the t-th time point as an input parameter, thereby predicting the well production at a time step in the future.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011385850.XA CN114575802A (en) | 2020-12-01 | 2020-12-01 | High water cut oil reservoir oil well yield prediction method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011385850.XA CN114575802A (en) | 2020-12-01 | 2020-12-01 | High water cut oil reservoir oil well yield prediction method based on machine learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114575802A true CN114575802A (en) | 2022-06-03 |
Family
ID=81768403
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011385850.XA Pending CN114575802A (en) | 2020-12-01 | 2020-12-01 | High water cut oil reservoir oil well yield prediction method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114575802A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114737928A (en) * | 2022-06-13 | 2022-07-12 | 中煤科工集团西安研究院有限公司 | Nuclear learning-based coalbed methane intelligent drainage and mining method and system |
-
2020
- 2020-12-01 CN CN202011385850.XA patent/CN114575802A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114737928A (en) * | 2022-06-13 | 2022-07-12 | 中煤科工集团西安研究院有限公司 | Nuclear learning-based coalbed methane intelligent drainage and mining method and system |
| CN114737928B (en) * | 2022-06-13 | 2022-09-06 | 中煤科工集团西安研究院有限公司 | Nuclear learning-based coalbed methane intelligent drainage and mining method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111144542B (en) | Oil well productivity prediction method, device and equipment | |
| CN112529341B (en) | Drilling well leakage probability prediction method based on naive Bayesian algorithm | |
| CN111985610B (en) | Oil pumping well pump efficiency prediction system and method based on time sequence data | |
| CN112989708A (en) | Well logging lithology identification method and system based on LSTM neural network | |
| WO2009134685A2 (en) | System and method for interpretation of well data | |
| CN115061219B (en) | Fracture type reservoir prediction identification method and system based on petroleum and natural gas detection | |
| CN112364560B (en) | Intelligent prediction method for working hours of mine rock drilling equipment | |
| CN112765894B (en) | K-LSTM-based aluminum electrolysis cell state prediction method | |
| CN115049627B (en) | Steel surface defect detection method and system based on domain self-adaptive depth migration network | |
| CN114575802A (en) | High water cut oil reservoir oil well yield prediction method based on machine learning | |
| CN116756679A (en) | Multi-source information fusion-based method for judging geological mode of down-the-hole mountain | |
| CN116882292A (en) | Lost circulation overflow early warning method based on LightGBM and anomaly detection algorithm | |
| CN115618987A (en) | Production well production data prediction method, device, equipment and storage medium | |
| CN113177578A (en) | Agricultural product quality classification method based on LSTM | |
| CN113033898A (en) | Electrical load prediction method and system based on K-means clustering and BI-LSTM neural network | |
| CN112766548A (en) | Order completion time prediction method based on GASA-BP neural network | |
| CN116663607A (en) | Method for constructing oilfield separate layer water injection prediction model based on LSTM | |
| CN115394381B (en) | High-entropy alloy hardness prediction method and device based on machine learning and two-step data expansion | |
| CN116562174A (en) | Method and device for determining water distribution and injection quantity | |
| CN116629686A (en) | Method and device for evaluating enterprise energy consumption data | |
| CN113468821B (en) | Decision regression algorithm-based slope abortion sand threshold determination method | |
| CN114266337A (en) | Intelligent tobacco leaf grading model based on residual error network and grading method using model | |
| CN114581666A (en) | Intelligent rock sample identification method and system based on image deep learning | |
| CN114299330A (en) | Seismic facies classification method | |
| CN112348275A (en) | Regional ecological environment change prediction method based on online incremental learning |
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 |