Prediction method for calculating oil well skin coefficient based on real-time monitoring data

By applying the BP neural network-time series model in the real-time monitoring data of oil wells, the problem of real-time monitoring and prediction of the epidermal coefficient of oil wells is solved, and efficient and accurate prediction of the epidermal coefficient is achieved.

CN119664332APending Publication Date: 2025-03-21CHINA PETROLEUM & CHEMICAL CORP +1
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202311224963.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-09-21
Publication Date
2025-03-21

AI Technical Summary

Technical Problem

The prior art is difficult to monitor and predict the skin coefficient of oil wells in real time, and the calculation method requires a large amount of basic data, which is difficult to obtain and time-consuming.

Method used

Using a BP neural network-time series method, an epidermal coefficient prediction model is established using oil well real-time monitoring data, and a neural network is trained through historical data to predict future epidermal coefficient changes.

Benefits of technology

Real-time calculation of the skin coefficient of the oil well and prediction of future changes dynamics are realized, which simplifies data demand, reduces calculation costs and improves prediction accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure FT_1
    Figure FT_1
  • Figure FT_2
    Figure FT_2
  • Figure FT_3
    Figure FT_3
Patent Text Reader

Abstract

The invention relates to the technical field of oil and gas development, in particular to a prediction method for calculating an oil well skin coefficient based on real-time monitoring data. Comprising the following steps: acquiring skin coefficient historical data in a time period, establishing a skin coefficient prediction model based on a BP neural network-time sequence, and taking actual oil well skin coefficient time sequence data as an input value to obtain a skin coefficient prediction value in a future time period; according to the method, real-time monitoring data in the oil well production process are fully utilized, modeling is carried out based on a mechanism, and a skin coefficient historical value is calculated; and a BP neural network-time sequence model is used to predict the future value of the skin coefficient. According to the method, calculation is simple, needed basic data are easy to obtain on site, calculation results are reliable, a large amount of time and labor cost can be saved for oil field enterprises, remarkable economic benefits are brought, and the method has high practical value and practical significance.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of oil and gas development, and in particular to a method for predicting the skin coefficient of an oil well based on real-time monitoring data. Background Art

[0002] The bottom hole resistance coefficient, also known as the skin coefficient, is a dimensionless coefficient that indicates the degree of perfection of the well. The bottom hole resistance coefficient is represented by S, which is equal to the natural logarithm of the ratio of the completion radius rw to the reduced radius rc of the well. If the S value is zero, it means that the well is perfect; if the S value is a positive value, it means that the well is imperfect; it can be expressed as S = ln (rw / rc), where: rw is the completion radius and rc is the reduced radius. When rw / rc = 1, the S value is 0, indicating that the well is perfect; when rw / rc <1, the S value is a negative value, indicating that the well is super-perfect; and when rw / rc > 1, the S value is a positive value, indicating that the well is imperfect. The bottom hole resistance coefficient can be quantitatively calculated based on the pressure recovery and pressure drop test data. If S>0, it means that there is a blockage at the bottom of the well, resulting in a throttling effect; if S<0, it means that the production increase measures have improved the permeability around the bottom of the well and reduced the bottom hole flow resistance.

[0003] The bottom hole resistance coefficient is an important technical indicator for evaluating the degree of pollution and reservoir damage in the near-wellbore area of ​​oil and gas wells. It is related to the accuracy of oil and gas layer damage evaluation and production capacity prediction, the optimization of oil and gas production methods and production efficiency, and the selection, design and effect of treatment measures.

[0004] Therefore, scientific and reasonable monitoring and prediction of the skin coefficient is of great significance for increasing and stabilizing production, reducing costs and increasing efficiency during oil and gas production.

[0005] However, the skin coefficient of oil wells cannot be monitored in real time at oil fields. The existing skin coefficient calculation methods require a lot of basic data, some data are difficult to obtain, and the calculation requires a certain amount of time and economic cost.

[0006] For example, the invention patent with patent number CN201910672740.2 discloses a method for calculating the total skin coefficient of a multi-layer reservoir. According to the seepage mechanics theory, an analytical method is used to obtain the relationship between the total skin coefficient and the stratified skin coefficient. When a multi-layer reservoir is produced together, the total skin coefficient is the harmonic mean of the stratified skin coefficient corrected by physical properties and the weighted formation coefficient, that is, further, wherein aj is the weighted coefficient of the formation coefficient, and bj is the physical property correction coefficient. Furthermore, the method for calculating the total skin coefficient of the present invention reveals the relationship between the total skin coefficient and the stratified skin coefficient of the multi-layer reservoir system. When each layer is tested but not the entire layer, the total skin coefficient of the multi-layer reservoir can be obtained, which provides support for the calculation of the combined production; when a certain layer is not tested and other layers and the total system are tested, the skin coefficient of the untested layer can be obtained, and the reservoir protection effect of the untested layer can be evaluated; saving testing costs and time.

[0007] The invention patent with patent number CN201910740455.X discloses a method for calculating the skin coefficient of non-uniform damage of fractured reservoirs, obtains basic parameters, discretizes reservoir sections at equal intervals, assigns basic parameters to discrete sections, establishes original reservoir physical property parameter set, reservoir physical property parameter set after damage, and non-uniform damage skin parameter set; takes discrete sections as objects, calculates original reservoir physical property parameters of discrete sections using wellbore profile parameters and logging parameters; calculates reservoir physical property parameters after damage of discrete sections, calculates matrix and fracture damage skin coefficient of discrete sections, calculates non-uniform damage skin coefficient of discrete sections, and comprehensive skin coefficient of the whole section. The present invention solves the problem in the prior art that the skin coefficient of the whole section cannot reflect the degree of non-uniform damage, nor the degree of damage to the reservoir matrix and fracture system, and realizes the purpose that the skin coefficient can be decomposed into a long section to reflect the characteristics of non-uniform damage, and the damage to the matrix and fracture system can be distinguished within the section.

[0008] Patent No. CN201910673562.5 discloses a method for calculating the heterogeneous skin coefficient of a layered reservoir. According to the seepage theory, a calculation formula for the heterogeneous skin coefficient Svk of the layered reservoir is derived, that is, further wherein N is the total number of layers. Further, the present invention provides a specific method for calculating the heterogeneous skin coefficient of the layered reservoir and a derivation process, proving that the skin coefficient generated by the heterogeneity of the layered reservoir is related to the physical properties of each layer. The stronger the interlayer heterogeneity, the greater the skin coefficient generated, but the influence is logarithmic, and the change in the degree of heterogeneity has less and less influence on the generated skin coefficient.

[0009] As well as the pressure drop well test analysis method mentioned in the journal "Application Case Analysis of Skin Coefficient Calculation and Decomposition" edited by Liu Quangang, Yang Bin, Song Aili, Zhou Jiyong and Sun Lin.

[0010] Regarding the calculation and prediction of the skin coefficient, in addition to the several commonly used methods mentioned in the above patents and journals, there are many other calculation methods, all of which have the problems of numerous calculation data, complicated calculation steps, and being unable to predict the skin coefficient.

[0011] After searching, there is no precedent for combining neural network with the calculation and prediction of skin coefficient.

[0012] BP (backpropagation) neural network is a concept proposed by scientists led by Rumelhart and McClelland in 1986. It is a multi-layer feedforward neural network trained according to the error back propagation algorithm and is one of the most widely used neural network models.

[0013] Artificial neural networks do not need to determine the mathematical equations of the mapping relationship between input and output in advance. They only learn certain rules through their own training and get the result closest to the expected output value when the input value is given. As an intelligent information processing system, the core of artificial neural networks to realize their functions is the algorithm. BP neural network is a multi-layer feedforward network trained by error back propagation (abbreviated as error back propagation). Its algorithm is called BP algorithm. Its basic idea is the gradient descent method, which uses gradient search technology to minimize the mean square error between the actual output value and the expected output value of the network.

[0014] The basic BP algorithm includes two processes: forward propagation of signals and back propagation of errors. That is, the calculation of error output is carried out from input to output, while the adjustment of weights and thresholds is carried out from output to input. During forward propagation, the input signal acts on the output node through the hidden layer, and after nonlinear transformation, an output signal is generated. If the actual output does not match the expected output, the error back propagation process is entered. Error back propagation is to propagate the output error through the hidden layer to the input layer layer by layer, and distribute the error to all units in each layer, and use the error signal obtained from each layer as the basis for adjusting the weights of each unit. By adjusting the connection strength between the input node and the hidden layer node, the connection strength between the hidden layer node and the output node, and the threshold, the error is reduced along the gradient direction. After repeated learning and training, the network parameters (weights and thresholds) corresponding to the minimum error are determined, and the training stops. At this time, the trained neural network can process the input information of similar samples and output the information with the minimum nonlinear transformation.

[0015] The BP network adds several layers (one or more layers) of neurons between the input layer and the output layer. These neurons are called hidden units. They have no direct connection with the outside world, but changes in their states can affect the relationship between the input and output. Each layer can have several nodes.

[0016] With the rapid development of digitalization, informatization and intelligence of oil and gas fields, the digital coverage rate of oil and gas production sites has reached more than 90%, the labor organization mode and production operation mode have undergone major changes, and various operation indicators have been greatly improved. Many real-time monitoring data of oil wells provided on site contain important information on oil and gas reservoir dynamics, but they have not been fully utilized, and there is a lack of research on the prediction of oil well skin coefficient.

[0017] Therefore, the present invention mainly solves the problems of "how to make full use of the historical data of real-time monitoring of oil wells to study its correlation with the skin coefficient?", "how to calculate the skin coefficient of oil wells and analyze its development law?" and "how to predict the future changes of the skin coefficient based on historical data?", forming a method for predicting the skin coefficient of oil wells based on real-time monitoring data. Summary of the invention

[0018] The purpose of the present invention is to provide a method for calculating the oil well skin factor prediction based on real-time monitoring data in view of the deficiencies in the prior art.

[0019] The problem that it is difficult to accurately obtain the true value of the skin coefficient on site and the future changes in the skin coefficient of the oil well cannot be predicted is solved.

[0020] The technical solution is as follows:

[0021] A method for predicting the skin coefficient of an oil well based on real-time monitoring data obtains historical data of the skin coefficient within a period of time, establishes a skin coefficient prediction model based on a BP neural network-time series, takes the actual oil well skin coefficient time series data as input, and obtains the predicted value of the skin coefficient within a period of time in the future.

[0022] Furthermore, the skin coefficient prediction model based on BP neural network-time series uses part of the skin coefficient historical data as a training set and part of the skin coefficient historical data as a test set, and its activation function is a Sigmoid function.

[0023] Further, the specific steps include:

[0024] S1: Calculate the real-time skin coefficient S(t) of the oil well at time t, where t=1, 2, 3, etc., and obtain the time series data of the real-time skin coefficient of the oil well;

[0025] S2: using the time series data to establish a skin coefficient prediction model based on BP neural network-time series, and obtaining a time series skin coefficient prediction value within a future period of time.

[0026] Furthermore, the step S2 of establishing a skin coefficient prediction model based on BP neural network-time series includes:

[0027] S21: data preprocessing;

[0028] S22: Obtain skin coefficient historical data;

[0029] S23: 80% of the historical data of the skin coefficient is taken as a training set, and the remaining 20% ​​of the historical data of the skin coefficient is taken as a test set;

[0030] S24: Establish a BP neural network, set the input layer, hidden layer and output layer, the network structure is 5-3-1, initialize the connection weight w between the neurons in the hidden layer and the output layer, the hidden layer threshold a and the output layer threshold b, where the initial values ​​of w, a and b are all random numbers;

[0031] S25: The activation function used by the BP neural network is the Sigmoid function, the maximum number of training times is set to 1500 times, the learning rate is 0.1, and the minimum error of the training target is set to 0.0001.

[0032] Furthermore, the step S2 of obtaining the predicted value of the skin coefficient of the time series in the future period includes:

[0033] The time series data of the actual oil well skin coefficient are preprocessed and used as input values. The skin coefficient prediction model based on BP neural network-time series is used to obtain the predicted value of the skin coefficient in the future.

[0034] Furthermore, the real-time skin coefficient S(t) of the oil well in step S1 is calculated by using real-time monitoring data, wellbore structure and fluid property parameters.

[0035] Furthermore, step S1 specifically includes:

[0036] S11: Obtain real-time monitoring data, wellbore structure and fluid property parameters;

[0037] S12: Establish a real-time skin coefficient calculation model for oil wells;

[0038] S13: Calculate the time series data of skin coefficient.

[0039] Furthermore, the real-time monitoring data in step S11 includes: the dynamic liquid level depth D pf (t), actual liquid production q t (t) and moisture content f w (t); Wellbore structure and fluid property parameters include: crude oil density ρ o , crude oil viscosity μ o , crude oil volume coefficient B o , saturation pressure p b , effective permeability of oil phase k o 、Oil well leakage radius r e , average reservoir pressure Effective thickness of oil layer h, radius of oil well r w .

[0040] Furthermore, the real-time skin coefficient calculation model of the oil well in step S12 is:

[0041]

[0042] Where J(t) is the liquid extraction index corresponding to time t, t / (d·MPa); t represents the number of production days, d; t = 1, 2, 3…; ρ o is the density of crude oil on the ground, t / m3; k o is the effective permeability of the oil phase, 10-3μm 2 ; h is the effective thickness of the oil layer, m; μ o is the viscosity of crude oil, mPa·s; B o is the volume coefficient of crude oil, m3 / m3; r e is the oil well leakage radius, m; r w is the radius of the oil well, m.

[0043] Furthermore, step S13 includes: using the acquired on-site oil well real-time monitoring data and the wellbore structure and fluid property parameters to substitute into formula (8) to calculate the skin coefficient, and obtaining the skin coefficient time series data S(1), S(2), S(3) ... corresponding to different production days.

[0044] The beneficial effects of the present invention are:

[0045] The present invention makes full use of the real-time monitoring data in the oil well production process, calculates the historical value of the skin coefficient based on the mechanism modeling, and then uses the BP neural network-time series model to predict the future value of the skin coefficient. The method is simple to calculate, the required basic data is easy to obtain on site, the calculation results are reliable, and it can save a lot of time and labor costs for various oil field enterprises, bring significant economic benefits, and has strong practical value and realistic significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 The present invention provides a flow chart for predicting the oil well skin coefficient based on real-time monitoring data of the oil well.

[0047] Figure 2 This is the fitting result of the oil well skin coefficient in Example 2 of the present invention.

[0048] Figure 3 This is the prediction result of the oil well skin coefficient in Example 2 of the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present invention. In addition, in the following description, the description of well-known structures and technologies is omitted to avoid unnecessary confusion of the concept of the present invention.

[0050] Embodiment 1:

[0051] A method for predicting the skin coefficient of an oil well based on real-time monitoring data obtains historical data of the skin coefficient within a period of time, establishes a skin coefficient prediction model based on a BP neural network-time series, takes the actual oil well skin coefficient time series data as input, and obtains the predicted value of the skin coefficient within a period of time in the future.

[0052] The skin coefficient prediction model based on BP neural network-time series uses part of the skin coefficient historical data as a training set and part of the skin coefficient historical data as a test set, and its activation function is a Sigmoid function.

[0053] The specific steps include:

[0054] S1: Calculate the real-time skin coefficient S(t) of the oil well at time t, where t=1, 2, 3, etc., and obtain the time series data of the real-time skin coefficient of the oil well;

[0055] S2: using the time series data to establish a skin coefficient prediction model based on BP neural network-time series, and obtaining a time series skin coefficient prediction value within a future period of time.

[0056] The step S2 of establishing a skin coefficient prediction model based on BP neural network-time series includes:

[0057] S21: data preprocessing;

[0058] S22: Obtain skin coefficient historical data;

[0059] S23: 80% of the historical data of the skin coefficient is taken as a training set, and the remaining 20% ​​of the historical data of the skin coefficient is taken as a test set;

[0060] S24: Establish a BP neural network, set the input layer, hidden layer and output layer, the network structure is 5-3-1, initialize the connection weight w between the neurons in the hidden layer and the output layer, the hidden layer threshold a and the output layer threshold b, where the initial values ​​of w, a and b are all random numbers;

[0061] S25: The activation function used by the BP neural network is the Sigmoid function, the maximum number of training times is set to 1500 times, the learning rate is 0.1, and the minimum error of the training target is set to 0.0001.

[0062] The step S2 of obtaining the predicted value of the skin coefficient of the time series in the future period includes:

[0063] The time series data of the actual oil well skin coefficient are preprocessed and used as input values. The skin coefficient prediction model based on BP neural network-time series is used to obtain the predicted value of the skin coefficient in the future.

[0064] In step S1, the real-time skin coefficient S(t) of the oil well is calculated by real-time monitoring data, wellbore structure and fluid property parameters.

[0065] Step S1 specifically includes:

[0066] S11: Obtain real-time monitoring data, wellbore structure and fluid property parameters;

[0067] S12: Establish a real-time skin coefficient calculation model for oil wells;

[0068] S13: Calculate the time series data of skin coefficient.

[0069] The real-time monitoring data in step S11 includes: the dynamic liquid level depth D pf (t), actual liquid production q t (t) and moisture content f w (t); Wellbore structure and fluid property parameters include: crude oil density ρ o , crude oil viscosity μ o , crude oil volume coefficient B o , saturation pressure p b , effective permeability of oil phase k o 、Oil well leakage radius r e , average reservoir pressure p r , effective thickness of oil layer h, radius of oil well r w .

[0070] The real-time skin coefficient calculation model of the oil well in step S12 is:

[0071]

[0072] Where J(t) is the liquid extraction index corresponding to time t, t / (d·MPa); t represents the number of production days, d; t = 1, 2, 3…; ρ o is the density of crude oil on the ground, t / m3; k o is the effective permeability of the oil phase, 10-3μm 2 ; h is the effective thickness of the oil layer, m; μ o is the viscosity of crude oil, mPa·s; B o is the volume coefficient of crude oil, m3 / m3; r eis the oil well leakage radius, m; r w is the radius of the oil well, m.

[0073] Step S13 includes: using the acquired on-site oil well real-time monitoring data and the wellbore structure and fluid property parameters to substitute into formula (8) to calculate the skin coefficient, and obtaining skin coefficient time series data S(1), S(2), S(3) ... corresponding to different production days.

[0074] Embodiment 2:

[0075] A method for predicting oil well skin factor based on real-time monitoring data, the specific steps are as follows:

[0076] Step 1: Obtain real-time monitoring data, wellbore structure and fluid property parameters for calculating the real-time skin coefficient of the oil well. The real-time monitoring data includes: dynamic liquid level depth D pf (t), actual liquid production q t (t) and moisture content f w (t); Wellbore structure and fluid property parameters include: crude oil density ρ o , crude oil viscosity μ o , crude oil volume coefficient B o , saturation pressure p b , effective permeability of oil phase k o 、Oil well leakage radius r e , average reservoir pressure Effective thickness of oil layer h, radius of oil well r w .

[0077] Step 2: Establish a real-time skin coefficient calculation model for oil wells:

[0078]

[0079] Where J(t) is the liquid extraction index corresponding to time t, t / (d·MPa); t represents the number of production days, d; t = 1, 2, 3…; ρ o is the density of crude oil on the ground, t / m3; k o is the effective permeability of the oil phase, 10-3μm 2 ; h is the effective thickness of the oil layer, m; μ o is the viscosity of crude oil, mPa·s; B o is the volume coefficient of crude oil, m3 / m3; r e is the oil well leakage radius, m; r w is the radius of the oil well, m.

[0080] Step 3: Use the acquired real-time monitoring data of the on-site oil wells and the wellbore structure and fluid property parameters to substitute into formula (8) to calculate the skin coefficient, and obtain the skin coefficient time series data S(1), S(2), S(3)... corresponding to different production days.

[0081] Step 4: Use the calculated time series data of the real-time skin coefficient of the oil well as the input layer and output layer for training to establish a skin coefficient prediction model based on the BP neural network-time series.

[0082] Including data preprocessing; selecting 80% of the historical data of the skin coefficient as the training set, and the remaining 20% ​​of the historical data of the skin coefficient as the test set; establishing a BP neural network, setting the input layer, hidden layer and output layer, the network structure is 5-3-1, initializing the connection weight w between the hidden layer and the output layer neurons, the hidden layer threshold a and the output layer threshold b, where the initial values ​​of w, a and b are all random numbers; the activation function used by the BP neural network is the Sigmoid function, the maximum number of training times is set to 1500 times, the learning rate is 0.1, and the minimum error of the training target is set to 0.0001.

[0083] Step 5: Preprocess the actual oil well skin coefficient time series data and use it as input value, use the skin coefficient prediction model based on BP neural network-time series to obtain the skin coefficient prediction value in the future period.

[0084] Embodiment three:

[0085] Take the No. 13 oil well in Shengli Oilfield as an example well for process calculation analysis:

[0086] The real-time monitoring data and wellbore structure and fluid property parameters of No. 13 oil well are obtained. The real-time monitoring data include: dynamic liquid level depth, actual liquid production and water content; the wellbore structure and fluid property parameters include: crude oil density, crude oil viscosity, crude oil volume coefficient, saturation pressure, effective permeability of oil phase, oil well drainage radius, average oil layer pressure, effective oil layer thickness, and oil well radius.

[0087] The established real-time skin coefficient calculation model for oil wells is used, combined with the real-time monitoring data of No. 13 oil well and the wellbore structure and fluid property parameters, to calculate the skin coefficient time series data corresponding to different production dates of No. 13 oil well.

[0088] The calculated time series data of the skin coefficient of the 13# oil well is used as the input layer and the output layer. The skin coefficient from April 14, 2022 to July 2, 2022 (a total of 80 days of data) is selected as the training set, and the skin coefficient from July 3, 2022 to July 22, 2022 (a total of 20 days of data) is selected as the test set. The skin coefficient prediction model of the 13# oil well based on the BP neural network-time series is obtained through training.

[0089] Using the 13# oil well skin coefficient prediction model based on BP neural network-time series, the predicted value of the skin coefficient in the next 20 days is calculated.

[0090] The calculation results of oil well No. 13 are as follows: Figure 2 and Figure 3 As shown, when the network structure is set to 5-3-1, the average relative error of historical data training and fitting is 5.6%; the average relative error between the predicted value and the actual value of the skin coefficient in the next 20 days is 2.4%; both the fitting and prediction accuracy are high.

[0091] The real-time monitoring data of 13 wells on site and the wellbore structure and fluid property parameters were selected to predict the skin coefficient of the oil wells. The average relative error between the predicted value and the actual value of the skin coefficient is shown in Table 1.

[0092] Table 1:

[0093]

[0094]

[0095] It can be seen from Table 1 that the average relative error between the predicted and actual values ​​of the skin coefficients of the 13 wells is 3.51%, the maximum relative error is 4.8%, and the minimum relative error is 2.3%, indicating that this method has high accuracy and practical value, and can provide theoretical and technical support for the real-time calculation of the skin coefficient of on-site oil wells and the dynamic prediction of future changes.

[0096] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims.

Claims

1. A method for predicting oil well skin factor based on real-time monitoring data, characterized in that: The historical data of skin coefficient within a period of time is obtained, and a skin coefficient prediction model based on BP neural network-time series is established. The actual oil well skin coefficient time series data is used as input value to obtain the skin coefficient prediction value within a period of time in the future.

2. The method for predicting oil well skin coefficient based on real-time monitoring data according to claim 1, characterized in that: The skin coefficient prediction model based on BP neural network-time series uses part of the skin coefficient historical data as a training set and part of the skin coefficient historical data as a test set, and its activation function is a Sigmoid function.

3. The method for predicting oil well skin coefficient based on real-time monitoring data according to claim 2, characterized in that: The specific steps include: S1: Calculate the real-time skin coefficient S(t) of the oil well at time t, where t=1, 2, 3, etc., and obtain the time series data of the real-time skin coefficient of the oil well; S2: using the time series data to establish a skin coefficient prediction model based on BP neural network-time series, and obtaining a time series skin coefficient prediction value within a future period of time.

4. The method for predicting oil well skin coefficient based on real-time monitoring data according to claim 3 is characterized in that: The step S2 of establishing a skin coefficient prediction model based on BP neural network-time series includes: S21: data preprocessing; S22: Obtain skin coefficient historical data; S23: 80% of the historical data of the skin coefficient is taken as a training set, and the remaining 20% ​​of the historical data of the skin coefficient is taken as a test set; S24: Establish a BP neural network, set the input layer, hidden layer and output layer, the network structure is 5-3-1, initialize the connection weight w between the neurons in the hidden layer and the output layer, the hidden layer threshold a and the output layer threshold b, where the initial values ​​of w, a and b are all random numbers; S25: The activation function used by the BP neural network is the Sigmoid function, the maximum number of training times is set to 1500 times, the learning rate is 0.1, and the minimum error of the training target is set to 0.0001.

5. The method for predicting oil well skin coefficient based on real-time monitoring data according to claim 4, characterized in that: The step S2 of obtaining the predicted value of the skin coefficient of the time series in the future period includes: The time series data of the actual oil well skin coefficient are preprocessed and used as input values. The skin coefficient prediction model based on BP neural network-time series is used to obtain the predicted value of the skin coefficient in the future.

6. A method for predicting oil well skin factor based on real-time monitoring data according to any one of claims 3 to 5, characterized in that: In step S1, the real-time skin coefficient S(t) of the oil well is calculated by real-time monitoring data, wellbore structure and fluid property parameters.

7. The method for predicting oil well skin factor based on real-time monitoring data according to claim 6, characterized in that: Step S1 specifically includes: S11: Obtain real-time monitoring data, wellbore structure and fluid property parameters; S12: Establish a real-time skin coefficient calculation model for oil wells; S13: Calculate the time series data of skin coefficient.

8. The method for predicting oil well skin factor based on real-time monitoring data according to claim 7, characterized in that: The real-time monitoring data in step S11 includes: the dynamic liquid level depth D pf (t), actual liquid production q t (t) and moisture content f w (t); Wellbore structure and fluid property parameters include: crude oil density ρ o 、Crude oil viscosity μ o , crude oil volume coefficient B o , saturation pressure p b , effective permeability of oil phase k o 、Oil well leakage radius r e , average reservoir pressure p r , effective thickness of oil layer h, radius of oil well r w .

9. The method for predicting oil well skin factor based on real-time monitoring data according to claim 8, characterized in that: The real-time skin coefficient calculation model of the oil well in step S12 is: Where J(t) is the liquid extraction index corresponding to time t, t / (d·MPa); t represents the number of production days, d; t = 1, 2, 3…; ρ o is the density of crude oil on the ground, t / m3; k o is the effective permeability of the oil phase, 10-3μm 2 ; h is the effective thickness of the oil layer, m; μ o is the crude oil viscosity, mPa·s; B o is the volume coefficient of crude oil, m3 / m3; r e is the oil well leakage radius, m; r w is the radius of the oil well, m.

10. The method for predicting oil well skin factor based on real-time monitoring data according to claim 9, characterized in that: Step S13 includes: using the acquired on-site oil well real-time monitoring data and the wellbore structure and fluid property parameters to substitute into formula (8) to calculate the skin coefficient, and obtaining skin coefficient time series data S(1), S(2), S(3) ... corresponding to different production days.

Citation Information

Patent Citations

  • A method for calculating the heterogeneous skin coefficient of layered oil reservoirs

    CN110378046B

  • Method for calculating non-uniform damage skin coefficients of fractured reservoir

    CN110454152A

  • Method for calculating total skin coefficient of multilayer oil reservoir

    CN110608036A

  • Offshore oil well productivity evaluation method

    CN113338904A

  • Methods and systems for gas condensate well performance prediction

    US20220090496A1