Modular-based Intelligent Operation and Maintenance System and Method for Oilfield Gathering and Transportation

Abstract

The present invention discloses a modular-based intelligent operation and maintenance system and method for oilfield gathering and transportation, which relates to the technical field of coordinated control of oilfield gathering and transportation operation and maintenance. The method includes the following steps: collecting data of oilfield gathering and transportation well stations under the jurisdiction of the region through sensors, establishing a digital twin model for predicting the daily oil production of well stations by combining the predicted value of the current crude oil reserve in the well station and the crude oil collection situation, and obtaining the oil collection prediction cycle; for the well stations under the jurisdiction of the transfer station, according to the oil collection prediction cycle of different well stations and combining the transfer capacity of the transfer station, allocating the well stations connected to the transfer station in the region, and at the same time using the AI algorithm to optimize the liquid volume, pressure, and temperature parameters in real time to achieve the coordinated operation of the transfer station and the well station in the process of oil production and transportation. The present invention facilitates the intelligent operation and maintenance and coordinated control of oilfield gathering and transportation in the region by establishing a digital twin model for well stations in the region, and enhances the practicability and functionality.

Description

Technical Field

[0001] The present invention relates to the technical field of coordinated control of oilfield gathering and transportation operation and maintenance. Specifically, it relates to a modular-based intelligent operation and maintenance system and method for oilfield gathering and transportation. Background Technique

[0002] In the process of oilfield gathering and transportation, in a region, a transfer station often corresponds to multiple well-site stations. Therefore, ensuring the efficient and stable operation of the transfer station and the well-site stations is crucial for the production of the entire oilfield. Therefore, it is necessary to coordinate and control the well-site stations and transfer stations of oilfield gathering and transportation in the region to ensure the smooth progress of oilfield production;

[0003] In the traditional oilfield gathering and transportation operation and maintenance system, data prediction mostly relies on a single model to make macroscopic adjustments to oilfield gathering and transportation. Simply relying on a physical model, although a basic framework can be constructed based on geological conditions, it is difficult to adapt to the complex changes in actual production in real time. And only using a data-driven model may lack compliance with the physical characteristics and geological laws of the oil reservoir, resulting in a large deviation between the prediction result and the actual situation. The limitations of this single model make the accuracy of the predicted data insufficient, unable to provide a reliable basis for production decisions, and unable to reasonably plan and allocate the oil volume processing cycle of the transfer stations within the subordinate regions, resulting in problems of low practicability and functionality.

[0004] In response to the problems in the related art, no effective solution has been proposed yet. Summary of the Invention

[0005] In response to the problems in the related art, the present invention proposes a modular-based intelligent operation and maintenance system and method for oilfield gathering and transportation to overcome the above technical problems existing in the existing related technologies.

[0006] To this end, the specific technical solutions adopted by the present invention are as follows:

[0007] A modular-based intelligent operation and maintenance method for oilfield gathering and transportation, the method includes the following steps:

[0008] S1. Collect data of well-site stations for oilfield gathering and transportation under the jurisdiction of the region through sensors, and establish a digital twin model for predicting the daily oil production of the well-site stations in combination with the predicted value of the current crude oil reserves of the well-site stations and the crude oil collection situation, and obtain the oil volume collection prediction period;

[0009] S2. For the well-site stations under the jurisdiction of the transfer station, according to the oil volume collection prediction periods of different well-site stations, and in combination with the transfer capacity of the transfer station, allocate the well-site stations connected to the transfer station in the region, and at the same time use the AI algorithm to optimize the liquid volume, pressure, and temperature parameters in real time to achieve the coordination of the oil production and transportation processes of the transfer station and the well-site stations.

[0010] As a preferred implementation manner, the S1 includes the following steps:

[0011] S11. Set sensors in the gathering stations between oil wells in the oilfields under the jurisdiction of the region, including flow sensors, pressure sensors, and temperature sensors, for collecting real-time flow rate, wellhead pressure, and temperature during the crude oil collection process at the gathering stations between wells.

[0012] S12. Through the combination of physical models and data-driven models, construct a digital twin model of the gathering stations between wells under the jurisdiction of the region. Use the physical model to provide a basic description of the oilfield gathering process, and use historical and real-time data through the data-driven model to optimize the parameters of the digital twin model and obtain the oil volume collection prediction period.

[0013] As a preferred implementation method, S12 includes the following steps:

[0014] S121. Collect the daily oil production, wellhead pressure, oil well production, and historical production time data of different gathering stations between wells in the region. Combine the geological data of different gathering stations between wells to construct a three-dimensional geological model of the oil reservoir, describe the distribution, structure, and properties of the reservoir, and simulate the fluid flow process in the oil reservoir through Eclipse to simulate the theoretical oil reservoir pressure and physical model oil production prediction values at different time periods.

[0015] S122. Based on the historical production data of the gathering stations between wells, construct a data-driven model to optimize the parameters of the physical model. Collect the historical production data of the gathering stations between wells and perform normalization processing, including historical daily oil production, oil well pressure, temperature, water cut, equipment operation duration, and valve opening. Calculate the Pearson correlation coefficient between each feature and the daily oil production. Its algorithm formula is:

[0016]

[0017] where x i is different eigenvalue, y i is the daily oil production, and respectively represent the mean values of the feature and the daily oil production, n represents the number of samples, and retain the features with |r| > 0.5 as the correlation features;

[0018] Construct a linear regression equation based on the correlation features as the data-driven model to obtain the daily oil production prediction value:

[0019] y = β0 + β1x1 + β2x2 +... + β p x p + ε;

[0020] where β0 is the intercept, β p represents the coefficients of different features, ε is the error term, y is the predicted value of the daily oil production, and x p represents the feature parameter value on the pth day;

[0021] Train the data-driven model using historical data and optimize the hyperparameters of the model. The specific steps are as follows:

[0022] Divide the historical data into a training set and a validation set. The training set contains 70% of the historical data, and the validation set contains 30% of the historical data. Use the mean squared error as the loss function:

[0023]

[0024] Among them, represent the predicted value at time t and the actual daily oil production at time t respectively, n represents the number of samples, and update the model coefficients iteratively through the gradient descent method

[0025]

[0026] Among them, α represents the learning rate. Validate the model through the validation set and evaluate it through the mean absolute error, represents the coefficient value at time t + 1;

[0027] S123. Combine the predicted oil production value of the comprehensive physical model and the predicted oil production value of the data-driven model based on the weight ratio to obtain the output value of the oil production of the digital twin model:

[0028] Y = ω1 × A + ω2 × B;

[0029] Among them, Y, A, and B represent the output value of the daily oil production of the digital twin model between current wells, the predicted value of the daily oil production of the physical model, and the predicted value of the daily oil production of the data-driven model respectively. ω1 and ω2 represent the weight ratio, and ω1 + ω2 = 1.

[0030] As a preferred implementation, the S123 includes the following steps:

[0031] S1231. Based on the output value Y of the daily oil production of the digital twin model between current wells, establish a set of output values of oil production within N days {Y1, Y2, Y3,..., Y N}, with the date as the abscissa and the output value of the daily oil production of the digital twin model between current wells as the ordinate, and draw a line chart of the oil production collection prediction cycle of the current well-interstation.

[0032] As a preferred implementation, the S2 includes the following steps:

[0033] S21. Collect the oil production collection prediction cycles of each well-interstation under the jurisdiction of the transfer station in the area Among them, e represents the date, u represents the number of well-interstations under the current transfer station, statistically obtain the oil production processing cycle of the current transfer station, and coordinate and optimize it in combination with the processing capacity of the subordinate transfer stations;

[0034] S22. For the well - to - station and transfer stations that cannot be coordinated, through the AI algorithm, under the premise of meeting the transfer capacity of the transfer station and the production requirements of each well - to - station, the oil production and transportation process is coordinated and optimized.

[0035] As a preferred implementation manner, the S21 includes the following steps:

[0036] S211. For the oil volume acquisition and prediction cycle of each well - to - station under the jurisdiction of the transfer station, accumulate to obtain the oil volume processing cycle of the transfer station:

[0037] Accumulate the daily oil volume prediction values of the well - to - stations under the jurisdiction of the transfer station within the cycle to obtain the daily oil volume processing prediction value of the transfer station. Where e represents the date and q is the number of the current transfer station;

[0038] S212. Based on the maximum oil volume processing value c of the current transfer station, count the abnormal dates:

[0039] When It represents that the oil volume processing of the transfer station is normal on the current date and no adjustment is required;

[0040] When It represents that the oil volume processing of the transfer station is abnormal on the current date and adjustment is required;

[0041] S213. Count the abnormal dates of the oil volume processing of all transfer stations. When the abnormal dates are consecutive dates, calculate the daily deficit value. At the same time, calculate the surplus value of the transfer stations with normal processing under its jurisdiction. Where f is the abnormal date. When is satisfied, re - plan the oilfield gathering pipeline for the current two transfer stations.

[0042] As a preferred implementation manner, the S22 includes the following steps:

[0043] S221. When the abnormal dates are non - consecutive dates, record the current transfer station as an uncoordinated transfer station and adjust the oil production and transportation parameters of the transfer station and the well - to - stations through a multi - layer perceptron. The steps are as follows:

[0044] Collect the difference data between the historical oil volume processing values of the transfer station and the oil volume acquisition values of the well - to - stations as the input of the input layer. After the non - linear transformation of the hidden layer, the output layer outputs the adjusted oil production and transportation parameters, including the adjusted value of the oil production rate of the well - to - station and the adjusted value of the valve opening of the transfer station. Use the historical data to train the model and adjust the weight and bias parameters of the model by minimizing the objective function;

[0045] S222. On the abnormal date, input the difference between the transfer station and the well - to - well station on that day into the multi - layer perceptron. The model outputs the adjusted oil production and transportation parameters, including the adjusted value of the oil production rate of the well - to - well station and the adjusted value of the valve opening of the transfer station.

[0046] The modular intelligent operation and maintenance system for oilfield gathering and transportation includes a data acquisition module, a digital twin model establishment module, a periodic prediction module, and a station allocation module:

[0047] The data acquisition module includes flow sensors, pressure sensors, and temperature sensors, which are used to collect real - time flow rate, wellhead pressure, and temperature during the crude oil collection process of the well - to - well station, and collect historical operation data of each well - to - well station, and transmit it to the subsequent modules;

[0048] The digital twin model establishment module constructs a digital twin model of the well - to - well stations under the jurisdiction of the region through the combination of physical models and data - driven models. The physical model provides a basic description of the oilfield gathering and transportation process, and the data - driven model uses historical and real - time data to optimize the parameters of the digital twin model;

[0049] The periodic prediction module, based on the output value of the daily oil production of the digital twin model of the current well - to - well station, establishes a set of output values of oil production within N days. Taking the date as the abscissa and the output value of the daily oil production of the digital twin model of the well - to - well station as the ordinate, draw a prediction period broken - line graph of the oil production of the current well - to - well station;

[0050] The station allocation module, for the well - to - well stations under the jurisdiction of the transfer station, according to the oil production prediction periods of different well - to - well stations, combined with the transfer capacity of the transfer station, allocates the well - to - well stations connected to the transfer station in the region, and at the same time uses AI algorithms to optimize the liquid volume, pressure, and temperature parameters in real - time to achieve the coordination of the oil production and transportation processes of the transfer station and the well - to - well stations.

[0051] The beneficial effects of the present invention are as follows:

[0052] 1. Based on the real - time parameters and historical data of the oilfield gathering and transportation transfer stations and well - to - well stations in the region, the present invention establishes a daily oil production prediction digital twin model for the well - to - well stations, establishes a prediction period for the daily oil production of the well - to - well stations under its jurisdiction, and predicts the operation conditions of the transfer stations, so as to conduct centralized control over the gathering and transportation of the transfer stations and well - to - well stations in the region, enhancing the functionality of the method.

[0053] 2. By combining the daily oil production prediction value of the physical model, considering the changes in geological parameters caused by the change of oil storage during the oilfield acquisition process, and integrating the daily oil production prediction value of the data-driven model, the present invention constructs a broken line graph of the oil volume acquisition prediction cycle between current wells and stations. The physical model provides a basic framework and constraints based on geological conditions to ensure that the prediction conforms to the physical characteristics and geological laws of the reservoir. The digital model can supplement and correct the physical model, and use the information in real-time data and historical data to adjust the prediction results to make them more in line with the actual production situation, enhancing the data accuracy of control adjustment between stations.

[0054] 3. By accumulating the oil volume acquisition prediction cycles of each well-to-station under the jurisdiction of the transfer station, the present invention clarifies the oil volume processing cycle of the transfer station itself. At the same time, the daily oil volume prediction values within the cycle are accumulated to obtain the daily processing prediction value, providing basic data for subsequent analysis to plan the production arrangement, resource reserve, etc. of the transfer station in advance. And for the anomalies on consecutive dates, the transfer stations are paired for processing, and for the anomalies on non-consecutive dates, the coordinated optimization of operation and maintenance control of the abnormal transfer stations and well-to-stations is carried out, which can eliminate the deviation caused by accidental anomalies and avoid unnecessary resource waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0056] Figure 1 is a flowchart of a modular-based intelligent operation and maintenance method for oilfield gathering and transportation according to an embodiment of the present invention;

[0057] Figure 2 is a flowchart of a modular-based intelligent operation and maintenance method for oilfield gathering and transportation according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] To further illustrate the embodiments, the present invention provides drawings. These drawings are part of the disclosure of the present invention, mainly used to illustrate the embodiments, and can be used to explain the operating principle of the embodiments in conjunction with the relevant descriptions in the specification. With reference to these contents, those of ordinary skill in the art should be able to understand other possible implementation manners and the advantages of the present invention. The components in the drawings are not drawn to scale, and similar component symbols are usually used to represent similar components.

[0059] According to an embodiment of the present invention, a modular-based intelligent operation and maintenance system and method for oilfield gathering and transportation are provided.

[0060] The present invention will be further described below in conjunction with the drawings and specific embodiments:

[0061] Example 1:

[0062] As Figure 1 shown, the modular-based intelligent operation and maintenance method for oilfield gathering and transportation according to the embodiments of the present invention includes the following steps:

[0063] S1. Collect data of oilfield gathering and transportation well stations under the jurisdiction of the region through sensors, and establish a digital twin model for predicting the daily oil production of well stations by combining the predicted value of the current crude oil reserve in the well station and the crude oil collection situation, so as to obtain the oil collection prediction cycle;

[0064] S11. Set sensors in the oilfield gathering and transportation well stations under the jurisdiction of the region, including flow sensors, pressure sensors, and temperature sensors, for collecting real-time flow rate, wellhead pressure, and temperature during the crude oil collection process of the well station;

[0065] S12. Construct a digital twin model of the well stations under the jurisdiction of the region through the combination of physical models and data-driven models. Provide a basic description of the oilfield gathering and transportation process through the physical model, and use historical and real-time data through the data-driven model to optimize the parameters of the digital twin model to obtain the oil collection prediction cycle;

[0066] S121. Collect daily oil production, wellhead pressure, oil well production, and oil well production historical time data of different well stations under the jurisdiction of the region, and combine the geological data of different well stations to construct a three-dimensional geological model of the oil reservoir, describe the distribution, structure, and properties of the reservoir, and simulate the fluid flow process in the oil reservoir through Eclipse to simulate the theoretical oil reservoir pressure and physical model oil production prediction values at different time periods;

[0067] It should be noted that the geological data of the well station includes oil reservoir geological characteristics, reservoir parameters, etc. Among them, the reservoir parameters include permeability, porosity, oil saturation, etc. When constructing the physical model of the well station, it is necessary to further improve the physical model through equations, describe the mass transfer of crude oil in the pipeline based on the mass conservation equation, describe the temperature change of crude oil in the pipeline based on the energy conservation equation, and describe the flow characteristics of crude oil in the pipeline through the Darcy-Weisbach equation.

[0068] S122. Based on the historical production data of the well station, construct a data-driven model to optimize the parameters of the physical model. Collect the historical production data of the well station and perform normalization processing, including historical daily oil production, oil well pressure, temperature, water cut, equipment operation duration, and valve opening, and calculate the Pearson correlation coefficient between each feature and the daily oil production. The algorithm formula is:

[0069]

[0070] where x iare different eigenvalue, y i is the daily oil production, and respectively represent the mean values of the feature and the daily oil production, n represents the number of samples, and the features with |r|>0.5 are retained as the correlation features;

[0071] Based on the correlation features, a linear regression equation is constructed as a data-driven model to obtain the predicted value of the daily oil production:

[0072] y = β0 + β1x1 + β2x2 +... + β p x p + ε;

[0073] Among them, β0 is the intercept, β p represents the coefficients of different features, ε is the error term, y is the predicted value of the daily oil production on the current day, and x p represents the feature parameter value on the pth day;

[0074] Use historical data to train the data-driven model and optimize the hyperparameters of the model. The specific steps are as follows:

[0075] Divide the historical data into a training set and a validation set. The training set contains 70% of the historical data, and the validation set contains 30% of the historical data. The mean square error is used as the loss function:

[0076]

[0077] Among them, respectively represent the predicted value at time t and the actual daily oil production at time t, n represents the number of samples, and the model coefficients are iteratively updated by the gradient descent method

[0078]

[0079] Among them, α represents the learning rate. The model is verified through the validation set and evaluated by the mean absolute error, represents the coefficient value at time t + 1;

[0080] It should be noted that the updated coefficient value is obtained by subtracting the product of the learning rate and the partial derivative of the mean square error with respect to the coefficient from the current coefficient value. Each coefficient is iteratively updated separately to gradually find the combination of coefficient values that minimizes the loss function MSE. The calculation formula for the mean absolute error is:

[0081]

[0082] The smaller the MAE, the smaller the average error between the predicted value and the true value of the model, and the better the prediction performance of the model. By combining the magnitude of the collected data to set the judgment threshold, the prediction performance of the model is further judged.

[0083] S123. Based on the predicted oil production of the comprehensive physical model and the predicted oil production of the data-driven model according to the weight ratio, obtain the output value of the oil production of the digital twin model:

[0084] Y = ω1×A + ω2×B;

[0085] Where Y, A, and B represent the daily output value of the oil production of the digital twin model between the current wells, the predicted daily oil production of the physical model, and the predicted daily oil production of the data-driven model respectively. ω1 and ω2 represent the weight ratio, and ω1 + ω2 = 1.

[0086] It should be noted that the weight ratios ω1 and ω2 are set to 0.6 and 0.4, and the weight ratio can also be adjusted according to the actual situation to meet the actual conditions of different well-to-station areas;

[0087] S1231. Based on the daily output value Y of the oil production of the digital twin model between the current wells, establish a set of oil production output values {Y1, Y2, Y3,..., Y N} within N days. Taking the date as the abscissa and the daily output value of the oil production of the digital twin model between the wells as the ordinate, draw a line graph of the oil production collection prediction period for the current well-to-station area.

[0088] It should be noted that by combining the predicted daily oil production of the physical model, the change of geological parameters caused by the change of oil storage during the oilfield collection process can be considered. By integrating the predicted daily oil production of the data-driven model, a line graph of the oil production collection prediction period for the current well-to-station area is constructed. The physical model provides a basic framework and constraints based on the geological situation to ensure that the prediction conforms to the physical characteristics and geological laws of the oil reservoir. The digital model can supplement and correct the physical model, and use the information in the real-time data and historical data to adjust the prediction results to make them more in line with the actual production situation.

[0089] Example 2:

[0090] S2. For the well-to-station areas under the jurisdiction of the transfer station, according to the oil production collection prediction periods of different well-to-station areas, combined with the transfer capacity of the transfer station, allocate the well-to-station areas connected to the transfer station in the region, and at the same time use the AI algorithm to optimize the liquid volume, pressure, and temperature parameters in real time to achieve the coordination of the oil production and transportation process between the transfer station and the well-to-station areas;

[0091] S21. Collect the oil production collection prediction periods of each well-to-station area under the jurisdiction of the transfer station in the region Among them, e represents the date, u represents the number of well-intermediate stations under the current transfer station. The oil volume processing cycle of the current transfer station is statistically obtained, and coordinated optimization is carried out in combination with the processing capacity of the subordinate transfer stations;

[0092] S21 includes the following steps:

[0093] S211. For the oil volume acquisition and prediction cycle of each well-intermediate station under the transfer station, accumulate to obtain the oil volume processing cycle of the transfer station:

[0094] Accumulate the predicted oil volume values of the well-intermediate stations under the transfer station on each day within the cycle to obtain the daily predicted oil volume processing value of the transfer station Among them, e represents the date, and q is the number of the current transfer station;

[0095] S212. Based on the maximum oil volume processing value c of the current transfer station, statistically determine the abnormal dates:

[0096] When It represents that the oil volume processing of the transfer station is normal on the current date, and no allocation is required;

[0097] When It represents that the oil volume processing of the transfer station is abnormal on the current date, and allocation is required;

[0098] S213. Statistically determine the abnormal dates of all transfer stations. When the abnormal dates are consecutive dates, calculate the daily deficit value At the same time, calculate the surplus value of the subordinate transfer stations with normal processing Among them, f is the abnormal date. When is satisfied, re-plan the oilfield gathering pipeline for the current two transfer stations.

[0099] It should be noted that by accumulating the oil volume acquisition and prediction cycles of each well-intermediate station under the transfer station, the oil volume processing cycle of the transfer station itself is clarified. At the same time, the daily processing prediction value is obtained by accumulating the daily oil volume prediction values within the cycle, providing basic data for subsequent analysis to plan the production arrangement, resource reserve, etc. of the transfer station in advance. Pairing and processing the transfer stations for the abnormalities on consecutive dates can eliminate the deviation caused by occasional abnormalities and avoid unnecessary resource waste;

[0100] S22. For the well-intermediate stations and transfer stations that cannot be allocated, through the AI algorithm, coordinate and optimize the oil production and transportation process on the premise of meeting the transfer capacity of the transfer station and the production requirements of each well-intermediate station;

[0101] S221. When the abnormal dates are non-consecutive dates, record the current transfer station as a non-allocated transfer station, and adjust the oil production and transportation parameters of the transfer station and the well-intermediate station through a multi-layer perceptron. The steps are as follows:

[0102] Collect the difference data between the historical oil volume processing values of the transfer station and the oil volume acquisition values of the inter-well stations as the input of the input layer. After the non-linear transformation of the hidden layer, the output layer outputs the adjusted oil production and transportation parameters, including the adjusted oil production rate value of the inter-well stations and the adjusted valve opening value of the transfer station. Use historical data to train the model and adjust the weight and bias parameters of the model by minimizing the objective function;

[0103] S222. On the abnormal date, input the difference between the transfer station and the subordinate inter-well stations on that day into the multi-layer perceptron. The model outputs the adjusted oil production and transportation parameters, including the adjusted oil production rate value of the inter-well stations and the adjusted valve opening value of the transfer station.

[0104] It should be noted that the number of neurons in the output layer is determined according to needs. If it is necessary to output the oil production rate of the inter-well stations, assuming there are 10 inter-well stations in total, the output layer can have 10 neurons, corresponding to the adjusted oil production rate values of each inter-well station respectively. If it is necessary to output the valve opening of the transfer station, the number of neurons in the output layer can be set according to the number of valves of the transfer station. The output value of the neurons in the output layer is the predicted oil production rate of the inter-well stations or the valve opening of the transfer station by the model.

[0105] Embodiment 3:

[0106] As Figure 2 shown, the modular intelligent operation and maintenance system for oilfield gathering and transportation includes a data acquisition module, a digital twin model establishment module, a periodic prediction module, and a station allocation module:

[0107] The data acquisition module includes flow sensors, pressure sensors, and temperature sensors, which are used to collect the real-time flow rate, wellhead pressure, and temperature during the crude oil acquisition process of the inter-well stations, and collect the historical operation data of each inter-well station and transmit it to the subsequent modules;

[0108] The digital twin model establishment module constructs a digital twin model of the inter-well stations under the jurisdiction of the region through the combination of a physical model and a data-driven model. The physical model provides a basic description of the oilfield gathering and transportation process, and the data-driven model uses historical and real-time data to optimize the parameters of the digital twin model;

[0109] The periodic prediction module establishes a set of oil production output values within N days based on the daily digital twin model oil production output value of the current inter-well station. Using the date as the abscissa and the daily digital twin model oil production output value of the inter-well station as the ordinate, draw the oil volume acquisition prediction period line chart of the current inter-well station;

[0110] The station allocation module, for the inter-well stations under the jurisdiction of the transfer station, according to the oil volume acquisition prediction period of different inter-well stations and in combination with the transfer capacity of the transfer station, allocates the inter-well stations connected to the transfer station in the region, and at the same time uses the AI algorithm to optimize the liquid volume, pressure, and temperature parameters in real time to achieve the coordination of the oil production and transportation processes of the transfer station and the inter-well stations.

[0111] In summary, based on the real-time parameters and historical data of oil gathering and transfer stations and inter-well stations in the oilfield within the region, the present invention establishes a daily oil production prediction digital twin model for inter-well stations, establishes a prediction cycle for the daily oil production of the subordinate inter-well stations, predicts the operation status of the transfer stations, and conducts centralized control over the oil gathering and transportation of the transfer stations and inter-well stations in the region, enhancing the functionality of the method;

[0112] By combining the predicted daily oil production value of the physical model, considering the change in geological parameters caused by the change in oil storage during the oilfield collection process, and integrating the predicted daily oil production value of the data-driven model, a broken line graph of the oil production collection prediction cycle of the current inter-well station is constructed. The physical model provides a basic framework and constraints based on the geological situation to ensure that the prediction conforms to the physical characteristics and geological laws of the oil reservoir. The digital model can supplement and correct the physical model, and use the information in the real-time data and historical data to adjust the prediction results to make them more in line with the actual production situation, enhancing the data accuracy of the control adjustment between stations. By accumulating the oil production collection prediction cycles of each inter-well station under the jurisdiction of the transfer station, the oil volume processing cycle of the transfer station itself is clarified. At the same time, the predicted daily oil production values within the cycle are accumulated to obtain the daily processing prediction value, providing basic data for subsequent analysis to plan the production arrangement, resource reserve, etc. of the transfer station in advance, and pairing the transfer stations for the anomalies on consecutive dates, and optimizing the operation and maintenance control of the oil production and transportation coordination of the abnormal transfer stations and inter-well stations for non-consecutive date anomalies, which can eliminate the deviation caused by accidental anomalies and avoid unnecessary resource waste.

[0113] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A modularized intelligent operation and maintenance method for oilfield gathering and transportation, characterized in that: The method comprises the following steps: S1. Collect data from the well stations of the oil fields in the region through sensors, combine the current predicted crude oil reserves of the well stations and the crude oil collection situation, establish a digital twin model for predicting daily oil production at the well stations, and obtain the oil collection prediction cycle; S11. Install sensors in the oilfield gathering and transportation well stations under the jurisdiction of the region, including flow sensors, pressure sensors, and temperature sensors, to collect real-time flow, wellhead pressure, and temperature during the crude oil collection process at the well stations; S12. By combining the physical model and the data-driven model, a digital twin model of the well stations under the jurisdiction of the region is constructed. The physical model provides a basic description of the oilfield gathering and transportation process. The data-driven model uses historical and real-time data to optimize the parameters of the digital twin model and obtain the oil volume collection prediction cycle. S121. Collect daily oil production, wellhead pressure, oil well production, and oil well production history data of different well stations in the region, and build a three-dimensional geological model of the reservoir in combination with geological data of different well stations to describe the distribution, structure, and properties of the reservoir. Use Eclipse to simulate the flow process of fluids in the reservoir, and simulate theoretical reservoir pressures and physical model oil production prediction values ​​for different time periods. S122. Based on the historical production data of the well station, a data-driven model is constructed to optimize the parameters of the physical model. The historical production data of the well station is collected and normalized, including the historical daily oil production, oil well pressure, temperature, water content, equipment operation time, valve opening, and the Pearson correlation coefficient between each feature and the daily oil production is calculated. The algorithm formula is: Among them, x i For different eigenvalues, y i is the daily oil production, and represent the mean of the feature and daily oil production, respectively, n represents the number of samples, and the features with |r|>0.5 are retained as correlation features; Based on the correlation characteristics, a linear regression equation is constructed as a data-driven model to obtain the predicted value of daily oil production: y=β0+β1x1+β2x2+...+β p x p +e; Among them, β0 is the intercept, β p represents the coefficients of different characteristics, ε is the error term, y is the predicted value of oil production on that day, and x p represents the characteristic parameter value of day p; Use historical data to train the data-driven model and optimize the model's hyperparameters. The specific steps are: The historical data is divided into a training set and a validation set, where the training set contains 70% of the historical data and the validation set contains 30% of the historical data. The mean square error is used as the loss function: in, Represent the predicted value at time t and the actual daily oil production at time t, respectively. n represents the number of samples. The model coefficients are iteratively updated by the gradient descent method. Among them, α represents the learning rate, the model is verified by the validation set, and the mean absolute error is used to evaluate. Represents the coefficient value at time t+1; S123. Based on the weight ratio, the predicted value of oil production from the comprehensive physical model and the predicted value of oil production from the data-driven model are used to obtain the output value of oil production from the digital twin model: Y = ω1 × A + ω2 × B; Among them, Y, A, and B represent the current daily oil production output value of the digital twin model of the well station, the daily oil production prediction value of the physical model, and the daily oil production prediction value of the data-driven model, respectively. ω1 and ω2 represent the weight ratio, ω1+ω2=1; S2. For the inter-well stations under the jurisdiction of the oil transfer station, according to the oil volume collection prediction cycle of different inter-well stations and the transfer capacity of the oil transfer station, the inter-well stations connected to the oil transfer stations in the region are deployed, and the AI ​​algorithm is used to optimize the liquid volume, pressure, and temperature parameters in real time to achieve coordination of the production and transportation process of the oil transfer station and the inter-well stations.

2. The modular-based intelligent operation and maintenance method for oilfield gathering and transportation according to claim 1 is characterized in that: The S123 comprises the following steps: S1231, based on the current daily digital twin model oil production output value Y of the well station, establish an oil production output value set {Y1, Y2, Y3, ..., Y N }, with the date as the horizontal axis and the daily oil production output value of the digital twin model of the inter-well station as the vertical axis, a line chart of the current inter-well station oil production collection prediction period is drawn.

3. The modular-based intelligent operation and maintenance method for oilfield gathering and transportation according to claim 1 is characterized in that: The S2 comprises the following steps: S21. Oil volume collection prediction cycle of each well station under the jurisdiction of the oil transfer station in the collection area Where e represents the date, u represents the number of inter-well stations under the current oil transfer station, and the current oil volume processing cycle of the oil transfer station is obtained by statistics, and coordinated optimization is carried out in combination with the processing capacity of the subordinate oil transfer stations; S22. For well stations and oil transfer stations that cannot be deployed, AI algorithms are used to coordinate and optimize the production and transportation process while meeting the transfer capacity of the oil transfer stations and the production requirements of each well station.

4. The modularized intelligent operation and maintenance method for oilfield gathering and transportation according to claim 3 is characterized in that: The S21 comprises the following steps: S211, the oil volume collection prediction cycle of each well station under the oil transfer station is accumulated to obtain the oil volume processing cycle of the oil transfer station: The predicted oil volume of each well station under the daily oil transfer station in the cycle is accumulated to obtain the predicted oil volume processing value of the daily oil transfer station Where e represents the date, and q is the number of the current oil transfer station; S212, based on the maximum value c of the oil volume processing of the current oil transfer station, count the abnormal date: when Indicates that the oil volume at the transfer station is normal on the current date and no allocation is required; when Indicates that the oil volume at the transfer station is abnormal on the current date and needs to be allocated; S213. Count the abnormal dates of oil volume processing at all oil transfer stations. When the abnormal dates are consecutive dates, calculate the daily deficit value. At the same time, the surplus value of the normal oil transfer stations under its jurisdiction is calculated Where f is the abnormal date, when it satisfies At the same time, the oilfield gathering and transportation pipelines of the two current oil transfer stations will be re-planned.

5. The modular-based intelligent operation and maintenance method for oilfield gathering and transportation according to claim 4 is characterized in that: The S22 comprises the following steps: S221, when the abnormal date is a discontinuous date, the current oil transfer station is recorded as an oil transfer station that cannot be deployed, and the production and transportation parameters of the oil transfer station and the inter-well station are adjusted through a multi-layer perceptron, and the steps are as follows: The difference data between the historical oil volume processing value of the oil transfer station and the oil volume collection value of the inter-well station is collected as the input of the input layer. After the nonlinear transformation of the hidden layer, the output layer outputs the adjusted production and transportation parameters, including the oil production rate adjustment value of the inter-well station and the valve opening adjustment value of the oil transfer station. The model is trained using historical data, and the weight and bias parameters of the model are adjusted by minimizing the objective function. S222. On the abnormal date, the difference between the oil transfer station and the subordinate inter-well stations on that day is input into the multi-layer perceptron, and the model outputs the adjusted production and transportation parameters, including the oil production rate adjustment value of the inter-well station and the valve opening adjustment value of the oil transfer station.

6. The modularized intelligent operation and maintenance system for oilfield gathering and transportation is characterized by: The system adopts the modularized intelligent operation and maintenance method for oilfield gathering and transportation as described in any one of claims 1 to 5, including a data acquisition module, a digital twin model establishment module, a cycle prediction module, and a site deployment module: The data acquisition module includes a flow sensor, a pressure sensor, and a temperature sensor, which are used to collect real-time flow, wellhead pressure, and temperature during the crude oil collection process of the well station, and collect historical operation data of each well station and transmit it to the subsequent module; The digital twin model building module constructs a digital twin model of the well station under the jurisdiction of the region through the combination of physical model and data-driven model, provides a basic description of the oilfield gathering and transportation process through the physical model, and optimizes the parameters of the digital twin model by using historical and real-time data through the data-driven model; The cycle prediction module establishes a set of oil production output values ​​within N days based on the current daily digital twin model oil production output value of the inter-well station, takes the date as the horizontal coordinate and the daily digital twin model oil production output value of the inter-well station as the vertical coordinate, and draws a line chart of the current inter-well station oil production collection prediction cycle; The site allocation module allocates the inter-well stations connected to the oil transfer stations in the region according to the oil volume collection prediction cycle of different inter-well stations and the transfer capacity of the oil transfer stations. At the same time, it uses AI algorithms to optimize the liquid volume, pressure, and temperature parameters in real time to achieve coordination of the production and transportation processes of the oil transfer stations and inter-well stations.

Images