Shale fracturing casing deformation machine learning prediction method based on mechanism model

Abstract

The invention relates to the technical field of shale oil casing deformation prediction, in particular to a shale fracturing casing deformation machine learning prediction method based on a mechanism model, and the method comprises the steps: obtaining a target to-be-processed simulation data set meeting an acquisition standard; preprocessing the target to-be-processed simulation data set, correcting each to-be-processed simulation feature set according to a preprocessing result to obtain a target actual simulation data set, and performing initial numerical simulation on each actual simulation feature set, target shale oil well reservoir slip casing variables under different influence factors are determined according to the initial numerical simulation result; obtaining influence factors which cannot be measured actually to obtain a plurality of invisible simulation feature sets, and performing inversion based on each invisible simulation feature set so as to optimize the initial numerical simulation result to obtain an actual numerical simulation result; and performing prediction training and prediction testing on the actual numerical simulation result based on the selected initial prediction standard to optimize the initial prediction standard so as to obtain an actual prediction standard.

Description

Technical Field

[0001] The present invention relates to the technical field of shale oil casing change prediction, and in particular to a shale fracturing casing change machine learning prediction method based on a mechanism model. Background Art

[0002] During the fracturing process of the horizontal reservoir section of a shale oil well, the induced stress at the tip of the hydraulic fracture changes the original ground stress state around the weak structural surface, thereby activating the weak structural surface within a certain probability, causing the weak structural surface to slip and fail, and triggering shear-type casing deformation. After the casing deformation occurs, the fracturing string cannot be lowered to the designated position, resulting in the abandonment of fracturing transformation in some sections of the well. Therefore, the casing deformation problem seriously affects the efficient development of shale oil. At present, for the casing change phenomenon caused by the fracturing process of shale oil wells, most of the numerical simulation methods are used to solve the casing variables caused by slippage. In the modeling process, in order to achieve more accurate results, three-dimensional modeling is often used to simulate the relative position relationship between the actual weak structural surface and the wellbore; but in the three-dimensional modeling process, there are the following problems: ① The relative position of the wellbore and the weak structural surface is not fixed, and the model needs to be rebuilt each time the casing variables are calculated, which is a cumbersome process; ② The solution process of the three-dimensional numerical model is complicated and the solution time is long; ③ When the initial and boundary conditions of the solution are changed, the model may not converge, resulting in failure to solve; ④ The geometric structure of the weak structural surface is complex, and the grid may not be constructed; In summary, although the use of numerical simulation to directly solve the casing change risk and casing variables under different engineering and geological parameters can provide more accurate results, it is impossible to perform large-scale parameter calculations, and the solution time is long and the error rate is high; therefore, it is particularly important to realize an efficient and universal shale oil well casing change prediction method.

[0003] Chinese patent application publication number CN114878340A discloses a numerical simulation method for shale hydraulic fracturing based on PFC discrete elements, which includes the following steps: Step 1: Establish an external wall unit, generate a particle unit in the wall, construct a circular area at the center, delete the particle unit, and establish a liquid injection wellbore; according to the stress-strain curve output by comparing and analyzing the uniaxial compression test results, the Brazilian splitting test results, the uniaxial compression simulation results, and the Brazilian splitting simulation results, calibrate the shale microscopic parameters; assign basic physical property parameters to the particle unit, bond adjacent particle units through parallel bonding units, and generate a discrete element numerical model of the shale reservoir; Step 2: Set adjacent particles The area enclosed by the units is the fluid domain, the vertical contact point of the particle units is set as the fluid pipeline, and a network flow model is established on the discrete element numerical model of the shale reservoir; Step three: apply confining pressure to the shale boundary wall, set the velocity of the fluid injected into the wellbore, and perform numerical simulation calculations of seepage-displacement-stress coupling; Step four: monitor the bonding state between the particle units in real time to determine whether the bonding unit is damaged. When the stress borne by the bonding unit is greater than the normal strength or tangential strength of the bonding unit, the bonding unit is in a damaged state and is considered to have microcracks; calculate the number of cracks and the expansion morphology of hydraulic fractures; after the calculation is completed, export and save the simulation results of the expansion morphology of hydraulic fractures in the shale reservoir.

[0004] It can be seen that the current shale oil casing change prediction technology cannot achieve efficient and universal shale oil well casing change prediction. Summary of the invention

[0005] To this end, the purpose of the present invention is to provide a shale fracturing casing change machine learning prediction method based on a mechanism model, which is used to overcome the problem that the current shale oil casing change prediction technology cannot achieve efficient and universal shale oil well casing change prediction.

[0006] To achieve the above object, the present invention provides a shale fracturing casing deformation machine learning prediction method based on a mechanism model, comprising: Based on the shale oil reservoir characteristics, a target to-be-processed simulation data set of a target shale oil reservoir meeting the acquisition criteria is obtained, wherein the target to-be-processed simulation data set includes a plurality of to-be-processed simulation feature sets; Preprocessing the target to-be-processed simulation data set to obtain a preprocessing result, and modifying each to-be-processed simulation feature set according to the preprocessing result to obtain a target actual simulation data set, wherein the target actual simulation data set includes a plurality of actual simulation feature sets; Performing initial numerical simulation on each actual simulation feature set to obtain initial numerical simulation results, and determining sliding sleeve variables of target shale oil well reservoirs under different influencing factors according to the initial numerical simulation results; Acquire influencing factors that cannot be measured in practice to obtain a plurality of invisible simulation feature sets, and perform inversion based on each invisible simulation feature set to optimize the initial numerical simulation result to obtain an actual numerical simulation result; Based on the selected initial prediction standard, the actual numerical simulation results are sequentially subjected to prediction training and prediction testing to optimize the initial prediction standard, so as to obtain the actual prediction standard.

[0007] Furthermore, the process of obtaining a target to-be-processed simulation data set of a target shale oil reservoir that meets the acquisition criteria includes: Based on the shale oil reservoir characteristics, obtaining each initial simulation feature set of the target shale oil reservoir; For any initial simulation feature set, obtaining actual acquisition status of the initial simulation feature set; Marking the data according to the actual acquisition situation; A single acquisition situation is determined based on the data marking result, and according to the single acquisition situation, it is determined to start an overall analysis mode, or to start a re-acquisition mode, or to start a pre-processing mode.

[0008] Furthermore, the process of determining a single collection situation based on the data marking result includes: Identify erroneous data, missing data and abnormal data in the actual acquisition situation, and clean and process the initial simulation feature set accordingly based on the identification result to obtain an intermediate simulation feature set; Marking the erroneous data, the missing data and the abnormal data by type, and obtaining the number of times each type is marked; The abnormality evaluation level of the initial simulation feature set is determined according to the type marking result, and a preliminary judgment is made on whether the initial simulation feature set meets the collection standard based on the abnormality evaluation level.

[0009] Furthermore, the process of determining a single collection situation based on the data marking result also includes: Based on the initial judgment result and the number of times of each type of marking, a collection evaluation value is calculated, and a secondary judgment is performed to determine whether the intermediate simulation feature set meets the collection standard according to the collection evaluation value; if it meets the standard, the intermediate simulation feature set is determined to be the simulation feature set to be processed; Among them, the type marking results include: error type, missing type and abnormal type.

[0010] Further, the process of determining to start the overall analysis mode, or to start the re-collection mode, or to start the pre-processing mode according to the single collection situation includes: Determine the number of actual qualified sets in the target initial simulation data set that meet the acquisition standard according to the secondary judgment result; Determining whether to start the overall analysis mode, or the re-collection mode, or the pre-processing mode according to the actual number of qualified sets combined with the preset number of standard qualified sets; Among them, each initial simulation feature set constitutes the target initial simulation data set.

[0011] Further, the collection qualified rate is calculated according to the actual number of qualified sets combined with the total number of sets; Determining whether to start the overall analysis mode, or the re-collection mode, or the pre-processing mode according to the first difference absolute value and the preset first evaluation value; Based on the absolute value of the first difference being less than or equal to the first evaluation value, it is determined that the overall analysis mode is turned on; based on the absolute value of the first difference being greater than the first evaluation value and the acquisition pass rate being greater than a preset standard pass rate, it is determined that the preprocessing mode is turned on; based on the absolute value of the first difference being greater than the first evaluation value and the acquisition pass rate being less than the standard pass rate, it is determined that the re-acquisition mode is turned on; The first absolute value of the difference is the absolute value of the difference between the acquisition qualified rate and the standard qualified rate.

[0012] Further, based on starting the overall analysis mode, the qualified number of each to-be-processed simulation feature set that meets the acquisition standard in the target initial simulation data set and the unqualified number of each abnormal simulation feature set that does not meet the acquisition standard in the target initial simulation data set are integrated; An overall abnormality score is determined according to the integration result, and the re-collection mode or the pre-processing mode is judged to be started according to the overall abnormality score.

[0013] Furthermore, the process of preprocessing the target simulated data set to be processed to obtain a preprocessing result includes: Analyze the first actual correlation between each to-be-processed simulation feature set, and perform feature screening on the target to-be-processed simulation data set according to the first actual correlation to obtain the target to-be-processed simulation data set; Wherein, the target data set to be simulated includes a plurality of feature sets to be simulated; For any feature set to be simulated, the secondary actual correlation between the feature parameters in the feature set to be simulated is analyzed, and the feature set to be simulated is screened according to the secondary actual correlation to obtain the actual simulation feature set.

[0014] Furthermore, the process of determining the sliding sleeve variables of the target shale oil well reservoir under different influencing factors according to the initial numerical simulation results includes: Determine actual geometric parameters and induced stress field distribution during hydraulic fracturing based on each actual simulation feature set; Based on the actual geometric parameters and the induced stress field distribution, determine the numerical range of the influencing factors causing the slip of the weak structural surface and the slip amount of the weak structural surface under different influencing factors; The sliding sleeve variable of the target shale oil well reservoir is determined based on the numerical range of the influencing factors of the slip of the weak structural surface and the slip amount of the weak structural surface under different influencing factors.

[0015] Furthermore, the process of optimizing the initial numerical simulation results by inverting based on each stealth simulation feature set to obtain actual numerical simulation results includes: Based on history matching, inverting each stealth simulation feature set to obtain the actual numerical simulation result; Comparing the actual numerical simulation results with the obtained actual numerical calculation results; The initial prediction standard is optimized according to the comparison result, or historical matching is performed again.

[0016] Compared with the prior art, the beneficial effects of the present invention are that, by collecting data based on the characteristics of shale oil reservoirs, the high quality and pertinence of the target simulation data set to be processed are ensured, laying a solid foundation for subsequent simulation and prediction; the accuracy and consistency of the data are improved, and the accuracy of subsequent numerical simulation is further improved; the sliding casing variables of shale oil well reservoirs under different influencing factors are considered, and various situations that may occur in the actual fracturing process can be fully simulated, which is helpful to understand the complexity and diversity of casing change phenomena; by optimizing the initial numerical simulation results, the problem of influencing factors that cannot be directly measured in actual engineering is solved, so that the simulation results are closer to the actual situation; by repeatedly optimizing the actual numerical simulation results, a reliable prediction standard is formed; not only the accuracy and efficiency of the prediction are improved, but also the intelligence and automation of the prediction process are realized; by accurately predicting the casing change risks and casing change variables that may occur during the fracturing of shale oil wells, it is helpful to formulate prevention and control measures in advance, reduce the failure of fracturing transformation and abandonment of well sections caused by casing change, thereby significantly reducing development costs and risks; through efficient numerical simulation and intelligent prediction optimization, strong support is provided for the efficient development of shale oil, which is helpful to improve the overall benefits and competitiveness of shale oil mining.

[0017] By dividing the influencing factors in detail based on the characteristics of shale oil reservoirs and obtaining each initial simulation feature set separately, the comprehensiveness and accuracy of the simulation data can be ensured; the refined data collection method helps to more realistically reflect the complexity and diversity of shale oil reservoirs and provide a reliable basis for subsequent analysis; by marking the actual acquisition of the initial simulation feature set, and flexibly choosing to start the overall analysis mode, re-collection mode or pre-processing mode according to the marking results, the data processing process is more flexible and efficient; the adaptive data processing mechanism can effectively deal with data quality issues in different situations and improve the efficiency and accuracy of data processing; by pre-determining the influencing factors that need to be numerically simulated and obtaining each initial simulation feature set in a targeted manner, it can Avoid blindly collecting and processing irrelevant data, thereby reducing the cost of data collection and processing; help reduce duplication of work and waste of resources caused by data quality issues; based on an accurate and comprehensive simulation data set, a more accurate and reliable shale fracturing casing deformation prediction model can be established; through the application of machine learning algorithms such as random forests, the potential laws and correlations in the data can be deeply explored, and the prediction accuracy and generalization ability of the model for shale fracturing casing deformation phenomena can be improved; through the data acquisition and processing method described in this embodiment, the casing deformation risk in the shale fracturing process can be more accurately predicted, providing a scientific basis for the design of fracturing schemes in actual production; help reduce fracturing transformation failures and cost waste caused by casing deformation, and improve the development efficiency and economic benefits of shale oil resources.

[0018] By identifying and processing erroneous data, missing data and abnormal data, the accuracy and completeness of the data are ensured; it helps to reduce analytical bias or wrong decisions caused by data quality problems; by introducing the calculation of abnormal evaluation levels and collection evaluation values, the quality of the data set is evaluated more finely; the hierarchical evaluation mechanism can give more accurate data quality judgments according to different error types and error degrees, thereby guiding the subsequent data processing or decision-making process; according to the specific total amount of data and the accuracy of collection, the method is adjusted to have high flexibility and adaptability, automatically identify and process errors and anomalies in the data, reduce the need for manual intervention, and improve the efficiency and accuracy of data processing; the automated evaluation process also reduces the subjectivity and inconsistency of human judgment; on the basis of the initial judgment, a secondary judgment is made by calculating the collection evaluation value, which further improves the accuracy of data quality evaluation; the multi-level and multi-step evaluation process helps to ensure that the final data set meets the strict collection standards; after strict quality evaluation and screening, the final data set has a higher credibility, providing a solid data foundation for subsequent data analysis and decision-making; through the refined data quality evaluation process and automated processing mechanism, the accuracy and credibility of the data set are effectively improved, providing strong support for data analysis and decision-making.

[0019] By calculating the collection pass rate and comparing it with the preset standard pass rate, the most appropriate subsequent processing mode can be intelligently selected; unnecessary repeated collection is avoided, thereby optimizing the utilization of data processing resources; for data sets that meet the collection standards, the preprocessing mode is directly turned on, and the preprocessing stage can be quickly entered, shortening the data processing cycle; for data sets that require overall analysis, overall processing is performed to improve the pertinence and efficiency of data processing; for data sets with a collection pass rate lower than the standard, the re-collection mode is selected to ensure that the data used in subsequent analysis has high quality and credibility; it helps to reduce analytical deviations or wrong decisions caused by data quality issues; it realizes the automatic selection of data processing modes, reduces the subjectivity and inconsistency of manual decision-making, and improves the accuracy and efficiency of decision-making; the automated processing flow reduces the need for human intervention and reduces the risk of data processing errors or omissions due to improper human operation; targeted processing of data according to different data processing modes can improve the accuracy and effectiveness of data processing, and provide more reliable data support for subsequent data analysis and decision-making.

[0020] By numbering and scoring qualified and unqualified simulated feature sets in detail, the overall abnormality of the data can be evaluated more comprehensively; the scoring method based on detailed data can better reflect the actual situation of the data than a single collection pass rate, thereby improving the accuracy of the subsequent processing mode selection; by calculating the overall abnormality score and intelligently selecting the subsequent data processing mode, the data processing process can be further optimized; for data sets with a low overall abnormality, the preprocessing mode is directly used for targeted processing; for data sets with a high overall abnormality, the re-collection mode is selected to ensure data quality; it helps to reduce unnecessary processing steps and improve data processing efficiency; for data sets with a high overall abnormality score, the re-collection mode can ensure that the data used in subsequent analysis has high quality and credibility; it helps to reduce analytical deviations or wrong decisions caused by data quality problems and improve the accuracy and reliability of data analysis; numbering qualified simulated feature sets and unqualified simulated feature sets helps to quickly locate and track problems in subsequent data processing or analysis; detailed records also help to analyze potential problems in the data collection process and provide a basis for subsequent collection process optimization.

[0021] By calculating the first actual correlation and the second actual correlation, and performing feature screening based on these correlations, redundant and irrelevant features can be effectively removed, and the features that have the greatest impact on the prediction target or research problem can be retained; this helps reduce data noise and improve the quality and purity of data; in machine learning or statistical modeling, using high-quality data sets for training and prediction can usually significantly improve the performance of the model; by removing duplicate related features through the preprocessing process, more focus can be placed on features that really affect the prediction results, thereby improving the accuracy and generalization ability of the model; by reducing the dimension of the data set through feature screening, the computational complexity and storage requirements can be reduced, making data processing and model training more efficient; in a data-based decision support system, preprocessed data can provide decision makers with more accurate and useful information; by removing duplicate related features, the decision-making process can be made more focused and efficient, thereby improving the quality and speed of decision-making; the quality of the data is significantly improved, while reducing the computational cost and improving the efficiency of decision-making.

[0022] By adopting machine learning methods, we have solved the problems of traditional shale oil well casing change prediction, such as the lack of actual data, high difficulty in modeling, complex and time-consuming calculation process, and the calculation results of casing variables being greatly affected by the degree of modeling refinement. This has achieved the goal of predicting casing variables quickly, conveniently, accurately, and with high universality, providing a new method for predicting shale oil well fracturing casing changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a flow chart of a machine learning prediction method for shale fracturing casing deformation based on a mechanism model according to an embodiment of the present invention; Figure 2 It is a flow chart of obtaining a target simulation data set to be processed in a shale fracturing casing change machine learning prediction method based on a mechanism model in an embodiment of the present invention; Figure 3 It is a flow chart of compliance judgment in a shale fracturing casing deformation machine learning prediction method based on a mechanism model in an embodiment of the present invention; Figure 4 This is a preprocessing flow chart of a shale fracturing casing deformation machine learning prediction method based on a mechanism model in an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0025] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0026] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0027] See also Figure 1 As shown, the present invention provides a machine learning prediction method for shale fracturing casing deformation based on a mechanism model, comprising: Step S100, based on the shale oil reservoir characteristics, obtaining a target to-be-processed simulation data set of a target shale oil reservoir that meets the acquisition criteria, wherein the target to-be-processed simulation data set includes a plurality of to-be-processed simulation feature sets; Step S200, preprocessing the target to-be-processed simulation data set to obtain a preprocessing result, and modifying each to-be-processed simulation feature set according to the preprocessing result to obtain a target actual simulation data set, wherein the target actual simulation data set includes a plurality of actual simulation feature sets; Step S300, performing initial numerical simulation on each actual simulation feature set to obtain initial numerical simulation results, and determining sliding sleeve variables of a target shale oil well reservoir under different influencing factors according to the initial numerical simulation results; Step S400, obtaining influencing factors that cannot be measured in practice to obtain a plurality of invisible simulation feature sets, and performing inversion based on each invisible simulation feature set to optimize the initial numerical simulation result to obtain an actual numerical simulation result; Step S500, based on the selected initial prediction standard, the actual numerical simulation results are sequentially subjected to prediction training and prediction testing to optimize the initial prediction standard, so as to obtain the actual prediction standard.

[0028] Specifically, the embodiment of the present invention ensures the high quality and pertinence of the target simulation data set to be processed by data collection based on the characteristics of shale oil reservoirs, laying a solid foundation for subsequent simulation and prediction; improves the accuracy and consistency of data, and further improves the accuracy of subsequent numerical simulation; considers the sliding casing variables of shale oil well reservoirs under different influencing factors, can comprehensively simulate various situations that may occur in the actual fracturing process, and helps to understand the complexity and diversity of casing change phenomena; by optimizing the initial numerical simulation results, the problem of influencing factors that cannot be directly measured in actual engineering is solved, so that the simulation results are closer to the actual situation; through repeated optimization of the actual numerical simulation results, a reliable prediction standard is formed; not only the accuracy and efficiency of the prediction are improved, but also the intelligence and automation of the prediction process are realized; by accurately predicting the casing change risks and casing change variables that may occur during the fracturing of shale oil wells, it is helpful to formulate prevention and control measures in advance, reduce the failure of fracturing transformation and abandonment of well sections caused by casing change, thereby significantly reducing development costs and risks; through efficient numerical simulation and intelligent prediction optimization, it provides strong support for the efficient development of shale oil, which helps to improve the overall benefits and competitiveness of shale oil mining.

[0029] See also Figure 2 As shown, it is a flow chart for obtaining a target simulation data set to be processed; wherein, the process of obtaining a target simulation data set to be processed of a target shale oil reservoir that meets the acquisition criteria includes: Step S110, based on the shale oil reservoir characteristics, obtaining each initial simulation feature set of the target shale oil reservoir; Step S120, for any initial simulation feature set, obtaining actual acquisition status of the initial simulation feature set; Step S130, marking data according to the actual acquisition situation; Step S140, determining a single acquisition situation based on the data marking result, and determining to start an overall analysis mode, or to start a re-acquisition mode, or to start a pre-processing mode according to the single acquisition situation.

[0030] Specifically, in this embodiment, according to the characteristics of the shale oil reservoir, the following influencing factors that need to be numerically simulated are determined; the influencing factors are divided into eight categories: ① fracturing construction parameters; ② fracturing fluid and proppant parameters; ③ formation stress environment parameters: ④ formation physical property parameters; ⑤ weak structural surface characteristic parameters; ⑥ casing parameters; ⑦ cement ring parameters; ⑧ wellbore structure parameters; therefore, the initial simulation feature sets of the target shale oil reservoir are obtained, including: ① initial simulation fracturing construction set; ② initial simulation fracturing fluid and proppant set; ③ initial simulation formation stress environment set: ④ initial simulation formation physical property set; ⑤ initial simulation weak structural surface characteristic set; ⑥ initial simulation casing set; ⑦ initial simulation cement ring set; ⑧ initial simulation wellbore structure set.

[0031] Specifically, the embodiments of the present invention can ensure the comprehensiveness and accuracy of simulation data by dividing the influencing factors in detail based on the characteristics of shale oil reservoirs and obtaining each initial simulation feature set respectively; the refined data collection method helps to more realistically reflect the complexity and diversity of shale oil reservoirs and provide a reliable basis for subsequent analysis; by marking the actual acquisition of the initial simulation feature set, and flexibly choosing to start the overall analysis mode, re-collection mode or pre-processing mode according to the marking results, the data processing process is made more flexible and efficient; the adaptive data processing mechanism can effectively deal with data quality problems in different situations and improve the efficiency and accuracy of data processing; by pre-determining the influencing factors that need to be numerically simulated, and obtaining each initial simulation feature set in a targeted manner The data set can avoid blindly collecting and processing irrelevant data, thereby reducing the cost of data collection and processing; it helps to reduce duplication of work and waste of resources caused by data quality issues; based on an accurate and comprehensive simulation data set, a more accurate and reliable shale fracturing casing deformation prediction model can be established; through the application of machine learning algorithms such as random forests, the potential laws and correlations in the data can be deeply explored, and the prediction accuracy and generalization ability of the model for shale fracturing casing deformation phenomena can be improved; through the data acquisition and processing method described in this embodiment, the casing deformation risk in the shale fracturing process can be more accurately predicted, providing a scientific basis for the design of fracturing schemes in actual production; it helps to reduce fracturing transformation failures and cost waste caused by casing deformation, and improve the development efficiency and economic benefits of shale oil resources.

[0032] Specifically, the process of determining a single collection situation based on the data marking result in this embodiment includes: Step S1411, identifying erroneous data, missing data and abnormal data in the actual acquisition situation, and performing corresponding cleaning and processing on the initial simulation feature set based on the identification result to obtain an intermediate simulation feature set; Step S1412, marking the type of the erroneous data, the missing data and the abnormal data, and obtaining the number of times each type is marked; Step S1413, determining the abnormality evaluation level of the initial simulation feature set according to the type marking result, and making a preliminary judgment on whether the initial simulation feature set meets the collection standard based on the abnormality evaluation level.

[0033] Specifically, the process of determining the abnormality evaluation level of the initial simulation feature set according to the recognition result and making a preliminary judgment on whether the initial simulation feature set meets the collection standard based on the abnormality evaluation level in this embodiment is as follows: Analyzing the number and type of items in the recognition results that meet a single determination condition; Determine whether the project types that meet the single determination condition include abnormal types. If yes, determine whether the number of projects is 1. If yes, determine that the abnormal evaluation level is level one, and determine the initial determination result as: the initial simulation feature set meets the collection standard. If no, determine that the abnormal evaluation level is level two, and determine the initial determination result as: start secondary determination. If the item types that meet the single judgment condition do not include abnormal types, determine whether the number of items is equal to 1, if yes, determine that the abnormal evaluation level is level 2, and determine the initial judgment result as: start secondary judgment, if no, determine that the abnormal evaluation level is level 3, and determine the initial judgment result as: the initial simulation feature set does not meet the collection standard; Among them, for any type of marking result, if the absolute value of the marking difference corresponding to any type of marking result is greater than the preset marking evaluation value and the corresponding number of type markings is greater than the preset standard marking number, then it is judged that any type of marking result meets the single judgment condition; the absolute value of the marking difference is the absolute value of the difference between the number of type markings and the standard marking number.

[0034] Specifically, in this embodiment, for any type marking result, the marking difference absolute value S1 is calculated according to the type marking times X and the standard marking times X0 of the type marking result; S1=|X-X0|; If S1≤S10, it is determined that the type marking result does not meet the single determination condition; If S1>S10 and X<X0, it is determined that the type marking result does not meet the single determination condition; If S1>S10 and X>X0, it is determined that the type marking result meets the single determination condition; Wherein, S10 is a preset marking evaluation value; In the specific implementation process, the standard marking times are set to 8; the standard evaluation value is set to 2; the specific settings are also affected by the total amount of data in the initial simulation feature set and the acquisition accuracy.

[0035] Specifically, the process of determining a single collection situation based on the data marking result in this embodiment also includes: Based on the initial judgment result and the number of times of each type of marking, the collection evaluation value is calculated, and the intermediate simulation feature set is judged to be in compliance with the collection standard for secondary judgment according to the collection evaluation value; if it is in compliance, the intermediate simulation feature set is judged to be the simulation feature set to be processed; if it is not in compliance, the intermediate simulation feature set is judged to be an abnormal simulation feature set; Among them, the type marking result includes: the error type, the missing type and the exception type.

[0036] See also Figure 3 As shown, it is a flow chart of compliance determination, wherein the process of calculating the acquisition evaluation value based on the initial determination result combined with the number of times of each type of marking, and determining whether the intermediate simulation feature set meets the acquisition standard for secondary determination based on the acquisition evaluation value is as follows: When the abnormality evaluation level is level 2, determining whether the intermediate simulation feature set meets the acquisition standard according to the second difference absolute value combined with a preset second evaluation value; Determine whether the absolute value of the second difference is less than or equal to the second evaluation value. If, it is less than or equal to, determine that the intermediate simulation feature set meets the acquisition standard. If, it is greater than, determine whether the acquisition evaluation value is less than a preset acquisition standard value. If, it is less than, determine that the intermediate simulation feature set meets the acquisition standard. If, it is greater than, determine that the intermediate simulation feature set does not meet the acquisition standard. The second absolute value of the difference is the absolute value of the difference between the acquisition evaluation value and the acquisition standard value.

[0037] Specifically, in this embodiment, the second difference absolute value S2 is calculated according to the acquisition evaluation value H1 and the preset acquisition standard value H0; S2=|H1-H0|; wherein S20 is a preset second evaluation value; When the initial judgment result is to start the secondary judgment, the acquisition evaluation value H1 is calculated according to the obtained first type mark number A1 of the error type, the second type mark number A2 of the missing type, and the third type mark number A3 of the abnormal type; H1=A1×b1+A2×b2+A3×b3; Wherein, b1 is the first calculation compensation parameter of the first type of marking times A1 for the collection evaluation value H1; b2 is the second calculation compensation parameter of the second type of marking times A2 for the collection evaluation value H1; b3 is the third calculation compensation parameter of the third type of marking times A3 for the collection evaluation value H1; During the specific implementation process, the acquisition standard value H0 is set to 9.6; the second evaluation value is set to 1.7; the first calculation compensation parameter b1 is set to 0.6; the second calculation compensation parameter b2 is set to 0.6; the third calculation compensation parameter b3 is set to 0.8; the specific settings are also affected by the total number of data in the initial simulation feature set and the acquisition accuracy.

[0038] Specifically, the embodiments of the present invention ensure the accuracy and completeness of data by identifying and processing erroneous data, missing data and abnormal data; help reduce analytical deviations or wrong decisions caused by data quality problems; by introducing the calculation of abnormal evaluation levels and collection evaluation values, a more refined evaluation of the quality of the data set is performed; the hierarchical evaluation mechanism can give more accurate data quality judgments based on different error types and error degrees, thereby guiding subsequent data processing or decision-making processes; the method can be adjusted according to the specific total amount of data and collection accuracy, so that the method has high flexibility and adaptability, automatically identifies and processes errors and anomalies in the data, and reduces the need for manual intervention. The automated evaluation process also reduces the subjectivity and inconsistency of human judgment. On the basis of the initial judgment, a secondary judgment is made by calculating the collection evaluation value, which further improves the accuracy of data quality assessment. The multi-level and multi-step evaluation process helps to ensure that the final data set meets the strict collection standards. After strict quality assessment and screening, the final data set has a higher credibility, providing a solid data foundation for subsequent data analysis and decision-making. Through the refined data quality assessment process and automated processing mechanism, the accuracy and credibility of the data set are effectively improved, providing strong support for data analysis and decision-making.

[0039] Specifically, the process of determining to start the overall analysis mode, or to start the re-collection mode, or to start the pre-processing mode according to the single collection situation in this embodiment includes: Step S141, determining the number of actual qualified sets in the target initial simulation data set that meet the acquisition standard according to the secondary judgment result; Step S142, determining whether to start the overall analysis mode, or start the re-collection mode, or start the pre-processing mode according to the actual number of qualified sets combined with the total number of sets; Among them, each initial simulation feature set constitutes the target initial simulation data set.

[0040] Specifically, in this embodiment, the acquisition qualified rate is calculated based on the actual number of qualified sets combined with the total number of sets; The collection qualified rate = the actual number of qualified sets / the total number of sets; Determining whether to start the overall analysis mode, or the re-collection mode, or the pre-processing mode according to the first difference absolute value and the preset first evaluation value; Based on the absolute value of the first difference being less than or equal to the first evaluation value, it is determined that the overall analysis mode is turned on; based on the absolute value of the first difference being greater than the first evaluation value and the acquisition pass rate being greater than a preset standard pass rate, it is determined that the preprocessing mode is turned on; based on the absolute value of the first difference being greater than the first evaluation value and the acquisition pass rate being less than the standard pass rate, it is determined that the re-acquisition mode is turned on; Wherein, the first difference absolute value is the absolute value of the difference between the acquisition qualified rate and the standard qualified rate; the first difference absolute value=|acquisition qualified rate-standard qualified rate|; In this embodiment, the standard qualification rate is set to 80%-85%; the first evaluation value is set to 3%-5%; the specific setting is also affected by the total amount of data in the initial simulation feature set and the acquisition accuracy.

[0041] Specifically, the embodiment of the present invention can intelligently select the most appropriate subsequent processing mode by calculating the collection pass rate and comparing it with the preset standard pass rate; avoid repeated collection when it is not necessary, thereby optimizing the utilization of data processing resources; for data sets that meet the collection standards, directly start the preprocessing mode, can quickly enter the preprocessing stage, and shorten the data processing cycle; for data sets that require overall analysis, perform overall processing to improve the pertinence and efficiency of data processing; for data sets with a collection pass rate lower than the standard, select a re-collection mode to ensure that the data used in subsequent analysis has high quality and credibility; it helps to reduce analysis deviations or wrong decisions caused by data quality problems; realizes the automatic selection of data processing modes, reduces the subjectivity and inconsistency of manual decision-making, and improves the accuracy and efficiency of decision-making; the automated processing flow reduces the need for human intervention and reduces the risk of data processing errors or omissions due to improper human operation; according to different data processing modes, targeted processing of data can improve the accuracy and effectiveness of data processing, and provide more reliable data support for subsequent data analysis and decision-making.

[0042] Specifically, in this embodiment, based on turning on the overall analysis mode, the qualified number of each to-be-processed simulation feature set that meets the acquisition standard in the target initial simulation data set and the unqualified number of each abnormal simulation feature set that does not meet the acquisition standard in the target initial simulation data set are integrated; An overall abnormality score is determined according to the integration result, and the re-collection mode or the pre-processing mode is judged to be started according to the overall abnormality score.

[0043] Specifically, the target initial simulation data set is integrated to obtain each set of to-be-processed simulation features that meet the acquisition standard and calculate the qualified score, and each set of abnormal simulation features that do not meet the acquisition standard and calculate the unqualified score, including: 1) Number each qualified simulation feature set to be processed, and record it as the first qualified set W11, the second qualified set W12, ... the mth qualified set W1m, 2) Number each unqualified abnormal simulation feature set, and record it as the first abnormal set W21, the second abnormal set W22, ... the nth abnormal set W2n, Determine the overall anomaly score K based on the integration results and set K= 1-

[0044] Wherein, m is the qualified number; n is the unqualified number; E1 is the qualified compensation parameter of the qualified number m to the overall abnormality score K; E2 is the unqualified compensation parameter of the unqualified number n to the overall abnormality score K; If K≥K0, it is determined to start the preprocessing mode. If K<K0, it is determined to start the re-collection mode; Among them, K0 is the calculation standard evaluation value.

[0045] In this embodiment, the qualified compensation parameter E1 is set to 0.7; the unqualified compensation parameter E2 is set to 0.8; the specific settings are also affected by the total amount of data in the initial simulation feature set and the acquisition accuracy.

[0046] Specifically, the embodiment of the present invention can more comprehensively evaluate the overall abnormality of the data by numbering and scoring qualified and unqualified simulation feature sets in detail; the scoring method based on detailed data can better reflect the actual situation of the data than a single collection pass rate, thereby improving the accuracy of the subsequent processing mode selection; by calculating the overall abnormality score and intelligently selecting the subsequent data processing mode, the data processing process can be further optimized; for data sets with a lower overall abnormality, the preprocessing mode is directly used for targeted processing; for data sets with a higher overall abnormality, the re-collection mode is selected to ensure data quality; it helps to reduce unnecessary processing steps and improve data processing efficiency; for data sets with a higher overall abnormality score, the re-collection mode can ensure that the data used in subsequent analysis has higher quality and credibility; it helps to reduce analysis deviations or wrong decisions caused by data quality problems and improve the accuracy and reliability of data analysis; numbering qualified simulation feature sets and unqualified simulation feature sets helps to quickly locate and track problems in subsequent data processing or analysis; detailed records also help to analyze potential problems in the data collection process and provide a basis for subsequent collection process optimization.

[0047] See also Figure 4 As shown, it is a preprocessing flow chart; specifically, the process of preprocessing the target simulated data set to be processed to obtain the preprocessing result in this embodiment includes: Step S210, analyzing the first actual correlation between each to-be-processed simulation feature set, and performing feature screening on the target to-be-processed simulation data set according to the first actual correlation to obtain the target to-be-processed simulation data set; Wherein, the target data set to be simulated includes a plurality of feature sets to be simulated; Step S220 , for any feature set to be simulated, analyzing the secondary actual correlation between the feature parameters in the feature set to be simulated, and performing feature screening on the feature set to be simulated according to the secondary actual correlation to obtain the actual simulated feature set.

[0048] Specifically, in this embodiment, the first actual correlation degree refers to the correlation or mutual influence degree between different simulated feature sets to be processed; by analyzing the first actual correlation degree, a feature set that can fully characterize the prediction target is screened out from the target initial simulation data set to form a target data set to be simulated; In this embodiment, the Pearson correlation coefficient method is used to calculate the primary actual correlation between each set of simulated features to be processed; The Pearson correlation coefficient is used to measure the linear correlation between two simulated feature sets to be processed, and its value range is -1 to 1; If the actual correlation between any two to-be-processed simulation feature sets is close to 1, it means that the linear relationship between the two to-be-processed simulation feature sets is stronger; if it is closer to 0, it means that the linear relationship between the two to-be-processed simulation feature sets is weaker; Calculating the actual correlation degree of each to-be-processed simulation feature set in pairs, so as to screen out the to-be-simulated feature sets with weak linear relationships, so as to obtain the target to-be-simulated data set; For any feature set to be simulated, the Pearson correlation coefficient method is used to calculate the quadratic actual correlation between each feature parameter in the feature set to be simulated; The actual correlation degree of each feature parameter is calculated twice in pairs to screen out the feature parameters with weak linear relationship, so as to obtain the actual simulation feature set.

[0049] The target actual simulation data set in this embodiment includes a plurality of actual simulation feature sets, each of which includes a plurality of feature parameters; ① Actual simulation fracturing operation set: 1.1 injection rate, m 3 / min; 1.2Total injection volume, m 3 ; 1.3 Construction pressure, MPa; 1.4 Construction temperature, ℃; 1.5 Cluster spacing, m; 1.6 Number of clusters.

[0050] ② Actual simulated fracturing fluid and proppant set: 2.1 Fracturing fluid viscosity, mPa·s; 2.2 Proppant concentration, kg / m 3 ; 2.3 Proppant particle size, m; 2.4 Proppant dosage, m 3 .

[0051] ③ Set of actual simulated formation stress environment: 3.1 Maximum horizontal principal stress, MPa; 3.2 Minimum horizontal principal stress, MPa; 3.3 Vertical stress, MPa; 3.4 Formation pore pressure, MPa.

[0052] ④ Actual simulated formation physical properties set: 4.1 Reservoir rock elastic modulus, GPa; 4.2 Reservoir rock Poisson's ratio; 4.3 Reservoir rock permeability, mD; 4.4 Reservoir rock porosity; 4.5 Reservoir rock density, kg / m 3 ; 4.6 Reservoir rock fracture energy release rate; 4.7 Formation temperature, ℃.

[0053] ⑤ The actual simulated weak structural surface characteristics set: 5.1 Weak structural surface inclination, °; ​​5.2 Weak structural surface length, m; 5.3 Curvature of the weak structural surface bend, 1 / m; 5.4 Position of the weak structural surface and the horizontal well, m; 5.5 Distance between the weak structural surface and the fracturing point, m; 5.6 Friction coefficient of the weak structural surface; 5.7 Fracture energy release rate of the weak structural surface.

[0054] ⑥ Actual simulated casing set: 6.1 casing diameter, m; 6.2 casing wall thickness, m; 6.3 casing elastic modulus, GPa; 6.4 casing Poisson's ratio.

[0055] ⑦ Actual simulated cement sheath set: 7.1 Cement sheath thickness, m; 7.2 Cement sheath elastic modulus, GPa; 7.3 Cement sheath Poisson’s ratio.

[0056] ⑧Actual simulated wellbore structure set: 8.1 well diameter, m; 8.2 well inclination, °; ​​8.3 azimuth, °.

[0057] Specifically, the embodiment of the present invention can effectively remove redundant and irrelevant features and retain the features that have the greatest impact on the prediction target or research problem by calculating the primary actual correlation degree and the secondary actual correlation degree and performing feature screening based on these correlation degrees; it helps to reduce data noise and improve the quality and purity of data; in machine learning or statistical modeling, using high-quality data sets for training and prediction can usually significantly improve the performance of the model; by removing repeated related features through the preprocessing process, more focus can be placed on those features that really affect the prediction results, thereby improving the accuracy and generalization ability of the model; by reducing the dimension of the data set through feature screening, the computational complexity and storage requirements can be reduced, making data processing and model training more efficient; in a data-based decision support system, the preprocessed data can provide decision makers with more accurate and useful information; by removing repeated related features, the decision-making process can be made more focused and efficient, thereby improving the quality and speed of decision-making; the quality of the data is significantly improved, while reducing the computational cost and improving the efficiency of decision-making.

[0058] Specifically, the process of determining the sliding sleeve variables of the target shale oil well reservoir under different influencing factors according to the initial numerical simulation results in this embodiment includes: Step S310, determining actual geometric parameters and induced stress field distribution during hydraulic fracturing based on each actual simulation feature set; Step S320, determining the numerical range of the influencing factors causing the slip of the weak structural surface and the slip amount of the weak structural surface under different influencing factors based on the actual geometric parameters and the induced stress field distribution; Step S330, determining the sliding sleeve variable of the target shale oil well reservoir based on the numerical range of the influencing factors of the slip of the weak structural surface and the slip amount of the weak structural surface under different influencing factors.

[0059] The specific process includes: (1) Establishing a numerical model for the expansion of shale reservoir fractures: Calculating the geometric shape, size, and orientation of the fractured body under different influencing factors; wherein the actual geometric parameters include: geometric shape, size, and orientation; at the same time, calculating the distribution of the induced stress field during hydraulic fracturing to provide a numerical basis for calculating the slip amount of the weak structural surface during the fracturing process; the process of establishing the numerical model for the expansion of shale reservoir fractures includes: 1) Establish a three-dimensional geometric model, and control the model size within the range of well depth + 2000m.

[0060] 2) Based on the coordinates of the well, the wellbore is constructed in the geometric model, and the location of the fracturing point is designed based on the cluster spacing and cluster number.

[0061] 3) Construct a grid and assign permeability and porosity attributes to the grid cells.

[0062] 4) Add material property parameters. Including: ① fracturing fluid viscosity, mPa·s; ② proppant particle size, m; ③ proppant concentration, kg / m 3 ; ④ elastic modulus of reservoir rock, GPa; ⑤ Poisson's ratio of reservoir rock; ⑥ permeability of reservoir rock, mD; ⑦ porosity of reservoir rock; ⑧ density of reservoir rock, kg / m 3 ; ⑨Reservoir rock fracture energy release rate.

[0063] 5) Add boundary conditions and initial values. Boundary conditions include: ① injection rate, m 3 / min; ② Construction pressure, MPa; ③ Construction temperature, °C; ④ Proppant dosage, m 3 ; The initial values ​​include: ① formation temperature, °C; ② maximum horizontal principal stress, MPa; ③ minimum horizontal principal stress, MPa; ④ vertical stress, MPa; ⑤ formation pore pressure, MPa.

[0064] 6) Set the solution time step, which is the ratio of the total injection volume to the injection rate.

[0065] 7) Solve the model.

[0066] (2) Establishing a numerical model of weak structural surface slip during fracturing: Calculate the slip amount of the weak structural surface during shale reservoir fracturing under different influencing factors; Use the Mohr-Coulomb criterion to identify the slip of the weak structural surface, judge the slip of the weak structural surface under the induced stress field of the shale reservoir calculated in step (1), and obtain the numerical range of the influencing factors causing the slip of the weak structural surface and the slip amount of the weak structural surface under different influencing factors; The process of establishing a numerical model of weak structural surface slip during fracturing includes: 1) Based on step (1), a parameterized surface is established to characterize the weak structural surface in the shale reservoir according to the geometric coordinates of the weak structural surface. The design parameters include: ① the inclination angle of the weak structural surface, °; ② the length of the weak structural surface, m; ③ the curvature of the bend of the weak structural surface, 1 / m; ④ the position between the weak structural surface and the horizontal well, m; ⑤ the distance between the weak structural surface and the fracturing point, m.

[0067] 2) Assign the material properties of the weak structural surface, the parameters include: ① the friction coefficient of the weak structural surface; ② the fracture energy release rate of the weak structural surface.

[0068] 3) Set the model boundary conditions and initial values, and the design parameters are consistent with the actual simulation data set of the target.

[0069] 4) The model is solved in a steady state to calculate the slip of the weak structural plane under the induced stress field obtained in step (1).

[0070] (3) Establish a numerical model of shale oil well reservoir sliding casing variation, and calculate the shale oil well casing variation under the influencing factors. Under the condition of the slip amount of the weak structural surface of the shale reservoir calculated in step (2) during the fracturing process, calculate the shear stress on the casing. When the shear stress exceeds the shear strength of the casing, the casing undergoes shear deformation, and the casing variation generated when the shear stress exceeds the shear strength is obtained. The process of establishing a numerical model of shale oil well reservoir sliding casing variation includes: 1) Establish a cuboid with a length, width and height of 5m×1m×1m as the casing-cement sheath-stratum combination geometric model, in which a tubular geometric body is added to the model according to the casing and cement sheath geometric parameters. The parameters involved include: ① casing diameter, m; ② casing wall thickness, m; ③ cement sheath thickness, m; 2) Construct a mesh and assign solid mechanics properties to the mesh elements.

[0071] 3) Add material attribute parameters, including: ① elastic modulus of casing, GPa; ② Poisson's ratio of casing; ③ elastic modulus of cement sheath, GPa; ④ Poisson's ratio of cement sheath; ⑤ elastic modulus of reservoir rock, GPa; ⑥ Poisson's ratio of reservoir rock; ⑦ density of reservoir rock, kg / m 3 .

[0072] 4) The slip amount of the weak structural surface calculated in step (2) is used as the displacement boundary condition of the shale oil well reservoir sliding sleeve variable numerical model.

[0073] 5) Solve the model and obtain the sliding sleeve variables of shale oil well reservoirs under different influencing factors.

[0074] Specifically, the process of optimizing the initial numerical simulation results by inversion based on each stealth simulation feature set in this embodiment to obtain the actual numerical simulation results includes: Step S410, inverting each stealth simulation feature set based on history matching to obtain the actual numerical simulation result; Step S420, comparing the actual numerical simulation result with the obtained actual numerical calculation result; Step S430: Determine to optimize the initial prediction standard according to the comparison result, or perform history matching again.

[0075] The specific process includes: (4) Through historical matching, invert the influencing factors that cannot be measured in practice to optimize the numerical model. The parameters that need to be inverted include: ① the energy release rate of reservoir rock fracture; ② the friction coefficient of weak structural surface; ③ the energy release rate of weak structural surface fracture. If there are other parameters without measurement results in the influencing factors in the actual simulation data set of the target in the actual project, it is also necessary to invert through historical matching to optimize the numerical model.

[0076] (5) Using random forest algorithm as the machine learning method, 80% of the variables under different influencing factors obtained in step (3) are used as the training set to establish a shale fracturing variable prediction model.

[0077] (6) Use 20% of the set of variables obtained in step (3) as a test set to optimize the machine learning algorithm.

[0078] (7) After optimization of step (6), a shale fracturing casing deformation machine learning prediction method based on the mechanism model is finally formed.

[0079] The embodiments of the present invention solve the problems of the traditional shale oil well casing change prediction such as the small amount of actual data, the great difficulty in modeling, the complex and time-consuming calculation process, and the casing variable calculation results being greatly affected by the degree of modeling sophistication by adopting a machine learning method. The goal of predicting casing variables quickly, conveniently, accurately, and with high universality is achieved, providing a new method for predicting shale oil well fracturing casing changes. Embodiment 1:

[0080] The structural position of a certain area is located at the junction of the Sichuan Basin and the Yunnan-Guizhou Plateau, between the low-steep structural area of ​​the ancient depression in southern Sichuan and the Loushan fold belt. It is affected by the westward extension of the eastern Sichuan fold-thrust belt in the north and controlled by the evolution of the Loushan fold belt in the south. Its structural characteristics are a structural complex that integrates the two. In this block, the casing deformation of the H platform is relatively serious. The formation of the platform is about 5 degrees downward along the extension direction of the horizontal section; the buried depth of the lower half of the actual drilling horizontal section is 2430~2600m, the formation pressure coefficient is about 1.8, and the dip angle of the lower half of the formation is about 3.0~5.0°. In the H platform, the shear deformation of the casing in the horizontal section of the three horizontal wells X1, X2 and X3 is particularly serious. The casing variable is predicted by the machine learning prediction method of shale fracturing casing deformation based on the mechanism model. The steps are as follows: (1) Establish a numerical model for fracture propagation in shale reservoirs. Solve the geometric shape, size and orientation of the reformed bodies produced by hydraulic fracturing of the three horizontal wells X1, X2 and X3 in the H platform, and calculate the distribution of the induced stress field during hydraulic fracturing.

[0081] A geometric model with the dimensions of 5000m × 5000m × 3000m in length × width × height was established. After the geometric modeling of the three wells X1, X2, and X3 was completed, the material properties and boundary conditions required for numerical solution were assigned according to the actual fracturing data. The parameter range used for the solution is shown in Table 1: Table 1 Parameter selection of numerical model for fracture extension in shale reservoirs

[0082] According to the parameters in Table 1, the geometric shape, size, and orientation of the hydraulic fracture reconstruction body during the fracturing process, as well as the distribution of the induced stress field in the shale reservoir after fracturing, are solved.

[0083] (2) Establish a numerical model of weak structural plane sliding during fracturing. The geometric distribution of weak structural planes and their characteristic parameter distributions are added to the numerical model of shale reservoir fracture propagation. The characteristic parameters of weak structural planes are shown in Table 2.

[0084] Table 2 Selection table of weak structural surface characteristic parameters

[0085] On the basis of the model including weak structural surface, combined with the induced stress field length distribution, the stress distribution at the weak structural surface is solved. The slip of the weak structural surface is judged according to the Mohr-Coulomb criterion, and the numerical range of the influencing factors causing the slip of the weak structural surface and the slip amount of the weak structural surface under different influencing factors are obtained. The slip amount calculation results are shown in Table 3.

[0086] Table 3 Slip amount of weak structural surface under different influencing factors

[0087] (3) Establish a numerical model of sliding casing in shale oil well reservoirs. First, a rectangular block with a length, width and height of 5m×1m×1m is established as a geometric model of the casing-cement ring-formation combination, in which a tubular geometric body is added to the model according to the geometric parameters of the casing and cement ring. The parameters used for the simulation of the casing and cement ring are shown in Table 4: Table 4 Casing and cement ring simulation parameters

[0088] Table 5 Casing variables under different casing and cement sheath parameters

[0089] (4) Through historical matching, the following parameters were inverted from the casing data of the three wells X1, X2, and X3 on the H platform: ① reservoir rock fracture energy release rate; ② weak structural surface friction coefficient; ③ weak structural surface fracture energy release rate. After inversion, the reservoir rock fracture energy release rate was 0.02, the weak structural surface friction coefficient was 0.42, and the weak structural surface fracture energy release rate was 0.003. After the model was corrected by historical matching, the model accuracy was verified. The accuracy verification results are shown in Table 6.

[0090] Table 6. Numerical model accuracy verification

[0091] (5) The random forest algorithm is used as the machine learning method. 80% of the variables under different influencing factors in Table (5) are used as the training set to establish a shale fracturing variable prediction model.

[0092] (6) 20% of the set variables obtained in Table (5) were used as the test set to optimize the machine learning algorithm, and finally a shale fracturing set variable machine learning prediction method based on the mechanism model was formed.

[0093] The calculation compensation parameters and the calculation adjustment parameters described in the present invention have two functions: one is to balance the left and right dimensions of the formula, and the other is to adjust the numerical results. No specific assignment is performed in this embodiment. In addition, the calculation formulas in this embodiment are used to intuitively reflect the adjustment relationship between the numerical values, such as positive correlation and negative correlation. Unless otherwise specified, the parameter values ​​that are not specifically limited to numerical values ​​are all positive.

[0094] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0095] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A machine learning prediction method for shale fracturing casing deformation based on a mechanism model, comprising: Based on the shale oil reservoir characteristics, a target to-be-processed simulation data set of a target shale oil reservoir meeting the acquisition criteria is obtained, wherein the target to-be-processed simulation data set includes a plurality of to-be-processed simulation feature sets; Preprocessing the target to-be-processed simulation data set to obtain a preprocessing result, and modifying each to-be-processed simulation feature set according to the preprocessing result to obtain a target actual simulation data set, wherein the target actual simulation data set includes a plurality of actual simulation feature sets; Performing initial numerical simulation on each actual simulation feature set to obtain initial numerical simulation results, and determining sliding sleeve variables of a target shale oil well reservoir under different influencing factors according to the initial numerical simulation results; Acquire influencing factors that cannot be measured in practice to obtain a plurality of invisible simulation feature sets, and perform inversion based on each invisible simulation feature set to optimize the initial numerical simulation result to obtain an actual numerical simulation result; Based on the selected initial prediction standard, the actual numerical simulation results are sequentially subjected to prediction training and prediction testing to optimize the initial prediction standard, so as to obtain the actual prediction standard.

2. The shale fracturing casing deformation machine learning prediction method based on the mechanism model according to claim 1 is characterized in that: The process of obtaining a target to-be-processed simulation data set of a target shale oil reservoir that meets the acquisition criteria includes: Based on the shale oil reservoir characteristics, obtaining each initial simulation feature set of the target shale oil reservoir; For any initial simulation feature set, obtaining actual acquisition status of the initial simulation feature set; Marking the data according to the actual acquisition situation; A single acquisition situation is determined based on the data marking result, and according to the single acquisition situation, it is determined to start an overall analysis mode, or to start a re-acquisition mode, or to start a pre-processing mode.

3. The shale fracturing casing deformation machine learning prediction method based on the mechanism model according to claim 2 is characterized in that: The process of determining a single acquisition situation based on the data marking result includes: Identify erroneous data, missing data and abnormal data in the actual acquisition situation, and clean and process the initial simulation feature set accordingly based on the identification result to obtain an intermediate simulation feature set; Marking the erroneous data, the missing data and the abnormal data by type, and obtaining the number of times each type is marked; The abnormality evaluation level of the initial simulation feature set is determined according to the type marking result, and a preliminary judgment is made on whether the initial simulation feature set meets the collection standard based on the abnormality evaluation level.

4. The shale fracturing casing deformation machine learning prediction method based on the mechanism model according to claim 3 is characterized in that: The process of determining a single acquisition situation based on the data marking result also includes: Based on the initial judgment result and the number of times of each type of marking, a collection evaluation value is calculated, and a secondary judgment is performed to determine whether the intermediate simulation feature set meets the collection standard according to the collection evaluation value; if it meets the standard, the intermediate simulation feature set is determined to be the simulation feature set to be processed; Among them, the type marking results include: error type, missing type and abnormal type.

5. The shale fracturing casing deformation machine learning prediction method based on the mechanism model according to claim 4 is characterized in that: The process of determining to start the overall analysis mode, or to start the re-collection mode, or to start the pre-processing mode according to the single collection situation includes: Determine the number of actual qualified sets in the target initial simulation data set that meet the acquisition standard according to the secondary judgment result; Determining whether to start the overall analysis mode, or the re-collection mode, or the pre-processing mode according to the actual number of qualified sets combined with the preset number of standard qualified sets; Among them, each initial simulation feature set constitutes the target initial simulation data set.

6. The shale fracturing casing deformation machine learning prediction method based on the mechanism model according to claim 5 is characterized in that: Calculate the collection pass rate according to the actual number of qualified sets combined with the total number of sets; Determining whether to start the overall analysis mode, or the re-collection mode, or the pre-processing mode according to the first difference absolute value and the preset first evaluation value; Based on the absolute value of the first difference being less than or equal to the first evaluation value, determining to start the overall analysis mode; Based on the fact that the absolute value of the first difference is greater than the first evaluation value and the acquisition qualification rate is greater than a preset standard qualification rate, determining to start the preprocessing mode; Based on the fact that the absolute value of the first difference is greater than the first evaluation value and the acquisition qualification rate is less than the standard qualification rate, determining to start the re-acquisition mode; The first absolute value of the difference is the absolute value of the difference between the acquisition qualified rate and the standard qualified rate.

7. The shale fracturing casing deformation machine learning prediction method based on the mechanism model according to claim 6 is characterized in that: Based on starting the overall analysis mode, integrating the qualified number of each to-be-processed simulation feature set that meets the acquisition standard in the target initial simulation data set and the unqualified number of each abnormal simulation feature set that does not meet the acquisition standard in the target initial simulation data set; An overall abnormality score is determined according to the integration result, and the re-collection mode or the pre-processing mode is judged to be started according to the overall abnormality score.

8. The shale fracturing casing deformation machine learning prediction method based on the mechanism model according to claim 7 is characterized in that: The process of preprocessing the target simulated data set to be processed to obtain a preprocessing result includes: Analyze the first actual correlation between each to-be-processed simulation feature set, and perform feature screening on the target to-be-processed simulation data set according to the first actual correlation to obtain the target to-be-processed simulation data set; Wherein, the target data set to be simulated includes a plurality of feature sets to be simulated; For any feature set to be simulated, the secondary actual correlation between the feature parameters in the feature set to be simulated is analyzed, and the feature set to be simulated is screened according to the secondary actual correlation to obtain the actual simulation feature set.

9. The shale fracturing casing deformation machine learning prediction method based on the mechanism model according to claim 8 is characterized in that: The process of determining the sliding sleeve variables of the target shale oil well reservoir under different influencing factors according to the initial numerical simulation results includes: Determine actual geometric parameters and induced stress field distribution during hydraulic fracturing based on each actual simulation feature set; Based on the actual geometric parameters and the induced stress field distribution, determine the numerical range of the influencing factors causing the slip of the weak structural surface and the slip amount of the weak structural surface under different influencing factors; The sliding sleeve variable of the target shale oil well reservoir is determined based on the numerical range of the influencing factors of the slip of the weak structural surface and the slip amount of the weak structural surface under different influencing factors.

10. The shale fracturing casing deformation machine learning prediction method based on the mechanism model according to claim 9 is characterized in that: The process of optimizing the initial numerical simulation results by inversion based on each stealth simulation feature set to obtain the actual numerical simulation results includes: Based on history matching, inverting each stealth simulation feature set to obtain the actual numerical simulation result; Comparing the actual numerical simulation results with the obtained actual numerical calculation results; The initial prediction standard is optimized according to the comparison result, or historical matching is performed again.

Images