A method for preparing a high-temperature, high-salinity oil reservoir profile control and water shut-off composite agent

By using machine learning algorithms to establish a temperature difference model in high-temperature and high-salinity reservoirs, and by monitoring viscosity changes in real time and adjusting the crosslinking agent ratio, the problem of premature gelation of the composite agent near the wellbore was solved, achieving precise sealing of deep high-permeability channels and improving the profile control and water shut-off effect.

CN123120133APending Publication Date: 2026-11-17HUANGGANG NORMAL UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611245906.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-17
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing profile control and water shut-off composite agents tend to gel prematurely near the wellbore in high-temperature and high-salinity reservoirs, making it difficult to penetrate deep into high-permeability channels, resulting in poor sealing effects and affecting oil recovery.

Method used

By establishing a model of the temperature difference between the ground and the formation through machine learning algorithms, the trigger threshold of the crosslinking reaction is determined, viscosity changes are monitored in real time, and the crosslinking agent ratio is adjusted to ensure that the composite agent accurately seals the deep channels.

Benefits of technology

It has achieved intelligent and precise control of the timing of cross-linking of composite agents, improving the success rate and recovery rate of profile control and water shut-off operations in high-temperature and high-salinity oil reservoirs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123120133A_ABST
    Figure CN123120133A_ABST
Patent Text Reader

Abstract

This application provides a method for preparing a high-temperature, high-salinity reservoir profile control and water shut-off composite agent, comprising: if the real-time viscosity increase data is greater than the premature gelation judgment threshold, generating a ground formulation adjustment command for correcting the crosslinking agent ratio of the composite agent, and obtaining second viscosity increase data reflecting the viscosity evolution of the composite agent after processing by the ground formulation adjustment command; if the flow state data matches the deep plugging state characteristics, outputting deep plugging success confirmation information, and obtaining the final profile control and water shut-off operation record corresponding to the deep plugging success confirmation information, wherein the final profile control and water shut-off operation record is used to record the plugging state of the composite agent in the reservoir.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method for preparing a composite agent for profile control and water shut-off in high-temperature and high-salinity oil reservoirs. Background Technology

[0002] As my country's major oilfields enter the mid-to-late stages of development, high-permeability zones formed within the reservoirs due to long-term water injection are becoming increasingly prominent. Injected water circulates ineffectively along these channels, leading to a continuous increase in the water cut of the produced fluid and hindering the effective utilization of remaining oil. Profile control and water shut-off technology, which injects chemical composites with sealing capabilities into the formation to block high-permeability channels and force subsequent fluids to divert to unaffected low-permeability areas, is a crucial means of improving oil recovery, especially in demanding reservoirs with high temperatures and salinity. However, existing profile control and water shut-off composite agents generally suffer from a contradiction in application: the composite agent usually begins its cross-linking reaction during the surface preparation stage, and the system viscosity continues to rise as it advances from the injection well into the deeper formation. As a result, the composite agent prematurely gels in the near-wellbore zone before it even reaches the deep, high-permeability channels that truly need to be blocked. The resulting plug is trapped around the wellbore, causing a sharp increase in injection pressure and making it difficult to advance to the target area. Looking further, two interrelated technical factors are at play behind this "premature gelation." First, the chemical activity of the crosslinking agent lacks controllable initiation conditions; once the system is prepared, the reaction occurs spontaneously and cannot be delayed over time. Second, during surface preparation and pipeline injection, there is a significant difference between ambient temperature and deep formation temperature, and traditional crosslinking agents do not convert this temperature difference into a usable trigger signal, causing the crosslinking reaction to proceed indiscriminately throughout the injection path. The former factor determines whether the reaction "can be stopped," while the latter determines whether the reaction "can be directionally activated." Together, they determine whether the composite agent can reach its designated location before exerting its effect. To illustrate with a specific scenario, in a deep oil reservoir with a depth exceeding 3,000 meters, a formation temperature approaching 120 degrees Celsius, and extremely high salinity, operators hope that the composite agent can penetrate the near-wellbore zone and reach high-permeability bands tens of meters away from the wellbore before beginning to gel. However, with existing formulations, the composite agent often solidifies within a few meters of the wellbore, leaving the deep channels unobstructed. This fails to fundamentally solve the water channeling problem, and the effects of multiple profile control operations are difficult to sustain. Therefore, how to ensure that the profile control and water shut-off compound maintains a stable low viscosity during the surface preparation and wellbore injection stages, and can be precisely awakened and initiate a cross-linking reaction by the formation's own temperature environment after reaching the deep part of the high-temperature and high-salt reservoir, has become the key issue determining the profile control and water shut-off effect and the success or failure of the operation in this type of reservoir. Summary of the Invention

[0003] This invention provides a method for preparing a high-temperature, high-salinity reservoir profile control and water shut-off composite agent, mainly comprising: The surface temperature data and formation temperature data of the target oil reservoir are obtained by monitoring the surface temperature and formation temperature data at the reservoir location. The surface temperature data and formation temperature data are processed by machine learning algorithm to obtain temperature difference characteristics that reflect the temperature gradient change from the surface to the formation. Based on the temperature difference characteristics, a corresponding trigger signal threshold is obtained. The trigger signal threshold is the critical temperature difference value for determining whether the composite agent has the conditions for initiating a crosslinking reaction. If the temperature difference characteristics are greater than the trigger signal threshold, the target crosslinking activity initiation conditions that meet the temperature activation requirements are determined. Based on the target crosslinking activity initiation conditions, the initial crosslinking activity data of the composite agent in the ground preparation stage is obtained. The initial crosslinking activity data is used to reflect the crosslinking reaction rate of the composite agent in the ground stirring and mixing stage. The initial crosslinking activity data is classified by a classification algorithm to obtain a crosslinking activity inhibition instruction for delaying the crosslinking reaction process. According to the crosslinking activity inhibition instruction, the real-time viscosity rise data of the composite agent in the near-wellbore oil layer area is obtained, and it is determined whether the real-time viscosity rise data is greater than the preset premature gelation judgment threshold. The premature gelation judgment threshold is a viscosity change critical value that distinguishes between normal viscosity evolution and premature gelation failure risk. If the real-time viscosity increase data is greater than the premature gelation judgment threshold, a ground formulation adjustment instruction for correcting the crosslinking agent ratio of the composite agent is generated, and a second viscosity increase data reflecting the viscosity evolution of the composite agent after being processed by the ground formulation adjustment instruction is obtained. Based on the second viscosity rise data, the flow state data of the composite agent in the deep channel reservoir area is obtained. The flow state data includes flow resistance information and flow rate information. It is determined whether the flow state data matches the preset deep blockage state characteristics. The deep blockage state characteristics are the flow response mode exhibited by the composite agent when blockage is formed in the deep channel. If the flow state data matches the deep plugging state characteristics, then the deep plugging success confirmation information is output, and the final profile control and water shut-off operation record corresponding to the deep plugging success confirmation information is obtained. The final profile control and water shut-off operation record is used to record the plugging state of the compound agent in the reservoir.

[0004] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention discloses a method for preparing a composite agent for profile control and water shut-off in high-temperature and high-salinity oil reservoirs. This invention addresses the core problem of large temperature gradients in deep oil reservoirs, which cause the composite agent to prematurely cross-link and gel in surface preparation or near-wellbore areas, hindering its ability to reach deep channels for effective sealing. By establishing a model of the difference between surface and formation temperatures using machine learning algorithms, the invention determines the trigger threshold for temperature-activated cross-linking reactions. Based on this, the initial cross-linking activity during surface preparation is classified and identified, and inhibition commands are generated to effectively delay the cross-linking reaction process. Simultaneously, near-wellbore viscosity rise data is monitored in real time. When the threshold for premature gelation is exceeded, the cross-linking agent ratio is automatically adjusted to ensure the composite agent maintains suitable fluidity to enter deep channels. By matching the flow resistance and rate response patterns in the deep sealing state characteristics, the invention accurately determines whether the composite agent has formed an effective seal in the target reservoir area. This invention achieves intelligent and precise control of the cross-linking timing of the composite agent, solving the problem of deep sealing in profile control and water shut-off in high-temperature and high-salinity oil reservoirs, and significantly improving the success rate and recovery rate of profile control and water shut-off operations. Attached Figure Description

[0005] Figure 1 This is a flowchart illustrating a method for preparing a high-temperature, high-salinity reservoir profile control and water shut-off composite agent according to the present invention.

[0006] Figure 2 This is a schematic diagram of a method for preparing a high-temperature and high-salinity reservoir profile control and water shut-off composite agent according to the present invention.

[0007] Figure 3 This is another schematic diagram of a method for preparing a high-temperature and high-salinity reservoir profile control and water shut-off composite agent according to the present invention. Detailed Implementation

[0008] To further understand the content of this invention, a detailed description of the invention is provided in conjunction with the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0009] like Figures 1-3 The preparation method of a high-temperature and high-salinity reservoir profile control and water shut-off composite agent in this embodiment may specifically include: S101. Obtain surface temperature data and formation temperature data of the target oil reservoir at the surface location and at the reservoir location, and process the surface temperature data and formation temperature data through machine learning algorithms to obtain temperature difference characteristics that reflect the temperature gradient change from the surface to the formation.

[0010] A first temperature value collected from surface monitoring points in the oil reservoir area and a second temperature value collected from reservoir monitoring points are obtained. An isolated forest algorithm is used to process the first and second temperature values, resulting in a third and fourth temperature value after outlier removal. Based on the third and fourth temperature values, a first depth value corresponding to the surface monitoring point and a second depth value corresponding to the reservoir monitoring point are obtained. The difference between the first and second depth values ​​is calculated to obtain a vertical depth difference. For this vertical depth difference, the temperature difference between the third and fourth temperature values ​​is obtained. A division operation is used to process this temperature difference and the vertical depth difference to determine an initial temperature gradient value. A support vector regression algorithm is used to process the initial temperature gradient value to obtain a smoothed target temperature gradient value. If the target temperature gradient value is greater than a preset gradient threshold, a formation thermal anomaly is identified, and a thermal anomaly identifier is obtained. Based on the thermal anomaly identifier, the spatial coordinate information corresponding to the target temperature gradient value is extracted. The spatial coordinate information and the target temperature gradient value are then processed by interpolation to obtain the temperature difference characteristics reflecting the temperature gradient change from the ground to the formation.

[0011] For example, in reservoir temperature monitoring, sensors acquire the first surface temperature value and the second reservoir temperature value. Due to the complex downhole environment, the data is susceptible to noise interference. The Isolation Forest algorithm is used to process this data. This algorithm constructs a random decision tree to partition the data space, quickly isolating anomalous data with few and distinct characteristics. Inputting raw data containing a normal value of 25 degrees Celsius and an anomalous jump value of 150 degrees Celsius, the Isolation Forest algorithm removes outliers caused by equipment malfunctions, outputting cleaned third and fourth temperature values, significantly improving the reliability of thermodynamic analysis data.

[0012] In one possible implementation, the first depth value of 10 meters at the surface layer and the second depth value of 2010 meters in the reservoir section are extracted, and the difference is used to obtain a vertical depth difference of 2000 meters. The temperature difference of 60 degrees Celsius between the third and fourth temperature values ​​is calculated. A division operation is used to process the temperature difference and the depth difference to determine the initial temperature gradient value as 30 degrees Celsius per kilometer. This quantitative calculation can intuitively reflect the vertical thermal conductivity characteristics of the strata.

[0013] It should be noted that due to the complex lithology of the strata, the initial temperature gradient values ​​often exhibit random fluctuations. Support vector regression is used to process the initial temperature gradient values, and a kernel function is used to map the nonlinear data to a high-dimensional space, finding the optimal fitting hyperplane within the tolerance range. The fluctuating initial gradient sequence is input, and the smoothed target temperature gradient value is output, effectively eliminating geological noise. If the target temperature gradient value exceeds a preset threshold of 35 degrees Celsius per kilometer, a formation thermal anomaly is identified and a thermal anomaly identifier is obtained. Based on this identifier, three-dimensional spatial coordinate information is extracted, and Kriging interpolation is used to calculate and process the coordinates and the target temperature gradient value. This allows for the deduction of the gradient distribution in unmonitored areas, obtaining temperature difference characteristics reflecting the temperature gradient change from the surface to the strata, providing accurate evidence for tracking the migration of underground thermal fluids.

[0014] S102. Based on the temperature difference characteristics, obtain the corresponding trigger signal threshold. The trigger signal threshold is the critical temperature difference value for determining whether the composite agent has the conditions for initiating a crosslinking reaction. If the temperature difference characteristics are greater than the trigger signal threshold, then the target crosslinking activity initiation conditions that meet the temperature activation requirements are determined.

[0015] Initial temperature data of the composite agent is acquired, and the initial temperature data is classified using a decision tree algorithm to obtain a first temperature classification set. Based on the first temperature classification set, the temperature difference under different classifications is extracted to obtain a first temperature difference value. For the first temperature difference value, a support vector machine algorithm is used for classification analysis to determine a second temperature difference value. Based on the second temperature difference value, a corresponding trigger value is obtained. If the second temperature difference value is greater than a preset trigger value, it is determined that a temperature difference critical value has been reached. When the temperature difference critical value is reached, the initial crosslinking degree of the composite agent is acquired, and a first crosslinking reaction period is determined based on the initial crosslinking degree. For the first crosslinking reaction period, a corresponding initial activity value is extracted. If the initial activity value is greater than a preset activity critical value, a target state is obtained. Based on the target state, a target crosslinking activity initiation period that meets the temperature activation requirements is determined.

[0016] Specifically, in reservoir injection and production operations, the composite agent injected into the formation is affected by the environment from the wellbore to the reservoir, causing changes in its thermodynamic state. Initial temperature data of the composite agent at different depths within the injection string is obtained, for example, multiple temperature samples are collected within the range of 50 to 90 degrees Celsius. These samples are then input into a pre-trained decision tree algorithm model, using temperature range and heating rate as node partitioning criteria, outputting a first temperature classification set containing both low-temperature inactive states and medium-temperature preheated states.

[0017] For example, based on the aforementioned first temperature classification set, the temperature difference between adjacent depth monitoring points is extracted to obtain a first temperature difference value. Using this first temperature difference value as an input feature, a support vector machine algorithm is employed for nonlinear mapping and hyperplane classification to identify a second temperature difference value dominated by ground thermal conduction.

[0018] It should be noted that a pre-set trigger value is obtained, for example, a trigger value of 15 degrees. The second temperature difference value is compared with this trigger value. When the second temperature difference value reaches 18 degrees, which is greater than the set trigger value of 15 degrees, it is determined that the environment in which the current compound agent is located has reached the critical temperature difference value.

[0019] Specifically, after confirming that the critical temperature difference value has been reached, the current initial crosslinking degree of the composite agent is obtained. This is achieved by detecting the proportion of molecular chain entanglement in the polymer within the composite agent, for example, an initial crosslinking degree of 5%. Based on this initial crosslinking degree of 5%, combined with the rheological property curve, it is determined that the composite agent is in the first crosslinking reaction phase.

[0020] For example, for a composite agent in the first crosslinking reaction stage, the initial activity value released by the crosslinking agent is extracted. An activity threshold of 40 is set. If the extracted initial activity value reaches 55, which is greater than the activity threshold of 40, the composite agent is confirmed to have entered the target state. Based on the fluid characteristics of this target state, the target crosslinking activity initiation period that meets the activation requirements of deep formation temperature is finally determined, completing the determination process for the entire composite agent reaction stage.

[0021] S103. Based on the target crosslinking activity initiation conditions, obtain the initial crosslinking activity data of the composite agent in the ground preparation stage. The initial crosslinking activity data is used to reflect the crosslinking reaction rate of the composite agent in the ground stirring and mixing stage. The initial crosslinking activity data is classified by a classification algorithm to obtain a crosslinking activity inhibition command for delaying the crosslinking reaction process.

[0022] Initial crosslinking activity data of the composite agent in the ground preparation section is obtained based on the target crosslinking activity initiation conditions. The crosslinking reaction rate of the composite agent in the ground mixing section is determined using this initial crosslinking activity data. A support vector machine classification algorithm is used to perform classification processing based on the crosslinking reaction rate to obtain an activity classification result. If the activity classification result is greater than a preset activity judgment value, a crosslinking activity inhibition command is obtained based on the activity classification result. The corresponding retardation period duration is determined based on the obtained crosslinking activity inhibition command. Reaction period control is performed using the retardation period duration to obtain the target mixture.

[0023] In one possible implementation, based on the pre-determined target crosslinking activity initiation conditions, initial crosslinking activity data of the composite agent are collected in the ground preparation section.

[0024] Specifically, the initial viscosity and free monomer concentration of the composite agent are obtained as initial crosslinking activity data through a detection device deployed within the preparation equipment. These initial viscosity and free monomer concentrations are then input into a preset reaction kinetic model, which outputs the crosslinking reaction rate of the composite agent in the ground mixing section. When the initial viscosity is 50 and the free monomer concentration is 20, the model outputs a crosslinking reaction rate of 15.

[0025] For example, a support vector machine (SVM) classification algorithm is used to perform classification processing on the obtained crosslinking reaction rate. The crosslinking reaction rate is used as the input feature of the SVM classification algorithm, mapped and divided through a pre-trained hyperplane, and the corresponding activity classification result is output. This activity classification result is represented by a specific activity level value, used to characterize the degree of crosslinking activity of the macromolecular chain segments within the composite agent during the current stirring stage.

[0026] It should be noted that after obtaining the activity classification result, it is compared with the preset activity judgment value. If the activity classification result is 4 and the preset activity judgment value is 3, since the activity classification result is greater than the activity judgment value, it indicates that the current crosslinking reaction is too fast, and a crosslinking activity inhibition command is generated. According to this crosslinking activity inhibition command, a preset duration mapping table is called to match the corresponding retardation period duration, such as setting the retardation period duration to 120 seconds. The reaction period is controlled by this retardation period duration, that is, the crosslinking delay agent is injected into the composite agent within 120 seconds and the operating parameters of the stirring equipment are adjusted, thereby obtaining the target mixture in the target crosslinking state.

[0027] S104. According to the crosslinking activity inhibition instruction, obtain the real-time viscosity increase data of the composite agent in the near-wellbore oil layer area, and determine whether the real-time viscosity increase data is greater than the preset premature gelation judgment threshold. The premature gelation judgment threshold is a viscosity change critical value that distinguishes between normal viscosity evolution and premature gelation failure risk.

[0028] Based on the crosslinking activity inhibition command, initial viscosity data of the composite agent in the near-wellbore zone is collected. The initial viscosity data is then processed using a Kalman filter algorithm to obtain smoothed viscosity data. Using the smoothed viscosity data, real-time viscosity rise data of the composite agent in the oil reservoir is calculated. This real-time viscosity rise data is compared to a preset premature gelation threshold. If the real-time viscosity rise data exceeds the premature gelation threshold, an early gelation warning signal is generated. Based on the early gelation warning signal, a support vector machine model is used for classification to determine the composite agent concentration adjustment command. This concentration adjustment command is then used to control the opening of the injection pump valve in the near-wellbore zone.

[0029] Specifically, upon receiving the crosslinking activity inhibition command, the near-wellbore online monitoring equipment is activated to collect the initial viscosity data of the composite agent. Due to the complex formation environment, the collected initial viscosity data usually contains random noise caused by fluid pulsation. This initial viscosity data is then passed as input to the Kalman filter algorithm. The Kalman filter algorithm uses the estimated value from the previous moment and the observed value from the current moment to eliminate high-frequency fluctuation interference through state prediction and measurement update processes, outputting smooth viscosity data. For example, if the input initial viscosity data fluctuates frequently between 15.5 and 18.2, after filtering, the output smooth viscosity data stabilizes at around 16.8.

[0030] For example, the viscosity change gradient per unit time is calculated based on continuously acquired smooth viscosity data and used as the real-time viscosity rise data of the composite agent in the oil reservoir. This real-time viscosity rise data is acquired and compared with a pre-set premature gelation threshold. Assume the premature gelation threshold is set at an increase of 2.5 per minute. When the calculated real-time viscosity rise data reaches 3.1 per minute, it meets the threshold condition, triggering and generating an early gelation warning signal. This warning signal reflects that the composite agent has begun a vigorous cross-linking reaction before reaching the target deep formation.

[0031] In one possible implementation, the generated early gelation warning signal and the current smoothed viscosity data are used as feature vectors and input into a pre-trained support vector machine (SVM) model. The SVM model maps the input features to a high-dimensional space using a kernel function and finds the optimal classification hyperplane to output the corresponding composite agent concentration adjustment command. For example, the model outputs a concentration adjustment command to reduce the crosslinking agent injection concentration by 15%. After obtaining this concentration adjustment command, an execution signal is directly sent to the injection pump controller near the wellbore to precisely reduce the injection pump valve opening from the current 80% to 65%, thus completing the dynamic intervention of the composite agent injection concentration.

[0032] S105. If the real-time viscosity increase data is greater than the premature gelation judgment threshold, a ground formulation adjustment command for correcting the crosslinking agent ratio of the composite agent is generated, and the second viscosity increase data reflecting the viscosity evolution of the composite agent after being processed by the ground formulation adjustment command is obtained.

[0033] Real-time viscosity rise data and premature gelation threshold are acquired. If the real-time viscosity rise data exceeds the premature gelation threshold, a ground formulation adjustment command is generated. Based on the ground formulation adjustment command, a random forest algorithm is used to classify and predict the preset composite agent crosslinking agent ratio to obtain a target ratio. The current composite agent crosslinking agent ratio is numerically replaced using the target ratio to obtain a corrected composite agent crosslinking agent ratio. Historical formulation records corresponding to the corrected composite agent crosslinking agent ratio are acquired, and corresponding second viscosity rise data is extracted from these historical formulation records. Based on the second viscosity rise data, a support vector machine algorithm is used for regression calculation to determine the viscosity evolution. If the viscosity evolution exceeds a preset evolution safety threshold, a crosslinking agent concentration reduction command is generated.

[0034] For example, the real-time viscosity rise data transmitted from the downhole monitoring equipment is obtained as 50 mPa·s per hour, while the preset premature gelation judgment threshold is obtained as 40 mPa·s per hour. Since 50 is greater than 40, it indicates that the currently injected compound has begun to thicken rapidly before reaching the deep oil layer, posing a risk of blockage in the near-wellbore area. At this time, a surface preparation adjustment command is triggered and generated, requiring the surface preparation station to change the current chemical agent mixing state.

[0035] In one possible implementation, upon receiving a surface formulation adjustment command, features such as the current formation temperature, injection pressure, and initial main agent concentration are extracted as input data and fed into a pre-trained random forest algorithm model. This model contains multiple decision trees and, through classification learning and a voting mechanism based on a large number of historical formulation schemes, outputs a target formulation ratio suitable for the current well conditions. For example, the predicted target formulation ratio of main agent to crosslinking agent is 100:3. Subsequently, the 100:5 formulation currently used by the surface equipment is replaced with 100:3, thereby obtaining the corrected composite agent-crosslinking agent formulation ratio.

[0036] It should be noted that, to verify the evolution trend of this modified ratio in actual geological environments, historical formulation records under the same conditions were retrieved from the operational database based on the 100:3 modified composite crosslinker ratio, and the corresponding second viscosity increase data sequence was extracted. This second viscosity increase data was used as input features, and a support vector machine (SVM) algorithm was employed for regression calculation. The SVM algorithm uses kernel function mapping to find the optimal hyperplane to fit the viscosity change trend, outputting a predicted viscosity evolution value of 80 mPa·s for a specific future time period.

[0037] For example, the calculated viscosity evolution of 80 mPa·s is compared with the preset evolution safety threshold of 70 mPa·s. Since 80 is greater than 70, it indicates that even with the corrected ratio, there is still a high risk of gelation and accumulation of the subsequent fluid in the near-wellbore area. Therefore, a directive to reduce the crosslinking agent concentration is generated, forcing the surface fluid preparation equipment to further reduce the amount of crosslinking agent injected in order to maintain the fluidity of the injected fluid.

[0038] S106. Based on the second viscosity rise data, obtain the flow state data of the composite agent in the deep channel reservoir area. The flow state data includes flow resistance information and flow rate information. Determine whether the flow state data matches the preset deep blockage state characteristics. The deep blockage state characteristics are the flow response mode exhibited by the composite agent when blockage is formed in the deep channel.

[0039] Based on the second viscosity rise data, the flow resistance and flow rate values ​​of the composite agent within the reservoir region are obtained. The flow resistance and flow rate values ​​are combined to obtain an initial flow state set. A support vector machine algorithm is used to extract features from the initial flow state set to obtain a target flow state set. Based on the target flow state set, the real-time flow response modulus of the composite agent within the channel network is extracted. A pre-established deep plugging state feature library is obtained, and the real-time flow response modulus is compared with a preset response modulus in the deep plugging state feature library to obtain a response matching degree. If the response matching degree is greater than a preset matching threshold, the target flow state set is determined to conform to the deep plugging state, and a deep plugging determination criterion is generated. Based on the deep plugging determination criterion, the final plugging state set of the composite agent within the reservoir region is determined.

[0040] In one possible implementation, based on the second viscosity rise data, combined with pressure monitoring equipment and flow meters within the reservoir area, the flow resistance and flow rate values ​​of the composite agent in the core pores are collected.

[0041] For example, when the second viscosity rise data shows a rapid increase in viscosity, the corresponding flow resistance value may reach 50 MPa, while the flow rate value drops to 2 cubic centimeters per minute. Vector concatenation of these physical quantities yields an initial flow state set, which reflects the preliminary transport characteristics of the composite agent in the complex pore-throat structure.

[0042] It should be noted that the initial flow state set often contains a large amount of formation environment noise, which needs to be extracted using the support vector machine algorithm. The initial flow state set is fed into a pre-trained support vector machine model as input data. The kernel function is used to map the low-dimensional flow data to a high-dimensional space, and the optimal hyperplane is found to separate the core features representing the gelation phase transition. The target flow state set after removing redundant interference is output.

[0043] For example, after processing by a support vector machine, the initial state set, which originally contained 100 dimensions, is reduced in dimensionality and 20 key feature dimensions are extracted to form the target flow state set.

[0044] In one possible implementation, the real-time flow response model of the composite agent within the channel network is extracted based on the target flow state set. This response model is essentially the rheological curve feature of the composite agent at a specific shear rate. Subsequently, a pre-established deep plugging state feature library is acquired, which stores preset response models from a large number of historical successful plugging cases. The real-time flow response model is then dynamically time-warped and compared with the preset response models to calculate the similarity in morphology and trend, thereby obtaining the response matching degree.

[0045] For example, if the curves of the two curves highly overlap, the calculated response matching degree can reach 90%.

[0046] It should be noted that if the response matching degree is greater than a preset matching threshold, for example, if the matching threshold is set to 85, then the target flow state set is determined to conform to the deep plugging state. At this time, a deep plugging determination label is generated. This label serves as a quantitative indicator for state confirmation and is used to determine the final plugging state set of the composite agent in the reservoir area, thereby completing the objective characterization of the composite agent's gelation and residence in the deep reservoir.

[0047] S107. If the flow state data matches the deep plugging state characteristics, output deep plugging success confirmation information, and obtain the final profile control and water shut-off operation record corresponding to the deep plugging success confirmation information. The final profile control and water shut-off operation record is used to record the plugging state of the compound agent in the reservoir.

[0048] The initial state value corresponding to the flow is obtained, and the initial state value is processed by mean filtering to obtain the standard state value. Based on the standard state value, the isolated forest algorithm is used to remove outliers to obtain the target state matrix. The target state matrix and the reference feature set corresponding to the preset sealing surface are obtained, and the cosine similarity between the target state matrix and the reference feature set is calculated to obtain the state matching degree. If the state matching degree is greater than the preset matching threshold, a deep sealing success confirmation code is generated. Based on the deep sealing success confirmation code, the corresponding information flow extraction item is generated. Through the information flow extraction item, the final profile control and water plugging operation order is retrieved from the preset database. Based on the final profile control and water plugging operation order, the actual sealing record table of the compound agent in the reservoir area is obtained by parsing.

[0049] In one possible implementation, the initial state values ​​are obtained from the mobile phase data collected during the migration of the composite agent in the deep pores of the reservoir. These initial state values ​​typically include high-frequency fluctuation data from the downhole environment. The initial state values ​​are then processed using a mean filter, with a sliding time window length of 5 acquisition cycles. The average value of the pressure and flow velocity data within the window is calculated, and a standard state value is output.

[0050] For example, if the continuously collected initial pressure state values ​​are 15 MPa, 17 MPa, and 16 MPa, the corresponding standard state values ​​are output after mean filtering. Then, based on these standard state values, an isolated forest algorithm is used to remove outliers. Using the standard state values ​​as input features, a forest model containing 100 isolated trees is constructed, and the outlier score for each data point is calculated. When the outlier score of a data point is greater than 65, it is identified as an outlier caused by formation micro-fractures and removed. Normal data is retained and reconstructed to obtain the target state matrix.

[0051] It should be noted that after obtaining the target state matrix and the preset reference feature set corresponding to the plugging surface, it is necessary to quantify the degree of closeness between the current flow state and the ideal plugging state. The reference feature set contains multidimensional state vectors when the composite agent forms a dense gel segment in a standard core tube. The cosine similarity between the target state matrix and the reference feature set is calculated, and the state matching degree is obtained by evaluating the cosine value of the angle between the two multidimensional vectors in the direction.

[0052] For example, if the calculated state matching degree is 88 and the preset matching threshold is 85, since 88 is greater than 85, it is determined that the composite agent has formed a residence in the deep high-permeability channel, and a deep sealing success confirmation code is generated.

[0053] In one possible implementation, a corresponding information stream extraction item is generated based on the deep plugging success confirmation code. This extraction item includes a specific timestamp and well number identifier. Using this information stream extraction item, the final profile control and water plugging operation order is retrieved from a pre-set oilfield production database. Based on the final profile control and water plugging operation order, an actual plugging record table of the compound agent within the reservoir area is obtained through parsing. This record table covers quantitative indicators such as a cumulative injection volume of 1200 cubic meters for a single well and a plugging radius of 15 meters.

[0054] The above embodiments are merely one of the preferred embodiments of the present invention and should not be used to limit the scope of protection of the present invention. Any modifications or refinements made to the main design concept and spirit of the present invention that are not of substantial significance, but solve the same technical problem as the present invention, should be included within the scope of protection of the present invention.

Claims

1. A method for preparing a high-temperature, high-salinity oil reservoir profile control and water shut-off composite agent, characterized in that, The method includes: The surface temperature data and formation temperature data of the target oil reservoir are obtained by monitoring the surface temperature and formation temperature data at the reservoir location. The surface temperature data and formation temperature data are processed by machine learning algorithm to obtain temperature difference characteristics that reflect the temperature gradient change from the surface to the formation. Based on the temperature difference characteristics, a corresponding trigger signal threshold is obtained. The trigger signal threshold is the critical temperature difference value for determining whether the composite agent has the conditions for initiating a crosslinking reaction. If the temperature difference characteristics are greater than the trigger signal threshold, the target crosslinking activity initiation conditions that meet the temperature activation requirements are determined. Based on the target crosslinking activity initiation conditions, the initial crosslinking activity data of the composite agent in the ground preparation stage is obtained. The initial crosslinking activity data is used to reflect the crosslinking reaction rate of the composite agent in the ground stirring and mixing stage. The initial crosslinking activity data is classified by a classification algorithm to obtain a crosslinking activity inhibition instruction for delaying the crosslinking reaction process. According to the crosslinking activity inhibition instruction, the real-time viscosity rise data of the composite agent in the near-wellbore oil layer area is obtained, and it is determined whether the real-time viscosity rise data is greater than the preset premature gelation judgment threshold. The premature gelation judgment threshold is a viscosity change critical value that distinguishes between normal viscosity evolution and premature gelation failure risk. If the real-time viscosity increase data is greater than the premature gelation judgment threshold, a ground formulation adjustment instruction for correcting the crosslinking agent ratio of the composite agent is generated, and a second viscosity increase data reflecting the viscosity evolution of the composite agent after being processed by the ground formulation adjustment instruction is obtained. Based on the second viscosity rise data, the flow state data of the composite agent in the deep channel reservoir area is obtained. The flow state data includes flow resistance information and flow rate information. It is determined whether the flow state data matches the preset deep blockage state characteristics. The deep blockage state characteristics are the flow response mode exhibited by the composite agent when blockage is formed in the deep channel. If the flow state data matches the deep plugging state characteristics, then the deep plugging success confirmation information is output, and the final profile control and water shut-off operation record corresponding to the deep plugging success confirmation information is obtained. The final profile control and water shut-off operation record is used to record the plugging state of the compound agent in the reservoir.

2. The preparation method of the high-temperature and high-salinity oil reservoir profile control and water shut-off composite agent according to claim 1, characterized in that, The process involves acquiring surface temperature data monitored at the target oil reservoir's surface location and formation temperature data monitored at the reservoir location. Then, a machine learning algorithm is used to process the surface temperature data and the formation temperature data to obtain temperature difference characteristics reflecting the temperature gradient change from the surface to the formation, including: The first temperature value collected from the surface monitoring point of the oil reservoir area and the second temperature value collected from the reservoir section monitoring point are obtained. The isolated forest algorithm is used to process the first temperature value and the second temperature value to obtain the third temperature value and the fourth temperature value after removing outliers. Based on the third temperature value and the fourth temperature value, the first depth value corresponding to the surface layer monitoring point and the second depth value corresponding to the reservoir section monitoring point are obtained. The vertical depth difference is obtained by subtracting the first depth value and the second depth value. For the vertical surface depth difference, the temperature difference between the third temperature value and the fourth temperature value is obtained, and the temperature difference and the vertical surface depth difference are processed by division to determine the initial temperature gradient value. The initial temperature gradient value is processed by the support vector regression algorithm to obtain the smoothed target temperature gradient value. If the target temperature gradient value is greater than the preset gradient threshold, it is determined that there is a formation thermal anomaly in the target temperature gradient value, and a thermal anomaly identifier is obtained. Based on the thermal anomaly identifier, the spatial coordinate information corresponding to the target temperature gradient value is extracted. The spatial coordinate information and the target temperature gradient value are then processed by interpolation to obtain the temperature difference characteristics reflecting the temperature gradient change from the ground to the formation.

3. The preparation method of the high-temperature and high-salinity oil reservoir profile control and water shut-off composite agent according to claim 1, characterized in that, The step involves obtaining a corresponding trigger signal threshold based on the temperature difference characteristic. The trigger signal threshold is a critical temperature difference value used to determine if the composite agent meets the conditions for initiating a crosslinking reaction. If the temperature difference characteristic is greater than the trigger signal threshold, then the target crosslinking activity initiation conditions that meet the temperature activation requirements are determined, including: The initial temperature data of the compound is obtained, and the initial temperature data is classified using a decision tree algorithm to obtain a first temperature classification set. Based on the first temperature classification set, extract the temperature difference under different classifications to obtain the first temperature difference value; For the first temperature difference value, a support vector machine algorithm is used for classification analysis to determine the second temperature difference value; Based on the second temperature difference value, obtain the corresponding trigger value. If the second temperature difference value is greater than the preset trigger value, it is determined that the temperature difference critical value has been reached. When the temperature difference critical value is reached, the initial degree of crosslinking of the composite agent is obtained, and the first crosslinking reaction period is determined based on the initial degree of crosslinking. For the first crosslinking reaction period, the corresponding initial activity value is extracted. If the initial activity value is greater than the preset activity threshold, the target state is obtained. Based on the target state, determine the target crosslinking activity initiation period that meets the temperature activation requirements.

4. The preparation method of the high-temperature and high-salinity oil reservoir profile control and water shut-off composite agent according to claim 1, characterized in that, The method involves obtaining initial crosslinking activity data of the composite agent during the ground preparation stage based on the target crosslinking activity initiation condition. This initial crosslinking activity data reflects the crosslinking reaction rate of the composite agent during the ground mixing stage. A classification algorithm is used to classify the initial crosslinking activity data to obtain crosslinking activity inhibition instructions for delaying the crosslinking reaction process, including: The initial cross-linking activity data of the composite agent in the ground preparation section were obtained according to the target cross-linking activity initiation conditions; The crosslinking reaction rate of the composite agent in the ground mixing section was determined using the initial crosslinking activity data. Based on the crosslinking reaction rate, a support vector machine classification algorithm is used to perform classification processing to obtain the activity classification result; If the activity classification result is greater than the preset activity determination value, a cross-linking activity inhibition instruction is obtained for the activity classification result; Obtain the crosslinking activity inhibition instruction and determine the corresponding delay period duration based on the crosslinking activity inhibition instruction; The target mixture is obtained by controlling the reaction period through the duration of the delay period.

5. The preparation method of the high-temperature and high-salinity oil reservoir profile control and water shut-off composite agent according to claim 1, characterized in that, The method involves acquiring real-time viscosity increase data of the composite agent in the near-wellbore oil reservoir region according to the crosslinking activity inhibition command, and determining whether the real-time viscosity increase data exceeds a preset premature gelation threshold. The premature gelation threshold is a critical viscosity change value that distinguishes between normal viscosity evolution and the risk of premature gelation failure, including: According to the cross-linking activity inhibition command, the initial viscosity data of the composite agent in the near-wellbore zone was collected; Initial viscosity data is obtained, and noise reduction is performed using the Kalman filter algorithm to obtain smooth viscosity data; By smoothing the viscosity data, the real-time viscosity rise of the compound in the oil layer area was calculated; Acquire real-time viscosity increase data and compare it with the preset premature gelation judgment threshold. If the real-time viscosity increase data is greater than the premature gelation judgment threshold, generate an early gelation warning signal. Based on the early gelation warning signal, a support vector machine model is used for classification to determine the concentration adjustment command of the compound agent; Obtain concentration adjustment commands to control the opening degree of injection pump valves in the near-wellbore zone.

6. The preparation method of the high-temperature and high-salinity oil reservoir profile control and water shut-off composite agent according to claim 1, characterized in that, If the real-time viscosity increase data is greater than the premature gelation determination threshold, a ground formulation adjustment command is generated to correct the crosslinking agent ratio of the composite agent, and second viscosity increase data reflecting the viscosity evolution of the composite agent after processing by the ground formulation adjustment command is obtained, including: Acquire real-time viscosity rise data and premature gelation judgment threshold. If the real-time viscosity rise data is greater than the premature gelation judgment threshold, generate a ground preparation adjustment command. According to the ground preparation adjustment instructions, the preset composite agent crosslinking agent ratio is classified and predicted using the random forest algorithm to obtain the target ratio; The current composite crosslinking agent ratio is numerically replaced by the target ratio to obtain the corrected composite crosslinking agent ratio. Obtain historical formulation records corresponding to the crosslinking agent ratio of the modified composite agent, and extract the corresponding second viscosity increase data from the historical formulation records; Based on the second viscosity increase data, a support vector machine algorithm is used to perform regression calculations to determine the viscosity evolution. If the viscosity evolution exceeds a preset evolution safety threshold, a command to reduce the crosslinking agent concentration is generated.

7. The preparation method of the high-temperature and high-salinity oil reservoir profile control and water shut-off composite agent according to claim 1, characterized in that, The process involves obtaining flow state data of the composite agent in the deep channel reservoir region based on the second viscosity rise data. This flow state data includes flow resistance information and flow rate information. The process then determines whether the flow state data matches preset deep plugging state characteristics. These deep plugging state characteristics are the flow response patterns exhibited by the composite agent when it forms a plug in the deep channel, including: Based on the second viscosity rise data, the flow resistance value and flow rate value of the composite agent in the reservoir area are obtained, and the initial flow state set is obtained by combining the flow resistance value and the flow rate value. The initial flow state set is used to extract features using the support vector machine algorithm to obtain the target flow state set; Based on the target flow state set, extract the real-time flow response model of the composite agent within the channel network; Obtain a pre-established deep occlusion state feature library, compare the real-time flow response model with the preset response model in the deep occlusion state feature library, and obtain the response matching degree; If the response matching degree is greater than the preset matching threshold, then the target flow state set is determined to conform to the deep blockage state, and a deep blockage determination label is generated. Based on the deep plugging determination criteria, the final plugging state set of the composite agent in the reservoir area is determined.

8. The preparation method of the high-temperature and high-salinity oil reservoir profile control and water shut-off composite agent according to claim 1, characterized in that, If the flow state data matches the deep plugging state characteristics, then a deep plugging success confirmation message is output, and the final profile control and water shut-off operation record corresponding to the deep plugging success confirmation message is obtained. The final profile control and water shut-off operation record is used to record the plugging state of the composite agent in the reservoir, including: Obtain the initial state value corresponding to the flow, and then process the initial state value by mean filtering to obtain the standard state value; Based on the standard state values, the isolated forest algorithm is used to remove outliers to obtain the target state matrix; Obtain the target state matrix and the reference feature set corresponding to the preset blocking surface, and calculate the cosine similarity between the target state matrix and the reference feature set to obtain the state matching degree; If the state matching degree is greater than the preset matching threshold, a deep sealing success confirmation code is generated; Based on the deep sealing success confirmation code, generate the corresponding information stream extraction item; The final profile control and water plugging operation order is retrieved from the preset database through the information flow extraction item; Based on the final profile control and water shut-off operation plan, the actual sealing record table of the composite agent in the reservoir area was obtained through analysis.