A reservoir facies permeability modeling method fusing multi-source data and agent constraints

Abstract

The application discloses a reservoir relative permeability modeling method fusing multi-source data and agent constraints, comprising the following steps: obtaining experimental data and historical curve images of a target reservoir; constructing a joint constraint model of relative permeability and water saturation based on the experimental data and the historical curve images; solving parameters of the joint constraint model by using a collaborative optimization algorithm; and generating a relative permeability curve of the target reservoir by using the joint constraint model after obtaining an optimal parameter set. In the application, the prediction of the relative permeability and the water saturation no longer depends on separate initial values, the continuity of the curve in a global range is maintained, and the prediction is in line with the actual fluid flow law of the reservoir, so that the prediction accuracy and engineering applicability are improved.

Description

Technical Field

[0001] This invention belongs to the field of oil and gas reservoir relative permeability modeling technology, specifically involving a reservoir relative permeability modeling method that integrates multi-source data and intelligent agent constraints. Background Technology

[0002] Reservoir relative permeability curves are core parameters describing the seepage characteristics of oil and water phases in porous media, and are of great significance for oilfield development prediction, water injection strategy optimization, and production capacity analysis. Traditional relative permeability modeling methods typically rely on core displacement experimental data, constructing curves using measured values ​​of the relative permeability of the oil and water phases under different water saturation conditions, and combining this with macroscopic reservoir production data for prediction. However, with the increasing heterogeneity and complex fluid flow patterns in reservoirs, prediction methods relying solely on experimental data or historical curves have limitations in terms of accuracy and continuity.

[0003] A search of application number 202111400726.0 reveals a machine learning-based method and system for predicting relative permeability curves. This method establishes a first relative permeability curve starting point model by using well logging curve data as input and the endpoint values ​​of water saturation as output. It then establishes a first relative permeability model by using the predicted initial water saturation values ​​from the well logging curve data and the first relative permeability curve starting point model as input and the relative permeability at different water saturations as output. Finally, a deep learning analysis and comprehensive prediction method for relative permeability curves is obtained. This method implicitly incorporates the control mechanism and parameters into the model, eliminating the need to establish a mathematical model of relative permeability or simulate nonlinear control mechanisms. This improves the efficiency of obtaining relative permeability curves, reduces the cost of relative permeability curve prediction, and increases prediction accuracy. Furthermore, it provides feasibility verification and technical support for applying artificial intelligence to inversion and imaging problems, thus serving as an effective tool for relative permeability prediction.

[0004] In the above technical solutions, the prediction of reservoir relative permeability curves typically adopts a step-by-step modeling approach. This involves first predicting the initial water saturation value, and then independently predicting the relative permeability of the oil and water phases based on this initial value. This method carries a significant risk of error propagation: if the prediction of the initial value in the first stage is flawed, it will directly lead to a systematic error in the prediction of relative permeability in the second stage. Furthermore, because the two models are trained independently, there is a lack of a collaborative optimization mechanism, and it is impossible to correct the deviation through joint constraints. In addition, relative permeability and water saturation are essentially a continuously related function, but existing technologies separate them into two independent targets: the "starting point" and "subsequent relative permeability," severing their physical correlation. This may result in discontinuous predicted curve shapes that do not conform to the actual fluid flow patterns in the reservoir, such as abnormal endpoint effects or deviations at intersection points, thus affecting the engineering applicability of relative permeability curves in oilfield development and reservoir analysis. Summary of the Invention

[0005] This invention is proposed to solve the problems of error propagation caused by step-by-step modeling and the disconnect between the physical correlation between relative permeability and water saturation in the prior art. Its purpose is to provide a reservoir relative permeability modeling method that integrates multi-source data and intelligent agent constraints.

[0006] This invention is achieved through the following technical solution: A reservoir revelocity permeability modeling method integrating multi-source data and agent constraints includes the following steps: S1. Obtain experimental data and historical curve images of the target reservoir; S2. Based on the experimental data and historical curve images of the target reservoir obtained in step S1, construct a joint constraint model of relative permeability and water saturation. S3. A multi-agent collaborative optimization algorithm is used to solve the parameters of the joint constraint model of relative permeability and water saturation, which is constrained by both experimental data and historical production curves, to obtain the optimal set of relative permeability curve parameters: S4. Based on the optimal relative permeability curve parameter set obtained in step S3, the relative permeability curve of the target reservoir is generated using the joint constraint model of relative permeability and water saturation constructed in step S2.

[0007] In the above technical solution, the experimental data includes the measured values ​​of the relative permeability of the oil phase and the relative permeability of the water phase at different water saturation levels. The historical curve images include historical variation curves of water cut and recovery rate during actual reservoir production. Key feature points of the historical variation curves are extracted using image digitization methods to form a feature dataset.

[0008] In the above technical solution, the value of water saturation ranges from bound water saturation to residual oil saturation, and includes the boundary value between bound water saturation and residual oil saturation.

[0009] In the above technical solution, the method for constructing the joint constraint model of relative permeability and water saturation specifically includes the following steps: S21. The relationship between relative permeability of the oil phase and water saturation, the relationship between relative permeability of the water phase and water saturation, and the relationship between water cut and recovery degree in historical curve images are used as input constraints for the joint constraint model of relative permeability and water saturation. The coupling relationship between relative permeability, water saturation, water cut and recovery degree is established by introducing physical correlation equations. The relationship between the relative permeability of the oil phase and the water saturation is expressed as follows: The relationship between the relative permeability of the aqueous phase and the water saturation is expressed as follows: In the formula: The relative permeability of the oil phase is dimensionless. The relative permeability of the aqueous phase is dimensionless. Real-time water saturation, dimensionless; Effective water saturation, dimensionless; This is the oil phase permeability proportionality coefficient, which is dimensionless. is the permeability proportionality coefficient of the aqueous phase, dimensionless; The index represents the relationship between relative permeability of the oil phase and water saturation, and is dimensionless. The exponent of the curve relating the relative permeability of the aqueous phase to water saturation is dimensionless. Among them, effective water saturation The calculation formula is: In the formula: Effective water saturation, dimensionless; To constrain water saturation, dimensionless; Residual oil saturation, dimensionless; Real-time water saturation, dimensionless; The physical correlation equation includes moisture content. relative permeability of oil phase Relative permeability of water phase Oil phase viscosity viscosity with water phase The functional relationship and the degree of extraction The functional relationship between water content and water saturation; The moisture content relative permeability of oil phase Relative permeability of water phase Oil phase viscosity viscosity with water phase The functional relationship is: In the formula: Moisture content, dimensionless; The relative permeability of the oil phase is dimensionless. The relative permeability of the aqueous phase is dimensionless. Oil phase viscosity, in units of ; Viscosity of the aqueous phase, in units of... ; The degree of extraction The functional relationship between water content and water saturation is: In the formula: The degree of extraction is dimensionless; Real-time water saturation, dimensionless; To constrain water saturation, dimensionless; Reservoir porosity, dimensionless; For integration variables; S22. Using the curves showing the relationship between relative permeability and water saturation in the oil phase and the water phase from the experimental data as local constraints, and the key feature points extracted from the historical curve images as global constraints; the local and global constraints of the joint constraint model of relative permeability and water saturation are combined to minimize the comprehensive error, and the function for minimizing the comprehensive error is expressed as: In the formula: For the parameter set of the joint constraint model, and These are weighting coefficients, which respectively control the relative importance of experimental data accuracy constraints and historical curve continuity constraints; For joint constraint models in The predicted relative permeability value of the oil phase; For the first Measured values ​​of relative permeability of the oil phase at each experimental point; For joint constraint models in The predicted moisture content value is given below. For the first Historical actual values ​​of moisture content at each feature point; S23. The combined formulas (1) to (6) constitute a joint constraint model of relative permeability and water saturation. S24. Utilizing fractal theory or the boundary point theory between capillary flow and thin film flow in unsaturated soils, the water content... relative permeability of oil phase Relative permeability of water phase Oil phase viscosity viscosity with water phase The functional relationship is divided into sub-functional relationships for the low moisture content stage and sub-functional relationships for the high moisture content stage. By introducing a boundary point function, the sub-model is made to operate at the moisture content boundary point. A smooth transition is achieved at the point.

[0010] In the above technical solution, the boundary point function is established based on the parameters of the soil-water characteristic curve SWCC, and is expressed as follows: In the formula: Represents the boundary point function; Indicates the moisture content boundary point; This is a smoothing coefficient used to control the steepness of the transition of the curve near the dividing point.

[0011] In the above technical solution, the moisture content dividing point The location is determined based on the parametric correlation of the soil-water characteristic curve, through water saturation. With matrix suction The relational model is used to determine the saturation level corresponding to the critical suction force for the transition from capillary flow to continuous flow. That is, the moisture content dividing point; The water saturation With matrix suction The relational model is represented as: In the formula: The water saturation level is dimensionless. This represents matrix suction, expressed in units of . These are empirical parameters for SWCC, and are dimensionless.

[0012] In the above technical solution, the parameter solution in step S3 specifically includes the following steps: S31. The multi-agent cooperative optimization algorithm uses the comprehensive error minimization function of the joint constraint model of relative permeability and water saturation as the optimization objective function; the optimization objective of the objective function is to find a function that minimizes the comprehensive error of the model ... Minimal parameter set ; The optimization objective function is expressed as: In the formula: For parameters to be optimized, Indicates experimental data points, Indicates the characteristic points of the historical curve; and These are the weighting factors for experimental data and historical curves, used to balance local accuracy and global matching degree; S32. Establish a two-layer structure of global population cooperation layer and local individual evolution layer. The global population cooperation layer is based on the genetic algorithm framework and is responsible for maintaining the diversity of the population and wide-area search. The local individual evolution layer is based on the particle swarm optimization mechanism and performs a refined search around the superior individuals selected by the genetic algorithm. S33. Randomize the test for multiple initial populations and perform variance analysis on the results; if the variance of the parameters converges... If the algorithm exhibits good convergence consistency and solution stability under the given reservoir conditions, it can be considered that the algorithm has good solution stability.

[0013] In the above technical solution, step S32 specifically includes the following steps: S321. In the global population cooperation layer, individuals, i.e., parameter vectors... Encoded in chromosome form, and evolving through selection, crossover, and mutation; Individual updates in a population are represented as follows: In the formula: These represent the selection, crossover, and mutation operators, respectively. The fitness function for each generation is defined as the fitness value. Used to evaluate the merits and demerits of individuals; S322. At the local individual evolution level, the particle swarm mechanism conducts local optimization near globally superior individuals. No. The velocity and position of each particle are updated according to the following formula: In the formula: For inertial weights, , For individual cognition and group learning factors, As a random factor, This is the optimal solution in the particle's history. This is the globally optimal solution; S323, the multi-agent cooperative optimization algorithm performs a cross-layer cooperative operation once every few generations: Specifically, this involves transferring a group of superior individuals from the global population cooperation layer to the local individual evolution layer for a refined search, and then using perturbation terms... The optimal solution is perturbed to maintain population diversity; The convergence condition of the algorithm is defined as follows: In the formula: A preset tolerance threshold is set; the algorithm stops when the objective function changes by less than this value for several consecutive iterations; the final optimal parameter set is obtained. As the steady-state solution of the joint constraint model.

[0014] In the above technical solution, step S4 specifically includes the following steps: S41. Based on the optimal relative permeability curve parameter set, the relative permeability of the oil phase and water saturation are directly calculated using continuous function form. and relative permeability and water saturation of the aqueous phase The correspondence is specifically represented as follows: in, In the formula: The effective water saturation in the optimal set of relative permeability curve parameters is dimensionless. The residual oil saturation in the optimal set of relative permeability curve parameters is dimensionless. The bound water saturation in the optimal set of parameters for the relative permeability curve is dimensionless. Real-time water saturation, dimensionless; The relative permeability of the oil phase is dimensionless, based on the optimal set of relative permeability curve parameters. The relative permeability of the aqueous phase is based on the optimal set of parameters for the relative permeability curve, and is dimensionless. is the oil phase permeability ratio coefficient in the optimal relative permeability curve parameter set, dimensionless; is the proportion coefficient of water phase permeability in the optimal set of parameters for the relative permeability curve, and is dimensionless. , is the exponent of the curve relating oil phase relative permeability to water saturation in the optimal relative permeability curve parameter set, and is dimensionless; , is the exponent of the curve relating the relative permeability of the aqueous phase to water saturation in the optimal set of relative permeability curve parameters, and is dimensionless; S42. Based on the optimal set of relative permeability curve parameters, the prediction results are mapped to the actual production curve through the functional relationship between water cut and extraction degree. The functional relationship between water rate and extraction degree based on the optimal relative permeability curve parameter set is as follows: In the formula: The relative permeability of the oil phase is dimensionless, based on the optimal set of relative permeability curve parameters. The relative permeability of the aqueous phase is based on the optimal set of parameters for the relative permeability curve, and is dimensionless. Oil phase viscosity, in units of ; Viscosity of the aqueous phase, in units of... ; S43. Result Verification: The generated relative permeability curves are substituted into numerical simulation software to simulate the dynamic changes in oil production, water cut, and recovery rate of the reservoir, and compared with the historical curves of actual production of the target reservoir.

[0015] The beneficial effects of this invention are: This invention provides a reservoir relative permeability modeling method that integrates multi-source data and intelligent agent constraints. By using a joint constraint model and a collaborative optimization algorithm, it achieves synchronous constraints between experimental data and historical production curves, making the relative permeability curve continuous in the range of low to high water cut and conforming to the actual fluid flow law of the reservoir, thereby improving prediction accuracy and engineering applicability.

[0016] This invention effectively solves the error propagation problem in traditional step-by-step modeling methods by jointly constructing a joint constraint model of the relative permeability curve using experimental data and historical curve images; in the method of this invention, the relative permeability... , With water saturation The prediction no longer relies on a single initial value, but is modeled holistically as a continuous function. Local accuracy is constrained by experimental data, and global dynamics are constrained by historical curve features, enabling collaborative optimization during the model building phase. Even if local prediction biases exist within a certain water saturation range, the optimization algorithm can iteratively adjust the parameters. Synchronous correction, while maintaining the continuity of the curve in the global range, significantly reduces the risk of error propagation and improves the predictive reliability of the relative permeability curve and its conformity with reservoir flow patterns.

[0017] This invention introduces a continuous function model and a physical correlation formula. By tightly coupling relative permeability with water saturation, a smooth transition of the curve is achieved in the low to high water cut range, while taking into account both film flow and continuous aqueous phase channel characteristics. The joint constraint model and collaborative optimization algorithm ensure that the predicted curve does not exhibit endpoint effects, abnormal crossover points, or discontinuous patterns, effectively maintaining physical rationality. Compared to existing technologies, this invention not only preserves the microscopic and macroscopic characteristics of reservoir multiphase flow but also achieves a balance between local accuracy and global continuity through a dynamic correction mechanism and historical curve feature input, significantly improving the accuracy and engineering applicability of the relative permeability curve. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the overall logical framework of the method of the present invention; Figure 2 This is a time sequence diagram of data acquisition and feature extraction in the method of this invention; Figure 3 This is a flowchart of the joint constraint model construction process in the method of this invention; Figure 4 This is a flowchart of the collaborative optimization process of the phase permeation curve in the method of this invention.

[0019] For those skilled in the art, other related figures can be obtained from the above figures without any creative effort. Detailed Implementation

[0020] To enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0021] A reservoir revelocity permeability modeling method integrating multi-source data and agent constraints includes the following steps: S1. Obtain experimental data and historical curve images of the target reservoir: The experimental data includes different water saturation levels. Relative permeability of the oil phase Measured values ​​and different water saturation levels Relative permeability of water phase below The measured value; The water saturation The scope covers the actual production range of the reservoir; water saturation The range is from bound water saturation To residual oil saturation And includes bound water saturation With residual oil saturation Boundary values; The experimental data were obtained through core displacement experiments; The historical curve images include the water cut during the actual production process of the reservoir. With extraction degree Historical change curve; The method for obtaining the historical curve image is as follows: First, collect historical data on water cut and recovery rate at different development stages during the actual production process of the target reservoir, and extract key feature points in its dynamic trend through image processing technology. The key feature point extraction method is as follows: key inflection points in the historical change curve are identified by image segmentation technology, and the long-term trend and short-term fluctuation features of the historical change curve are extracted by time series analysis; the key inflection points and trend features are used as global constraint inputs for the shape of the relative permeation curve in the joint constraint model; Historical change curves are derived from production monitoring databases or reservoir dynamic analysis reports. Since historical data is mostly stored in image format, this invention extracts key feature points using image digitization methods. And form a feature dataset, key feature points It reflects the breakthrough of water drive, the inflection point of recovery degree and other key dynamic stages, so as to unify the expression of local seepage information of experimental data and macroscopic dynamic characteristics of production history curve.

[0022] S2. Based on the experimental data and historical curve images of the target reservoir obtained in step S1, a joint constrained model (JCM) of relative permeability and water saturation is constructed to achieve a fusion modeling of local experimental accuracy and global dynamic continuity: The joint constraint model of relative permeability and water saturation takes the continuous functional relationship between relative permeability and water saturation as its core, and expresses the characteristics of relative permeability through parameterized functions; The method for constructing the joint constraint model of relative permeability and water saturation specifically includes the following steps: S21. The relationship between relative permeability of the oil phase and water saturation, the relationship between relative permeability of the water phase and water saturation, and the relationship between water cut and recovery degree in historical curve images are used as input constraints for the joint constraint model of relative permeability and water saturation. The coupling relationship between relative permeability, water saturation, water cut and recovery degree is established by introducing physical correlation equations. The relationship between the relative permeability of the oil phase and the water saturation is expressed as follows: The relationship between the relative permeability of the aqueous phase and the water saturation is expressed as follows: In the formula: The relative permeability of the oil phase is dimensionless. The relative permeability of the aqueous phase is dimensionless. Real-time water saturation, dimensionless; Effective water saturation, dimensionless; This is the oil phase permeability proportionality coefficient, which is dimensionless. is the permeability proportionality coefficient of the aqueous phase, dimensionless; The index represents the relationship between relative permeability of the oil phase and water saturation, and is dimensionless. The curve exponent is the relationship between relative permeability of the aqueous phase and water saturation. It is dimensionless and is used to control the rate of change of the curve shape in the low and high water content stages. Among them, effective water saturation The calculation formula is: In the formula: Effective water saturation, dimensionless; To constrain water saturation, dimensionless; Residual oil saturation, dimensionless; Real-time water saturation, dimensionless; The physical correlation equation is established based on multiphase flow theory, and the physical correlation equation includes water content. relative permeability of oil phase Relative permeability of water phase Oil phase viscosity viscosity with water phase The functional relationship (Equation 4) and the degree of extraction The functional relationship between water saturation and water content (Equation 5); The moisture content relative permeability of oil phase Relative permeability of water phase Oil phase viscosity viscosity with water phase The functional relationship is: In the formula: Moisture content, dimensionless; The relative permeability of the oil phase is dimensionless. The relative permeability of the aqueous phase is dimensionless. Oil phase viscosity, in units of ; Viscosity of the aqueous phase, in units of... ; As can be seen from the above formula (4), the moisture content The variation pattern is not only affected by water saturation The changes are influenced by the oil-water two-phase physical properties and the relationship between relative permeability; The degree of extraction The functional relationship (physical mapping relationship) between water saturation and water content can be expressed using an empirical or numerical fitting function as follows: In the formula: The degree of extraction is dimensionless; Real-time water saturation, dimensionless; To constrain water saturation, dimensionless; Reservoir porosity, dimensionless; For integration variables; As can be seen from the above formula (5), the extraction degree The functional relationship with water saturation further reflects the macroscopic production characteristics of the reservoir and embodies the cumulative recovery capacity of the reservoir at different water-cut stages. S22. In the process of constructing the joint constraint model of relative permeability and water saturation, the relationship curve between the relative permeability of the oil phase and water saturation in the experimental data ( ) and the curves showing the relationship between relative permeability of the water phase and water saturation ( As a local constraint, key feature points extracted from historical curve images are used. As a global constraint; The goal of the joint constraint model is to ensure that the predicted relative permeability changes continuously while maintaining the characteristic of continuous change in relative permeability. Curve and historical curve at key feature points Maintain consistency; The constraints of the joint constraint model of relative permeability and water saturation can be formally expressed as a comprehensive error minimization problem, which is functionally expressed as: In the formula: For the parameter set of the joint constraint model, and These are weighting coefficients, which respectively control the relative importance of experimental data accuracy constraints and historical curve continuity constraints; For joint constraint models in The predicted relative permeability value of the oil phase; For the first Measured values ​​of relative permeability of the oil phase at each experimental point; For joint constraint models in The predicted moisture content value is given below. For the first Historical actual values ​​of moisture content at each feature point; Through the constraints of the joint constraint model (Formula 6), the joint constraint model can mathematically guarantee the continuous differentiability of the relative permeability function, while maintaining the consistency between the oil-water two-phase flow law and the actual production behavior in a physical sense. Ultimately, the constructed joint constraint model of relative permeability and water saturation (formulas 1 to 6 together constitute the joint constraint model) formally realizes the cross-scale fusion from microscopic seepage experiments to macroscopic dynamic curves, providing a solid theoretical foundation and constraint framework for subsequent parameter solving based on collaborative optimization algorithms. S23. In the process of constructing the joint constraint model of relative permeability and water saturation, in order to more accurately characterize the seepage characteristics of multiphase fluids in the reservoir at different water saturation stages, fractal theory or the boundary point theory of capillary flow and film flow in unsaturated soil is used to integrate the reservoir's relative permeability prediction model (water content) with the water saturation level. relative permeability of oil phase Relative permeability of water phase Oil phase viscosity viscosity with water phase The functional relationship (Equation 4) is divided into a sub-model for the low moisture content stage and a sub-model for the high moisture content stage. By introducing a correction coefficient or a boundary point function, the sub-model can achieve a smooth transition at the moisture content boundary point. In the low water cut stage, the reservoir is mainly controlled by film flow and capillary action, with the relative permeability of the water phase being relatively low. The growth rate is relatively slow, while the relative penetration rate of the oil phase is relatively high. The downward trend is significantly affected by the wettability of the pore surface and the distribution of bound water; while in the high water content stage, the flow mechanism changes to continuous phase permeation flow, leading to Rising sharply and The decline has stabilized; The boundary point function is established based on the parameters of the soil-water characteristic curve SWCC, and is expressed as: In the formula: Represents the boundary point function; Indicates the moisture content boundary point; This is a smoothing coefficient used to control the steepness of the curve's transition near the dividing point; By By incorporating a joint constraint model that integrates overall relative permeability and water saturation, the model can... The low moisture content sub-model dominates, while The high water content sub-model dominates, thus preserving the staged seepage characteristics while ensuring the continuity of the curve; Moisture content boundary point Location determination is based on the parametric correlation of the Soil-Water Characteristic Curve (SWCC), through water saturation. With matrix suction The relational model is used to determine the saturation level corresponding to the critical suction force for the transition from capillary flow to continuous flow. The water content boundary point is determined, and the center position of the boundary point function is determined. By setting the boundary point function, the joint constraint model can be made hierarchical and adaptive in a physical sense. It can reflect the fractal characteristics of the micro-pore structure and maintain the smoothness and continuity of the relative permeability curve on the macro scale, providing a solid physical constraint basis for subsequent parameter optimization and curve generation. The water saturation With matrix suction The relational model is represented as: In the formula: The water saturation level is dimensionless. Matrix suction, unit: ; These are empirical parameters for SWCC, and are dimensionless. S3. The Collaborative Optimization Algorithm (COA) is used to solve the parameters of the joint constraint model of relative permeability and water saturation, which is constrained by both experimental data and historical production curves, to obtain the optimal set of relative permeability curve parameters. The parameter solution specifically includes the following steps: S31. The multi-agent cooperative optimization algorithm uses the constraint function (comprehensive error minimization function) of the joint constraint model of relative permeability and water saturation as the optimization objective function; the optimization objective of the objective function is to find a way to minimize the combined error. Minimal parameter set ; The objective function is to minimize the error between the predicted relative permeation curve and the experimental data at local points, while minimizing the difference between the predicted curve and the historical curve image in terms of global dynamic features. The optimization objective function is expressed as: In the formula: For parameters to be optimized, Indicates experimental data points, Indicates the characteristic points of the historical curve; and These are the weighting factors for experimental data and historical curves, used to balance local accuracy and global matching degree; Formula 9 is the representation of the solution obtained by the algorithm in Formula 6; S32. The core idea of ​​the collaborative optimization algorithm is to establish a two-layer structure of a global population cooperation layer and a local individual evolution layer. The global population cooperation layer is based on the Genetic Algorithm (GA) framework and is responsible for maintaining the diversity of the population and wide-area search. The local individual evolution layer is based on the Particle Swarm Optimization (PSO) mechanism and performs a refined search around the superior individuals selected by the genetic algorithm to accelerate convergence and improve accuracy. S321. In the global population cooperation layer, individuals, i.e., parameter vectors... Encoded in chromosome form, and evolving through selection, crossover, and mutation operations; Individual updates in a population are represented as follows: In the formula: These represent the selection, crossover, and mutation operators, respectively. The fitness function for each generation is defined as the fitness value. Used to evaluate the merits and demerits of individuals; S322. At the local individual evolution level, the particle swarm mechanism conducts local optimization near globally superior individuals. No. The velocity and position of each particle are updated according to the following formula: In the formula: For inertial weights, , For individual cognition and group learning factors, As a random factor, This is the optimal solution in the particle's history. To find the global optimal solution, the multi-agent collaborative optimization algorithm can adaptively adjust the search step size in the parameter space through this dynamic update mechanism, taking into account both global exploration and local convergence. S323. To avoid getting trapped in local optima, the multi-agent cooperative optimization algorithm performs a cross-layer cooperative operation every few generations: Specifically, this involves transferring a group of superior individuals from the global population cooperation layer to the local individual evolution layer for a refined search, and then using perturbation terms... The optimal solution is perturbed to maintain population diversity; The convergence condition of the multi-agent cooperative optimization algorithm is defined as follows: In the formula: A preset tolerance threshold is set; the algorithm stops when the objective function changes by less than this value for several consecutive iterations; the final optimal parameter set is obtained. As the steady-state solution of the joint constraint model, it ensures that the model achieves an optimal balance in both experimental data fitting and historical curve matching. S33. To verify the stability and robustness of the algorithm, multiple initial populations were randomly run, and variance analysis was performed on the results; if the variance of the parameters converges... If so, the algorithm can be considered to have good convergence consistency and solution stability under the reservoir conditions; If the variance of the parameter converges The optimization strategy was adjusted and optimized in a targeted manner from three dimensions: internal mechanism of the algorithm, parameter initialization, and constraints. The parameter solution process was re-executed until the variance met the threshold requirement. Multi-agent cooperative optimization (MAO) is a hybrid optimization method based on physical mechanisms. MAO iteratively adjusts parameters to ensure the predicted curve aligns with experimental data and historical curves across a range from low to high water content, maintaining continuity and physical plausibility. MAO combines the global search capabilities and local convergence characteristics of Genetic Algorithm (GA) and Particle Swarm Optimization (PSO), achieving stable solutions in high-dimensional nonlinear objective spaces through a multi-layered cooperative mechanism. MAO includes, but is not limited to, Genetic Algorithm (GA), PSO, or Bayesian optimization (BO). The multi-agent collaborative optimization algorithm, by jointly adjusting model parameters and constraints, ensures that the prediction results of relative permeability and water saturation remain continuous in both low and high water content ranges, and also satisfies the fluid flow law of the actual reservoir. The multi-agent cooperative optimization algorithm introduces a dynamic correction mechanism during the iteration process. When the optimized relative permeability curve deviates locally from the experimental data and historical curve images in a certain water saturation range, the multi-agent cooperative optimization algorithm compensates for the parameters through the dynamic correction mechanism, while maintaining the continuity constraint of the model in the global range from water content to high water content. The dynamic correction mechanism adjusts the parameters related to pore structure in the model. That is, the parameter compensation process introduces the correlation between the change in void ratio and matrix suction. It uses the reservoir pore-capillary characteristics to fine-tune the local curves, so that the curves can improve the local fitting accuracy without destroying the overall trend and multiphase flow law. This achieves synergistic optimization of local and global constraints and ensures physical rationality. The dynamic correction mechanism is triggered based on the determination of local deviations by the comprehensive error objective function in step S3. The parameter adjustment targets are the parameters related to pore structure in the core formula of relative permeability in step S2. Physical rationality is constrained by the correlation between void ratio and matrix suction established by the soil-water characteristic curve formula in step S2. It is a fine-tuning strategy for local deviations of the relative permeability curve in the algorithm iteration.

[0023] The parameter adjustment process of the multi-agent collaborative optimization algorithm includes dynamically introducing weight coefficients (α and β in the optimization objective function) of experimental data and historical curve images during the optimization process to balance the requirements of local accuracy and global continuity; the weight coefficients α and β are adaptively adjusted according to the heterogeneity characteristics of the reservoir.

[0024] The introduction of the collaborative optimization strategy in this application not only significantly improves the efficiency and accuracy of parameter inversion, but also realizes the adaptive fusion of seepage experimental information and production dynamic information in a physical sense, providing a robust parameter basis for the subsequent generation of relative permeability curves. S4. Generate the relative permeability curve of the target reservoir: Obtain the optimal relative permeability curve parameter set in step S3. Then, the relative permeability curve of the target reservoir is generated using the joint constraint model of relative permeability and water saturation constructed in step S2. S41. Based on the optimal relative permeability curve parameter set, the relative permeability of the oil phase and water saturation are directly calculated using continuous function form. and relative permeability and water saturation of the aqueous phase The correspondence is specifically represented as follows: in, In the formula: The effective water saturation in the optimal set of relative permeability curve parameters is dimensionless. The residual oil saturation in the optimal set of relative permeability curve parameters is dimensionless. The bound water saturation in the optimal set of parameters for the relative permeability curve is dimensionless. Real-time water saturation, dimensionless; The relative permeability of the oil phase is dimensionless, based on the optimal set of relative permeability curve parameters. The relative permeability of the aqueous phase is based on the optimal set of parameters for the relative permeability curve, and is dimensionless. is the oil phase permeability ratio coefficient in the optimal relative permeability curve parameter set, dimensionless; is the proportion coefficient of water phase permeability in the optimal set of parameters for the relative permeability curve, and is dimensionless. , is the exponent of the curve relating oil phase relative permeability to water saturation in the optimal relative permeability curve parameter set, and is dimensionless; , is the exponent of the curve relating the relative permeability of the aqueous phase to water saturation in the optimal set of relative permeability curve parameters, and is dimensionless; Based on the above oil phase relative permeability and water saturation and relative permeability and water saturation of the aqueous phase The continuous functional form of the corresponding relationship (Equations 14 and 15) ensures a smooth transition of the phase permeability curve between the low and high water cut stages, and the curve gradient conforms to the actual multiphase flow law of the reservoir: In the low water content stage, the relative permeability of the oil phase With water saturation The change is slow (the gradient of change is small), and the curve shape is consistent with the characteristics of thin film flow or capillary flow in the experimental data, indicating that the oil phase still dominates the flow. During the high water content stage, the relative permeability of the water phase With water saturation The change gradient is large, showing a rapid increase, and the curve shape is consistent with the characteristics of continuous phase permeation channels in the experimental data, reflecting the gradual establishment of the aqueous phase channel. S42. Based on the optimal relative permeability curve parameter set, the prediction results are mapped to the actual production curve through the functional relationship between water cut and recovery degree, which is used to dynamically verify the matching of reservoir oil production, water cut and recovery degree. The functional relationship between water rate and extraction degree based on the optimal relative permeability curve parameter set is as follows: In the formula: The relative permeability of the oil phase is dimensionless, based on the optimal set of relative permeability curve parameters. The relative permeability of the aqueous phase is based on the optimal set of parameters for the relative permeability curve, and is dimensionless. Oil phase viscosity, in units of ; Viscosity of the aqueous phase, in units of... ; The method of this invention ensures performance at low moisture content stages. Follow The change is slow, reflecting the characteristics of film flow; in the high water content stage Rapidly increasing the number of channels reflects the characteristics of a continuous aqueous phase, while avoiding the problems of error propagation and curve discontinuity caused by step-by-step modeling.

[0025] S43. Result Verification: The generated relative permeability curves are substituted into numerical simulation software to simulate the dynamic changes in oil production, water cut, and recovery rate of the reservoir, and compared with the historical curves of actual production of the target reservoir.

[0026] The numerical simulation software can be Eclipse (Schlumberger) or Petrel RE (Schlumberger Reservoir Numerical Simulation Module). Example 1

[0027] like Figures 1-4 As shown, a reservoir revelocity permeability modeling method integrating multi-source data and agent constraints includes the following steps: S1. Obtain experimental data and historical curve images of the target reservoir: The experimental data came from core displacement experiments, by controlling different water saturation levels. Displacement process under certain conditions, and determination of relative permeability of the oil phase. relative permeability with water The measured values, and the water saturation range involved in the experiment need to cover the actual production range of the reservoir, from the bound water saturation To residual oil saturation ,For example = 0.20, = 0.25. To ensure data continuity and reliability, at least 10 to 15 sets of data points should be acquired within this interval to form... and The experimental scatter set; At the same time, obtain the water cut during the actual production process of the reservoir. With extraction degree Historical change curves, typically derived from production monitoring databases or reservoir dynamic analysis reports, are often stored as images. This invention utilizes image recognition and time series extraction algorithms to digitize these curve images. Curve contours are extracted through image edge detection, and polynomial fitting and time series smoothing analysis are combined to decompose the long-term trend and short-term fluctuation characteristics of the curves. This identifies key feature points, including waterdrive breakthrough points corresponding to sudden increases in water cut and inflection points in the relationship between recovery rate and water cut curves. For example, when… = 0.35 appears The phenomenon of a sudden increase from 30% to 50% is identified as the breakthrough stage of water-driven hydroelectricity, ultimately forming a feature dataset. As a global constraint input for the shape of the relative permeability curve in subsequent models, it enables the local seepage information of the experimental point to be uniformly expressed with the macroscopic characteristics of the production dynamic curve.

[0028] S2. Based on the experimental data and historical curve images of the target reservoir obtained in step S1, a joint constrained model (JCM) of relative permeability and water saturation is constructed to achieve a fusion modeling of local experimental accuracy and global dynamic continuity: The joint constraint model of relative permeability and water saturation takes the continuous functional relationship between relative permeability and water saturation as its core, and expresses the characteristics of relative permeability through parameterized functions; S21. The relationship between relative permeability of the oil phase and water saturation, the relationship between relative permeability of the water phase and water saturation, and the relationship between water cut and recovery degree in historical curve images are used as input constraints for the joint constraint model of relative permeability and water saturation. The coupling relationship between relative permeability, water saturation, water cut and recovery degree is established by introducing physical correlation equations. The relationship between the relative permeability of the oil phase and the water saturation is expressed as follows: The relationship between the relative permeability of the aqueous phase and the water saturation is expressed as follows: In the formula: The relative permeability of the oil phase is dimensionless. The relative permeability of the aqueous phase is dimensionless. Real-time water saturation, dimensionless; Effective water saturation, dimensionless; This is the oil phase permeability proportionality coefficient, which is dimensionless. is the permeability proportionality coefficient of the aqueous phase, dimensionless; The index represents the relationship between relative permeability of the oil phase and water saturation, and is dimensionless. The curve exponent is the relationship between relative permeability of the aqueous phase and water saturation. It is dimensionless and is used to control the rate of change of the curve shape in the low and high water content stages. Among them, effective water saturation The calculation formula is: In the formula: Effective water saturation, dimensionless; To constrain water saturation, dimensionless; Residual oil saturation, dimensionless; Real-time water saturation, dimensionless; The physical correlation equation is established based on multiphase flow theory, and the physical correlation equation includes water content. relative permeability of oil phase Relative permeability of water phase Oil phase viscosity viscosity with water phase The functional relationship (Equation 4) and the degree of extraction The functional relationship between water saturation and water content (Equation 5); The moisture content relative permeability of oil phase Relative permeability of water phase Oil phase viscosity viscosity with water phase The functional relationship is: In the formula: Moisture content, dimensionless; The relative permeability of the oil phase is dimensionless. The relative permeability of the aqueous phase is dimensionless. Oil phase viscosity, in units of ; Viscosity of the aqueous phase, in units of... ; To further reflect the macroscopic production characteristics of the reservoir, the degree of recovery will be... A physical mapping relationship is established with water saturation, which can be expressed using an empirical or numerical fitting function as follows: In the formula: The degree of extraction is dimensionless; Real-time water saturation, dimensionless; To constrain water saturation, dimensionless; Reservoir porosity, dimensionless; For integration variables; S22. In the process of constructing the joint constraint model of relative permeability and water saturation, the relationship curve between the relative permeability of the oil phase and water saturation in the experimental data ( ) and the curves showing the relationship between relative permeability of the water phase and water saturation ( As a local constraint, key feature points extracted from historical curve images are used. As a global constraint; The goal of the joint constraint model is to ensure that the predicted relative permeability changes continuously while maintaining the characteristic of continuous change in relative permeability. Curve and historical curve at key feature points Maintain consistency; The constraints of the joint constraint model of relative permeability and water saturation can be formally expressed as a problem of minimizing the comprehensive error, and the function for minimizing the comprehensive error is expressed as: In the formula: For the parameter set of the joint constraint model, and These are weighting coefficients, which respectively control the relative importance of experimental data accuracy constraints and historical curve continuity constraints; For joint constraint models in The predicted relative permeability value of the oil phase; For the first Measured values ​​of relative permeability of the oil phase at each experimental point; For joint constraint models in The predicted moisture content value is given below. For the first Historical actual values ​​of moisture content at each feature point; S23. In the process of constructing the joint constraint model of relative permeability and water saturation, in order to more accurately characterize the seepage characteristics of multiphase fluids in the reservoir at different water saturation stages, the relative permeability prediction model of the reservoir is divided into a sub-model for the low water saturation stage and a sub-model for the high water saturation stage by using fractal theory or the boundary point theory of capillary flow and film flow in unsaturated soil. By introducing correction coefficients or boundary point functions, the sub-models achieve a smooth transition at the water saturation boundary point. In the low water cut stage, the reservoir is mainly controlled by film flow and capillary action, with the relative permeability of the water phase being relatively low. The growth rate is relatively slow, while the relative penetration rate of the oil phase is relatively high. The downward trend is significantly affected by the wettability of the pore surface and the distribution of bound water; while in the high water content stage, the flow mechanism changes to continuous phase permeation flow, leading to Rising sharply and The decrease tends to stabilize; to accurately describe the above phase transition behavior, this method divides the joint constraint model into a low water content sub-model and a high water content sub-model, and introduces a boundary point function at the junction of the two. Used to achieve the curve at the moisture content boundary point The smooth transition at the boundary point is expressed in the form of the boundary point function as follows: In the formula: Represents the boundary point function; Indicates the moisture content boundary point; This is a smoothing coefficient used to control the steepness of the curve's transition near the dividing point; By By incorporating a joint constraint model that integrates overall relative permeability and water saturation, the model can... The low moisture content sub-model dominates, while The high water content sub-model dominates, thus preserving the staged seepage characteristics while ensuring the continuity of the curve; Moisture content boundary point The location is determined based on the parametric correlation of the Soil-Water Characteristic Curve (SWCC), through water saturation. With matrix suction The relational model is used to determine the saturation level corresponding to the critical suction force for the transition from capillary flow to continuous flow. The water content boundary point is determined, and the center position of the boundary point function is determined. By setting the boundary point function, the model can be made to have hierarchy and adaptability in a physical sense. It can reflect the fractal characteristics of the micro-pore structure and maintain the smoothness and continuity of the relative permeability curve on a macro scale, providing a solid physical constraint basis for subsequent parameter optimization and curve generation. The water saturation With matrix suction The relational model is represented as: In the formula: The water saturation level is dimensionless. Matrix suction, unit: ; These are empirical parameters for SWCC, dimensionless; using this relationship, the saturation level corresponding to the critical suction for the transition from capillary flow to continuous flow can be calculated. This allows the center position of the boundary point function to be determined. Through the above settings, the model can be made to have hierarchy and adaptability in a physical sense. It can reflect the fractal characteristics of the micro-pore structure and maintain the smoothness and continuity of the phase permeation curve on a macro scale, providing a solid physical constraint basis for subsequent parameter optimization and curve generation.

[0029] S3. The Collaborative Optimization Algorithm (COA) is used to solve the parameters of the joint constraint model of relative permeability and water saturation, which is constrained by both experimental data and historical production curves, to obtain the optimal set of relative permeability curve parameters. The multi-agent cooperative optimization algorithm iteratively adjusts parameters to ensure that the predicted curve is consistent with experimental data and historical curves within the range of low to high water content, maintaining continuity and physical rationality. The multi-agent cooperative optimization algorithm combines the global search capability and local convergence characteristics of the genetic algorithm (GA) and particle swarm optimization (PSO), and achieves stable solutions in high-dimensional nonlinear objective spaces through a multi-layer cooperative mechanism. The solution process specifically includes the following steps: S31. The multi-agent cooperative optimization algorithm uses the constraint function (comprehensive error minimization function) of the joint constraint model of relative permeability and water saturation as the optimization objective function; the optimization objective of the objective function is to find a way to minimize the combined error. Minimal parameter set ; The objective function is to minimize the error between the predicted relative permeation curve and the experimental data at local points, while minimizing the difference between the predicted curve and the historical curve image in terms of global dynamic features. The optimization objective function is expressed as: In the formula: For parameters to be optimized, Indicates experimental data points, Indicates the characteristic points of the historical curve; and These are the weighting factors for experimental data and historical curves, used to balance local accuracy and global matching degree; S32. The core idea of ​​the collaborative optimization algorithm is to establish a two-layer structure of a global population cooperation layer and a local individual evolution layer. The global population cooperation layer is based on the Genetic Algorithm (GA) framework and is responsible for maintaining the diversity of the population and wide-area search. The local individual evolution layer is based on the Particle Swarm Optimization (PSO) mechanism and performs a refined search around the superior individuals selected by the genetic algorithm to accelerate convergence and improve accuracy. S321. In the global population cooperation layer, individuals, i.e., parameter vectors... Encoded in chromosome form, and evolving through selection, crossover, and mutation operations; Individual updates in a population are represented as follows: In the formula: These represent the selection, crossover, and mutation operators, respectively. The fitness function for each generation is defined as the fitness value. Used to evaluate the merits and demerits of individuals; S322. At the local individual evolution level, the particle swarm mechanism conducts local optimization near globally superior individuals. No. The velocity and position of each particle are updated according to the following formula: In the formula: For inertial weights, , For individual cognition and group learning factors, As a random factor, This is the optimal solution in the particle's history. To find the global optimal solution, the multi-agent collaborative optimization algorithm can adaptively adjust the search step size in the parameter space through this dynamic update mechanism, taking into account both global exploration and local convergence. S323. To avoid getting trapped in local optima, the multi-agent cooperative optimization algorithm performs a cross-layer cooperative operation every few generations: Specifically, this involves transferring a group of superior individuals from the global population cooperation layer to the local individual evolution layer for a refined search, and then using perturbation terms... The optimal solution is perturbed to maintain population diversity; The convergence condition of the multi-agent cooperative optimization algorithm is defined as follows: In the formula: A preset tolerance threshold is set; the algorithm stops when the objective function changes by less than this value for several consecutive iterations; the final optimal parameter set is obtained. As the steady-state solution of the joint constraint model, it ensures that the model achieves an optimal balance in both experimental data fitting and historical curve matching. S33. To verify the stability and robustness of the algorithm, multiple initial populations were randomly run, and variance analysis was performed on the results; if the variance of the parameters converges... If so, the algorithm can be considered to have good convergence consistency and solution stability under the reservoir conditions; S4. Generate the relative permeability curve of the target reservoir: Obtain the optimal relative permeability curve parameter set in step S3. Then, the relative permeability curve of the target reservoir is generated using the joint constraint model of relative permeability and water saturation constructed in step S2. S41. Based on the optimal relative permeability curve parameter set, the relative permeability of the oil phase and water saturation are directly calculated using continuous function form. and relative permeability and water saturation of the aqueous phase The correspondence is specifically represented as follows: in, In the formula: The effective water saturation in the optimal set of relative permeability curve parameters is dimensionless. The residual oil saturation in the optimal set of relative permeability curve parameters is dimensionless. The bound water saturation in the optimal set of parameters for the relative permeability curve is dimensionless. Real-time water saturation, dimensionless; The relative permeability of the oil phase is dimensionless, based on the optimal set of relative permeability curve parameters. The relative permeability of the aqueous phase is based on the optimal set of parameters for the relative permeability curve, and is dimensionless. is the oil phase permeability ratio coefficient in the optimal relative permeability curve parameter set, dimensionless; is the proportion coefficient of water phase permeability in the optimal set of parameters for the relative permeability curve, and is dimensionless. , is the exponent of the curve relating oil phase relative permeability to water saturation in the optimal relative permeability curve parameter set, and is dimensionless; , is the exponent of the curve relating the relative permeability of the aqueous phase to water saturation in the optimal set of relative permeability curve parameters, and is dimensionless; This continuous function form ensures a smooth transition of the phase permeability curve between the low and high water cut stages, and the curve gradient conforms to the actual multiphase flow characteristics of the reservoir. In the low water content stage, the relative permeability of the oil phase With water saturation The change is slow (the gradient of change is small), and the curve shape is consistent with the characteristics of thin film flow or capillary flow in the experimental data, indicating that the oil phase still dominates the flow. During the high water content stage, the relative permeability of the water phase With water saturation The change gradient is large, showing a rapid increase, and the curve shape is consistent with the characteristics of continuous phase permeation channels in the experimental data, reflecting the gradual establishment of the aqueous phase channel. S42. To verify the physical rationality and production applicability of the curve, based on the optimal relative permeability curve parameter set, the prediction results are mapped to the actual production curve through the functional relationship between water cut and recovery degree, so as to dynamically verify the matching of reservoir oil production, water cut and recovery degree. The functional relationship between water rate and extraction degree based on the optimal relative permeability curve parameter set is as follows: In the formula: The relative permeability of the oil phase is dimensionless, based on the optimal set of relative permeability curve parameters. The relative permeability of the aqueous phase is based on the optimal set of parameters for the relative permeability curve, and is dimensionless. Oil phase viscosity, in units of ; Viscosity of the aqueous phase, in units of... ; This function (Equation 17) can map the experimentally obtained relative permeability relationship to the actual production curve, thereby obtaining the predicted water cut-extraction degree. If the predicted curve and the historical curve are consistent at key stages, it indicates that the generated relative permeability curve can accurately reflect the dynamic characteristics of the reservoir. The method ensures that during the low moisture content stage Follow The change is slow, reflecting the characteristics of film flow; in the high water content stage Rapidly increasing the number of channels reflects the characteristics of a continuous aqueous phase, while avoiding the problems of error propagation and curve discontinuity caused by step-by-step modeling.

[0030] S43. Result Verification: The generated relative permeability curves are substituted into numerical simulation software to simulate the dynamic changes in oil production, water cut, and recovery rate of the reservoir, and compared with the historical curves of actual production of the target reservoir.

[0031] When anomalies are found in the verification results, such as the extraction degree exceeding the empirical threshold when the moisture content reaches 98%, iterative corrections are made by adjusting model parameters or optimizing algorithm weights until the deviation between the simulated curve and the actual production data in indicators such as moisture content, extraction degree, and endpoint effect is less than the set error. The final output is then determined. and The curves maintain continuity and physical consistency throughout the entire range and can be directly used as input parameters for reservoir numerical simulation and development scheme optimization.

[0032] The applicant declares that the above description is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Those skilled in the art should understand that any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention fall within the protection and disclosure scope of the present invention.

Claims

1. A reservoir revelocity permeability modeling method integrating multi-source data and agent constraints, characterized in that: Includes the following steps: S1. Obtain experimental data and historical curve images of the target reservoir; S2. Based on the experimental data and historical curve images of the target reservoir obtained in step S1, construct a joint constraint model of relative permeability and water saturation. S3. A multi-agent collaborative optimization algorithm is used to solve the parameters of the joint constraint model of relative permeability and water saturation, which is constrained by both experimental data and historical production curves, to obtain the optimal set of relative permeability curve parameters: S4. Based on the optimal relative permeability curve parameter set obtained in step S3, the relative permeability curve of the target reservoir is generated using the joint constraint model of relative permeability and water saturation constructed in step S2.

2. The reservoir revelocity permeability modeling method integrating multi-source data and agent constraints according to claim 1, characterized in that: The experimental data include the measured values ​​of the relative permeability of the oil phase and the relative permeability of the water phase at different water saturation levels. The historical curve images include historical variation curves of water cut and recovery rate during actual reservoir production. Key feature points of the historical variation curves are extracted using image digitization methods to form a feature dataset.

3. The reservoir revelocity permeability modeling method integrating multi-source data and agent constraints according to claim 2, characterized in that: The range of the water saturation value is from bound water saturation to residual oil saturation, and includes the boundary value between bound water saturation and residual oil saturation.

4. The reservoir revelocity permeability modeling method integrating multi-source data and agent constraints according to claim 1, characterized in that: The method for constructing the joint constraint model of relative permeability and water saturation specifically includes the following steps: S21. The relationship between relative permeability of the oil phase and water saturation, the relationship between relative permeability of the water phase and water saturation, and the relationship between water cut and recovery degree in historical curve images are used as input constraints for the joint constraint model of relative permeability and water saturation. The coupling relationship between relative permeability, water saturation, water cut and recovery degree is established by introducing physical correlation equations. The relationship between the relative permeability of the oil phase and the water saturation is expressed as follows: The relationship between the relative permeability of the aqueous phase and the water saturation is expressed as follows: In the formula: The relative permeability of the oil phase is dimensionless. The relative permeability of the aqueous phase is dimensionless. Real-time water saturation, dimensionless; Effective water saturation, dimensionless; This is the oil phase permeability proportionality coefficient, which is dimensionless. is the permeability proportionality coefficient of the aqueous phase, dimensionless; The index represents the relationship between relative permeability of the oil phase and water saturation, and is dimensionless. The exponent of the curve relating the relative permeability of the aqueous phase to water saturation is dimensionless. Among them, effective water saturation The calculation formula is: In the formula: Effective water saturation, dimensionless; To constrain water saturation, dimensionless; Residual oil saturation, dimensionless; Real-time water saturation, dimensionless; The physical correlation equation includes moisture content. relative permeability of oil phase Relative permeability of water phase Oil phase viscosity viscosity with water phase The functional relationship and the degree of extraction The functional relationship between water content and water saturation; The moisture content relative permeability of oil phase Relative permeability of water phase Oil phase viscosity viscosity with water phase The functional relationship is: In the formula: Moisture content, dimensionless; The relative permeability of the oil phase is dimensionless. The relative permeability of the aqueous phase is dimensionless. Oil phase viscosity, in units of ; Viscosity of the aqueous phase, in units of... ; The degree of extraction The functional relationship between water content and water saturation is: In the formula: The degree of extraction is dimensionless; Real-time water saturation, dimensionless; To constrain water saturation, dimensionless; Reservoir porosity, dimensionless; For integration variables; S22. Using the curves showing the relationship between relative permeability and water saturation in the oil phase and the water phase from the experimental data as local constraints, and the key feature points extracted from the historical curve images as global constraints; the local and global constraints of the joint constraint model of relative permeability and water saturation are combined to minimize the comprehensive error, and the function for minimizing the comprehensive error is expressed as: In the formula: For the parameter set of the joint constraint model, and These are weighting coefficients, which respectively control the relative importance of experimental data accuracy constraints and historical curve continuity constraints; For joint constraint models in The predicted relative permeability value of the oil phase; For the first Measured values ​​of relative permeability of the oil phase at each experimental point; For joint constraint models in The predicted moisture content value is given below. For the first Historical actual values ​​of moisture content at each feature point; S23. The combined formulas (1) to (6) constitute a joint constraint model of relative permeability and water saturation. S24. Utilizing fractal theory or the boundary point theory between capillary flow and thin film flow in unsaturated soils, the water content... relative permeability of oil phase Relative permeability of water phase Oil phase viscosity viscosity with water phase The functional relationship is divided into sub-functional relationships for the low moisture content stage and sub-functional relationships for the high moisture content stage. By introducing a boundary point function, the sub-model is made to operate at the moisture content boundary point. A smooth transition is achieved at the point.

5. The reservoir reactivity modeling method integrating multi-source data and agent constraints according to claim 4, characterized in that: The boundary point function is established based on the parameters of the soil-water characteristic curve SWCC, and is expressed as: In the formula: Represents the boundary point function; Indicates the moisture content boundary point; This is a smoothing coefficient used to control the steepness of the transition of the curve near the dividing point.

6. The reservoir reactivity modeling method integrating multi-source data and agent constraints according to claim 4, characterized in that: The moisture content dividing point The location is determined based on the parametric correlation of the soil-water characteristic curve, through water saturation. With matrix suction The relational model is used to determine the saturation level corresponding to the critical suction force for the transition from capillary flow to continuous flow. That is, the moisture content dividing point; The water saturation With matrix suction The relational model is represented as: In the formula: The water saturation level is dimensionless. This represents matrix suction, expressed in units of . These are empirical parameters for SWCC, and are dimensionless.

7. The reservoir revelocity permeability modeling method integrating multi-source data and agent constraints according to claim 1, characterized in that: The parameter solution in step S3 specifically includes the following steps: S31. The multi-agent cooperative optimization algorithm uses the comprehensive error minimization function of the joint constraint model of relative permeability and water saturation as the optimization objective function; the optimization objective of the objective function is to find a function that minimizes the comprehensive error of the model ... Minimal parameter set ; The optimization objective function is expressed as: In the formula: For parameters to be optimized, Indicates experimental data points, Indicates the characteristic points of the historical curve; and These are the weighting factors for experimental data and historical curves, used to balance local accuracy and global matching degree; S32. Establish a two-layer structure of global population cooperation layer and local individual evolution layer. The global population cooperation layer is based on the genetic algorithm framework and is responsible for maintaining the diversity of the population and wide-area search. The local individual evolution layer is based on the particle swarm optimization mechanism and performs a refined search around the superior individuals selected by the genetic algorithm. S33. Randomize the test for multiple initial populations and perform variance analysis on the results; if the variance of the parameters converges... If the algorithm exhibits good convergence consistency and solution stability under the given reservoir conditions, it can be considered that the algorithm has good solution stability.

8. The reservoir reactivity modeling method integrating multi-source data and agent constraints according to claim 7, characterized in that: Step S32 specifically includes the following steps: S321. In the global population cooperation layer, individuals, i.e., parameter vectors... Encoded in chromosome form, and evolving through selection, crossover, and mutation; Individual updates in a population are represented as follows: In the formula: These represent the selection, crossover, and mutation operators, respectively. The fitness function for each generation is defined as the fitness value. Used to evaluate the merits and demerits of individuals; S322. At the local individual evolution level, the particle swarm mechanism conducts local optimization near globally superior individuals. No. The velocity and position of each particle are updated according to the following formula: In the formula: For inertial weights, , For individual cognition and group learning factors, As a random factor, This is the optimal solution in the particle's history. This is the globally optimal solution; S323, the multi-agent cooperative optimization algorithm performs a cross-layer cooperative operation once every few generations: Specifically, this involves transferring a group of superior individuals from the global population cooperation layer to the local individual evolution layer for a refined search, and then using perturbation terms... The optimal solution is perturbed to maintain population diversity; The convergence condition of the algorithm is defined as follows: In the formula: A preset tolerance threshold is set; the algorithm stops when the objective function changes by less than this value for several consecutive iterations; the final optimal parameter set is obtained. As the steady-state solution of the joint constraint model.

9. The reservoir reactivity modeling method integrating multi-source data and agent constraints according to claim 1, characterized in that: Step S4 specifically includes the following steps: S41. Based on the optimal relative permeability curve parameter set, the relative permeability of the oil phase and water saturation are directly calculated using continuous function form. and relative permeability and water saturation of the aqueous phase The correspondence is specifically represented as follows: in, In the formula: The effective water saturation in the optimal set of relative permeability curve parameters is dimensionless. The residual oil saturation in the optimal set of relative permeability curve parameters is dimensionless. The bound water saturation in the optimal set of parameters for the relative permeability curve is dimensionless. Real-time water saturation, dimensionless; The relative permeability of the oil phase is dimensionless, based on the optimal set of relative permeability curve parameters. The relative permeability of the aqueous phase is based on the optimal set of parameters for the relative permeability curve, and is dimensionless. is the oil phase permeability ratio coefficient in the optimal relative permeability curve parameter set, dimensionless; is the proportion coefficient of water phase permeability in the optimal set of parameters for the relative permeability curve, and is dimensionless. , is the exponent of the curve relating oil phase relative permeability to water saturation in the optimal relative permeability curve parameter set, and is dimensionless; , is the exponent of the curve relating the relative permeability of the aqueous phase to water saturation in the optimal set of relative permeability curve parameters, and is dimensionless; S42. Based on the optimal set of relative permeability curve parameters, the prediction results are mapped to the actual production curve through the functional relationship between water cut and extraction degree. The functional relationship between water rate and extraction degree based on the optimal relative permeability curve parameter set is as follows: In the formula: The relative permeability of the oil phase is dimensionless, based on the optimal set of relative permeability curve parameters. The relative permeability of the aqueous phase is based on the optimal set of parameters for the relative permeability curve, and is dimensionless. Oil phase viscosity, in units of ; Viscosity of the aqueous phase, in units of... ; S43. Result Verification: The generated relative permeability curves are substituted into numerical simulation software to simulate the dynamic changes in oil production, water cut, and recovery rate of the reservoir, and compared with the historical curves of actual production of the target reservoir.

Images