A reservoir ground stress and pore pressure identification method, system, device and medium based on multi-well DFIT structured semantic features

Abstract

The application discloses a reservoir stress and pore pressure identification method and system based on multi-well DFIT structured semantic features, electronic equipment and a computer readable storage medium. The method acquires and preprocesses DFIT test data of multiple well times, extracts structured semantic features such as pre-closing pressure response, post-closing pressure response, fitting error, well basic information and physical constraint parameters; on this basis, multi-well correlation features and feature input vectors are constructed, and through a reservoir stress and pore pressure identification calculation model, the closing pressure, pore pressure, minimum horizontal principal stress, maximum horizontal principal stress orientation or value interval and result confidence are output; then the output results are subjected to consistency verification, abnormality identification and sample set updating. The application is not limited to a single well closure identification chart, a certain pressure decay formula or a certain specific algorithm implementation, but through multi-well DFIT data standardization, structured semantic feature construction, multi-well correlation constraint and calculation result verification and updating, the consistency, traceability and engineering reusability of multi-well DFIT interpretation are improved.

Description

Technical Field

[0001] This invention belongs to the technical field of reservoir stress and pore pressure interpretation and reservoir geomechanical evaluation. Specifically, it relates to a method, system, electronic device and computer-readable storage medium for reservoir stress and pore pressure identification based on multi-well DFIT structured semantic features.

[0002] This invention is applicable to tight sandstone reservoirs, shale reservoirs, unconventional oil and gas reservoirs, gas storage reservoirs, low-permeability reservoirs, and other engineering scenarios that require evaluation of closure pressure, pore pressure, minimum horizontal principal stress, and regional stress characteristics using diagnostic fracture injection test data. Background Technology

[0003] DFIT, or Diagnostic Fracture Injection Test, is a commonly used field testing method in oil and gas reservoir development, fracturing design, and reservoir geomechanical evaluation. Through small-scale injection, well shut-in after pump shutdown, and pressure decay response analysis, it is possible to obtain the closure pressure, minimum horizontal principal stress, pore pressure, flow regime, reservoir pressure response, and basic parameters required for fracturing design.

[0004] Current DFIT interpretation typically revolves around single-well test curves. Interpreters comprehensively assess closure pressure, pore pressure, and related stress parameters based on pressure decay curves, derivative curves, pre-closure response, post-closure response, flow regime changes, and fitting results. This method can yield effective results when the single-well data is of good quality, the curve characteristics are clear, and the interpreters are experienced.

[0005] However, in engineering scenarios where multi-well, multi-layer, and multi-phase test data accumulate rapidly, the traditional single-well interpretation process has significant shortcomings. First, the pressure units, time bases, sampling intervals, test phase divisions, wellhead and bottomhole pressure fields, quality identifiers, and interpretation result formats are not uniform across different wells, making it difficult to directly compare, verify, and utilize multi-well data in a regionalized manner.

[0006] Secondly, the determination of closure pressure, pore pressure, and flow regime often relies on the experience of the interpreters. For wells with high noise, unclear closure characteristics, insufficient post-closure flow regime, insufficient testing time window, or multiple interpretation results, different personnel, different software, or different processing procedures may give different interpretation results, resulting in insufficient consistency and repeatability of interpretations.

[0007] Furthermore, existing methods typically rely on manual interpretation of pressure decay maps, derivative plots, and fitting results, lacking structured representation of information such as curve morphology, curvature changes, effective fitting time windows, residual distribution, multi-start point stability, adjacent well consistency, and the degree to which physical constraints are met. A significant amount of information valuable for interpreting reliability remains confined to maps or text reports, making it difficult to systematically access.

[0008] Furthermore, historically interpreted DFIT samples are often scattered across reports, tables, maps, or individual interpretation documents, lacking an updatable sample set and a unified index. Interpretive experience already verified by experts is difficult to distill into reusable data for subsequent well interpretations, resulting in repeated interpretations and difficulty in quality tracking of data within the same region, stratigraphic level, and pressure system.

[0009] It should also be noted that reservoir stress and pore pressure exhibit significant spatial continuity, stratigraphic correlation, and pressure system constraints. Interpreting solely based on a single DFIT curve from the target well may lead to problems such as abrupt changes in results from adjacent wells, discontinuities in pore pressure within the same pressure system, abnormal stress trends with depth within the same stratigraphic level, and the neglect of historical interpretation results from adjacent wells.

[0010] Therefore, it is necessary to propose an identification method for multi-well DFIT data, which unifies the standardization of raw test data, extraction of structured semantic features, construction of multi-well correlation features, expression of physical constraint parameters, solution of computational models, verification of results and updating of historical samples. Without limiting a single-well interpretation chart, a pressure decay formula or a specific algorithm implementation, this method can improve the consistency, traceability and engineering reusability of multi-well DFIT interpretation results. Summary of the Invention

[0011] The purpose of this invention is to provide a method, system, electronic device, and computer-readable storage medium for identifying reservoir stress and pore pressure based on multi-well DFIT structured semantic features, in order to solve problems in the prior art such as inconsistent standards for multi-well DFIT data, strong reliance on manual interpretation of single wells, difficulty in reusing historical samples, insufficient embedding of adjacent wells and physical constraints, and lack of consistency verification and anomaly review basis for interpretation results.

[0012] The core of this invention does not lie in protecting a single-well closure pressure identification chart, a G-function or derivative plot criterion, a pressure decay analytical formula, a pore pressure calculation formula, or a specific algorithm implementation. What this invention aims to protect is the overall engineering technical solution formed by organizing multi-well DFIT data, expressing structured semantic features, constructing multi-well correlation features, fusing physical constraint parameters, solving computational models, and verifying and updating results.

[0013] To achieve the above objectives, the present invention provides a method for identifying reservoir stress and pore pressure based on multi-well DFIT structured semantic features, comprising the following steps.

[0014] S1, Data Acquisition and Preprocessing. Acquire DFIT test data, well foundation information, stratigraphic information, construction information, well logging interpretation information, production or injection history information, and existing interpretation results from multiple well runs; perform unit unification, time benchmark unification, stage division, outlier identification, missing value identification, and quality evaluation on the DFIT test data to generate standardized data that can be used for subsequent calculations.

[0015] S2, Structured Semantic Feature Extraction. Based on standardized data, pre-closure pressure response features, post-closure pressure response features, and fitting error features are extracted. Combined with well basic information features and physical constraint parameters, a set of computable, indexable, and verifiable structured semantic features is formed.

[0016] S3. Construction of Multi-Well Correlation Features. Based on the spatial distance, vertical distance, stratigraphic similarity, pressure system similarity, lithological similarity, differences in production or injection history, and differences in existing interpretation results between the target well and its neighboring wells, multi-well correlation features are constructed to constrain the spatial continuity and engineering rationality of the target well identification results.

[0017] S4, Feature Input Vector Construction. The structured semantic features, multi-well correlation features, and physical constraint parameters are standardized, missing values ​​are handled, features are combined, weights are assigned, and dimensions are encoded to form the feature input vector.

[0018] S5, Solve the computational model. Input the feature input vector into the reservoir stress and pore pressure identification computational model, and output one or more of the following: closure pressure, pore pressure, minimum horizontal principal stress, maximum horizontal principal stress orientation or value range, flow regime category, and confidence level.

[0019] S6, Result Verification and Update. Physical consistency verification, adjacent well consistency verification, anomaly deviation identification, and review priority ranking are performed on the output results; confirmed identification results are updated to the historical sample set, interpretation result database, or regional results database.

[0020] This invention also provides a reservoir stress and pore pressure identification system based on multi-well DFIT structured semantic features, including a database module, a data processing unit, a feature extraction unit, a multi-well association unit, a calculation model unit, an output verification unit, and a result update unit. Each unit can be deployed in the same computing device or distributed according to the business process of database, interpretation calculation, result output, and engineering review.

[0021] The present invention also provides an electronic device, including a processor and a memory, wherein the memory stores a computer program, which, when executed by the processor, implements the above-described method.

[0022] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method.

[0023] Compared with existing technologies, this invention has at least the following beneficial effects: First, through a multi-well DFIT structured database and standardized preprocessing, DFIT data from different wells, different stratigraphic levels, and different test batches have a unified data entry point, quality identification, and result tracking path; Second, through structured semantic features such as pre-closure pressure response, post-closure pressure response, and fitting error, traditional manual interpretation experience is transformed into computable features; Third, through multi-well correlation features and physical constraint parameters, the impact of local misjudgments of single-well curves on regional results is reduced; Fourth, through the mapping between feature input vectors and multi-parameter output vectors, the closure pressure, pore pressure, minimum horizontal principal stress, and maximum horizontal principal stress orientation or value range can be identified within a unified framework; Fifth, through result verification, anomaly identification, and sample update mechanisms, a traceable parameter basis is provided for batch DFIT interpretation, fracturing design, reservoir evaluation, gas storage operation pressure control, and regional geomechanical modeling. Attached Figure Description

[0024] Figure 1 The flowchart of the reservoir stress and pore pressure identification method based on multi-well DFIT structured semantic features provided by the present invention is shown.

[0025] Figure 2 The structure diagram of the multi-well DFIT structured semantic database provided by this invention.

[0026] Figure 3 A schematic diagram of the structured parameter calculation module provided by this invention.

[0027] Figure 4 The structural diagram of the automatic reservoir stress and pore pressure identification system provided by the present invention is shown.

[0028] Figure 5 This is a schematic diagram illustrating the parameter construction stage and engineering application stage provided by the present invention.

[0029] Figure 6 This is a schematic diagram illustrating the mapping relationship between multi-well feature inputs and multi-parameter outputs provided by the present invention. Detailed Implementation

[0030] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings and specific embodiments. The described embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit the scope of protection of the present invention. Other implementation methods obtained by those skilled in the art based on the embodiments of the present invention without departing from the technical concept of the present invention should all fall within the scope of protection of the present invention.

[0031] It should be noted that the DFIT referred to in this invention includes diagnostic fracture injection testing, micro-fracture diagnostic testing, pressure drop testing, and equivalent or similar field testing methods. Any method that can generate injection, pump shutdown, well shut-in, and pressure decay responses, and can be used for reservoir closure pressure, pore pressure, or stress parameter identification, can be included within the data processing scope of this invention.

[0032] It should also be noted that the structured semantic features described in this invention are not simply a set of curve numerical sampling points, but rather convert the curve morphology, stage response, candidate interval, fitting error, flow regime identification, wellbore and stratigraphic background, multi-well correlation and physical constraint parameters with engineering significance in the DFIT interpretation into storable, indexable, comparable and computable feature fields or feature vectors.

[0033] like Figure 1 As shown in the figure, this embodiment provides a method for identifying reservoir stress and pore pressure based on multi-well DFIT structured semantic features. The method is executed according to the following process: data acquisition and preprocessing, feature extraction, construction of multi-well associated features, construction of feature input vectors, solving the computational model, and result verification and updating.

[0034] During the data acquisition and preprocessing phase, the system acquires DFIT test data from multiple wells. For each well, the system records the test ID, well ID, test section, test date, wellhead pressure, bottomhole pressure, displacement, time, temperature, pumping stage, pump shutdown stage, shut-in stage, flowback or blowout stage, and test remarks. When preprocessing the data, the system first standardizes the pressure, displacement, and time units; then, it identifies the test stage based on pumping records, displacement changes, and pressure responses; and finally, it identifies outliers, missing data, significant drift points, and data from non-test sections.

[0035] During the feature extraction stage, the system extracts pre-closure pressure response features, post-closure pressure response features, and fitting error features. Pre-closure pressure response features characterize the pressure drop and derivative changes from pump shutdown to fracture closure, including the pressure drop rate, curve slope changes, curvature changes, closure candidate intervals, derivative inflection points, pressure response stability, and noise levels. Post-closure pressure response features characterize reservoir pressure decay, late-stage flow regimes, quasi-radial flow, or boundary responses after closure. Fitting error features describe the residual distribution, effective time window, convergence state, and multi-start point stability during candidate model calculations.

[0036] In the multi-well correlation feature construction stage, the system selects neighboring wells from the same block, the same stratigraphy, the same pressure system, or adjacent structural locations, and calculates the distance between the target well and neighboring wells, stratigraphic similarity, burial depth difference, lithological similarity, pressure system similarity, differences in production or injection history, and differences in existing interpretation results. Through this step, the single-well interpretation results of the target well no longer exist in isolation, but are judged within the context of multi-well and regional geomechanical background.

[0037] During the feature input vector construction phase, the system standardizes the pre-closure pressure response features, post-closure pressure response features, fitting error features, well foundation information features, adjacent well interpretation features, and physical constraint parameters. For missing features, the system identifies or replaces them based on the cause of the missing feature, its degree of absence, and the feature's importance. For features with different dimensions, the system normalizes or standardizes them. For categorical features, the system encodes them. For constraint features with clear physical meaning, the system retains their engineering dimensions or converts them into constraint identifiers.

[0038] During the computational model solution phase, the system solves for the output results based on the feature input vector. The output results may include closure pressure, pore pressure, minimum horizontal principal stress, maximum horizontal principal stress orientation or range, flow regime, confidence level, and anomaly verification priority. For results related to the maximum horizontal principal stress, this invention preferably outputs candidate intervals or orientation ranges, rather than outputting a unique definite value when sufficient constraints are lacking.

[0039] During the result verification and update phase, the system performs multiple verifications on the identification results. Verification includes the consistency between the closure pressure and the minimum horizontal principal stress, the consistency between the pore pressure and the test pressure response, the continuity of pore pressure within the same pressure system, the rationality of the stress variation trend with burial depth within the same formation, whether the differences in interpretation results from adjacent wells exceed a threshold, and whether the output results fall within a preset physical range. If the result passes verification, it is updated to the result database; if the result is abnormal, an anomaly flag and review priority are generated.

[0040] like Figure 2 As shown, this embodiment provides a multi-well DFIT structured semantic database. This database is not a simple file repository, but a structured data system built around DFIT interpretation operations, including at least a well basic information table, a DFIT test information table, a DFIT interpretation result table, a multi-well relationship table, a physical constraint parameter table, and a result update record table.

[0041] The well basic information table records information such as well ID, well name, block, coordinates, well type, completion date, well depth, measuring point depth, wellbore trajectory, test layer, layer position, sub-layer number, lithology, burial depth, thickness, porosity, permeability, and completion method. This table provides a basic index for multi-well correlation and layer constraints.

[0042] The DFIT test information table records information such as test ID, well ID, test section, test date, test depth, fluid type, test instrument, pump discharge rate, injection volume, wellhead pressure, bottomhole pressure, pump shutdown time, shut-in time, temperature record, and location of raw data files. This table is linked to the well basic information table via the well ID.

[0043] The DFIT interpretation results table records the result ID, test ID, closure pressure, pore pressure, minimum horizontal principal stress, location or range of maximum horizontal principal stress, flow regime, interpretation date, interpreter or calculation model version, confidence level, and remarks. This table allows each interpretation result to be traced back to the corresponding test, data version, and calculation process.

[0044] The multi-well relationship table records the relationships between the target well and adjacent wells, including well ID, adjacent well ID, horizontal distance, vertical distance, stratigraphic relationship, azimuth angle, pressure system relationship, structural relationship, and the basis for selecting adjacent wells. Through this table, the system can generate contextual features of adjacent wells oriented towards the target well.

[0045] The physical constraint parameter table is used to record constraint ID, flow type, lithology category, physical property range, correspondence between closure pressure and minimum horizontal principal stress, pore pressure range, minimum horizontal principal stress direction constraint, maximum horizontal principal stress candidate range constraint, adjacent well continuity constraint, and other engineering boundary conditions.

[0046] The results update record table is used to record the result ID, update time, updater or system version, update type, results before update, results after update, update basis, and remarks. This table enables the historical sample set to be continuously updated with the addition of new wells, manual review results, and regional interpretation results.

[0047] like Figure 3 As shown, this embodiment provides a structured parameter calculation module. The input features of this module include pre-closure features, post-closure features, fitting error features, well foundation information features, adjacent well interpretation features, and physical constraint parameters. The output vector of this module includes closure pressure, pore pressure, minimum horizontal principal stress, maximum horizontal principal stress azimuth or value range, and confidence level.

[0048] The structured parameter calculation module includes a feature processing and combination unit and a mapping calculation unit. The feature processing and combination unit is used to perform feature standardization, missing value handling, feature combination, and weight construction. The mapping calculation unit is used to perform calculations based on the mapping relationship between the input vector and the output vector.

[0049] In one implementation, the mapping calculation unit can adopt a rule-based mapping method. For example, when the closed candidate interval is stable, the derivative response is clear, the flow state after closure is sufficient, and the results of neighboring wells are consistent, a higher confidence level is assigned; when the closed candidate interval is scattered, the residual is large, the effective time window is insufficient, or the neighboring wells deviate significantly, the confidence level is reduced and a check mark is generated.

[0050] In another implementation, the mapping calculation unit can employ statistical regression, sample similarity matching, piecewise functions, ensemble calculation, or parameterized calculation models. Regardless of the specific implementation method used, it should be subject to verification by physical constraint parameters and multi-well correlation characteristics, and the minimum curve fitting error of a single well should not be used as the sole criterion.

[0051] In this embodiment, the confidence level is not simply the goodness of fit, but a reliability index of the result formed by comprehensively considering data quality, curve recognition clarity, fitting residuals, multi-start point stability, neighboring well consistency, historical sample similarity, and the degree of physical constraint satisfaction.

[0052] like Figure 4 As shown, this embodiment provides an automatic reservoir stress and pore pressure identification system. The system includes a data layer, a functional layer, and a support layer. The data layer includes a multi-well DFIT structured database; the functional layer includes a data processing unit, a feature extraction unit, a multi-well correlation unit, a calculation model unit, an output verification unit, and a result update unit; the support layer includes parameter management, physical constraint management, consistency rule management, and log and audit management.

[0053] The data processing unit is used to access, clean, standardize, and quality-label multi-well DFIT test data. The feature extraction unit is used to extract pre-closure pressure response features, post-closure pressure response features, and fitting error features from the standardized test data. The multi-well correlation unit is used to construct multi-well correlation features for the target well based on well location, stratigraphy, pressure system, lithology, burial depth, and existing interpretation results.

[0054] The computational model unit outputs identification results based on the feature input vector and physical constraint parameters. The output verification unit checks whether the output results meet physical consistency, multi-well consistency, and engineering boundary conditions. The result update unit writes the confirmed output results into the result database and synchronously updates the historical sample set and regional results.

[0055] Parameter management is used to manage computational model parameters, feature weights, thresholds, and version information. Physical constraint management is used to maintain flow type, lithology category, physical property range, pore pressure range, stress trend constraints, and adjacent well continuity constraints. Consistency rule management is used to set abnormal deviation thresholds and review priority rules. Log and audit management is used to record the entire process of data access, feature generation, model calculation, result updates, and manual review.

[0056] like Figure 5 As shown, this invention preferably divides the formation and use of the computational model into a parameter construction stage and an engineering application stage. The parameter construction stage is based on historical samples, while the engineering application stage focuses on the interpretation of target wells. By distinguishing between these two stages, it is possible to avoid directly using unverified target well results as reliable samples, while allowing manually verified interpretation results to be gradually established.

[0057] During the parameter construction phase, the system collects historical DFIT test data, extracts structured semantic features and multi-well correlation features, constructs feature input vectors, and determines the output vector based on verified historical interpretation results, engineering confirmation results, or historical interpretation outcomes. Subsequently, the system determines the mapping relationship between the feature input vector and the output vector, forming a computational model that can be used for engineering applications.

[0058] During the engineering application phase, the system receives DFIT data from the target well and completes preprocessing, feature extraction, multi-well correlation, feature input vector construction, and computational model solving. The output results, after consistency verification, can be used as the interpretation results for the target well; if anomalies are found, they enter a manual or expert review process. The verified results can be updated to the historical sample set for subsequent model parameter updates and regional result corrections.

[0059] The above-described phase division enables the invention to be continuously updated. When new DFIT test data, manual verification results, adjacent well interpretation results, or changes in regional geomechanical understanding occur, the system can update the sample set, physical constraint parameters, or mapping relationships, thereby improving the stability and engineering applicability of subsequent interpretation results.

[0060] like Figure 6 As shown, this embodiment provides a mapping relationship between multi-well feature inputs and multi-parameter outputs. Let the multi-well feature matrix be X, the physical constraint parameter vector be P, and the multi-parameter result vector be Y. Then, the system obtains the recognition result through the mapping relationship Y=f(X,P). This expression is only used to illustrate the computational relationship between inputs, constraints, and outputs, and does not limit the specific algorithm form.

[0061] Each row of the multi-well feature matrix X corresponds to a well sequence or a test segment, and each column corresponds to a structured semantic feature, such as pressure response features before closure, pressure response features after closure, fitting error features, well basic information features, and neighboring well interpretation features. The physical constraint parameter vector P is used to express constraints such as flow regime, lithology, physical property range, the correspondence between closure pressure and minimum horizontal principal stress, pore pressure range, and regional stress trend.

[0062] The multi-parameter result vector Y includes closure pressure, pore pressure, minimum horizontal principal stress, location or range of maximum horizontal principal stress, confidence level, and other engineering interpretation results. For each output parameter, the system can record its source characteristics, constraint satisfaction, and confidence level, making the results interpretable and traceable.

[0063] In one implementation, if the target well identification result simultaneously satisfies the following conditions: clear single-well curve characteristics, small fitting error, stable multi-start point calculation, continuous results from adjacent wells, reasonable trend at the same stratigraphic level, and a high degree of satisfaction with physical constraints, then the system outputs a high confidence level. If the target well has scattered closed candidate intervals, insufficient late-stage flow patterns, large deviations from adjacent wells, or violations of physical constraints, then the system outputs a low confidence level and generates a review priority.

[0064] This embodiment illustrates the process of identifying and updating abnormal wells. For each target well, after outputting the identification result, the system calculates the degree of deviation between this result and the results of adjacent wells, regional trends, and physical constraints. The degree of deviation can be represented by absolute difference, relative difference, standardized deviation value, residual trend at the same stratigraphic level, or pressure system continuity error.

[0065] When the deviation is below a preset threshold and the data quality meets the requirements, the system identifies the target well as an acceptable result and updates it to the results database. When the deviation exceeds the preset threshold, the data quality is insufficient, the fit is unstable, or the physical constraints are violated, the system identifies the target well as an abnormal well and generates a review priority based on the project importance, the degree of deviation, and the confidence level.

[0066] After manual review, the system records the review comments, correction results, and update basis. If the review confirms the calculation results are reliable, the well number is included in the historical sample set; if the review confirms the results need correction, the corrected results are retained along with the original calculation results, and the reason for the correction is recorded. Through this mechanism, the system can avoid erroneous samples from contaminating the historical sample set, while continuously incorporating reliable interpretation results.

[0067] This embodiment also provides an electronic device. The electronic device includes a processor, a memory, a communication interface, and a data interface. The memory stores a computer program, which, when executed by the processor, implements the steps described in any of the above embodiments. The electronic device can be a server, workstation, cloud computing node, industrial computer, laptop computer, or other device with data processing capabilities.

[0068] The data interface is used to receive DFIT test data, well foundation information, stratigraphic information, well logging interpretation results, production or injection history, and existing interpretation results. The communication interface is used to exchange data with database systems, geological modeling systems, fracturing design systems, gas storage operation management systems, or engineering decision-making systems.

[0069] This embodiment also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements the method steps described in any of the above embodiments. The computer-readable storage medium includes, but is not limited to, hard disks, solid-state drives, read-only memory, random access memory, flash memory, mobile storage devices, or cloud storage media.

[0070] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the scope of protection of the present invention. For those skilled in the art, any equivalent substitutions, combinations, adjustments or improvements made to the above embodiments without departing from the technical concept of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for identifying reservoir stress and pore pressure based on multi-well DFIT structured semantic features, characterized in that, Includes the following steps: S1. Obtain DFIT test data from multiple wells and corresponding well foundation information, and preprocess the DFIT test data; S2. Extract structured semantic features based on the preprocessed DFIT test data, the structured semantic features including at least pre-closure pressure response features, post-closure pressure response features, and fitting error features; S3. Construct multi-well correlation features based on well foundation information, neighboring well interpretation results, and physical connections between the target well and adjacent wells; S4. Perform feature processing and vector construction on the structured semantic features, multi-well correlation features, and physical constraint parameters to generate feature input vectors; S5. Input the feature input vectors into the reservoir stress and pore pressure identification calculation model for solution, and output identification results, the identification results including one or more of the following: closure pressure, pore pressure, minimum horizontal principal stress, and maximum horizontal principal stress orientation or value range; S6. Perform consistency verification and anomaly identification on the identification results, and update the confirmed identification results to the historical sample set or result database.

2. The method according to claim 1, characterized in that, The DFIT test data includes one or more of the following: bottomhole pressure data, wellhead pressure data, discharge data, time data, temperature data, pumping stage records, pump shutdown stage records, shut-in stage records, and flowback or blowout stage records; the well basic information includes one or more of the following: well number, well type, wellbore trajectory, measuring point depth, test section, layer, sub-layer number, burial depth, thickness, lithology, porosity, permeability, completion method, construction parameters, logging interpretation results, and production history or injection history.

3. The method according to claim 1, characterized in that, The preprocessing includes one or more of the following: pressure unit unification, displacement unit unification, time base unification, sampling interval unification, wellhead pressure and bottom hole pressure field matching, pump injection stage and shut-in stage division, anomaly point identification, missing value identification, pressure drift correction, temperature influence identification, data integrity evaluation, and data quality identification generation.

4. The method according to claim 1, characterized in that, The pressure response characteristics before closure include one or more of the following: early pressure drop rate after pump shutdown, pressure decay curve shape, pressure derivative curve shape, curvature change characteristics, closure candidate interval characteristics, pressure response stability characteristics, pre-closure flow response characteristics, pre-closure noise characteristics, and pre-closure interpretability characteristics.

5. The method according to claim 1, characterized in that, The pressure response characteristics after closure include one or more of the following: pressure decay rate after closure, derivative shape after closure, late flow response characteristics, quasi-radial flow response indicator, linear flow response indicator, boundary effect indicator, pressure recovery or relief trend characteristics, effective fitting time window characteristics, and interpretation reliability characteristics after closure.

6. The method according to claim 1, characterized in that, The fitting error features include one or more of the following: fitting parameters, objective function value, residual mean, residual variance, residual distribution shape, convergence state, number of iterations, parameter boundary triggering state, abnormal termination state, effective fitting time window, and multi-starting point stability index corresponding to at least one candidate explanatory model; the fitting error features are not used as a single explanatory conclusion, but as input features for judging the reliability of the identification results.

7. The method according to claim 1, characterized in that, The multi-well correlation characteristics include one or more of the following: planar distance, vertical distance, stratigraphic similarity, pressure system similarity, lithological similarity, burial depth difference, historical difference in production pressure drop, historical difference in injection pressure rise, difference in closure pressure of adjacent wells, difference in pore pressure of adjacent wells, difference in minimum horizontal principal stress of adjacent wells, and difference in the orientation or value range of maximum horizontal principal stress of adjacent wells.

8. The method according to claim 1, characterized in that, The physical constraint parameters include one or more of the following: flow type, lithology category, physical property range, correspondence between closure pressure and minimum horizontal principal stress, pore pressure range, pressure continuity constraint within the same pressure system, stress change trend constraint within the same stratum, burial depth trend constraint, and abnormal deviation threshold.

9. The method according to claim 1, characterized in that, The feature input vector is formed by structured semantic features, multi-well association features, well basic information features, and physical constraint parameters after feature standardization, missing value processing, feature combination, weight assignment, and dimension encoding; wherein, the weight assignment is determined based on data quality identification, test integrity, neighbor well similarity, and degree of physical constraint satisfaction.

10. The method according to claim 1, characterized in that, The reservoir stress and pore pressure identification calculation model is formed and used through the parameter construction stage and the engineering application stage; In the parameter construction phase, historical DFIT test data and verified historical interpretation results are collected, structured semantic features and multi-well correlation features of historical samples are extracted, feature input vectors are constructed, and the mapping relationship between the input vector and the output result vector is determined. In the engineering application phase, the target well DFIT data is input into the mapping relationship for calculation, and the output results are verified for consistency and updated.

11. The method according to claim 10, characterized in that, The mapping relationship is not limited to a specific algorithm and can be implemented using rule mapping, statistical regression, sample similarity matching, piecewise functions, ensemble calculation, parameterized calculation models, or combinations thereof. The mapping relationship is verified by satisfying physical constraints and multi-well consistency, rather than using the minimum curve fitting error of a single well as the sole criterion.

12. The method according to claim 1, characterized in that, The identification results also include one or more of the following: flow regime category, interpretation confidence level, priority of abnormal well verification, result update status, and regional reservoir stress and pore pressure results; the interpretation confidence level is determined comprehensively based on data quality identifiers, fitting error characteristics, multi-start point stability indicators, consistency degree of adjacent wells, similarity of historical samples, and degree of physical constraint satisfaction.

13. The method according to claim 12, characterized in that, The priority of reviewing abnormal wells is determined by comprehensively considering the interpretation confidence level, the fitted abnormal state, the deviation degree of adjacent wells, the degree of violation of physical constraints, the engineering importance, and the result update status. When the target well identification result deviates from the result of wells in the same layer, pressure system, or adjacent wells by more than a preset threshold, the target well is marked as a well to be reviewed.

14. The method according to claim 1, characterized in that, The results database includes a well foundation information table, a DFIT test information table, a DFIT interpretation results table, a multi-well relationship table, a physical constraint parameter table, and a results update record table. The DFIT interpretation results table records at least the result ID, test ID, closure pressure, pore pressure, minimum horizontal principal stress, maximum horizontal principal stress azimuth or value range, interpretation date, interpreter or calculation version, confidence level, and remarks. The results update record table records at least the results before update, the results after update, the update time, the update type, and the basis for the update.

15. The method according to claim 1, characterized in that, The results of reservoir stress and pore pressure in the region include the distribution of closure pressure, pore pressure, minimum horizontal principal stress, maximum horizontal principal stress azimuth or value range, flow regime distribution, interpretation confidence level distribution, and a checklist of abnormal wells for different well locations, different stratigraphic levels or different pressure systems within the target block.

16. A reservoir stress and pore pressure identification system based on multi-well DFIT structured semantic features, characterized in that, include: The database module is used to store a multi-well DFIT structured database; The data processing unit is used to acquire and preprocess multi-well DFIT test data; The feature extraction unit is used to extract the pressure response features before closure, the pressure response features after closure, and the fitting error features. The multi-well association unit is used to construct the multi-well association features between the target well and adjacent wells; the calculation model unit is used to solve the reservoir stress and pore pressure identification results based on the feature input vector. The output verification unit is used to perform consistency verification, anomaly identification, and review priority generation on the recognition results. The result update unit is used to update the confirmed identification results to the historical sample set or result database.

17. An electronic device, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program that, when executed by the processor, implements the method according to any one of claims 1 to 15.

18. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method according to any one of claims 1 to 15.

Images