Reservoir fluid identification method, device, equipment and medium

Abstract

The invention discloses a reservoir fluid identification method, device and equipment and a medium, and relates to the technical field of oil and gas field exploration. The invention provides a multi-level nested reservoir fluid identification method based on geological mode constraint for double-medium tight sandstone. For different reservoir types, a preset algorithm is utilized to fuse the comprehensive logging parameters and the proposed fluid sensitive factors suitable for the double-medium tight sandstone, and fluid discrimination parameters suitable for different reservoir types are constructed. On the basis, reservoir types are divided by comprehensively considering the coupling action result of the matrix and the tectonic fracture, the difference of fluid characteristics of different reservoir types is determined, and based on the classification constraint of a geological mode, the different reservoir types are divided, and the fluid type of the tight sandstone reservoir is identified by using an integrated identification model. Logging curve data for fluid identification are easy to obtain, the fluid identification precision is high, the identification efficiency is high, and the method can be widely applied to reservoir fluid identification of double-medium tight sandstone.

Description

Technical Field

[0001] The present application relates to the technical field of oil and gas field exploration, and in particular to a reservoir fluid identification method, device, equipment and medium. Background Art

[0002] Reservoir fluid identification refers to the process of analyzing and determining the type of fluid in the oil and gas reservoir through various technical means during oil and gas exploration and development. Reservoir fluids usually include crude oil, natural gas, water, etc., which may exist in the reservoir in different proportions and states.

[0003] Well logging interpretation of reservoir fluids aims to identify reservoir fluid types and response characteristics by analyzing conventional logging curves. However, for unconventional oil and gas reservoirs such as dual-medium tight sandstone gas reservoirs, the complex pore structure and logging response are very different, making the recognition accuracy of conventional methods such as curve overlap method and intersection plate method low. Although array acoustic logging and nuclear magnetic resonance logging have improved accuracy, they are costly and difficult to promote.

[0004] In view of the above problems, how to solve the current conventional analysis methods for reservoir fluid identification, which have low identification accuracy and high cost, is an urgent problem to be solved by technicians in this field. Summary of the invention

[0005] The purpose of the present application is to provide a reservoir fluid identification method, device, equipment and medium to solve the problem of low identification accuracy and high cost of the current conventional analysis method for reservoir fluid identification.

[0006] In order to solve the above technical problems, the present application provides a reservoir fluid identification method, comprising:

[0007] Obtaining well logging data of reservoirs in the study area and determining the inferred fluid type of the reservoirs;

[0008] Acquiring the original resistivity of the reservoir, and performing ground stress resistivity correction and fracture resistivity correction on the original resistivity to obtain a corrected resistivity of the reservoir;

[0009] constructing a fluid sensitivity factor according to the corrected resistivity, and generating a plurality of data sets representing different reservoir types according to the well logging data, the corrected resistivity and the fluid sensitivity factor;

[0010] Generating fluid discrimination parameters applicable to different reservoir types according to each of the data sets and a preset algorithm;

[0011] Inputting the logging curve data, the inferred fluid type and the fluid discrimination parameter into an integrated recognition model to output the fluid type of the reservoir in the study area;

[0012] Among them, the integrated identification model is a model for predicting reservoir fluid types constructed based on multiple integrated machine learning algorithms.

[0013] On the one hand, before obtaining the original resistivity of the reservoir and after obtaining the logging curve data of the reservoir in the study area, the method further includes:

[0014] Acquiring core data and analytical materials of the reservoir, and performing depth offset correction on the logging curve data according to the core data and the analytical materials;

[0015] Eliminating erroneous data and / or peak data in the logging curve data;

[0016] Processing abnormal values ​​and / or missing values ​​in the well logging curve data;

[0017] The logging curve data of different order of magnitude sequences are converted into the same order of magnitude sequences conforming to the standard normal distribution through Z-score normalization, and the logging curve data uniformly tested in the whole well or the logging curve data tested in different formation sections are excluded to pre-process the logging curve data.

[0018] On the other hand, the original resistivity is corrected for ground stress resistivity and fracture resistivity, including:

[0019] Obtaining formation factors at normal temperature and pressure, porosity at normal temperature and pressure, formation water resistivity, original cementation index, corrected cementation index, formation factors under overburden pressure, and overburden porosity of the reservoir;

[0020] Determine the empirical coefficient of the work area according to the formation factors at normal temperature and pressure, the porosity at normal temperature and pressure, the formation water resistivity, the original cementation index, the corrected cementation index, the formation factors under overburden pressure, the overburden porosity and Archie's formula;

[0021] Performing geostress resistivity correction on the original resistivity according to the empirical coefficient of the work area to obtain an initial corrected resistivity;

[0022] Determine the fracture development layer section of the reservoir according to the imaging logging data in the logging curve data, and determine the fracture resistivity of the fracture development layer section;

[0023] The fracture parameters of the fracture development layer are evaluated by using the dual lateral component method, the fractures with different inclination angles are decomposed into two fracture components, vertical and horizontal, and the lateral resistivity model under the wellbore environment is established according to the differential form of Ohm's law; the lateral resistivity model is solved by the deep and shallow dual lateral simultaneous equations to obtain the fracture inclination angle, fracture aperture and fracture porosity;

[0024] Fracture resistivity correction is performed according to the initial corrected resistivity, the fracture resistivity, the fracture dip, the fracture aperture and the fracture porosity to obtain the corrected resistivity of the reservoir.

[0025] On the other hand, constructing the fluid sensitivity factor according to the corrected resistivity comprises:

[0026] Obtaining a maximum principal stress at a formation level and a minimum principal stress at a formation level of the reservoir;

[0027] The fluid sensitivity factor is constructed according to the corrected resistivity, the maximum principal stress at the formation level, and the minimum principal stress at the formation level.

[0028] On the other hand, generating a plurality of data sets representing different reservoir types according to the well logging data, the corrected resistivity and the fluid sensitivity factor comprises:

[0029] Acquiring macroscopic lithology data and microscopic rock physical structure parameter data of the reservoir;

[0030] Determine the macroscopic lithology difference and rock physical structure difference of different reservoirs according to the macroscopic lithology data, and determine the microscopic pore structure difference of different reservoirs according to the microscopic rock physical structure parameter data;

[0031] According to the macroscopic lithology difference, the rock physical structure difference and the microscopic pore structure difference, the porosity and permeability in the microscopic rock physical structure parameter data are divided to establish a reservoir type classification standard;

[0032] Calibrate the logging curve data, the corrected resistivity and the fluid sensitivity factor according to the reservoir type classification standard, and perform semi-supervised cluster analysis on the calibrated data by using the DBSCAN algorithm to obtain a cluster analysis result;

[0033] The logging curve data, the corrected resistivity and the fluid sensitivity factor are divided into a plurality of data sets according to the cluster analysis result.

[0034] On the other hand, generating the fluid discrimination parameters applicable to different reservoir types according to each of the data sets and the preset algorithm includes:

[0035] Inputting the well log data, the corrected resistivity, the fluid sensitivity factor and each of the data sets into a Fisher algorithm model to determine an overall average and a classification average;

[0036] Determine an overall scatter matrix, an intra-class dispersion matrix, and an inter-class dispersion matrix according to the overall average value and the classification average value;

[0037] Calculating an eigenvector according to the eigenvalue, the intra-class discreteness matrix and the inter-class discreteness matrix, and determining the eigenvector corresponding to the maximum eigenvalue;

[0038] Generate an eigenvector matrix according to the eigenvector corresponding to the maximum eigenvalue;

[0039] The fluid discrimination parameter is determined according to the eigenvector matrix and each of the data sets.

[0040] On the other hand, the construction process of the integrated recognition model includes:

[0041] Get training set sample data and initialize weights;

[0042] Input the training set sample data and the initialization weights into the full convolutional neural network algorithm module, the extreme gradient boosting tree algorithm module, the categorical feature gradient boosting algorithm module, the decision tree algorithm module and the random forest algorithm module, respectively, to obtain the recognition results output by each algorithm module;

[0043] Input each of the recognition results into a weak classifier, and determine the error between each of the recognition results and the corresponding true result and the classifier weight through the weak classifier;

[0044] Adjust the initialization weight according to the classifier weight to obtain a new initialization weight, and return to the step of inputting the training set sample data and the initialization weight into the full convolutional neural network algorithm module, the extreme gradient boosting tree algorithm module, the categorical feature gradient boosting algorithm module, the decision tree algorithm module, and the random forest algorithm module, respectively, until the minimum error is determined;

[0045] A strong learner is constructed according to the weights of each classifier and the weighted combination strategy, so as to output a final recognition result through the strong learner.

[0046] In order to solve the above technical problems, the present application also provides a reservoir fluid identification device, comprising:

[0047] An acquisition module, used to acquire well logging data of reservoirs in a study area and determine the inferred fluid type of the reservoirs;

[0048] A correction module, used for acquiring the original resistivity of the reservoir, and performing ground stress resistivity correction and fracture resistivity correction on the original resistivity to obtain a corrected resistivity of the reservoir;

[0049] A first generating module is used to construct a fluid sensitivity factor according to the corrected resistivity, and to generate a plurality of data sets representing different reservoir types according to the logging curve data, the corrected resistivity and the fluid sensitivity factor;

[0050] A second generating module, used for generating fluid discrimination parameters applicable to different reservoir types according to each of the data sets and a preset algorithm;

[0051] An identification module, for inputting the logging curve data, the inferred fluid type and the fluid discrimination parameter into an integrated identification model to output the fluid type of the reservoir in the study area;

[0052] Among them, the integrated identification model is a model for predicting reservoir fluid types constructed based on multiple integrated machine learning algorithms.

[0053] In order to solve the above technical problems, the present application also provides a reservoir fluid identification device, comprising:

[0054] Memory for storing computer programs;

[0055] A processor is used to implement the steps of the above-mentioned reservoir fluid identification method when executing the computer program.

[0056] In order to solve the above technical problems, the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above reservoir fluid identification method are implemented.

[0057] The reservoir fluid identification method provided in the present application obtains the logging curve data of the reservoir in the study area and determines the inferred fluid type of the reservoir; obtains the original resistivity of the reservoir, and corrects the original resistivity for geostress resistivity and fracture resistivity to obtain the corrected resistivity of the reservoir; constructs a fluid sensitivity factor based on the corrected resistivity, and generates multiple data sets characterizing different reservoir types based on the logging curve data, the corrected resistivity and the fluid sensitivity factor; generates fluid discrimination parameters suitable for different reservoir types based on each data set and a preset algorithm; inputs the logging curve data, the inferred fluid type and the fluid discrimination parameter into an integrated identification model to output the fluid type of the reservoir in the study area; wherein the integrated identification model is a model for predicting reservoir fluid types constructed based on a variety of integrated machine learning algorithms. It can be seen that this scheme proposes a multi-level nested reservoir fluid identification method based on geological model constraints for dual-medium type tight sandstone. For different reservoir types, the preset algorithm is used to fuse the comprehensive logging parameters and the proposed fluid sensitivity factor suitable for dual-medium type tight sandstone to construct fluid discrimination parameters suitable for different reservoir types. On this basis, the reservoir types are divided by comprehensively considering the effects of the coupling between matrix and structural fractures, the differences in fluid properties of different reservoir types are clarified, and different reservoir types are classified based on the classification constraints of geological models, and the identification of fluid types in tight sandstone reservoirs is completed using an integrated identification model. The logging curve data used for fluid identification is easy to obtain, and the fluid identification accuracy and efficiency are high, which can be widely used in the identification of reservoir fluids in dual-medium tight sandstone.

[0058] In addition, the present application also provides a reservoir fluid identification device, equipment and medium, with the same effects as above. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0060] Figure 1 A flow chart of a reservoir fluid identification method provided in an embodiment of the present application;

[0061] Figure 2 A schematic diagram of reservoir fluid identification provided in an embodiment of the present application;

[0062] Figure 3 A schematic diagram of the preprocessing of well logging curve data provided in an embodiment of the present application;

[0063] Figure 4 A schematic diagram of the reservoir original resistivity correction provided in an embodiment of the present application;

[0064] Figure 5 A schematic diagram of a process for constructing a reservoir type sub-model based on geological knowledge provided in an embodiment of the present application;

[0065] Figure 6 A schematic diagram of a Fisher-based curve reconstruction technology process provided in an embodiment of the present application;

[0066] Figure 7 A schematic diagram of the integrated recognition model construction and optimization process provided in the embodiment of the present application;

[0067] Figure 8 A schematic diagram of the integrated algorithm technology flow based on Boosting provided in the embodiment of the present application;

[0068] Fig. 9 Porosity and permeability scatter plots of different reservoir types provided for embodiments of the present application;

[0069] Fig.10 A comparison chart of the accuracy of blind well cross-check of the algorithm model provided in the embodiment of the present application;

[0070] Fig.11 A comparison chart of the importance of algorithm model features provided in the embodiments of the present application;

[0071] Fig.12 A schematic diagram of the blind well identification effect of the algorithm model provided in the embodiment of the present application;

[0072] Fig.13 A schematic diagram of a reservoir fluid identification device provided in an embodiment of the present application;

[0073] Fig.14 A schematic diagram of a reservoir fluid identification device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0074] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0075] The core of this application is to provide a reservoir fluid identification method, device, equipment and medium to solve the problem of low identification accuracy and high cost of the current conventional analysis method for reservoir fluid identification.

[0076] In order to enable those skilled in the art to better understand the present application, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0077] Figure 1 A flow chart of a reservoir fluid identification method provided in an embodiment of the present application. Figure 1 As shown, the method includes:

[0078] S10: Obtain well logging data of the reservoir in the study area and determine the inferred fluid type of the reservoir;

[0079] Figure 2 This is a schematic diagram of reservoir fluid identification provided in the embodiment of the present application. Figure 2 As shown in the figure, first of all, it is necessary to determine the possible fluid types in the study area based on the gas test data in the study area. Since the single well test time is short and cannot fully reflect the actual situation of the underground reservoir fluid, it is necessary to combine the production data for comprehensive judgment, and finally determine the main fluid type of the underground reservoir in the study area, that is, the inferred fluid type. Among them, the production data include single well test gas and water data, single well daily production, and cumulative gas and water production data.

[0080] At the same time, it is necessary to obtain the logging curve data of the reservoir in the study area, including gamma ray data, deep lateral resistivity data, shallow lateral resistivity data, density data, sonic time difference data and neutron porosity data.

[0081] S11: obtaining the original resistivity of the reservoir, and performing ground stress resistivity correction and fracture resistivity correction on the original resistivity to obtain the corrected resistivity of the reservoir.

[0082] Furthermore, when the stress changes, it is assumed that the resistivity change of the dual-medium reservoir mainly depends on the contraction of the small pore throats, and the small pores are mainly filled with bound water; there are a large number of fine particles in the small pores or throats, then the resistivity change under stress change can be obtained according to the Archie formula, and it can be obtained that the change in the physical properties of the dense sandstone is determined by the pore structure coefficient. At the same time, the fluid in the pores is squeezed out due to the stress, causing the resistivity to increase. This is extremely obvious in the reservoir tested as a dry layer in the gas test. At the same time, it will cause the resistivity curve value of some water layers to increase abnormally, giving the illusion of a high-resistance water layer. Therefore, it is necessary to perform geostress resistivity correction on the original resistivity of the reservoir.

[0083] In addition, for dual-medium tight sandstone, the reservoir fractures are relatively developed. When formation water with a certain degree of mineralization or low-resistivity liquids such as drilling mud and mud filtrate invade and fill the fractures and pores, the resistivity will be significantly reduced. The more developed the fractures, the greater the impact, resulting in a decrease in the resistivity of the gas-bearing layer, which interferes with the logging interpretation. Therefore, it is necessary to correct the resistivity of the fracture-developed layer of the dual-medium reservoir on the basis of the correction of the ground stress.

[0084] It should be noted that in this embodiment, there is no limitation on the specific process of performing geostress resistivity correction and fracture resistivity correction on the original resistivity, which depends on the specific implementation situation.

[0085] S12: construct a fluid sensitivity factor based on the corrected resistivity, and generate multiple data sets representing different reservoir types based on the well logging data, the corrected resistivity and the fluid sensitivity factor.

[0086] Furthermore, in this embodiment, a fluid identification sensitive factor based on ground stress and resistivity is proposed, which can accurately identify the reservoir fluid type of dual-medium tight sandstone. The construction principle is mainly that for dual-medium tight sandstone, the matrix porosity, permeability, fracture porosity and opening of its reservoir gradually decrease with the increase of effective stress and horizontal stress difference, while the pore fluid and gas will cause the elastic limit and strength to decrease, and the toughness will be enhanced, resulting in a further decrease in stress difference. In addition, when the difference in reservoir rock skeleton is not large, the difference in fluid contained in the rock skeleton will lead to a significant difference in resistivity, among which the gas conductivity is much lower than the conductivity of water, so the resistivity measured by logging of pure gas layer is much higher than that of pure water layer. In this embodiment, there is no restriction on the specific construction method of the fluid sensitive factor.

[0087] At the same time, multiple data sets representing different reservoir types are generated based on logging curve data, corrected resistivity and fluid sensitivity factors, thereby realizing the construction of a reservoir type sub-model based on geological knowledge. It should be noted that the reservoir type classification based on geological knowledge as a whole is a soft constraint. The data set belongs to the classification type established based on comprehensive macro and micro differences, and each reservoir type includes multiple data sets; the establishment of this sub-model is only convenient for constructing a high-precision and high-stability sub-model with a relatively simple data structure and smaller variance. The specific process of data set construction is not limited in this embodiment.

[0088] S13: Generate fluid discrimination parameters suitable for different reservoir types according to each data set and a preset algorithm.

[0089] Furthermore, based on the divided multiple data sets and different reservoir types, the preset algorithm is used to fuse the comprehensive logging parameters and the fluid identification sensitivity factor, and the low-correlation variables with little impact on the results are removed, and finally the fluid discrimination parameters suitable for different reservoir types are calculated by fusion, and the fluid discrimination parameters are used as the input of the identification model. In this embodiment, there is no restriction on the generation process of the fluid discrimination parameters.

[0090] S14: Input the logging curve data, the inferred fluid type and the fluid discrimination parameters into the integrated identification model to output the fluid type of the reservoir in the study area.

[0091] Among them, the integrated identification model is a model constructed based on multiple integrated machine learning algorithms to predict reservoir fluid types.

[0092] Finally, based on multiple data sets constructed based on geological knowledge of reservoir types, an integrated approach is further used to identify reservoir fluids in a more refined manner. Based on multiple data sets constructed based on geological knowledge of reservoir types, only the sampling points predicted to be reservoirs are used as input, and an integrated machine learning algorithm is used to perform deep learning on the fluids of each subdivided reservoir to identify the fluid distribution characteristics under different reservoir types. In this way, the identification of fluid types is achieved.

[0093] It should be noted that the integrated identification model is a model for predicting reservoir fluid types based on multiple integrated machine learning algorithms; in this embodiment, there is no restriction on the construction process of the integrated identification model, which depends on the specific implementation situation.

[0094] In this embodiment, a multi-level nested reservoir fluid identification method based on geological model constraints for dual-medium tight sandstone is proposed. For different reservoir types, the preset algorithm is used to fuse the comprehensive logging parameters and the proposed fluid sensitivity factors suitable for dual-medium tight sandstone, and fluid discrimination parameters suitable for different reservoir types are constructed. On this basis, the reservoir types are divided by comprehensively considering the effects of the coupling between the matrix and structural fractures, and the differences in fluid properties of different reservoir types are clarified. Based on the classification constraints of the geological model, different reservoir types are divided, and the integrated identification model is used to complete the identification of tight sandstone reservoir fluid types. The logging curve data used for fluid identification is easy to obtain, and the fluid identification accuracy and efficiency are high. It can be widely used for reservoir fluid identification of dual-medium tight sandstone.

[0095] Figure 3 Schematic diagram of well logging data preprocessing provided in the embodiment of the present application. Based on the above embodiment, in some embodiments, before obtaining the original resistivity of the reservoir, after obtaining the well logging data of the reservoir in the study area, such as Figure 3 As shown, it also includes:

[0096] S15: Obtaining core data and analytical test materials of the reservoir, and performing depth offset correction on the well logging curve data according to the core data and analytical test materials;

[0097] S16: Eliminate erroneous data and / or peak data in the logging curve data;

[0098] S17: Processing abnormal values ​​and / or missing values ​​in the well logging curve data;

[0099] S18: The logging curve data of different order of magnitude sequences are converted into the same order of magnitude sequences conforming to the standard normal distribution through Z-score normalization, and the logging curve data uniformly tested in the whole well or the logging curve data tested in different formation sections are excluded to pre-process the logging curve data.

[0100] Due to geological factors and the influence of logging instruments, the logging curve data has depth deviation; at the same time, the range of the same logging data in different wells is different, which will affect the subsequent recognition accuracy. In order to eliminate the influence of the logging instrument itself on the logging data and ensure the reliability of the data, the logging curve data needs to be preprocessed.

[0101] Specifically, all logging curve data were corrected for depth offset using core data and analytical test data; the logging curve data were visually inspected to remove obvious data errors and / or spikes. The outliers and missing values ​​of the logging curve data were further processed, and the curves were baseline corrected and smoothed. Z-score normalization was used to convert the logging curve data of different order of magnitude sequences into the same order of magnitude sequences that conform to the standard normal distribution. Further, for all logging curve data processed by order of magnitude sequence, by screening out the logging curve data that were uniformly tested throughout the well or tested in different formation sections, errors can be effectively avoided and the true logging response of the main fluid types in the underground reservoir of the study area can be determined.

[0102] Figure 4 The schematic diagram of the reservoir original resistivity correction provided in the embodiment of the present application. Based on the above embodiment, in some embodiments, such as Figure 4 As shown, the original resistivity is corrected for ground stress resistivity and fracture resistivity, including:

[0103] S111: Obtaining the reservoir's formation factors at normal temperature and pressure, porosity at normal temperature and pressure, formation water resistivity, original cementation index, corrected cementation index, formation factors under overburden pressure, and overburden porosity;

[0104] S112: Determine the empirical coefficient of the work area based on formation factors at normal temperature and pressure, porosity at normal temperature and pressure, formation water resistivity, original cementation index, corrected cementation index, formation factors under overburden pressure, overburden porosity and Archie's formula;

[0105] S113: performing ground stress resistivity correction on the original resistivity according to the empirical coefficient of the work area to obtain an initial corrected resistivity;

[0106] S114: determining a fracture development layer section of the reservoir according to the imaging logging data in the logging curve data, and determining the fracture resistivity of the fracture development layer section;

[0107] S115: Use the dual lateral component method to evaluate the fracture parameters of the fracture development layer, decompose the fractures with different inclination angles into two fracture components, vertical and horizontal, and establish the lateral resistivity model in the wellbore environment based on the differential form of Ohm's law; solve the lateral resistivity model by the deep and shallow dual lateral simultaneous equations to obtain the fracture inclination, fracture aperture and fracture porosity;

[0108] S116: Perform fracture resistivity correction based on the initial corrected resistivity, fracture resistivity, fracture dip, fracture aperture and fracture porosity to obtain a corrected resistivity of the reservoir.

[0109] When the stress changes, it is assumed that the resistivity change of the dual-medium reservoir mainly depends on the contraction of the small pore throats, and the small pores are mainly filled with bound water; there are a large number of fine particles in the small pores or throats. The resistivity change formula under stress change can be obtained by deforming the Archie formula. Therefore, the change in the physical properties of dense sandstone is determined by the pore structure coefficient. At the same time, the fluid in the pores is squeezed out due to stress, causing the resistivity to increase. This is very obvious in the reservoir tested as a dry layer during the gas test. At the same time, it will cause the resistivity curve value of some water layers to increase abnormally, giving the illusion of a high-resistance water layer. Therefore, the empirical coefficient of the work area is introduced. .

[0110] Specifically, the reservoir's formation factors at normal temperature and pressure, porosity at normal temperature and pressure, formation water resistivity, original cementation index, corrected cementation index, formation factors under overburden pressure, and overburden porosity are obtained; based on the formation factors at normal temperature and pressure, porosity at normal temperature and pressure, formation water resistivity, original cementation index, corrected cementation index, formation factors under overburden pressure, overburden porosity, and Archie's formula, the empirical coefficient of the work area is determined. Archie's formula is as follows:

[0111] ;

[0112] The resistivity change formula is as follows:

[0113] ;

[0114] Furthermore, the geostress-corrected resistivity formula of the study area can be established based on the empirical coefficient of the work area. The geostress-corrected resistivity formula is performed on the original resistivity according to the empirical coefficient of the work area. The specific corrected resistivity formula is as follows:

[0115] ;

[0116] Among them, F is the formation factor at normal temperature and pressure; It is the underlying stratum factor; is the formation water resistivity, (Ω.m); is the porosity at normal temperature and pressure, %; is the overburden porosity, %; is the original cementation index, dimensionless; is the corrected cementation index under strong stress, dimensionless; is the original resistivity, (Ω.m); is the initial corrected resistivity after correction of geostress resistivity, (Ω.m); It is the empirical coefficient of the work area and is dimensionless.

[0117] It should be noted that the corrected cementation index under strong stress Factors related to the overburden formation It can be obtained through rock electrical experiments under high temperature and high pressure conditions. It can be obtained through overburden porosity experiment.

[0118] In addition, for dual-medium tight sandstone, the reservoir fractures are relatively developed. When formation water with a certain degree of mineralization or low-resistivity liquids such as drilling mud and mud filtrate invade and fill the fractures and pores, the resistivity will be significantly reduced. The more developed the fractures, the greater the impact, resulting in a decrease in the resistivity of the gas-bearing layer, which interferes with the logging interpretation. Therefore, it is necessary to correct the resistivity of the fracture-developed layer of the dual-medium reservoir on the basis of the correction of the ground stress.

[0119] Specifically, the fracture development layer section of the reservoir is determined based on the imaging logging data in the logging curve data, and the fracture resistivity of the fracture development layer section is determined. The dual lateral component method is further used to evaluate the fracture parameters of the fracture development layer, and the fractures with different inclination angles are decomposed into two fracture components, vertical and horizontal. The lateral resistivity model in the wellbore environment is established based on the differential form of Ohm's law. The lateral resistivity model is solved by the deep and shallow dual lateral simultaneous equations to obtain the fracture inclination angle. , crack opening and fracture porosity In the specific implementation, the imaging logging results can be used to verify and compare the fracture parameters and calculate the average error. If the error is greater than 30%, the wellbore environmental parameters are adjusted and the calculation model is rebuilt. The initial corrected resistivity after ground stress correction is converted into and crack resistivity The difference in logarithms of is used to establish a common relationship with the crack parameters obtained using the double lateral component method, and the formula is as follows:

[0120] ;

[0121] Finally, the above common relationship formula is deformed, and the fracture resistivity is corrected according to the initial corrected resistivity, fracture resistivity, fracture dip, fracture aperture and fracture porosity to obtain the corrected resistivity of the reservoir. The specific formula is as follows:

[0122] ;

[0123] in, is the initial corrected resistivity after ground stress correction, (Ω.m); is the final corrected resistivity, (Ω.m); is the crack resistivity, (Ω.m); is the crack inclination, °; is the crack opening, μm; is the fracture porosity, %; a, b, c, e are the multivariate fitting coefficients, dimensionless.

[0124] In summary, the in-situ stress resistivity correction and crack resistivity correction of the original resistivity are realized.

[0125] Based on the above embodiments, in some embodiments, constructing a fluid sensitivity factor according to the corrected resistivity includes:

[0126] S121: obtaining the maximum principal stress and the minimum principal stress at the formation level of the reservoir;

[0127] S122: Construct a fluid sensitivity factor based on the corrected resistivity, the maximum principal stress at the formation level, and the minimum principal stress at the formation level.

[0128] For dual-medium tight sandstone, the matrix porosity, permeability, and fracture porosity and opening of its reservoir gradually decrease with the increase of effective stress and horizontal stress difference, while the pore fluid and gas will cause the elastic limit and strength to decrease, and the enhanced toughness will lead to a further decrease in stress difference. In addition, when the difference in reservoir rock skeleton is not large, the difference in fluid contained in the rock skeleton will lead to significant difference in resistivity, among which the gas conductivity is much lower than the water conductivity. Therefore, the resistivity measured by logging in pure gas layer is much higher than that in pure water layer. Combining the above principles, the ratio method is used to amplify the gas-containing response. A fluid sensitivity factor suitable for dual-medium tight sandstone is constructed, and the formula is as follows:

[0129] ;

[0130] in, To correct for resistivity, is the stress difference of the formation, MPa; SDXM is the maximum principal stress of the formation, MPa; SDYM is the minimum principal stress of the formation, MPa.

[0131] Figure 5 A schematic diagram of a process for constructing a reservoir type submodel based on geological knowledge provided in an embodiment of the present application. Based on the above embodiments, in some embodiments, such as Figure 5As shown, multiple data sets representing different reservoir types are generated based on well logging data, corrected resistivity and fluid sensitivity factors, including:

[0132] S123: Obtaining macroscopic lithology data and microscopic rock physical structure parameter data of the reservoir;

[0133] S124: determining the macroscopic lithology differences and rock physical structure differences of different reservoirs based on the macroscopic lithology data, and determining the microscopic pore structure differences of different reservoirs based on the microscopic rock physical structure parameter data;

[0134] S125: Based on the macroscopic lithology differences, rock physical structure differences and microscopic pore structure differences, the porosity and permeability in the microscopic rock physical structure parameter data are divided to establish the reservoir type classification standard;

[0135] S126: Calibrate the logging curve data, corrected resistivity and fluid sensitivity factor according to the reservoir type classification standard, and perform semi-supervised cluster analysis on the calibrated data through the DBSCAN algorithm to obtain cluster analysis results;

[0136] S127: Divide the logging curve data, corrected resistivity and fluid sensitivity factor into multiple data sets according to the cluster analysis result.

[0137] Specifically, we first obtain macroscopic lithology data of reservoirs through geological observation methods such as cores, thin sections, and electron microscope images, and at the same time obtain microscopic rock physical structure parameter data through porosity and permeability. We determine the macroscopic lithology differences and rock physical structure differences of different reservoirs based on the macroscopic lithology data, and determine the microscopic pore structure differences of different reservoirs based on the microscopic rock physical structure parameter data.

[0138] Furthermore, based on the differences in macroscopic lithology, rock physical structure and microscopic pore structure, the porosity and permeability in the microscopic rock physical structure parameter data are divided, and the pore structures with similar porosity and permeability relationships are grouped together. By comprehensively considering the macroscopic and microscopic differences, a classification standard for dual-medium tight sandstone reservoir types is established.

[0139] Then, a semi-supervised cluster analysis was performed. The logging data, corrected resistivity and fluid sensitivity factor were calibrated according to the reservoir type classification standard. After data cleaning, the calibrated data were used as input data. The density-based spatial clustering of applications with noise (DBSCAN) algorithm was used to perform a semi-supervised cluster analysis on the calibrated data to obtain the cluster analysis results. The logging data, corrected resistivity and fluid sensitivity factor were further divided into multiple data sets based on the cluster analysis results. Each data set represents a potential reservoir type in the major category. It should be noted that the reservoir type classification based on geological knowledge as a soft constraint, the data set belongs to the classification type established based on comprehensive macroscopic and microscopic differences, and each reservoir type includes multiple data sets. The establishment of this sub-model is only convenient for constructing a high-precision and high-stability sub-model with a relatively simple data structure and smaller variance.

[0140] Figure 6 A flow chart of a Fisher-based curve reconstruction technology provided in an embodiment of the present application. Based on the above embodiments, in some embodiments, such as Figure 6 As shown in the figure, fluid discrimination parameters suitable for different reservoir types are generated based on each data set and preset algorithm, including:

[0141] S131: input the well logging data, the corrected resistivity, the fluid sensitivity factor and each data set into the Fisher algorithm model to determine the overall average and the classification average;

[0142] S132: Determine the overall scatter matrix, the intra-class dispersion matrix, and the inter-class dispersion matrix according to the overall average value and the classification average value;

[0143] S133: Calculate eigenvectors according to the eigenvalues, the intra-class discreteness matrix, and the inter-class discreteness matrix, and determine the eigenvector corresponding to the maximum eigenvalue;

[0144] S134: Generate an eigenvector matrix according to the eigenvector corresponding to the maximum eigenvalue;

[0145] S135: Determine fluid discrimination parameters according to the eigenvector matrix and each data set.

[0146] In this embodiment, based on the divided data set, for different reservoir types, the Fisher linear discriminant analysis (Fisher) algorithm is used to fuse the comprehensive logging parameters and fluid identification sensitivity factors, and remove the low-correlation variables that have a small impact on the results, and finally the fluid discrimination parameters suitable for different reservoir types are calculated by fusion.

[0147] Specifically, the logging curve data, corrected resistivity, fluid sensitivity factor and each data set (X1, X2, X2, ..., X k ) is input into the Fisher algorithm model to determine the overall mean and the category mean Based on the overall average and the category mean Determine the overall scatter matrix , intra-class dispersion matrix and the inter-class dispersion matrix .

[0148] The eigenvector is calculated based on the eigenvalue, the intra-class discreteness matrix and the inter-class discreteness matrix. The generalized characteristic matrix formula is as follows:

[0149] ;

[0150] in, is the characteristic value, is the feature vector.

[0151] Further determine the eigenvector corresponding to the maximum eigenvalue, and generate the eigenvector matrix based on the eigenvector corresponding to the maximum eigenvalue .

[0152] Finally, according to the eigenvector matrix and each data set (X1, X2, X2, ..., X k ) Determine the fluid discrimination parameters, the formula is as follows:

[0153] ;

[0154] in, is the fluid discrimination parameter, i∈[1,2,3,……,k].

[0155] Figure 7 The schematic diagram of the integrated recognition model construction and optimization process provided in the embodiment of the present application. Based on the above embodiment, in some embodiments, such as Figure 7 As shown in Figure 1, the construction process of the integrated recognition model includes:

[0156] S141: Obtain training set sample data and initialize weights;

[0157] S142: inputting the training set sample data and the initialization weights into the full convolutional neural network algorithm module, the extreme gradient boosting tree algorithm module, the category feature gradient boosting algorithm module, the decision tree algorithm module and the random forest algorithm module respectively, so as to obtain the recognition results output by each algorithm module respectively;

[0158] S143: inputting each recognition result into a weak classifier, and determining the error between each recognition result and the corresponding true result and the classifier weight through the weak classifier;

[0159] S144: adjusting the initialization weight according to the classifier weight to obtain a new initialization weight, and returning to the step of inputting the training set sample data and the initialization weight into the full convolutional neural network algorithm module, the extreme gradient boosting tree algorithm module, the categorical feature gradient boosting algorithm module, the decision tree algorithm module, and the random forest algorithm module, respectively, until the minimum error is determined;

[0160] S145: constructing a strong learner according to the weights of each classifier and the weighted combination strategy, so as to output a final recognition result through the strong learner.

[0161] In this embodiment, multiple sub-models, i.e., multiple data sets, are constructed based on geological knowledge, and an integrated method is further used to identify reservoir fluids in a more refined manner. Based on multiple sub-models constructed based on geological knowledge, only the sampling points predicted to be reservoirs are used as input, and an integrated machine learning algorithm is used to perform deep learning on the fluids of each subdivided reservoir to identify the fluid distribution characteristics under different reservoir types. The model construction process is described in detail below:

[0162] Figure 8 The following is a schematic diagram of the integrated algorithm technology flow based on Boosting provided in the embodiment of the present application. Figure 8 As shown in the figure, first, the training set sample data and initialization weight ω are obtained, and the training set sample data and initialization weight ω are respectively input into the Fully Connected Neural Network (FCN) algorithm module, the extreme gradient boosting tree (eXtreme Gradient Boosting, XGBoost) algorithm module, the category feature gradient boosting (CatBoost) algorithm module, the decision tree (Decision Tree, DT) algorithm module and the random forest (Random Forest, RF) algorithm module to obtain the recognition results output by each algorithm module. Each recognition result is input into the weak classifier Ci, and the error r between each recognition result and the corresponding true result is determined by the weak classifier. i and classifier weight α i . According to the classifier weight α i Adjust the initialization weight ω to get the new initialization weight ω i, return to the step of inputting the training set sample data and the initialization weights into the full convolutional neural network algorithm module, the extreme gradient boosting tree algorithm module, the category feature gradient boosting algorithm module, the decision tree algorithm module and the random forest algorithm module respectively, until the minimum error is determined. Finally, a strong learner is constructed according to the weights of each classifier and the weighted combination strategy to output the final recognition result through the strong learner. It should be noted that the weights constructed for different data sets (i.e., different reservoir types) are different, and the integrated recognition models formed are also different.

[0163] In addition, since there are multiple algorithm modules in the integrated recognition model, the model parameters are huge, and manual parameter adjustment is inefficient and ineffective. In the specific implementation, the Bayesian hyperparameter intelligent optimization algorithm can be used to optimize the model parameters. Specifically, the established integrated recognition model is fine-tuned through cross-validation and Bayesian hyperparameter optimization to improve the accuracy and robustness of fluid identification. At the same time, in order to evaluate the actual application effect of the model, the algorithm model is tested in blind wells that did not participate in the training in the application study area to ensure that the model meets the application requirements. Finally, all identification results of all levels are integrated to form a comprehensive understanding of the reservoir fluid, ensuring that the fluid type of each reservoir unit can be accurately identified.

[0164] In summary, the construction of the integrated recognition model is achieved.

[0165] The present invention is described in detail below with reference to specific embodiments:

[0166] The reservoir fluid identification was carried out based on the tight sandstone gas reservoir data of a certain place, which specifically includes the following steps:

[0167] (I) Analysis of fluid types in the study area. The possible fluid types in the study area were determined based on the gas test data in the study area. Since the single well test time is short and cannot fully reflect the actual situation of the underground reservoir fluid, a comprehensive judgment was made based on the production data, including the single well test gas and water data, the single well daily production, and the cumulative gas and water production data. Finally, the main fluid types of the underground reservoir in the study area were determined, including gas layer, poor gas layer, gas and water layer, water layer, and dry layer.

[0168] (II) Well logging curve preprocessing: The well logging curve data were preprocessed by depth offset correction, visual inspection, outlier and missing value processing, baseline correction and smoothing, and Z-score standardization, and samples that were uniformly tested in the whole well or tested in different formation sections were excluded through screening.

[0169] (III) Correction of resistivity parameters. The resistivity parameters of the study area are corrected for ground stress and crack effects. The ground stress correction formula and crack effect correction formula constructed in this embodiment are as follows:

[0170] ;

[0171] ;

[0172] in, is the porosity at normal temperature and pressure, %; is the overburden porosity, %; is the original cementation index, dimensionless; is the corrected cementation index under strong stress, dimensionless; is the original formation resistivity, (Ω.m); is the formation resistivity after correction for ground stress, (Ω.m); is the final corrected resistivity, (Ω.m); is the formation resistivity at the fracture, (Ω.m); is the crack inclination, °; is the crack opening, μm; is the fracture porosity, %; a, b, c, e are the multivariate fitting coefficients, dimensionless.

[0173] (IV) Construction of a new fluid sensitivity factor. First, based on well logging, gas testing, gas logging and geological data, the different response characteristics of fluids in different reservoir types are analyzed in detail. The geological knowledge and the rock physical response of the fluid are used to suppress the interference signal, highlight the contribution of the gas-bearing fluid, and preliminarily construct the fluid identification sensitive factor. This application uses the ratio method to amplify the gas-bearing response, and for the first time proposes a fluid identification sensitive factor A1 based on ground stress and resistivity for dual-medium tight sandstone. In addition, this embodiment also introduces two new logging curve characteristics, K1 and K2, based on the data of the study area. The formula is as follows:

[0174] ;

[0175] ;

[0176] Among them, R D is the deep lateral formation resistivity, (Ω.m); R S is the shallow lateral formation resistivity, (Ω.m).

[0177] (V) Construction of reservoir type sub-model based on geological knowledge. Using geological observation methods such as cores, thin sections, electron microscope images, and imaging logging data, the macroscopic lithology differences and rock physical structure differences of different reservoirs are determined. On this basis, the core analysis test data such as porosity and permeability experiments are used to clarify the differences in microscopic pore structures of different reservoirs. Based on the relationship between different pore and throat combinations as the basis for classification, a classification standard for dual-medium tight sandstone reservoir types is established. Finally, a semi-supervised cluster analysis is performed using the DBSCAN algorithm, and the fluid identification data set is divided into multiple sub-data sets according to the semi-supervised results, and each sub-data set represents a potential reservoir type in the large category. Fig. 9 The porosity and permeability scatter plots of different reservoir types provided in the embodiments of this application. Fig. 9 As shown, based on the above method, this study divides the reservoir types in the study area into four types: fracture reservoirs, matrix pore reservoirs, microfracture-pore composite reservoirs and non-reservoir types, among which non-reservoir reservoirs are not identified.

[0178] (VI) Fluid discrimination parameters constructed based on Fisher algorithm. The data set is divided based on the divided reservoir types. For different reservoir types, the Fisher algorithm is used to fuse comprehensive logging parameters and fluid identification sensitive factors, and remove low-correlation variables that have little impact on the results. Finally, the fluid discrimination parameters suitable for different reservoir types are calculated by fusion. The fluid discrimination parameters finally constructed for fracture-type reservoirs, matrix pore-type reservoirs, and microfracture-pore composite reservoirs are F1, F2, and F3, and the formula is as follows:

[0179] ;

[0180] ;

[0181] ;

[0182] (VII) Integrated identification of reservoir fluids. Based on multiple sub-models constructed based on geological knowledge of reservoir types, an integrated method is further used to identify reservoir fluids in a more refined manner. Based on multiple sub-models constructed based on geological knowledge of reservoir types, only the sampling points predicted to be reservoirs are used as input, and an integrated machine learning algorithm is used to perform deep learning on the fluids of each subdivided reservoir to identify the fluid distribution characteristics under different reservoir types. The data input in this embodiment includes R D , R S, AC, CNL, GR, SDXM, SDYM, K1, K2, A1. This example model is mainly constructed using five algorithm modules: FCN, XGBoost, CatBoost, DT, and RF. This time, different integrated model weights were constructed based on the constraints of geological models for fracture reservoirs, pore reservoirs, and microfracture-pore composite reservoirs, forming three models.

[0183] Fig.10 This is a comparison chart of the accuracy of blind well cross-check of the algorithm model provided in the embodiment of this application. Fig.10 As shown in the figure, through the cross-check of blind wells in the study area, the average recognition accuracy of the final model reached 85.2%, among which the highest recognition accuracy could reach more than 90%, effectively improving the recognition accuracy. Fig.11 This is a comparison chart of the importance of the algorithm model features provided in the embodiment of this application. Fig.11 As shown in the figure, the XGBoost algorithm module is used to output the feature importance, which proves the effectiveness of the proposed recognition factor.

[0184] 8. Model output and quality control. Fig.12 This is a schematic diagram of the blind well identification effect of the algorithm model provided in the embodiment of the present application. Fig.12 As shown, the output of the qualified fluid identification integrated algorithm model is used to complete the reservoir fluid logging interpretation of a single well.

[0185] In the above embodiments, reservoir fluid identification is described in detail, and the present application also provides corresponding embodiments of a reservoir fluid identification device.

[0186] Fig.13 This is a schematic diagram of a reservoir fluid identification device provided in an embodiment of the present application. Fig.13 As shown, the device comprises:

[0187] The acquisition module 10 is used to acquire the well logging data of the reservoir in the study area and determine the inferred fluid type of the reservoir;

[0188] The correction module 11 is used to obtain the original resistivity of the reservoir, and perform ground stress resistivity correction and fracture resistivity correction on the original resistivity to obtain the corrected resistivity of the reservoir;

[0189] A first generating module 12 is used to construct a fluid sensitivity factor according to the corrected resistivity, and to generate a plurality of data sets representing different reservoir types according to the logging curve data, the corrected resistivity and the fluid sensitivity factor;

[0190] The second generating module 13 is used to generate fluid discrimination parameters suitable for different reservoir types according to each data set and a preset algorithm;

[0191] The identification module 14 is used to input the logging curve data, the inferred fluid type and the fluid discrimination parameters into the integrated identification model to output the fluid type of the reservoir in the study area;

[0192] Among them, the integrated identification model is a model constructed based on multiple integrated machine learning algorithms to predict reservoir fluid types.

[0193] Since the embodiments of the apparatus part correspond to the embodiments of the method part, please refer to the description of the embodiments of the method part for the embodiments of the apparatus part, which will not be repeated here.

[0194] Fig.14 This is a schematic diagram of a reservoir fluid identification device provided in an embodiment of the present application. Fig.14 As shown, reservoir fluid identification includes:

[0195] A memory 20, used for storing computer programs;

[0196] The processor 21 is used to implement the steps of the reservoir fluid identification method mentioned in the above embodiment when executing the computer program.

[0197] The reservoir fluid identification device provided in this embodiment may include but is not limited to a smart phone, a tablet computer, a laptop computer or a desktop computer.

[0198] Among them, the processor 21 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 21 can be implemented in at least one hardware form of a digital signal processor (DSP), a field programmable gate array (FPGA), and a programmable logic array (PLA). The processor 21 may also include a main processor and a coprocessor. The main processor is a processor for processing data in an awake state, also known as a central processing unit (CPU); the coprocessor is a low-power processor for processing data in a standby state. In some embodiments, the processor 21 may be integrated with a graphics processing unit (GPU), and the GPU is responsible for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor 21 may also include an artificial intelligence (AI) processor, which is used to process computing operations related to machine learning.

[0199] The memory 20 may include one or more computer-readable storage media, which may be non-transitory. The memory 20 may also include a high-speed random access memory, and a non-volatile memory, such as one or more disk storage devices, flash memory storage devices. In this embodiment, the memory 20 is at least used to store the following computer program 201, wherein, after the computer program is loaded and executed by the processor 21, it can implement the relevant steps of the reservoir fluid identification method disclosed in any of the aforementioned embodiments. In addition, the resources stored in the memory 20 may also include an operating system 202 and data 203, etc., and the storage method may be temporary storage or permanent storage. Among them, the operating system 202 may include Windows, Unix, Linux, etc. Data 203 may include but is not limited to data related to the reservoir fluid identification method.

[0200] In some embodiments, the reservoir fluid identification device may further include a display screen 22 , an input / output interface 23 , a communication interface 24 , a power source 25 , and a communication bus 26 .

[0201] Those skilled in the art will understand that Fig.14 The structure shown in the figure does not constitute a limitation of the reservoir fluid identification device, and may include more or less components than shown in the figure.

[0202] Finally, the present application also provides an embodiment corresponding to a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps recorded in the above method embodiment are implemented.

[0203] It is understandable that if the method in the above embodiment is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0204] The above is a detailed introduction to a reservoir fluid identification method, device, equipment and medium provided by the present application. The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same and similar parts between the embodiments can be referenced to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the scope of protection of the present application.

[0205] It should also be noted that, in this specification, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

Claims

1. A reservoir fluid identification method, characterized in that: include: Obtaining well logging data of reservoirs in the study area and determining the inferred fluid type of the reservoirs; Acquiring the original resistivity of the reservoir, and performing ground stress resistivity correction and fracture resistivity correction on the original resistivity to obtain a corrected resistivity of the reservoir; constructing a fluid sensitivity factor according to the corrected resistivity, and generating a plurality of data sets representing different reservoir types according to the well logging data, the corrected resistivity and the fluid sensitivity factor; Generating fluid discrimination parameters applicable to different reservoir types according to each of the data sets and a preset algorithm; Inputting the logging curve data, the inferred fluid type and the fluid discrimination parameter into an integrated identification model to output the fluid type of the reservoir in the study area; Among them, the integrated identification model is a model for predicting reservoir fluid types constructed based on multiple integrated machine learning algorithms.

2. The reservoir fluid identification method according to claim 1, characterized in that: Before obtaining the original resistivity of the reservoir and after obtaining the logging curve data of the reservoir in the study area, the method further includes: Acquiring core data and analytical materials of the reservoir, and performing depth offset correction on the logging curve data according to the core data and the analytical materials; Eliminating erroneous data and / or peak data in the logging curve data; Processing abnormal values ​​and / or missing values ​​in the well logging curve data; The logging curve data of different order of magnitude sequences are converted into the same order of magnitude sequences conforming to the standard normal distribution through Z-score normalization, and the logging curve data uniformly tested in the whole well or the logging curve data tested in different formation sections are excluded to pre-process the logging curve data.

3. The reservoir fluid identification method according to claim 1, characterized in that: Performing ground stress resistivity correction and fracture resistivity correction on the original resistivity includes: Obtaining formation factors at normal temperature and pressure, porosity at normal temperature and pressure, formation water resistivity, original cementation index, corrected cementation index, formation factors under overburden pressure, and overburden porosity of the reservoir; Determine the empirical coefficient of the work area according to the formation factors at normal temperature and pressure, the porosity at normal temperature and pressure, the formation water resistivity, the original cementation index, the corrected cementation index, the formation factors under overburden pressure, the overburden porosity and Archie's formula; Performing geostress resistivity correction on the original resistivity according to the empirical coefficient of the work area to obtain an initial corrected resistivity; Determine the fracture development layer section of the reservoir according to the imaging logging data in the logging curve data, and determine the fracture resistivity of the fracture development layer section; The fracture parameters of the fracture development layer are evaluated by using the dual lateral component method, the fractures with different inclination angles are decomposed into two fracture components, vertical and horizontal, and the lateral resistivity model under the wellbore environment is established according to the differential form of Ohm's law; the lateral resistivity model is solved by the deep and shallow dual lateral simultaneous equations to obtain the fracture inclination angle, fracture aperture and fracture porosity; Fracture resistivity correction is performed according to the initial corrected resistivity, the fracture resistivity, the fracture dip, the fracture aperture and the fracture porosity to obtain the corrected resistivity of the reservoir.

4. The reservoir fluid identification method according to claim 1, characterized in that: Constructing the fluid sensitivity factor according to the corrected resistivity includes: Obtaining a maximum principal stress at a formation level and a minimum principal stress at a formation level of the reservoir; The fluid sensitivity factor is constructed according to the corrected resistivity, the maximum principal stress at the formation level, and the minimum principal stress at the formation level.

5. The reservoir fluid identification method according to claim 1, characterized in that: Generating a plurality of data sets representing different reservoir types according to the well logging data, the corrected resistivity and the fluid sensitivity factor comprises: Acquiring macroscopic lithology data and microscopic rock physical structure parameter data of the reservoir; Determine the macroscopic lithology difference and rock physical structure difference of different reservoirs according to the macroscopic lithology data, and determine the microscopic pore structure difference of different reservoirs according to the microscopic rock physical structure parameter data; According to the macroscopic lithology difference, the rock physical structure difference and the microscopic pore structure difference, the porosity and permeability in the microscopic rock physical structure parameter data are divided to establish a reservoir type classification standard; Calibrate the logging curve data, the corrected resistivity and the fluid sensitivity factor according to the reservoir type classification standard, and perform semi-supervised cluster analysis on the calibrated data by using the DBSCAN algorithm to obtain a cluster analysis result; The logging curve data, the corrected resistivity and the fluid sensitivity factor are divided into a plurality of data sets according to the cluster analysis result.

6. The reservoir fluid identification method according to claim 1, characterized in that: Generating the fluid discrimination parameters applicable to different reservoir types according to each of the data sets and the preset algorithm includes: Inputting the well log data, the corrected resistivity, the fluid sensitivity factor and each of the data sets into a Fisher algorithm model to determine an overall average and a classification average; Determine an overall scatter matrix, an intra-class dispersion matrix, and an inter-class dispersion matrix according to the overall average value and the classification average value; Calculating an eigenvector according to the eigenvalue, the intra-class discreteness matrix and the inter-class discreteness matrix, and determining the eigenvector corresponding to the maximum eigenvalue; Generate an eigenvector matrix according to the eigenvector corresponding to the maximum eigenvalue; The fluid discrimination parameter is determined according to the eigenvector matrix and each of the data sets.

7. The reservoir fluid identification method according to any one of claims 1 to 6, characterized in that: The construction process of the integrated recognition model includes: Get training set sample data and initialize weights; Input the training set sample data and the initialization weights into the full convolutional neural network algorithm module, the extreme gradient boosting tree algorithm module, the categorical feature gradient boosting algorithm module, the decision tree algorithm module and the random forest algorithm module, respectively, to obtain the recognition results output by each algorithm module; Input each of the recognition results into a weak classifier, and determine the error between each of the recognition results and the corresponding true result and the classifier weight through the weak classifier; Adjust the initialization weight according to the classifier weight to obtain a new initialization weight, and return to the step of inputting the training set sample data and the initialization weight into the full convolutional neural network algorithm module, the extreme gradient boosting tree algorithm module, the categorical feature gradient boosting algorithm module, the decision tree algorithm module, and the random forest algorithm module, respectively, until the minimum error is determined; A strong learner is constructed according to the weights of each classifier and the weighted combination strategy, so as to output a final recognition result through the strong learner.

8. A reservoir fluid identification device, characterized in that: include: An acquisition module, used to acquire well logging data of reservoirs in a study area and determine the inferred fluid type of the reservoirs; A correction module, used for acquiring the original resistivity of the reservoir, and performing ground stress resistivity correction and fracture resistivity correction on the original resistivity to obtain a corrected resistivity of the reservoir; A first generating module is used to construct a fluid sensitivity factor according to the corrected resistivity, and to generate a plurality of data sets representing different reservoir types according to the logging curve data, the corrected resistivity and the fluid sensitivity factor; A second generating module, used for generating fluid discrimination parameters applicable to different reservoir types according to each of the data sets and a preset algorithm; An identification module, for inputting the logging curve data, the inferred fluid type and the fluid discrimination parameter into an integrated identification model to output the fluid type of the reservoir in the study area; Among them, the integrated identification model is a model for predicting reservoir fluid types constructed based on multiple integrated machine learning algorithms.

9. A reservoir fluid identification device, characterized in that: include: Memory for storing computer programs; A processor, configured to implement the steps of the reservoir fluid identification method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the reservoir fluid identification method according to any one of claims 1 to 7 are implemented.

Images