Continuous characterization method and system for pore structure of low porosity and permeability reservoirs

Abstract

The present invention relates to a method and system for continuously characterizing the pore structure of low-porosity and permeability reservoirs, belonging to the field of geophysical well logging. The method comprises the following steps: collecting core sample data from a target layer in a study area and well logging data from a target well section in the study area; preprocessing the well logging data to obtain preprocessed well logging data; extracting microscopic parameters from the core sample data, performing correlation analysis on the microscopic parameters using factor analysis to determine the physical significance of the pore structure factor, the pore structure comprehensive factor, and the pore structure factor; selecting sensitive well logging data of the pore structure factor and the pore structure comprehensive factor from the preprocessed well logging data, and establishing an SVR+factor analysis method comprehensive factor characterization model based on the SVR model; and applying the SVR+factor analysis method comprehensive factor characterization model to well logging data from the entire well section in the study area to obtain continuous quantitative characterization of the pore structure of the entire well section. This method achieves continuous quantitative characterization of the pore structure and improves characterization accuracy.

Description

Technical Field

[0001] The present invention belongs to the field of geophysical well logging, and in particular relates to a method and system for continuously characterizing the pore structure of a low-porosity and permeability reservoir. Background Art

[0002] Low-porosity, low-permeability reservoirs are characterized by complex pore structures, strong heterogeneity, and poor pore-throat connectivity. Quantitative characterization of these reservoirs is a key technical bottleneck in oil and gas development. Existing technologies primarily rely on rock physics experiments (such as cast thin sections, nuclear magnetic resonance (NMR), and high-pressure mercury injection) to obtain microscopic parameters. However, these technologies suffer from the following issues: 1) A single experiment only reflects local features of the pore structure (e.g., cast thin sections focus on geometry, NMR reflects pore-throat connectivity, and mercury injection experiments characterize pore-throat distribution). The complex correlations between multiple parameters make unified quantitative evaluation difficult. 2) Rock physics experiments are expensive and sample sizes are limited. Traditional statistical methods (such as multivariate regression) struggle to capture the complex nonlinear relationship between pore structure and logging responses, resulting in insufficient model generalization. 3) Existing technologies rely on discrete core data, making it impossible to continuously predict pore structure across the entire wellbore, limiting the timeliness and accuracy of reservoir evaluation.

[0003] In recent years, machine learning has provided new insights for quantitative characterization of pore structure. However, single models often overlook the physical meaning of parameters, resulting in poor interpretability of predictions. To address this issue, this paper proposes a method that integrates multi-source experimental data with machine learning. This method uses factor analysis to extract comprehensive characterization factors with clear physical meaning. Support Vector Regression (SVR) is then used to achieve high-precision modeling with small sample sizes. This method enables continuous quantitative characterization of the pore structure of low-porosity, low-permeability reservoirs across the entire wellbore. This approach aims to address the limitations of single methods, insufficient data fusion, and difficulties in continuous evaluation, providing technical support for the effective development of low-porosity, low-permeability reservoirs. Summary of the Invention

[0004] In response to the problems existing in the prior art, the present invention provides a method and system for continuous characterization of the pore structure of low-porosity and low-permeability reservoirs, which can solve the problem of quantitative characterization and evaluation of the microscopic pore structure of dense low-porosity and low-permeability reservoirs and improve the accuracy of pore structure characterization of heterogeneous reservoirs.

[0005] In a first aspect, the present invention provides a method for continuously characterizing the pore structure of a low porosity and permeability reservoir, comprising the following steps:

[0006] Data collection steps: Collect core sample data of target layers in the study area and well logging curve data of target well sections in the study area.

[0007] Data preprocessing step: standardizing and normalizing the well logging curve data to obtain preprocessed well logging curve data;

[0008] Parameter extraction step: extracting microscopic parameters from the core sample data;

[0009] Factor analysis step: using factor analysis to conduct correlation analysis on the microscopic parameters to determine the physical meanings of the pore structure factor, the pore structure comprehensive factor and the pore structure factor;

[0010] SVR model establishment step: selecting sensitive logging curve data of pore structure factor and pore structure comprehensive factor from the pre-processed logging curve data, and establishing an SVR+factor analysis method comprehensive factor characterization model based on the SVR model;

[0011] Continuous quantitative characterization step: The SVR+factor analysis method comprehensive factor characterization model is applied to the logging data of the entire well section of the study area to obtain a continuous quantitative characterization of the pore structure of the entire well section.

[0012] In some embodiments, during the data collection step, the core sample data includes:

[0013] Casting thin section data, NMR T2 spectrum data and mercury injection curve experimental data.

[0014] In some embodiments, during the data collection step, the well logging data includes:

[0015] Natural gamma, deep lateral resistivity, shallow lateral resistivity, flushing zone resistivity, resistivity while drilling, compressional wave time difference, compensated neutron, compensated density, and caliper curve.

[0016] In some embodiments, normalizing the well logging curve data includes:

[0017] Wells with complete logging curves, systematic coring data, and complete experimental analysis data are selected as standard wells;

[0018] Compare and analyze geological layers to obtain standard layers;

[0019] According to the standard wells and standard layers, the frequency histogram method is used to perform standardization processing and correction on the logging curve data of each well in the study area.

[0020] In some embodiments, normalizing the well logging curve data includes:

[0021] Using the min-max normalization method, the standardized logging curve data is converted into data between 0 and 1;

[0022] ;

[0023] Where, x i is the logging curve data at a certain depth point,x max is the maximum value of the logging curve data of the target well section, x min It is the minimum value of the logging curve data in the target well section.

[0024] In some embodiments, in the parameter extraction step, the specific steps of extracting microscopic parameters from the core sample data are:

[0025] The casting thin section micro-parameter extraction steps include: performing threshold segmentation and morphological filtering on the casting thin section data to obtain a threshold segmentation image and a morphological filter image; performing edge recognition and detection on the pores and throats in the threshold segmentation image and the morphological filter image using a point counting statistical method to obtain the pore throat distribution in the image, and calculating the casting thin section micro-parameters: pore aspect ratio, roundness, shape factor, tortuosity, surface porosity, and pore throat average diameter ratio;

[0026] ;

[0027] ;

[0028] ;

[0029] ;

[0030] ;

[0031] ;

[0032] Where, represents the pore aspect ratio, Indicates roundness, represents the shape factor, Indicates the tortuosity, represents the face rate, represents the average pore throat diameter ratio, represents the maximum Feret diameter of the pore, represents the minimum Feret diameter of the pore, is the pore area, is the pore perimeter, is the porosity, is the air permeability, is the mean pore throat radius, is the total pore area observed in the thin section, is the total area of ​​the thin slice observation field, is the average pore diameter determined on the casting thin section, is the average throat diameter determined on the casting thin section;

[0033] According to the microscopic parameters of the casting thin section and in combination with the qualitative analysis of the casting thin section, the pore structure is classified based on the casting thin section to obtain a classification result based on the casting thin section;

[0034] The steps of extracting microscopic parameters of NMR T2 spectrum are as follows: using NMR T2 spectrum data, measuring the transverse relaxation distribution of core samples under saturation and centrifugal conditions, establishing NMR T2 distribution spectra under saturation and centrifugal conditions, and obtaining NMR T2 spectrum microscopic parameters: saturation Geometric mean ,saturation Arithmetic mean , centrifugal Geometric mean , centrifugal Arithmetic mean , irreducible water saturation 、 Cutoff value ;

[0035] According to the microscopic parameters of the nuclear magnetic resonance T2 spectrum and in combination with the morphology of the nuclear magnetic resonance T2 distribution spectrum, the pore structure is classified based on the nuclear magnetic resonance T2 spectrum to obtain a classification result based on the nuclear magnetic resonance T2 spectrum;

[0036] Mercury injection curve experiment micro-parameter extraction steps: Using mercury injection curve experimental data, calculate the mercury injection curve experimental micro-parameters: median pressure, median radius, pore throat radius mean, exit efficiency, maximum mercury injection saturation, displacement pressure, maximum connected pore throat radius, sorting coefficient, skewness, structure coefficient, homogeneity coefficient, variation coefficient, characteristic structure parameters:

[0037] ;

[0038] ;

[0039] ;

[0040] ;

[0041] ;

[0042] ;

[0043] ;

[0044] ;

[0045] ;

[0046] ;

[0047] Where, is the median pressure, is the median radius, For exit efficiency, is the maximum mercury saturation, is the displacement pressure, is the maximum connected pore throat radius, is the sorting coefficient, is the skewness, is the structural coefficient, is the homogeneity coefficient, is the coefficient of variation, is the characteristic structure parameter, is the mean pore throat radius, is the mercury saturation at a certain point, is the cumulative mercury saturation at the highest pressure of the experiment, is the mercury saturation remaining in the pores when the mercury is withdrawn to the initial pressure. is the pore throat radius at a certain point; To correspond to Mercury saturation in a certain range, is the porosity, is the air permeability;

[0048] According to the microscopic parameters of the mercury injection experiment and the morphology of the mercury injection curve, the pore structure is classified based on the mercury injection curve experiment to obtain a classification result based on the mercury injection curve experiment.

[0049] In some embodiments, the factor analysis step: performing correlation analysis on the microscopic parameters using factor analysis to determine the physical meanings corresponding to the pore structure factor, the pore structure comprehensive factor, and the pore structure factor, includes:

[0050] KMO and Baetlett tests were performed on the microscopic parameters, and the pore structure factor was determined by combining the scree plot and total variance explained plot;

[0051] The pore structure factor is analyzed by using a factor analysis method to obtain a loading matrix of the pore structure factor;

[0052] Performing factor rotation on the load matrix to obtain the load coefficient of the pore structure factor;

[0053] According to the load coefficient, the correlation between different pore structure factors and pore structure microscopic parameters is analyzed, the physical meaning corresponding to the pore structure factor is determined, and a pore structure factor calculation model is established;

[0054] According to the correlation between the different pore structure factors and the pore structure microscopic parameters, the weights of the different pore structure factors are calculated to obtain a pore structure comprehensive factor calculation model;

[0055] The pore structure comprehensive factor is obtained by using the pore structure comprehensive factor calculation model, and the core sample with consistent classification results based on the casting thin section, the nuclear magnetic resonance T2 spectrum, and the mercury injection curve experiment is selected to calibrate the pore structure comprehensive factor to obtain a calibrated pore structure comprehensive factor;

[0056] Core samples with inconsistent classification results based on casting thin sections, nuclear magnetic resonance T2 spectrum, and mercury injection curve experiments are comprehensively classified according to the calibrated pore structure comprehensive factor.

[0057] In some embodiments, in the SVR model establishment step, sensitive logging curve data of the pore structure factor and the pore structure comprehensive factor are selected from the pre-processed logging curve data, and the SVR+factor analysis method comprehensive factor characterization model is established based on the SVR model. The specific steps are:

[0058] Sensitive logging curve data selection step: using a linear correlation analysis method to perform correlation analysis on the pore structure factor, the pore structure comprehensive factor and the pre-processed logging curve data, and selecting sensitive logging curve data of the pore structure factor and the pore structure comprehensive factor;

[0059] The hyperparameter optimization step of the SVR support vector regression model is as follows: the sensitive logging curve data is input into the SVR support vector regression model, and the hyperparameters of the pore structure factor calculation model and the pore structure comprehensive factor calculation model are adjusted by using a grid search method to obtain the optimal parameters of the hyperparameters;

[0060] Dataset preparation and model construction: Based on the pore structure comprehensive factor and the sensitive logging curve data corresponding to different pore structure comprehensive factors, an SVR support vector regression data label set was established. The data set was divided into a training set and a test set according to a set ratio. The pore structure factor calculation model was trained. According to the factor analysis steps, the SVR+factor analysis method comprehensive factor characterization model was obtained.

[0061] In some embodiments, in the continuous quantitative characterization step, applying the SVR+factor analysis method comprehensive factor characterization model to the well logging data of the entire well section of the study area to obtain the continuous quantitative characterization of the pore structure of the entire well section includes:

[0062] The sensitive logging curve data are substituted into the comprehensive factor characterization model of the SVR+ factor analysis method to calculate the comprehensive factor of the pore structure of the entire well section of the target well in the study area, obtain the continuous classification results of the pore structure of the entire well section of the target well in the study area, and realize the continuous quantitative characterization of the pore structure of the entire well section.

[0063] In a second aspect, the present invention provides a system for continuously characterizing the pore structure of a low porosity and permeability reservoir, which is used to implement the method for continuously characterizing the pore structure of a low porosity and permeability reservoir described in the first aspect of the present invention, comprising:

[0064] Data collection module: collects core sample data of target layers in the study area and logging curve data of target well sections in the study area.

[0065] Data preprocessing module: standardizes and normalizes the well logging curve data to obtain preprocessed well logging curve data;

[0066] Parameter extraction module: extracting microscopic parameters from the core sample data;

[0067] Factor analysis module: using factor analysis to perform correlation analysis on the microscopic parameters to determine the pore structure factors and the physical meanings corresponding to the factors;

[0068] SVR model building module: selects sensitive logging curve data of pore structure factor and pore structure comprehensive factor from the pre-processed logging curve data, and establishes an SVR+factor analysis method comprehensive factor characterization model based on the SVR model;

[0069] Continuous quantitative characterization module: The SVR+factor analysis method comprehensive factor characterization model is applied to the logging data of the entire well section of the study area to obtain a continuous quantitative characterization of the pore structure of the entire well section.

[0070] Compared with the prior art, the advantages and positive effects of the present invention are:

[0071] (1) The present invention uses factor analysis to perform correlation analysis on the microscopic parameters, determines the physical meanings of the pore structure factor, the comprehensive pore structure factor, and the pore structure factor, and expands the characterization dimension from a single parameter to a multi-factor system, significantly enhancing the ability to interpret the complex pore system of low-porosity and low-permeability reservoirs, and achieving the coordinated characterization of microstructure and macroscopic physical properties;

[0072] (2) The present invention selects sensitive logging curve data of pore structure factor and pore structure comprehensive factor from pre-processed logging curve data based on SVR model, and establishes SVR+factor analysis method comprehensive factor characterization model to express the nonlinear relationship between logging curve and pore structure factor, pore structure comprehensive factor, and the error is significantly reduced compared with the traditional multiple regression model;

[0073] (3) The present invention applies the SVR+factor analysis method comprehensive factor characterization model to the logging data of the entire well section of the study area to obtain a continuous quantitative characterization of the pore structure of the entire well section. Only conventional logging data combined with core sample data are needed to complete the pore structure characterization of the entire well section, which significantly reduces the cost; the SVR+factor analysis method comprehensive factor characterization model can directly perform continuous calculation and classification of the pore structure of a single well, which is easy to operate and highly practical. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 A flow chart of a method for continuously characterizing the pore structure of a low porosity and permeability reservoir provided by an embodiment of the present invention;

[0075] Figure 2 This is a flow chart for normalizing well logging curve data provided by an embodiment of the present invention;

[0076] Figure 3 Schematic diagram of the logging response characteristics of the standard layer of the LDX-E well provided by an embodiment of the present invention;

[0077] Figure 4 1 is a schematic diagram of the comparison of the histogram of the natural gamma ray curve and the three-porosity curve before and after normalization of the LDX-C well provided by an embodiment of the present invention;

[0078] Figure 5 This is a flow chart of microscopic parameter extraction of a casting thin slice provided by an embodiment of the present invention;

[0079] Figure 6 Schematic diagram of the results of threshold segmentation and morphological filtering of a cast thin slice provided by an embodiment of the present invention;

[0080] Figure 7 Schematic diagram of the nuclear magnetic resonance T2 spectrum characteristics of four types of pore structures provided by the embodiments of the present invention;

[0081] Figure 8 Schematic diagram of mercury intrusion curve characteristics of four types of pore structures provided by embodiments of the present invention;

[0082] Figure 9 is a flowchart of factor analysis steps provided by an embodiment of the present invention;

[0083] Figure 10 This is a schematic diagram of the basis for selecting the pore structure factor quantity provided by an embodiment of the present invention;

[0084] Figure 11 is a thermodynamic diagram of the pore structure factor load matrix provided by an embodiment of the present invention;

[0085] Figure 12 is the comprehensive factor of different pore structures provided by the embodiment of the present invention Statistical diagram;

[0086] Figure 13 is a flow chart of the steps for establishing an SVR model provided by an embodiment of the present invention;

[0087] Figure 14 1 is a radar diagram of correlation coefficients between well logging data and pore structure factors and pore structure comprehensive factors provided by an embodiment of the present invention;

[0088] Figure 15 The pore structure comprehensive factors calculated by the three methods provided in the embodiment of the present invention are Schematic diagram;

[0089] Figure 16 Schematic diagram of the pore structure prediction results and logging results of the LDX-E well provided by an embodiment of the present invention;

[0090] Figure 17 A schematic diagram of a system for continuously characterizing the pore structure of a low-porosity and permeability reservoir provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0091] In order to make the purpose, technical solutions and advantages of this application more clearly understood, the present application is described and illustrated below in conjunction with the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely used to explain this application and are not intended to limit this application. Based on the embodiments provided in this application, all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of this application.

[0092] The present invention provides a method and system for continuously characterizing the pore structure of low-porosity and permeability reservoirs. The method collects core sample data from target layers in a study area, as well as well logging data from target well sections in the study area, and standardizes and normalizes the well logging data. Microscopic parameters are extracted from the core sample data. Correlation analysis is performed on the microscopic parameters using factor analysis to determine pore structure factors and their corresponding physical meanings. Sensitive well logging data for the pore structure factor and the comprehensive pore structure factor are selected as input, and a comprehensive factor characterization model (SVR+factor analysis) is established using the SVR model. The comprehensive factor characterization model (SVR+factor analysis) is applied to well logging data from all well sections in the study area to obtain a continuous quantitative characterization of the pore structure. The following describes the method and system for continuously characterizing the pore structure of low-porosity and permeability reservoirs in detail, with reference to the accompanying drawings.

[0093] like Figure 1 As shown, the first embodiment of the present invention provides a method for continuously characterizing the pore structure of a low porosity and permeability reservoir, the steps of which are:

[0094] S101, data collection step: collecting core sample data of target layers in the study area and well logging curve data of target well sections in the study area.

[0095] Specifically, the core sample data includes: casting thin section data, nuclear magnetic resonance T2 spectrum data and mercury injection curve experimental data.

[0096] A feasible method is to directly collect the casting thin section, nuclear magnetic resonance T2 spectrum and mercury injection curve experimental test results data of the core samples of the target layer in the study area.

[0097] Another feasible method is to carry out experimental measurements of cast thin sections, nuclear magnetic resonance T2 spectra and mercury injection curves of core samples in the target layer to obtain corresponding experimental data.

[0098] Specifically, the logging curve data includes: natural gamma ray GR, deep lateral resistivity R D , shallow lateral resistivity R S , flushing zone resistivity R xo , resistivity while drilling, longitudinal wave time difference AC or DTC, compensated neutron CNL, compensated density DEN, caliper curve CAL.

[0099] The resistivity while drilling includes the phase resistivity while drilling and the attenuation resistivity while drilling at different lateral detection depths.

[0100] For example, the while-drilling phase resistivity P16H, P16L, P22H, P22L, P28H, P28L, P34H, P34L, P40H, P40L, and the while-drilling attenuation resistivity A16H, A16L, A22H, A22L, A28H, A28L, A34H, A34L, A40H, A40L.

[0101] S102, data preprocessing step: standardizing and normalizing the well logging curve data to obtain preprocessed well logging curve data;

[0102] In one embodiment, the well logging data is normalized, such as Figure 2 As shown, the specific steps include:

[0103] S201, selecting wells with complete logging curves, systematic coring data, and complete experimental analysis data as standard wells.

[0104] S202, performing comparative analysis on geological strata to obtain standard layers.

[0105] S203, based on the standard wells and standard layers, the well logging curve data of each well in the study area are standardized and corrected using a frequency histogram method.

[0106] For example, Figure 3As shown, Well LDX-E in the study area was selected as the reference well. This well has complete logging curves with no missing values, systematic coring data, and complete analytical data. After comparative analysis of geological strata, the 4009-4025 m interval was selected as the reference layer. This interval has continuous and stable lithology, a large thickness, and is located at the top of the second member of the Huangliu Formation. This layer facilitates tracking and comparison across the entire area, and exhibits a gentle change in logging response. The frequency histogram method was used to standardize and correct the natural gamma ray (GR), acoustic travel time (AC), compensated neutron (CNL), and compensated density (DEN) curves for each well in the study area.

[0107] Further, such as Figure 4 As shown in the figure, taking the LDX-C well as an example, from the histogram before and after the natural gamma ray (GR) standardization (a), the histogram before and after the acoustic wave time difference (AC) standardization (b), the histogram before and after the compensated neutron (CNL) standardization (c), and the histogram before and after the compensated density (DEN) standardization (d), it can be seen that the distribution trend of the normalized curve is consistent with that of the standard well, and is basically normally distributed, indicating that the standardization is reliable.

[0108] In one embodiment, normalizing the well logging curve data includes: using a min-max normalization method to convert the standardized well logging curve data into data between 0 and 1.

[0109] Specifically, the min-max normalization calculation formula is:

[0110] ;

[0111] Where, x i is the logging curve data at a certain depth point, x max is the maximum value of the logging curve data of the target well section, x min It is the minimum value of the logging curve data in the target well section.

[0112] S103: Parameter extraction step, extracting microscopic parameters from the core sample data.

[0113] Specifically, such as Figure 5 As shown, the following steps are included:

[0114] S301, casting thin section micro-parameter extraction step: threshold segmentation and morphological filtering are performed on the casting thin section data to obtain a threshold segmentation image and a morphological filter image; edge recognition and detection of pores and throats in the threshold segmentation image and the morphological filter image are performed using a point counting statistical method to obtain the pore and throat distribution in the image, and the casting thin section micro-parameters are calculated: pore aspect ratio, roundness, shape factor, tortuosity, surface porosity, and pore-throat average diameter ratio.

[0115] Specifically, the calculation formula for the microscopic parameters of the casting thin film is:

[0116] ;

[0117] ;

[0118] ;

[0119] ;

[0120] ;

[0121] ;

[0122] Where, represents the pore aspect ratio, Indicates roundness, represents the shape factor, Indicates the tortuosity, represents the face rate, represents the average pore throat diameter ratio, represents the maximum Feret diameter of the pore, represents the minimum Feret diameter of the pore, is the pore area, is the pore perimeter, is the porosity, is the air permeability, is the mean pore throat radius, is the total pore area observed in the thin section, is the total area of ​​the thin slice observation field, is the average pore diameter determined on the casting thin section, is the average throat diameter determined on the casting thin section;

[0123] For example, Figure 6 As shown in the figure, the grayscale image of the casting slice (b) is obtained according to the original image of the casting slice (a). After the threshold segmentation and morphological filtering of the casting slice, the threshold segmentation image (c) and the morphological filtering image (d) are obtained. The point counting statistical method is used to Figure 6 The pores and throats in the threshold segmentation image (c) and the morphological filter image (d) are identified and detected by edge recognition to obtain the pore and throat distribution in the image, and the microscopic parameters of the casting thin section are calculated.

[0124] S302 , classifying the pore structure based on the casting thin slice according to the microscopic parameters of the casting thin slice and combining with the qualitative analysis of the casting thin slice, to obtain a classification result based on the casting thin slice.

[0125] For example, after calculating the microscopic parameters of the casting thin section based on the casting thin section data, the pore structure is divided into 4 categories based on the qualitative analysis of the casting thin section. The pore structure gradually deteriorates from Class I pores to Class IV pores. The pore throat coordination number of all samples of Class I pore structure is between 0.56 and 0.73, and the tortuosity is Between 1.38 and 1.67, the face rate The average pore throat diameter ratio is between 10.82% and 14.68%. The pore throat coordination numbers of all samples of type II pore structure are between 0.42 and 0.68, and the tortuosity is between 1.59 and 1.91. Between 1.44 and 1.95, the face rate The average pore throat diameter ratio is between 8.19% and 11.67%. The pore throat coordination number of all samples of type III pore structure is between 0.22 and 0.46, and the tortuosity is between 2.42 and 3.55. Between 1.51 and 2.13, the face rate The average pore throat diameter ratio is between 4.28% and 9.11%. The pore throat coordination numbers of all samples of type IV pore structure are between 0.12 and 0.28, and the tortuosity is between 2.57 and 4.34. Between 1.64 and 2.92, the face rate The average pore throat diameter ratio is between 1.15% and 5.42%. Between 2.83 and 4.92.

[0126] S303, NMR T2 spectrum micro-parameter extraction step: using NMR T2 spectrum data, measure the transverse relaxation distribution of the core sample under saturation and centrifugal conditions, establish NMR T2 distribution spectra under saturation and centrifugal conditions, and obtain NMR T2 spectrum micro-parameters: saturation Geometric mean ,saturation Arithmetic mean , centrifugal Geometric mean , centrifugal Arithmetic mean , irreducible water saturation 、 Cutoff value .

[0127] Among them, saturation Geometric mean ,saturation Arithmetic mean , centrifugal Geometric mean , centrifugal Arithmetic mean It is mainly used to reflect the pore size of the sample. The larger the value, the more macropores there are. The comparative analysis of the geometric mean and the arithmetic mean can reflect the degree of skewness of the distribution of large and small pores. It is calculated by the ratio of the porosity of the core sample after centrifugation to the total porosity of the core sample, which mainly reflects the pore throat connectivity; Cutoff value The method of obtaining is that when the cumulative porosity component of the nuclear magnetic resonance T2 spectrum in the saturated state is equal to the total cumulative porosity component of the nuclear magnetic resonance T2 spectrum in the centrifugal state, the corresponding The value is Cutoff value , which can be used to reflect the pore throat distribution.

[0128] S304 , performing NMR T2 spectrum-based classification on the pore structure according to the NMR T2 spectrum microscopic parameters and the morphology of the NMR T2 distribution spectrum to obtain a NMR T2 spectrum-based classification result.

[0129] For example, Figure 7 As shown in the figure, the pore structure of the study area is classified according to the characteristics of the microscopic parameters of the nuclear magnetic resonance T2 spectrum and the morphology of the nuclear magnetic resonance T2 distribution spectrum. From the T2 spectrum characteristic diagram of the type I pore structure (a) and the T2 spectrum characteristic diagram of the type I pore structure before and after centrifugation (b), it can be seen that the T2 spectrum of the type I pore structure is a bimodal distribution, and the peak representing the small pores and the peak representing the large pores are more than 1:2. Irreduced water saturation of type I pore structure Between 23.98% and 58.75%, Cutoff value Between 3.15 ms and 39.62 ms, saturation Geometric mean Between 20.7 ms and 33.54 ms, saturation Arithmetic mean At 39.55 ms~69.27 ms, centrifugal Geometric mean Between 8.35 ms and 11.81 ms, centrifugal Arithmetic mean Between 14.86 ms and 26.20 ms. The yellow area in the T2 spectrum characteristic diagram (b) of the Type I pore structure before and after centrifugation shows that there is a large amount of movable fluid in the large pores of the Type I pore structure sample, and the pore structure is relatively simple. From the T2 spectrum characteristic diagram (c) of the Type II pore structure and the T2 spectrum characteristic diagram (d) of the Type II pore structure before and after centrifugation, it can be seen that the T2 spectrum of the Type II pore structure has a bimodal distribution, and the peak representing the small pores and the peak representing the large pores are below 1:2 and above 1:1. Irreduced water saturation of Type II pore structure Between 21.19% and 70.31%, Cutoff value Between 3.69 ms and 41.32 ms, saturation Geometric mean Between 9.17 ms and 19.96 ms, saturation Arithmetic mean Between 26.73 ms and 66.27 ms, the centrifugal Geometric mean Between 3.82 ms and 8.15 ms, centrifugal Arithmetic mean Between 6.74 ms and 11.56 ms. The yellow area in the T2 spectrum characteristic diagram (d) of the Type II pore structure before and after centrifugation shows that the movable fluid in the Type II pore structure sample is less than that in the Type I pore structure sample, indicating a better pore structure. From the T2 spectrum characteristic diagram (e) of the Type III pore structure and the T2 spectrum characteristic diagram (f) of the Type II pore structure before and after centrifugation, it can be seen that the T2 spectrum of the Type III pore structure has a bimodal distribution, with the peak representing small pores and the peak representing large pores at a ratio of 1:1 or less than 1:1. Irreduced water saturation of Type III pore structure Between 37.24% and 91.6%, Cutoff value Between 3.73 ms and 73.82 ms, saturation Geometric mean Between 6.32 ms and 14.29 ms, saturation Arithmetic mean Between 9.26 ms and 62.07 ms, the centrifugal Geometric mean Between 4.91 ms and 12.08 ms, the centrifugal Arithmetic mean Between 3.20 ms and 36.67 ms. The yellow area in the T2 spectrum characteristic diagram (f) of the Class III pore structure before and after centrifugation shows that the movable fluid in the Class III pore structure sample is significantly less than that in the Class I and Class II, the overall content is low, and the pore structure is complex. From the T2 spectrum characteristic diagram (g) of the Class IV pore structure and the T2 spectrum characteristic diagram (h) of the Class II pore structure before and after centrifugation, it can be seen that the T2 spectrum of the Class IV pore structure is a single-peak distribution, and the peak mainly appears at the position representing the small pore. Irreduced water saturation of the Class IV pore structure Between 65.95% and 97.55%, Cutoff value Between 7.71 ms and 109 ms, saturation Geometric mean Between 1.22 ms and 6.89 ms, saturation Arithmetic mean Between 12.7 ms and 63.38 ms, centrifugal Geometric mean Between 1.06 ms and 6.04 ms, centrifugal Arithmetic mean The T2 spectrum characteristic diagram (h) of the Type IV pore structure before and after centrifugation shows that the Type IV pore structure sample has very little movable fluid and a complex pore structure.

[0130] S305, mercury injection curve experiment micro-parameter extraction step: using the mercury injection curve experimental data, calculate the mercury injection curve experimental micro-parameters: median pressure, median radius, mean pore throat radius, exit efficiency, maximum mercury injection saturation, displacement pressure, maximum connected pore throat radius, sorting coefficient, skewness, structure coefficient, homogeneity coefficient, coefficient of variation, and characteristic structure parameters.

[0131] Specifically, the specific calculation formula for the microscopic parameters of the mercury intrusion curve experiment is:

[0132] ;

[0133] ;

[0134] ;

[0135] ;

[0136] ;

[0137] ;

[0138] ;

[0139] ;

[0140] ;

[0141] ;

[0142] Where, is the median pressure, is the median radius, For exit efficiency, is the maximum mercury saturation, is the displacement pressure, is the maximum connected pore throat radius, is the sorting coefficient, is the skewness, is the structural coefficient, is the homogeneity coefficient, is the coefficient of variation, is the characteristic structure parameter, is the mean pore throat radius, is the mercury saturation at a certain point, is the cumulative mercury saturation at the highest pressure of the experiment, is the mercury saturation remaining in the pores when the mercury is withdrawn to the initial pressure. is the pore throat radius at a certain point; To correspond to Mercury saturation in a certain range, is the porosity, is the air permeability.

[0143] S306 , classifying the pore structure based on the mercury intrusion curve experiment according to the microscopic parameters of the mercury intrusion experiment and the morphology of the mercury intrusion curve, and obtaining a classification result based on the mercury intrusion curve experiment.

[0144] For example, Figure 8 As shown in the figure, according to the shape of mercury injection curve and the mercury injection microscopic parameters that characterize the pore structure, the pore structure is divided into four categories. The mercury injection curve of type I pore structure (a) is "concave", and the shape is low, flat and long. Mercury saturation is between 0.12 MPa and 0.25 MPa. Maximum mercury saturation The exit efficiency is high between 85.83% and 90.13%, and the pore throat radius is mainly distributed between 1.007μm and 3.967μm, with medium pore throats being the main ones. The mercury injection curve of type II pore structure (b) is "S-shaped", with the characteristics of low, steep and long displacement pressure. Mercury saturation is between 0.06Mpa and 0.4Mpa. Maximum mercury saturation Between 72.65% and 84.31%, the exit efficiency is low, the pore throat radius is mainly distributed between 0.63μm and 2.5μm, and is dominated by small and medium pore throats; the mercury injection curve of type III pore structure (c) is "S-shaped", with the characteristics of high, flat and long, and the displacement pressure is Mercury saturation is between 0.19 MPa and 1.06 MPa. Maximum mercury saturation The exit efficiency is between 68.68% and 83.53%, and the pore throat radius is mainly distributed between 0.25 μm and 1 μm, with small and medium pore throats being the main ones. The mercury injection curve of type IV pore structure (d) is "concave", with the characteristics of high, steep and short shape. The displacement pressure Mercury saturation is between 0.26 MPa and 4.4 MPa. Maximum mercury saturation Between 52.17% and 77.31%, the exit efficiency is low, and the pore throat radius is mainly distributed between 0.016μm and 0.392μm, with small pore throats being the main ones.

[0145] S104, factor analysis step: using factor analysis method to perform correlation analysis on the microscopic parameters to determine the pore structure factor, the pore structure comprehensive factor and the physical meaning corresponding to the pore structure factor.

[0146] Specifically, such as Figure 9 As shown, the following steps are included:

[0147] S401, performing KMO and Baetlett tests on the microscopic parameters, and determining the pore structure factor in combination with the scree plot and the total variance explanation plot.

[0148] S402: Analyze the pore structure factor using a factor analysis method to obtain a load matrix of the pore structure factor.

[0149] Specifically, according to the calculation formula, the correlation coefficient matrix of the micro parameters is calculated.

[0150] Among them, it is assumed that the indicator variables for factor analysis are indivual: , the evaluation objects are , The first evaluation object The value of the indicator is , the values ​​of each indicator Convert to standardized indicators , then the calculation formula is:

[0151] ;

[0152] Where:

[0153] ;

[0154] ;

[0155] in, 、 Respectively The sample mean and sample standard deviation of each indicator;

[0156] ;

[0157] Where, , , It is The indicator and The correlation coefficient of the indicators.

[0158] Furthermore, the eigenvalues ​​and corresponding eigenvectors of the correlation coefficient matrix are calculated, and the load matrix of the pore structure factor is obtained according to the eigenvalues ​​and corresponding eigenvectors:

[0159] ;

[0160] Where:

[0161] ;

[0162] in, 、 … is the eigenvalue, , is the corresponding eigenvector.

[0163] Furthermore, the contribution rate of each pore structure factor is calculated according to the load matrix of the pore structure factor, and m main microscopic parameters are selected.

[0164] S403, performing factor rotation on the load matrix to obtain the load coefficient of the pore structure factor.

[0165] Specifically, the load matrix is ​​rotated to obtain a rotated load matrix:

[0166] ;

[0167] in, for The first m columns of is an orthogonal matrix.

[0168] Furthermore, the pore structure factor model is constructed:

[0169] .

[0170] Furthermore, the pore structure factor score is calculated to obtain the load coefficient of the pore structure factor;

[0171] Among them, the single pore structure factor score function is obtained by regression method:

[0172] ;

[0173] remember

[0174] ;

[0175] Then we have:

[0176] .

[0177] S404: Analyze the correlation between different pore structure factors and pore structure microscopic parameters based on the load coefficient, determine the physical meaning of the pore structure factors, and establish a pore structure factor calculation model.

[0178] S405 , calculating the weights of different pore structure factors according to the correlations between the different pore structure factors and the pore structure microscopic parameters, and obtaining a pore structure comprehensive factor calculation model.

[0179] S406, using the pore structure comprehensive factor calculation model to obtain a pore structure comprehensive factor, selecting core samples that are consistent with the classification results based on the casting thin section, the classification results based on the nuclear magnetic resonance T2 spectrum, and the classification results based on the mercury injection curve experiment to calibrate the pore structure comprehensive factor, and obtain a calibrated pore structure comprehensive factor.

[0180] S407 , comprehensively classifying core samples whose classification results based on casting thin sections, nuclear magnetic resonance T2 spectrum, and mercury injection curve experiments are inconsistent according to the calibrated pore structure comprehensive factor.

[0181] In an exemplary embodiment, the factor analysis steps described in the embodiment of the present application are performed on the casting thin section, nuclear magnetic resonance T2 spectrum, and mercury intrusion curve experimental microscopic parameters, such as Figure 10 As shown in the scree plot (a) and the total variance explained plot (b), it can be seen that when the number of factors is 4, the eigenvalue is 1.527 and the cumulative variance contribution rate is 96.1%. Four pore structure factors are selected for analysis.

[0182] Furthermore, after obtaining the load coefficient of the pore structure factor, the load matrix heat map of the four pore structure factors: factor 1, factor 2, factor 3, and factor 4 is established, as shown in Figure 11 As shown in the figure, the redder or yellower the color, the greater the correlation between the pore structure factor and the microscopic parameters, while orange indicates a smaller correlation. Figure 11 It can be seen that the factor 1 is related to the saturation Geometric mean, saturation Arithmetic mean, centrifugal Geometric mean, centrifugal The arithmetic mean, displacement pressure, median pressure, median radius, maximum connected pore throat radius, and mean pore throat radius are highly correlated with microscopic parameters reflecting pore throat size. Factor 2 is highly correlated with microscopic parameters reflecting pore throat connectivity, such as irreducible water saturation, maximum mercury injection saturation, exit efficiency, pore throat coordination number, tortuosity, structural coefficient, and average pore throat diameter ratio. Factor 3 is highly correlated with The cutoff value, sorting coefficient, coefficient of variation, characteristic structure coefficient and other microscopic parameters reflecting pore throat distribution are highly correlated; factor 4 is highly correlated with shape factor, pore aspect ratio, roundness and other microscopic parameters reflecting pore throat geometry.

[0183] Furthermore, according to the correlation between different pore structure factors and microscopic parameters, the physical meanings of different pore structure factors are determined: Factor 1 is defined as the pore throat size factor , factor 2 is the pore throat connectivity factor , factor 3 is the pore throat distribution factor , factor 4 is the pore throat geometry factor According to the correlation between different factors and microscopic parameters, the weights of different microscopic parameters are calculated and the pore structure factor calculation model is established:

[0184] ;

[0185] ;

[0186] ;

[0187] .

[0188] Furthermore, according to the correlation between different pore structure factors and microscopic parameters, the 、 、 、 The weight of pore structure comprehensive factor calculation model is obtained:

[0189] .

[0190] Furthermore, the classification results of all core samples based on cast thin slices, nuclear magnetic resonance T2 spectrum classification results, mercury injection curve experimental classification results and pore structure comprehensive factors were counted. , calibrated with core samples of three experimental pore structure classifications .like Figure 12 As shown, it can be seen that the core samples of the three experiments are classified as type I pore structure All of them are greater than 10.24, which is classified as the core sample of type II pore structure. Core samples with a pore structure between 6.15 and 9.94 are classified as type III Core samples with a pore structure between 3 and 5.82 are classified as type IV According to the principle of conservative estimation, the pore structure of type I is calibrated. Greater than 10.24, type II pore structure Between 6.15 and 10.24, the type III pore structure Between 2.98 and 6.15, the IV type pore structure Less than 2.98, according to the calibration The scope is to conduct comprehensive classification of cores whose classification results are inconsistent based on thin film of casting, classification results based on nuclear magnetic resonance T2 spectrum and classification results based on mercury injection curve experiment.

[0191] S105, SVR model establishment step: selecting sensitive logging curve data of pore structure factor and pore structure comprehensive factor from the pre-processed logging curve data, and establishing an SVR+factor analysis method comprehensive factor characterization model based on the SVR model.

[0192] SVR, short for Support Vector Regression, is a regression analysis method based on support vector machines (SVMs) and widely used in forecasting and pattern recognition. SVR predicts continuous variables by finding an optimal hyperplane in high-dimensional space to maximize the distance between data points and the hyperplane.

[0193] Specifically, such as Figure 13 As shown, the following steps are included:

[0194] S501, sensitive logging curve data selection step: using linear correlation analysis method to perform correlation analysis on the pore structure factor, pore structure comprehensive factor and pre-processed logging curve data, and selecting sensitive logging curve data of the pore structure factor and pore structure comprehensive factor.

[0195] For example, the linear correlation analysis method is used to directly select the well logging curves that are sensitive to the changes in different pore structure factors as input. The natural gamma ray GR, acoustic wave time difference AC, compensated neutron CNL, compensated density DEN, wellbore CAL, while-drilling phase shift resistivity P40H and P16H, while-drilling attenuation resistivity A40H and A16H are calculated. 、 、 、 、 The correlation of the values ​​and the obtained correlation coefficient Do radar chart analysis, such as Figure 14 As shown, the red dotted line in the figure is the boundary, that is, As the boundary, the sensitive logging curve data of pore structure factor and pore structure comprehensive factor are selected. Select GR, P40H, P16H, A40H, AC, CNL, DEN; Select GR, CAL, P40H, P16H, A40H, A16H, AC, CNL, DEN; Select GR, P16H, A40H, A16H, CNL; Choose GR, P40H, P16H, AC, CNL; Select GR, CAL, P40H, P16H, A40H, A16H, AC, CNL, DEN.

[0196] S502, hyperparameter optimization step of the SVR support vector regression model: input the sensitive logging curve data into the SVR support vector regression model, and use the grid search method to adjust the hyperparameters of the pore structure factor calculation model and the pore structure comprehensive factor calculation model respectively to obtain the optimal parameters of the hyperparameters.

[0197] For example, the grid search method is used to search 、 、 、 、 Calculation model adjustment hyperparameters, pore throat size factor The penalty coefficient C, the degree of deviation And the RBF kernel function coefficient 10000, 0.15 and 0.03 respectively; pore throat connectivity factor The penalty coefficient C, the degree of deviation And the RBF kernel function coefficient are 10000, 0.15 and 0.03 respectively; pore throat distribution factor The penalty coefficient C, the degree of deviation And the RBF kernel function coefficient are 10000, 0.10 and 0.025 respectively; the pore throat geometry factor The penalty coefficient C, the degree of deviation And the RBF kernel function coefficient 1000, 0.10 and 0.03 respectively; comprehensive pore structure factor The penalty coefficient C, the degree of deviation And the RBF kernel function coefficient They are 10000, 0.15 and 0.015 respectively.

[0198] S503, data set preparation and model construction: Based on the pore structure comprehensive factor and combined with the sensitive logging curve data corresponding to different pore structure comprehensive factors, an SVR support vector regression data label set is established, the data set is divided into a training set and a test set according to a set ratio, the pore structure factor calculation model is trained, and according to the factor analysis steps, the SVR+factor analysis method comprehensive factor characterization model is obtained.

[0199] In order to verify the effectiveness of the comprehensive factor characterization model of the SVR+ factor analysis method in this embodiment, the pore structure comprehensive factor calculated by the multiple regression method, the SVR support vector regression method and the method of the embodiment of the application was analyzed. A comparative analysis was conducted, and the results were as follows Figure 15 As shown in the figure (a) of the multiple regression method calculation effect, it can be seen that the multiple regression method calculates Correlation coefficient =0.8; From the SVR support vector regression method calculation effect diagram (b), it can be seen that the SVR support vector regression method calculates Correlation coefficient =0.93; From the calculation effect diagram (c) of the embodiment of the present application, it can be seen that the calculation result obtained by the embodiment of the present application is Correlation coefficient =0.96, and the calculation accuracy is significantly better than the other two methods, which verifies the effectiveness of the comprehensive factor characterization model of the SVR+factor analysis method in this embodiment.

[0200] S106, continuous quantitative characterization step: applying the SVR+factor analysis method comprehensive factor characterization model to the logging data of the entire well section of the study area to obtain a continuous quantitative characterization of the pore structure of the entire well section.

[0201] Specifically, the sensitive logging curve data is substituted into the comprehensive factor characterization model of the SVR+ factor analysis method to calculate the comprehensive factor of the pore structure of the entire well section of the target well in the study area, obtain the continuous classification results of the pore structure of the entire well section of the target well in the study area, and realize the continuous quantitative characterization of the pore structure of the entire well section.

[0202] In order to verify the accuracy of the continuous quantitative characterization of pore structure in the embodiment of the present application, the pore structure comprehensive factor calculated by the multiple regression method, the SVR support vector regression method and the method of the embodiment of the present application was analyzed. A comparative analysis was conducted, and the results were as follows Figure 16As shown, it can be seen that the pore structure prediction of the embodiment method of the present application corresponds to the pore structure comprehensive classification result, and has a good performance in thin layer pore structure identification; the pore structure prediction of the SVR support vector machine regression method and the pore structure comprehensive classification result misclassify the Class II pore structure into Class III at 4094.4 m~4095.8 m, which corresponds well; while the pore structure prediction of the multivariate regression method and the pore structure comprehensive classification result have a large deviation, which verifies the accuracy of the continuous quantitative characterization of the pore structure of the embodiment of the present application.

[0203] like Figure 17 As shown, the second embodiment of the present invention provides a system for continuously characterizing the pore structure of a low porosity and permeability reservoir, which is used to realize the continuous characterization of the pore structure of a low porosity and permeability reservoir described in the first aspect of the present invention, comprising:

[0204] Data collection module 10: collects core sample data of target layers in the study area, as well as well logging curve data of target well sections of each well in the study area.

[0205] The data preprocessing module 20 standardizes and normalizes the well logging curve data to obtain preprocessed well logging curve data.

[0206] Parameter extraction module 30: extracts microscopic parameters from the core sample data.

[0207] Factor analysis module 40: performs correlation analysis on the microscopic parameters using factor analysis method to determine the pore structure factors and the physical meanings corresponding to the factors.

[0208] SVR model building module 50: Selecting sensitive logging curve data of pore structure factor and pore structure comprehensive factor from the pre-processed logging curve data, and building an SVR+factor analysis method comprehensive factor characterization model based on the SVR model.

[0209] Continuous quantitative characterization module 60: applying the SVR+factor analysis method comprehensive factor characterization model to the logging data of the entire well section of the study area to obtain a continuous quantitative characterization of the pore structure of the entire well section.

[0210] The above embodiments are used to explain the present invention rather than to limit the present invention. Any modifications and changes made to the present invention within the spirit of the present invention and the protection scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A method for continuous characterization of pore structure of low porosity and permeability reservoirs, characterized in that: The steps are: Data collection steps: Collect core sample data of target layers in the study area and well logging curve data of target well sections in the study area; Data preprocessing step: standardizing and normalizing the well logging curve data to obtain preprocessed well logging curve data; Parameter extraction step: extracting microscopic parameters from the core sample data; Factor analysis step: using factor analysis to conduct correlation analysis on the microscopic parameters to determine the physical meanings of the pore structure factor, the pore structure comprehensive factor and the pore structure factor; SVR model establishment step: selecting sensitive logging curve data of pore structure factor and pore structure comprehensive factor from the pre-processed logging curve data, and establishing an SVR+factor analysis method comprehensive factor characterization model based on the SVR model; Continuous quantitative characterization step: The SVR+factor analysis method comprehensive factor characterization model is applied to the logging data of the entire well section of the study area to obtain a continuous quantitative characterization of the pore structure of the entire well section.

2. The method for continuous characterization of pore structure of low porosity and permeability reservoir according to claim 1, characterized in that: In the data collection step, the core sample data includes: Casting thin section data, NMR T2 spectrum data and mercury injection curve experimental data.

3. The method for continuous characterization of pore structure of low porosity and permeability reservoir according to claim 1, characterized in that: In the data collection step, the well logging curve data includes: Natural gamma, deep lateral resistivity, shallow lateral resistivity, flushing zone resistivity, resistivity while drilling, compressional wave time difference, compensated neutron, compensated density, and caliper curve.

4. The method for continuous characterization of pore structure of low porosity and permeability reservoir according to claim 1, characterized in that: The step of normalizing the well logging curve data includes: Wells with complete logging curves, systematic coring data, and complete experimental analysis data are selected as standard wells; Compare and analyze geological layers to obtain standard layers; According to the standard wells and standard layers, the frequency histogram method is used to perform standardization processing and correction on the logging curve data of each well in the study area.

5. The method for continuous characterization of pore structure of low porosity and permeability reservoir according to claim 1, characterized in that: Normalizing the well logging curve data includes: Using the min-max normalization method, the standardized logging curve data is converted into data between 0 and 1; ； Where, After normalization Well logging data at depth points, yes Well logging data at depth points, is the maximum value of the logging curve data of the target well section, It is the minimum value of the logging curve data in the target well section.

6. The method for continuous characterization of pore structure of low porosity and permeability reservoir according to claim 1, characterized in that: In the parameter extraction step, the specific steps of extracting microscopic parameters from the core sample data are as follows: The casting thin section micro-parameter extraction steps include: performing threshold segmentation and morphological filtering on the casting thin section data to obtain a threshold segmentation image and a morphological filter image; performing edge recognition and detection on the pores and throats in the threshold segmentation image and the morphological filter image using a point counting statistical method to obtain the pore throat distribution in the image, and calculating the casting thin section micro-parameters: pore aspect ratio, roundness, shape factor, tortuosity, surface porosity, and pore throat average diameter ratio; ； ； ； ； ； ； Where, represents the pore aspect ratio, Indicates roundness, represents the shape factor, Indicates the tortuosity, represents the face rate, represents the average pore throat diameter ratio, represents the maximum Feret diameter of the pore, represents the minimum Feret diameter of the pore, is the pore area, is the pore perimeter, is the porosity, is the air permeability, is the mean pore throat radius, is the total pore area observed in the thin section, is the total area of ​​the thin slice observation field, is the average pore diameter determined on the casting thin section, is the average throat diameter determined on the casting thin section; According to the microscopic parameters of the casting thin section and in combination with the qualitative analysis of the casting thin section, the pore structure is classified based on the casting thin section to obtain a classification result based on the casting thin section; The steps of extracting microscopic parameters of NMR T2 spectrum are as follows: using NMR T2 spectrum data, measuring the transverse relaxation distribution of core samples under saturation and centrifugal conditions, establishing NMR T2 distribution spectra under saturation and centrifugal conditions, and obtaining NMR T2 spectrum microscopic parameters: saturation Geometric mean ,saturation Arithmetic mean , centrifugal Geometric mean , centrifugal Arithmetic mean , irreducible water saturation 、 Cutoff value ; According to the microscopic parameters of the nuclear magnetic resonance T2 spectrum and in combination with the morphology of the nuclear magnetic resonance T2 distribution spectrum, the pore structure is classified based on the nuclear magnetic resonance T2 spectrum to obtain a classification result based on the nuclear magnetic resonance T2 spectrum; Mercury injection curve experiment micro-parameter extraction steps: Using mercury injection curve experimental data, calculate the mercury injection curve experimental micro-parameters: median pressure, median radius, pore throat radius mean, exit efficiency, maximum mercury injection saturation, displacement pressure, maximum connected pore throat radius, sorting coefficient, skewness, structure coefficient, homogeneity coefficient, variation coefficient, characteristic structure parameters: ； ； ； ； ； ； ； ； ； ； Where, is the median pressure, is the median radius, For exit efficiency, is the maximum mercury saturation, is the displacement pressure, is the maximum connected pore throat radius, is the sorting coefficient, is the skewness, is the structural coefficient, is the homogeneity coefficient, is the coefficient of variation, is the characteristic structure parameter, is the mean pore throat radius, is the mercury saturation at a certain point, is the cumulative mercury saturation at the highest pressure of the experiment, is the mercury saturation remaining in the pores when the mercury is withdrawn to the initial pressure. is the pore throat radius at a certain point; To correspond to Mercury saturation in a certain range, is the porosity, is the air permeability; According to the microscopic parameters of the mercury injection experiment and the morphology of the mercury injection curve, the pore structure is classified based on the mercury injection curve experiment to obtain a classification result based on the mercury injection curve experiment.

7. The method for continuous characterization of pore structure of low porosity and permeability reservoir according to claim 1, characterized in that: The factor analysis step: using factor analysis to perform correlation analysis on the microscopic parameters to determine the pore structure factor, the pore structure comprehensive factor and the physical meaning corresponding to the pore structure factor, including: KMO and Baetlett tests were performed on the microscopic parameters, and the pore structure factor was determined by combining the scree plot and total variance explained plot; The pore structure factor is analyzed by using a factor analysis method to obtain a loading matrix of the pore structure factor; Performing factor rotation on the load matrix to obtain the load coefficient of the pore structure factor; According to the load coefficient, the correlation between different pore structure factors and pore structure microscopic parameters is analyzed, the physical meaning corresponding to the pore structure factor is determined, and a pore structure factor calculation model is established; According to the correlation between the different pore structure factors and the pore structure microscopic parameters, the weights of the different pore structure factors are calculated to obtain a pore structure comprehensive factor calculation model; The pore structure comprehensive factor is obtained by using the pore structure comprehensive factor calculation model, and the core sample with consistent classification results based on the casting thin section, the nuclear magnetic resonance T2 spectrum, and the mercury injection curve experiment is selected to calibrate the pore structure comprehensive factor to obtain a calibrated pore structure comprehensive factor; Core samples with inconsistent classification results based on casting thin sections, nuclear magnetic resonance T2 spectrum, and mercury injection curve experiments are comprehensively classified according to the calibrated pore structure comprehensive factor.

8. The method for continuous characterization of pore structure of low porosity and permeability reservoir according to claim 1, characterized in that: In the SVR model establishment step, sensitive logging curve data of the pore structure factor and the pore structure comprehensive factor are selected from the pre-processed logging curve data, and the SVR+factor analysis method comprehensive factor characterization model is established based on the SVR model. The specific steps are as follows: Sensitive logging curve data selection step: using a linear correlation analysis method to perform correlation analysis on the pore structure factor, the pore structure comprehensive factor and the pre-processed logging curve data, and selecting sensitive logging curve data of the pore structure factor and the pore structure comprehensive factor; The hyperparameter optimization step of the SVR support vector regression model is as follows: the sensitive logging curve data is input into the SVR support vector regression model, and the hyperparameters of the pore structure factor calculation model and the pore structure comprehensive factor calculation model are adjusted by using a grid search method to obtain the optimal parameters of the hyperparameters; Dataset preparation and model construction: Based on the pore structure comprehensive factor and the sensitive logging curve data corresponding to different pore structure comprehensive factors, an SVR support vector regression data label set is established. The data set is divided into a training set and a test set according to a set ratio. The pore structure factor calculation model is trained. The trained pore structure factor calculation model is subjected to the factor analysis steps described above to obtain the SVR+factor analysis method comprehensive factor characterization model.

9. The method for continuous characterization of pore structure of low porosity and permeability reservoir according to claim 1, characterized in that: In the continuous quantitative characterization step, the SVR+factor analysis method comprehensive factor characterization model is applied to the well logging data of the entire well section in the study area to obtain the continuous quantitative characterization of the pore structure of the entire well section, including: The sensitive logging curve data are substituted into the comprehensive factor characterization model of the SVR+ factor analysis method to calculate the comprehensive factor of the pore structure of the entire well section of the target well in the study area, obtain the continuous classification results of the pore structure of the entire well section of the target well in the study area, and realize the continuous quantitative characterization of the pore structure of the entire well section.

10. A system for continuously characterizing the pore structure of a low porosity and permeability reservoir, used to implement the method for continuously characterizing the pore structure of a low porosity and permeability reservoir according to any one of claims 1 to 9, characterized in that: include: Data collection module: collects core sample data of target layers in the study area and logging curve data of target well sections in the study area. Data preprocessing module: standardizes and normalizes the well logging curve data to obtain preprocessed well logging curve data; Parameter extraction module: extracting microscopic parameters from the core sample data; Factor analysis module: using factor analysis to perform correlation analysis on the microscopic parameters to determine the pore structure factors and the physical meanings corresponding to the factors; SVR model building module: selects sensitive logging curve data of pore structure factor and pore structure comprehensive factor from the pre-processed logging curve data, and establishes an SVR+factor analysis method comprehensive factor characterization model based on the SVR model; Continuous quantitative characterization module: The SVR+factor analysis method comprehensive factor characterization model is applied to the logging data of the entire well section of the study area to obtain a continuous quantitative characterization of the pore structure of the entire well section.

Images