A deep learning quantitative characterization method and system for CT scanning oil layer oil content

Abstract

The present invention belongs to the technical field of oil and gas exploration and development, and provides a deep learning quantitative characterization method and system for CT scanning oil content in oil reservoirs. The method determines CT scanning parameters through a digital intelligence model of the oil reservoir to obtain a three-dimensional grayscale image of the oil reservoir, and establishes a standardized three-dimensional digital model by comparing it with the electron microscope image; based on this, a deep learning standard model is created using a convolutional neural network to extract parameters such as oil content and oil saturation; based on sample experiments and analysis, judgment criteria are determined, and evaluation criteria and digital three-dimensional models of different types of oil reservoirs are constructed to screen out advantageous target layers. The system includes data acquisition, model construction, parameter extraction, and evaluation and decision-making modules for implementing this method. The present invention solves the problem of inaccurate measurements in existing systems, improves the authenticity and accuracy of quantitative predictions of oil content in oil reservoirs, enhances the efficiency of oil exploration and development, and reduces costs.

Description

Technical Field

[0001] The present invention belongs to the technical field of oil and gas exploration and development, and specifically relates to a deep learning quantitative characterization method and system for CT scanning oil content in oil layers. Background Art

[0002] The quantitative characterization of oil content in oil reservoirs has always been a world-class technical problem that has long been needed to be solved in oil and gas exploration and development projects but has not been fundamentally solved.

[0003] Currently, the main technologies for characterizing oil content in reservoirs are manual laboratory measurements using plunger cores and qualitative testing using scanning electron microscopes. Manual laboratory measurements using plunger cores can only obtain random, non-continuous, scattered values ​​of oil content and are unable to detect the three-dimensional spatial distribution of crude oil in the reservoir. Scanning electron microscopes can only detect the distribution of crude oil in reservoirs below micrometers. Furthermore, due to the limited field of view of scanning electron microscopes, they can only depict the distribution characteristics of crude oil in reservoirs within a very small range of two-dimensional slices, but cannot depict the spatial variation of crude oil distribution in the reservoir. Current methods can only obtain random probability values ​​and cannot effectively characterize the true three-dimensional spatial distribution of crude oil in the reservoir. This results in low accuracy in the distribution of crude oil in the reservoir and its oil content, which in turn leads to technical issues with low accuracy in oil and gas exploration and development in the reservoir. Summary of the Invention

[0004] The present invention aims to overcome the shortcomings of the aforementioned background technology by providing a deep learning quantitative characterization method for reservoir oil content using CT scanning. This method addresses the technical problem of inaccurate measurements in the prior art, which results in low accuracy in the spatial status of crude oil in the reservoir, and in turn, low accuracy in predictions of oil and gas exploration and development in the reservoir. This invention improves the authenticity and accuracy of quantitative predictions of reservoir oil content, thereby increasing the efficiency and reducing the cost of oil exploration and development.

[0005] In order to solve the above technical problems, the present invention is implemented through the following technical solutions:

[0006] In a first aspect, the present invention provides a deep learning quantitative characterization method for oil content in oil reservoirs using CT scanning, comprising:

[0007] Obtain a three-dimensional grayscale image of the oil layer through CT scanning, and then use the three-dimensional grayscale image of the oil layer to establish a standardized three-dimensional digital model of the oil content of the oil layer;

[0008] Based on the standardized 3D digital model of oil-bearing properties of oil reservoirs, a deep learning standard model of oil-bearing properties of oil reservoirs is created through deep learning of 3D grayscale images of oil reservoirs.

[0009] Based on the deep learning standard model of oil-bearing properties of oil reservoirs, a deep learning algorithm is applied to extract the digital result model of oil-bearing properties of oil reservoirs, and the oil content and oil saturation are obtained as the descriptive parameters of oil-bearing properties of oil reservoirs.

[0010] Obtaining judgment criteria for oil content and oil saturation, parameters characterizing oil content of the oil layer, and determining evaluation criteria for different types of oil layers based on the judgment criteria and the oil content and oil saturation of the oil layer;

[0011] Determine the evaluation criteria of the oil layer based on the oil content and oil saturation of the different types of oil layers and construct digital three-dimensional models of the oil content of the different types of oil layers;

[0012] Based on the digital three-dimensional oil-bearing models of different types of oil layers and the evaluation criteria of different types of oil layers, the dominant target intervals of the oil layers are evaluated and determined.

[0013] As a further optimization solution of the present invention, the method of obtaining a three-dimensional grayscale image of the oil layer by CT scanning of the oil layer includes:

[0014] The oil layer digital intelligence model is applied to determine the oil layer CT scanning parameters, and the oil layer is CT scanned according to the CT scanning parameters to obtain a three-dimensional grayscale image of the oil layer.

[0015] As a further optimization solution of the present invention, the method for constructing the oil reservoir digital intelligence model is:

[0016] Based on information data of different geological conditions and different types of oil layers, this information data is preprocessed, feature extracted and analyzed, and trained using machine learning algorithms to construct a digital intelligence model of the oil layer.

[0017] As a further optimization solution of the present invention, the method of establishing a three-dimensional digital model of the oil layer using the three-dimensional grayscale image of the oil layer includes:

[0018] Using image registration algorithm, the field emission scanning electron microscope grayscale image and the focused ion beam scanning electron microscope oil layer grayscale image are compared with the three-dimensional grayscale image of the oil layer obtained by CT scanning. By comparing the feature information in the image, the three-dimensional grayscale image of the oil layer obtained by CT scanning is calibrated and verified. Based on the comparison results, a standardized three-dimensional digital model of the oil content of the oil layer scanned by CT is established.

[0019] As a further optimization solution of the present invention, the method of creating a deep learning standard model of oil content of an oil layer based on the standardized three-dimensional digital model of oil content of the oil layer through deep learning of the three-dimensional grayscale image of the oil layer includes:

[0020] Based on the standardized three-dimensional digital model of oil-bearing properties of oil reservoirs, a convolutional neural network is adopted as the deep learning model architecture. During the model training process, the three-dimensional grayscale images of oil reservoirs obtained by CT scanning are used as training data. The model is trained for multiple rounds, and the parameters of the model are continuously adjusted through the back-propagation algorithm to create a deep learning standard model of oil-bearing properties of oil reservoirs.

[0021] As a further optimization solution of the present invention, the judgment criteria for obtaining the oil content and oil saturation of the oil layer characterization parameters include:

[0022] Different types of oil layer samples are selected, and a variety of experimental methods are used to determine the true values ​​of the oil content and oil saturation of the samples. Combined with the geological characteristics and production data of the oil layers, a comprehensive analysis is conducted to determine the judgment criteria for the oil content and oil saturation of different types of oil layers.

[0023] As a further optimization solution of the present invention, the evaluation and preferential determination of the dominant target intervals of the oil layer based on the digital three-dimensional oil-bearing models of different types of oil layers and the evaluation criteria of different types of oil layers includes:

[0024] Based on the digital 3D oil-bearing models of different types of oil layers and the evaluation criteria for different types of oil layers, a comprehensive evaluation method is used to comprehensively evaluate the oil layers, determine the weight of each factor, and screen out advantageous target layers by calculating the comprehensive evaluation score.

[0025] In a second aspect, the present invention provides a deep learning quantitative characterization system for oil content of oil layers scanned by CT, which is used to implement the deep learning quantitative characterization method for oil content of oil layers scanned by CT, and the system comprises:

[0026] The data acquisition module is used to apply the oil reservoir digital intelligence model to determine the oil reservoir CT scanning parameters, and perform CT scanning on the oil reservoir based on these parameters to obtain a three-dimensional grayscale image of the oil reservoir. At the same time, it collects field emission scanning electron microscope grayscale images, focused ion beam scanning electron microscope oil reservoir grayscale images, and various experimental data of different types of oil reservoir samples;

[0027] The model building module is used to use an image registration algorithm to compare the grayscale images of field emission scanning electron microscopes and focused ion beam scanning electron microscope oil layer grayscale images with the three-dimensional grayscale images of the oil layer obtained by CT scanning, and to establish a standardized three-dimensional digital model of the oil content of the oil layer after calibration and verification; it is used to use the standardized three-dimensional digital model of the oil content of the oil layer as a basis, adopt a convolutional neural network architecture, use the three-dimensional grayscale images of the oil layer obtained by CT scanning to perform multiple rounds of training, and adjust parameters through the back propagation algorithm to create a deep learning standard model of the oil content of the oil layer; it is also used to construct digital three-dimensional models of the oil content of different types of oil layers based on the experimental data and analysis results of different types of oil layer samples, and determine the evaluation criteria for different types of oil layers;

[0028] The parameter extraction module is used to extract the digital result model of oil content of the oil layer based on the deep learning standard model of oil content of the oil layer, and obtain the oil content description parameters such as oil content and oil saturation;

[0029] The evaluation and decision-making module is used to conduct a comprehensive evaluation of the oil layers based on the digital three-dimensional models of oil content and evaluation standards of different types of oil layers, determine the weights of various factors, calculate the comprehensive evaluation scores, screen out the dominant target layers, and output the evaluation results and recommended dominant target layer information.

[0030] In a third aspect, the present invention provides an electronic device comprising a memory, a processor and a computer program, wherein the computer program is stored in the memory and configured to be executed by the processor to implement the deep learning quantitative characterization method for oil content in CT scan oil layers.

[0031] In a fourth aspect, the present invention provides a computer-readable storage medium, in which a computer program is stored. The computer program is executed by a processor to implement the deep learning quantitative characterization method for oil content of CT scanning oil layers.

[0032] Compared with the existing technology, the deep learning quantitative characterization method and system for CT scanning oil-bearing properties of oil reservoirs proposed in this invention have the following significant advantages and beneficial effects:

[0033] The present invention determines CT scanning parameters by constructing a digital intelligent model of the oil layer and obtains a high-precision three-dimensional grayscale image of the oil layer, providing a reliable data basis for subsequent analysis and improving the accuracy of the constructed three-dimensional digital model of the oil content of the oil layer, thereby greatly improving the accuracy and efficiency of oil and gas exploration and development predictions. By using a variety of electron microscope images for comparison and calibration, a standardized three-dimensional digital model is established, and deep learning technology is used to create a deep learning standard model, which can more accurately extract the oil content characteristics of the oil layer and improve the comprehensiveness and reliability of the optimization model for evaluating the oil content development characteristics of the oil layer. The oil layer evaluation standards and digital three-dimensional models constructed based on accurate parameters such as oil content and oil saturation can more accurately determine the dominant target oil layer segments, thereby effectively improving the efficiency of oil and gas exploration and development and reducing the cost of oil and gas exploration and development. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0035] Figure 1 Grayscale image of oil layer development obtained based on CT scanning.

[0036] Figure 2 To obtain a three-dimensional digital quantitative model map of oil content in oil layers through deep learning (the red high-probability area is the oil-rich zone).

[0037] Figure 3 To obtain digital quantitative result curve of oil content in oil layers through deep learning.

[0038] Figure 4 To obtain digital quantitative result curve of oil saturation of oil layer through deep learning. DETAILED DESCRIPTION

[0039] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0040] The flowcharts used in the present invention illustrate operations implemented according to some embodiments of the present invention. It should be understood that the operations in the flowcharts may be implemented out of sequence, and steps that have no logical contextual relationship may be reversed or performed simultaneously. Those skilled in the art, guided by the present invention, may add one or more additional operations to the flowcharts or remove one or more operations from the flowcharts. These functional entities may be implemented in software.

[0041] References to "embodiments" in this disclosure mean that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the disclosure. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments. Example 1

[0042] The present invention provides a deep learning quantitative characterization method for oil content of oil layers using CT scanning, the method comprising:

[0043] S1. Obtaining a three-dimensional grayscale image of the oil layer through CT scanning of the oil layer

[0044] The oil reservoir digital intelligence model is constructed based on a large amount of oil reservoir sample data. This sample data covers information on different geological conditions and oil reservoir types, including lithology, pore structure, fluid properties, and other aspects of the reservoir. Data cleaning and feature engineering techniques are used to preprocess the collected data, remove outliers and noise, and extract features that are critical for determining CT scan parameters. The preprocessed data is divided into training and test sets, and the model is trained using a random forest algorithm. During training, algorithm parameters (such as the number of decision trees and maximum depth) are continuously adjusted to improve model accuracy. The trained model is then validated using the test set to ensure that it can accurately predict CT scan parameters for different oil reservoirs, thereby constructing an oil reservoir digital intelligence model that accurately reflects the relationship between oil reservoir characteristics and CT scan parameters.

[0045] According to the determined CT scanning parameters, high-precision CT scanning equipment is used to scan the oil layer, such as a microfocus industrial CT (model example: ZEISS Xradia 620 Versa), with a resolution of 1μm and a scanning layer thickness of 10μm, covering the entire scale of the oil layer (centimeter to meter level). Before scanning, the equipment is calibrated and debugged to ensure scanning accuracy. Set scanning parameters such as voltage, current, scanning layer thickness, etc., and perform a full-scale scan of the oil layer according to the set parameters to obtain a clear and accurate three-dimensional grayscale image of the oil layer. During the scanning process, strictly control the scanning environment and keep the temperature and humidity constant to ensure the stability and accuracy of the scanning results. Figure 1 shown.

[0046] S2. Use the three-dimensional grayscale image of the oil layer to establish a standardized three-dimensional digital model of the oil content of the oil layer

[0047] Grayscale images from field emission scanning electron microscopy (FESEM, 5nm resolution) and focused ion beam scanning electron microscopy (FIB-SEM) were compared with three-dimensional grayscale images of the oil reservoir obtained from CT scans. FESEM, with its high resolution, clearly reveals the microscopic pore structure and crude oil distribution details of the oil reservoir; FIB-SEM allows for high-precision analysis of specific areas. During the comparison process, image registration algorithms (such as the feature point-based SIFT algorithm) were used to align the different FESEM images with the CT scan images through feature point matching. The CT scan images were calibrated and verified by comparing the image features (such as pore shape, size, and crude oil distribution). Based on the comparison results, three-dimensional modeling techniques (such as voxel-based modeling) were used to construct a standardized three-dimensional digital model of the oil reservoir's oil content, based on the CT scans. This model accurately reflects the reservoir's true structure and oil content.

[0048] S3. Based on the standardized three-dimensional digital model of oil-bearing properties of oil reservoirs, a deep learning standard model of oil-bearing properties of oil reservoirs is created through deep learning of three-dimensional grayscale images of oil reservoirs.

[0049] Based on a standardized 3D digital model of oil reservoir oil content, deep learning technology is used to analyze 3D grayscale images of oil reservoirs. A convolutional neural network (CNN) is used as the deep learning model architecture, which has powerful image feature extraction capabilities. During model training, a large number of 3D grayscale images of oil reservoirs are used as training data, and the model undergoes multiple rounds of training. The backpropagation algorithm continuously adjusts the model parameters (such as convolution kernel weights and biases) to enable the model to accurately identify oil content features in the images, thereby creating a standard deep learning model of oil reservoir oil content.

[0050] S4. Based on the deep learning standard model of oil-bearing properties of oil reservoirs, a deep learning algorithm is applied to extract the digital result model of oil-bearing properties of oil reservoirs, and the oil content and oil saturation are obtained as the descriptive parameters of oil-bearing properties of oil reservoirs;

[0051] Obtaining judgment criteria for oil content and oil saturation, parameters characterizing oil content of the oil layer, and determining evaluation criteria for different types of oil layers based on the judgment criteria and the oil content and oil saturation of the oil layer;

[0052] The oil content and oil saturation of the different types of oil layers are used to determine the evaluation criteria for the oil layers and to construct digital three-dimensional oil content models for the different types of oil layers.

[0053] Among them, the three-dimensional grayscale image of the oil layer is input into the trained oil layer oil content deep learning standard model, and the oil layer oil content digital result model is obtained through the forward propagation calculation of the model, such as Figure 2 As shown in the figure, a three-dimensional digital quantitative model of oil content of the oil layer is obtained through deep learning, and then the oil content description parameter of the oil layer is obtained from the model (such as oil content). Figure 3 Continuous quantitative data of oil content in the oil layer shown), oil saturation (such as Figure 4 As shown, the oil saturation of the oil layer is continuously quantitatively measured).

[0054] To determine the criteria for oil content and oil saturation, several representative reservoir samples were selected and accurately measured in the laboratory using core flooding experiments and nuclear magnetic resonance (NMR) experiments. Simultaneously, the samples' geological characteristics, production data, and other information were analyzed to establish a relationship model between oil content, oil saturation, and these factors. Based on this relationship model and extensive experimental data, criteria for determining oil content and oil saturation for different reservoir types were determined. Based on these criteria and the actual reservoir conditions, evaluation criteria for different reservoir types were determined, such as categorizing reservoirs into high-quality, medium-quality, and low-quality reservoirs. Using 3D modeling technology, digital 3D models of the oil content of different reservoir types were constructed. These models visually demonstrate the oil content of each reservoir type, providing strong support for subsequent reservoir evaluation.

[0055] S5. Based on the digital three-dimensional oil-bearing models of different types of oil layers and the evaluation criteria of different types of oil layers, evaluate and preferentially determine the dominant target intervals of the oil layers.

[0056] Based on the digital 3D models and evaluation criteria for different oil reservoir types, an evaluation index system was established, including factors such as oil content, oil saturation, reservoir thickness, and permeability. A comprehensive evaluation method (such as a combination of the analytic hierarchy process (AHP) and the fuzzy comprehensive evaluation method) was employed to comprehensively evaluate the reservoirs. The AHP method was used to determine the weights of each indicator, and a judgment matrix was constructed using expert scoring methods to calculate the relative weights of each indicator. The fuzzy comprehensive evaluation method was then used to evaluate the reservoirs, fuzzifying the values ​​of each indicator and constructing a fuzzy relationship matrix. A comprehensive evaluation score was calculated based on the weights and the fuzzy relationship matrix. The reservoirs were then ranked by score to identify preferred target intervals, providing an accurate basis for target selection in oil and gas exploration and development. Example 2

[0057] Based on a general inventive concept, an embodiment of the present application provides a deep learning quantitative characterization system for oil content in oil reservoirs using CT scanning, the system comprising:

[0058] The data acquisition module is used to apply the oil reservoir digital intelligence model to determine the oil reservoir CT scanning parameters, and perform CT scanning on the oil reservoir based on these parameters to obtain a three-dimensional grayscale image of the oil reservoir. At the same time, it collects field emission scanning electron microscope grayscale images, focused ion beam scanning electron microscope oil reservoir grayscale images, and various experimental data of different types of oil reservoir samples;

[0059] The model building module is used to use an image registration algorithm to compare the grayscale images of field emission scanning electron microscopes and focused ion beam scanning electron microscope oil layer grayscale images with the three-dimensional grayscale images of the oil layer obtained by CT scanning, and to establish a standardized three-dimensional digital model of the oil content of the oil layer after calibration and verification; it is used to use the standardized three-dimensional digital model of the oil content of the oil layer as a basis, adopt a convolutional neural network architecture, use the three-dimensional grayscale images of the oil layer obtained by CT scanning to perform multiple rounds of training, and adjust parameters through the back propagation algorithm to create a deep learning standard model of the oil content of the oil layer; it is also used to construct digital three-dimensional models of the oil content of different types of oil layers based on the experimental data and analysis results of different types of oil layer samples, and determine the evaluation criteria for different types of oil layers;

[0060] The parameter extraction module is used to extract the digital result model of oil content of the oil layer based on the deep learning standard model of oil content of the oil layer, and obtain the oil content description parameters such as oil content and oil saturation;

[0061] The evaluation and decision-making module is used to conduct a comprehensive evaluation of the oil layers based on the digital three-dimensional models of oil content and evaluation standards of different types of oil layers, determine the weights of various factors, calculate the comprehensive evaluation scores, screen out the dominant target layers, and output the evaluation results and recommended dominant target layer information. Example 3

[0062] Based on a general inventive concept, an embodiment of the present application provides an electronic device, including a memory, a processor and a computer program, wherein the computer program is stored in the memory and is configured to be executed by the processor to implement the deep learning quantitative characterization method of CT scanning oil layer oil content. Example 4

[0063] Based on a general inventive concept, an embodiment of the present application provides a computer-readable storage medium, in which a computer program is stored. The computer program is executed by a processor to implement the deep learning quantitative characterization method for oil content in CT scanning oil layers.

[0064] In summary, the deep learning quantitative characterization method and system for oil content in oil reservoirs using CT scanning provided by the embodiments of the present invention have the following significant advantages:

[0065] Using a digital model of the oil reservoir, CT scanning parameters are determined to obtain a three-dimensional grayscale image of the reservoir. Grayscale images are then compared, calibrated, and verified using field emission scanning electron microscopy and focused ion beam scanning electron microscopy. This results in a standardized three-dimensional digital model of the reservoir's oil content, which accurately reflects the reservoir's true structure and oil content characteristics. Based on this standardized three-dimensional digital model, deep learning using a convolutional neural network was employed to create a standard deep learning model that accurately identifies oil content characteristics in the image. This model was then used to extract parameters such as oil content and oil saturation, significantly improving the accuracy of parameter extraction compared to traditional methods.

[0066] By selecting samples from different oil reservoir types and employing a variety of experimental methods to determine the true values ​​of oil content and oil saturation, and combining geological characteristics with production data to determine judgment criteria, different oil reservoir types are classified and a digital 3D model of oil content is constructed, enabling accurate evaluation of the reservoirs. This precise evaluation provides a more targeted basis for exploration and development, and helps formulate rational exploration and development plans. Based on the digital 3D models and evaluation criteria for different oil reservoir types, a hierarchical analysis method is combined with a fuzzy comprehensive evaluation method. Weights are determined for multiple factors, including oil content, oil saturation, reservoir thickness, and permeability. A comprehensive evaluation score is calculated to identify advantageous target intervals, accurately locate oil and gas exploration and development, and improve exploration and development efficiency.

[0067] Traditional methods provide inaccurate reservoir information, leading to significant ineffective exploration and development efforts. This invention, by providing accurate reservoir oil content data and precise target interval screening, avoids excessive investment in low-quality reservoirs or non-renewable zones, reducing unnecessary exploration and development investment and ultimately lowering exploration costs. Accurate oil content parameters and reservoir evaluation enable more rational production planning, improve oil and gas extraction efficiency, avoid resource waste, and further reduce development costs.

[0068] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

[0069] The contents not described in detail in the specification of the present invention belong to the prior art known to those skilled in the art.

Claims

1. A deep learning quantitative characterization method for oil content in oil reservoirs using CT scanning, characterized in that: include: Obtain a three-dimensional grayscale image of the oil layer through CT scanning, and then use the three-dimensional grayscale image of the oil layer to establish a standardized three-dimensional digital model of the oil content of the oil layer; Based on the standardized 3D digital model of oil-bearing properties of oil reservoirs, a deep learning standard model of oil-bearing properties of oil reservoirs is created through deep learning of 3D grayscale images of oil reservoirs. Based on the deep learning standard model of oil-bearing properties of oil reservoirs, a deep learning algorithm is applied to extract the digital result model of oil-bearing properties of oil reservoirs, and the oil content and oil saturation are obtained as the descriptive parameters of oil-bearing properties of oil reservoirs. Obtaining judgment criteria for oil content and oil saturation, parameters characterizing oil content of the oil layer, and determining evaluation criteria for different types of oil layers based on the judgment criteria and the oil content and oil saturation of the oil layer; Determine the evaluation criteria of the oil layer based on the oil content and oil saturation of the different types of oil layers and construct digital three-dimensional models of the oil content of the different types of oil layers; Based on the digital 3D oil-bearing models of different types of oil layers and the evaluation criteria of different types of oil layers, the dominant target intervals of the oil layers are evaluated and determined; The method of establishing a three-dimensional digital model of an oil layer by using the three-dimensional grayscale image of the oil layer comprises: Using an image registration algorithm, the grayscale images of the oil layer obtained by field emission scanning electron microscopy and focused ion beam scanning electron microscopy are compared with the three-dimensional grayscale images of the oil layer obtained by CT scanning. By comparing the feature information in the images, the three-dimensional grayscale images of the oil layer obtained by CT scanning are calibrated and verified. Based on the comparison results, a standardized three-dimensional digital model of the oil content of the oil layer obtained by CT scanning is established. The method of creating a deep learning standard model of oil content of an oil layer based on the standardized three-dimensional digital model of oil content of the oil layer by deep learning of the three-dimensional grayscale image of the oil layer includes: Based on the standardized three-dimensional digital model of oil-bearing properties of oil reservoirs, a convolutional neural network is adopted as the deep learning model architecture. During the model training process, the three-dimensional grayscale images of oil reservoirs obtained by CT scanning are used as training data. The model is trained for multiple rounds, and the parameters of the model are continuously adjusted through the back-propagation algorithm to create a deep learning standard model of oil-bearing properties of oil reservoirs.

2. The deep learning quantitative characterization method for oil content of oil layers by CT scanning according to claim 1 is characterized in that: The method of obtaining a three-dimensional grayscale image of the oil layer by CT scanning of the oil layer includes: The oil layer digital intelligence model is applied to determine the oil layer CT scanning parameters, and the oil layer is CT scanned according to the CT scanning parameters to obtain a three-dimensional grayscale image of the oil layer.

3. The deep learning quantitative characterization method for oil content of oil layers by CT scanning according to claim 2 is characterized in that: The method for constructing the oil reservoir digital intelligence model is as follows: Based on information data of different geological conditions and different types of oil layers, this information data is preprocessed, feature extracted and analyzed, and trained using machine learning algorithms to construct a digital intelligence model of the oil layer.

4. The deep learning quantitative characterization method for oil content of oil layers using CT scanning according to claim 1 is characterized in that: The judgment criteria for obtaining the oil content and oil saturation of the oil layer characterization parameters include: Different types of oil layer samples are selected, and a variety of experimental methods are used to determine the true values ​​of the oil content and oil saturation of the samples. Combined with the geological characteristics and production data of the oil layers, a comprehensive analysis is conducted to determine the judgment criteria for the oil content and oil saturation of different types of oil layers.

5. The deep learning quantitative characterization method for oil content of oil layers by CT scanning according to claim 1 is characterized in that: The method of evaluating and preferentially determining the dominant target intervals of the oil layer based on the digital three-dimensional oil-bearing models of different types of oil layers and the evaluation criteria of different types of oil layers includes: Based on the digital 3D oil-bearing models of different types of oil layers and the evaluation criteria for different types of oil layers, a comprehensive evaluation method is used to comprehensively evaluate the oil layers, determine the weight of each factor, and screen out advantageous target layers by calculating the comprehensive evaluation score.

6. A deep learning quantitative characterization system for oil content in CT scan oil reservoirs, characterized by: A system for implementing the deep learning quantitative characterization method for oil content of CT scan oil layers according to any one of claims 1 to 5, comprising: The data acquisition module is used to apply the oil reservoir digital intelligence model to determine the oil reservoir CT scanning parameters, and perform CT scanning on the oil reservoir based on these parameters to obtain a three-dimensional grayscale image of the oil reservoir. At the same time, it collects field emission scanning electron microscope grayscale images, focused ion beam scanning electron microscope oil reservoir grayscale images, and various experimental data of different types of oil reservoir samples; The model building module is used to use an image registration algorithm to compare the grayscale images of field emission scanning electron microscopes and focused ion beam scanning electron microscope oil layer grayscale images with the three-dimensional grayscale images of the oil layer obtained by CT scanning, and to establish a standardized three-dimensional digital model of the oil content of the oil layer after calibration and verification; it is used to use the standardized three-dimensional digital model of the oil content of the oil layer as a basis, adopt a convolutional neural network architecture, use the three-dimensional grayscale images of the oil layer obtained by CT scanning to perform multiple rounds of training, and adjust parameters through the back propagation algorithm to create a deep learning standard model of the oil content of the oil layer; it is also used to construct digital three-dimensional models of the oil content of different types of oil layers based on the experimental data and analysis results of different types of oil layer samples, and determine the evaluation criteria for different types of oil layers; The parameter extraction module is used to extract the digital result model of oil content of the oil layer based on the deep learning standard model of oil content of the oil layer, and obtain the oil content description parameters such as oil content and oil saturation; The evaluation and decision-making module is used to conduct a comprehensive evaluation of the oil layers based on the digital three-dimensional models of oil content and evaluation standards of different types of oil layers, determine the weights of various factors, calculate the comprehensive evaluation scores, screen out the dominant target layers, and output the evaluation results and recommended dominant target layer information.

7. An electronic device comprising a memory, a processor, and a computer program, characterized in that: The computer program is stored in the memory and is configured to be executed by the processor to implement the deep learning quantitative characterization method for oil content of CT scan oil layers according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which is executed by a processor to implement the deep learning quantitative characterization method for oil content of CT scan oil layers according to any one of claims 1 to 5.

Images