A fracture fine interpretation method and product of a full-core CT enhanced electric imaging logging

Abstract

This application provides a method and product for fine interpretation of fractures in whole-core CT-enhanced electro-optical imaging logging. The method includes: acquiring a trained fracture identification model; training the fracture identification model based on a first electro-optical imaging image of the target wellbore and a CT scan image of the target core; the target core and the target wellbore belong to the same well section; inputting a second electro-optical imaging image of the target wellbore into the trained fracture identification model to obtain the fracture identification result of the target wellbore. This application utilizes a fracture identification model trained based on the first electro-optical imaging image of the target wellbore and the CT scan image of the target core to identify fractures in the second electro-optical imaging image of the target wellbore, thereby improving the accuracy of fracture identification in the wellbore.

Description

Technical Field

[0001] This application relates to the field of petroleum technology, specifically to a method and product for fine interpretation of fractures in whole-core CT-enhanced electro-imaging logging. Background Technology

[0002] In underground engineering fields such as oil and gas exploration and development, geothermal development, and carbon sequestration, accurate characterization of reservoir geological features is a core prerequisite for scientific decision-making. Natural fractures, as key channels for underground fluid migration, directly determine reservoir permeability, productivity, and wellbore stability through their state, density, and distribution.

[0003] Currently, the mainstream method for obtaining downhole fracture information is to scan the wellbore using wellbore electrical imaging logging technology to obtain images, and then manually interpret the images for fracture identification. However, manual interpretation is highly subjective and limited by logging resolution and interference from the complex downhole environment, making it easy to miss many small or weak feature fractures, resulting in poor accuracy in wellbore fracture identification. Summary of the Invention

[0004] The main objective of this application is to propose a method and product for fine interpretation of fractures in whole-core CT enhanced electrical imaging logging, aiming to improve the accuracy of identifying wellbore fractures.

[0005] This application provides a method for fine interpretation of fractures in whole-core CT-enhanced electro-optical logging, comprising: acquiring a trained fracture identification model; the fracture identification model being trained based on a first electro-optical imaging image of the target wellbore and a CT scan image of the target core; the target core and the target wellbore belonging to the same well section; and inputting a second electro-optical imaging image of the target wellbore into the trained fracture identification model to obtain the fracture identification result of the target wellbore.

[0006] In one embodiment, the CT scan image carries a pixel-level crack ground truth mask; obtaining the trained crack recognition model includes: performing image registration between the first electrical imaging image and the CT scan image to obtain a training dataset; the training dataset includes multiple training samples; each training sample includes the first electrical imaging image and the pixel-level crack ground truth mask of the first electrical imaging image; training the crack recognition model using the training dataset until the training stopping condition is met to obtain the trained crack recognition model.

[0007] In one embodiment, there are multiple first electrical imaging images and multiple CT scan images. The step of image registration between the first electrical imaging images and the CT scan images to obtain a training dataset includes: for each first electrical imaging image, based on the logging depth corresponding to the first electrical imaging image and the core sampling depth corresponding to each of the multiple CT scan images, determining a target CT scan image in the multiple CT scan images that matches the first electrical imaging image in depth; based on the matching feature point pairs of the first electrical imaging image and the target CT scan image in common geological features, determining the pixel coordinate transformation matrix of the first electrical imaging image and the target CT scan image; based on the pixel coordinate transformation matrix, mapping the pixel-level fracture ground truth mask of the target CT scan image to the first electrical imaging image to obtain the pixel-level fracture ground truth mask of the first electrical imaging image.

[0008] In one embodiment, training the crack recognition model using the training dataset until the training stopping condition is met to obtain a trained crack recognition model includes: for each training sample, performing the following steps: inputting the training sample into the crack recognition model to obtain a pixel-level crack prediction mask of the first electro-optical imaging image; determining the target loss function value of the crack recognition model based on the pixel-level crack prediction mask and the pixel-level crack ground truth mask of the first electro-optical imaging image; the target loss function value is a weighted sum of the binary cross-entropy loss function value and the Dessian loss function value; if the target loss function value does not meet the training stopping condition, adjusting the model parameters of the crack recognition model to obtain an updated crack recognition model, and training the updated crack recognition model using the next training sample until the training stopping condition is met to obtain a trained crack recognition model.

[0009] In one embodiment, the fracture identification model includes a backbone network with an encoder and a decoder; the encoder and the decoder perform feature fusion through skip connections; the step of inputting a second electrical imaging image of the target wellbore into the trained fracture identification model to obtain a fracture identification result of the target wellbore includes: inputting the second electrical imaging image into the trained fracture identification model, extracting features from the second electrical imaging image through the encoder to obtain a first image feature; fusing the first image feature with the second image feature through the skip connections to obtain a fused image feature; the second image feature is the image feature extracted by the decoder; and reconstructing features based on the fused image feature through the decoder to output the fracture identification result of the target wellbore.

[0010] In one embodiment, the step of inputting the second electrical imaging image of the target well wall into the trained fracture recognition model to obtain the fracture recognition result of the target well wall includes: inputting the second electrical imaging image of the target well wall into the trained fracture recognition model to obtain a pixel-level fracture mask of the second electrical imaging image; and determining the fracture recognition result based on the pixel-level fracture mask of the second electrical imaging image.

[0011] In one embodiment, determining the crack identification result based on the pixel-level crack mask of the second electrical imaging image includes: generating a pixel-level crack mask map of the target well wall based on the pixel-level crack mask of the second electrical imaging image; performing connected component analysis on the crack mask map to determine the independent crack regions in the crack mask map; and marking the location and counting the number of crack regions to obtain the crack identification result of the target well wall.

[0012] This application embodiment also provides a fracture fine interpretation device for whole-core CT-enhanced electro-optical imaging logging. The device includes an acquisition module and an identification module. The acquisition module is used to acquire a trained fracture identification model. The fracture identification model is trained based on a first electro-optical imaging image of the target wellbore and a CT scan image of the target core. The target core and the target wellbore belong to the same well section. The identification module is used to input a second electro-optical imaging image of the target wellbore into the trained fracture identification model to obtain the fracture identification result of the target wellbore.

[0013] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0014] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0015] This application provides a method and product for fine interpretation of fractures in whole-core CT-enhanced electro-imaging logging. By using a fracture identification model trained based on a first electro-imaging image of the target wellbore and a CT scan image of the target core, fracture identification is performed on a second electro-imaging image of the target wellbore, resulting in fracture identification results for the target wellbore. This method can improve the accuracy of identifying fractures in the wellbore. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the method for fine interpretation of fractures in whole-core CT-enhanced electrical imaging logging provided in this application.

[0017] Figure 2This is a schematic diagram of the architecture of the fracture fine interpretation system for whole-core CT enhanced electrical imaging logging provided in this application.

[0018] Figure 3 This is a schematic diagram of the training process for the wellbore crack recognition model provided in this application.

[0019] Figure 4 This is a schematic diagram of the fracture fine interpretation device for whole-core CT enhanced electrical imaging logging provided in this application.

[0020] Figure 5 This is a schematic diagram of the structure of the electronic device provided in this application. Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0022] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the digit " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0023] The fracture fine interpretation method for whole-core CT-enhanced electro-optical imaging logging provided in this application embodiment can be applied to electronic devices or the software of electronic devices. The electronic device can be a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, or as a server cluster or distributed system composed of multiple physical servers. The software can be an application that implements the fracture fine interpretation method for whole-core CT-enhanced electro-optical imaging logging, but is not limited to the above forms.

[0024] The following detailed description, in conjunction with the accompanying drawings, of the method for fine interpretation of fractures in whole-core CT-enhanced electrical imaging logging provided in this application, through specific embodiments, will be provided in detail.

[0025] Please see Figure 1The present application provides a method for fine interpretation of fractures in whole-core CT-enhanced electrical imaging logging, which may include:

[0026] Step S101: Obtain the trained fracture recognition model; the fracture recognition model is trained based on the first electrical imaging image of the target wellbore and the CT scan image of the target core; the target core and the target wellbore belong to the same well section; Optionally, the first electrical imaging image of the target wellbore can be an image obtained by scanning the wellbore using wellbore electrical imaging logging technology; the CT scan image of the target core can be an image obtained by scanning the core sample using computed tomography (CT) technology.

[0027] In practice, the fracture identification model can be trained using the first electrical imaging image of the target wellbore and the CT scan image of the target core to obtain the trained fracture identification model. The trained fracture identification model can be stored, and when it is necessary to use the trained fracture identification model to identify fractures in the wellbore, the trained fracture identification model can be directly called.

[0028] Step S102: Input the second electrical imaging image of the target well wall into the trained crack recognition model to obtain the crack recognition result of the target well wall.

[0029] Optionally, the second electrical imaging image of the target wellbore can be an image obtained by scanning the wellbore using wellbore electrical imaging logging technology; the second electrical imaging image of the target wellbore and the aforementioned first electrical imaging image can be the same electrical imaging image or different electrical imaging images.

[0030] Optionally, the crack recognition result can be presented as a pixel-level crack mask or as a recognition result including the location and number of cracks. For specific implementation details, please refer to the relevant description below, which will not be described here.

[0031] This application embodiment utilizes a crack recognition model trained based on a first electrical imaging image of the target well wall and a CT scan image of the target core to perform crack recognition on a second electrical imaging image of the target well wall, thereby obtaining crack recognition results for the target well wall and improving the accuracy of crack recognition in the well wall.

[0032] Optionally, the aforementioned CT scan image carries a pixel-level crack truth mask. Optionally, the pixel-level crack truth mask can be understood as a crack label for each pixel in the CT scan image, which can be used to characterize whether each pixel in the CT scan image belongs to a crack region. For example, if the mask corresponding to pixel A is 1, it can indicate that pixel A belongs to a crack region; if the mask corresponding to pixel B is 0, it can indicate that pixel B does not belong to a crack region.

[0033] In practice, due to the extremely high resolution of CT scan images, the CT scan image of the target rock core can be segmented based on a pre-set image grayscale threshold to obtain the pixel-level crack ground truth mask of the CT scan image of the target rock core.

[0034] In one embodiment, obtaining the trained crack recognition model in step S101 above includes: Image registration is performed between the first electrical imaging image and the CT scan image to obtain a training dataset; the training dataset includes multiple training samples; each training sample includes the first electrical imaging image and a pixel-level crack ground truth mask of the first electrical imaging image; The crack recognition model is trained using the training dataset until the training stopping condition is met, resulting in a fully trained crack recognition model.

[0035] In practical implementation, a transformation relationship between the pixel coordinates of the first electrical imaging image and the pixel coordinates of the CT scan image can be established based on the common geological features corresponding to the first electrical imaging image and the CT scan image. Then, this transformation relationship can be used to map the pixel-level crack ground truth mask of the CT scan image to the first electrical imaging image, thus obtaining the pixel-level crack ground truth mask of the first electrical imaging image. For specific implementation details, please refer to the relevant descriptions below, which will not be described here. Then, the crack recognition model can be trained using each training sample in the training dataset until the training stopping condition is met, resulting in a trained crack recognition model.

[0036] This application embodiment obtains a training dataset by performing image registration between a first electrical imaging image and a CT scan image; the training dataset includes multiple training samples; each training sample includes a first electrical imaging image and a pixel-level crack ground truth mask of the first electrical imaging image; and the crack recognition model is trained using the training dataset until the training stopping condition is met, resulting in a trained crack recognition model, which can improve the training effect of the crack recognition model, thereby ensuring the accuracy of identifying well wall cracks using the trained crack recognition model.

[0037] Optionally, there can be multiple first electrical imaging images and multiple CT scan images. Optionally, the multiple first electrical imaging images can be electrical imaging images of the wellbore at different logging depths in the well section; the multiple CT scan images can be CT scan images of core samples taken from different sampling depths in the well section.

[0038] In one embodiment, the above-described image registration of the first electrical imaging image and the CT scan image to obtain a training dataset includes: For each first electrical imaging image, based on the logging depth corresponding to the first electrical imaging image and the core sampling depth corresponding to each of the multiple CT scan images, a target CT scan image that matches the first electrical imaging image in depth is determined among the multiple CT scan images. Based on the matching feature point pairs of the first electrical imaging image and the target CT scan image on common geological features, the pixel coordinate transformation matrix of the first electrical imaging image and the target CT scan image is determined. Based on the pixel coordinate transformation matrix, the pixel-level crack ground truth mask of the target CT scan image is mapped onto the first electrical imaging image to obtain the pixel-level crack ground truth mask of the first electrical imaging image.

[0039] In practice, for each first electrical imaging image, a target CT scan image matching the core sampling depth and logging depth can be selected from multiple CT scan images based on its corresponding logging depth. Optionally, a target CT scan image matching the first electrical imaging image in depth can be determined based on the relationship between the deviation between the core sampling depth and the logging depth and a preset deviation range. The preset deviation range can be obtained through calibration. If the deviation between the core sampling depth and the logging depth is within the preset deviation range, it can be characterized that the target core corresponding to the core sampling depth and the target wellbore corresponding to the logging depth belong to the same formation layer.

[0040] Next, by comparing and analyzing the first electrical imaging image and the target CT scan image, common geological features (such as obvious stratification, large cracks, etc.) that are shared by both can be identified. Then, at least one pair of matching feature points (such as crack endpoints, crack inflection points, etc.) can be selected from the image regions corresponding to the common geological features in the first electrical imaging image and the target CT scan image, respectively, as registration anchor points. Then, an image registration algorithm can be used to calculate the pixel coordinate transformation matrix between the first electrical imaging image and the target CT scan image, with the relationship between the coordinates of the matching feature point pairs as a constraint. Finally, based on the pixel coordinate transformation matrix, the crack ground truth mask of each pixel in the target CT scan image can be accurately mapped to the corresponding pixel in the first electrical imaging image to obtain the pixel-level crack ground truth mask of the first electrical imaging image.

[0041] Optionally, the aforementioned pixel coordinate transformation matrix can characterize at least one of the transformation relationships among translation, rotation, and scaling between pixel coordinates. Optionally, the target CT scan image can be a three-dimensional image, and the first electrophysiological imaging image can be a two-dimensional image; before comparing and analyzing the first electrophysiological imaging image and the target CT scan image, the target CT scan image can be virtually unfolded to obtain a two-dimensional target CT scan image; then, the first electrophysiological imaging image and the two-dimensional target CT scan image can be compared and analyzed.

[0042] This embodiment of the application determines a target CT scan image that matches the first electrical imaging image in depth among multiple CT scan images by using the logging depth corresponding to the first electrical imaging image and the core sampling depth corresponding to the CT scan image. This achieves coarse-grained depth alignment between the first electrical imaging image and the CT scan image, ensuring that they macroscopically correspond to the same stratum. Then, based on the matching feature point pairs of the first electrical imaging image and the target CT scan image in common geological features, a pixel coordinate transformation matrix is ​​determined. This enables sub-centimeter-level precise spatial registration of pixels in the low-resolution first electrical imaging image and the high-resolution target CT scan image. Finally, using the pixel coordinate transformation matrix, the pixel-level fracture ground truth mask of the target CT scan image can be efficiently and accurately mapped to the corresponding pixels in the first electrical imaging image. This provides a high-confidence ground truth label for the fracture recognition model training, thereby improving the training effect of the fracture recognition model.

[0043] In one embodiment, the above-mentioned method of training the crack recognition model using a training dataset until the training stopping condition is met to obtain a trained crack recognition model includes: For each training sample, perform the following steps: The training samples are input into the crack recognition model to obtain the pixel-level crack prediction mask of the first electrical imaging image; Based on the pixel-level crack prediction mask and the pixel-level crack ground truth mask of the first electrical imaging image, the target loss function value of the crack recognition model is determined; the target loss function value is the weighted sum of the binary cross-entropy loss function value and the Dess loss function value. If the target loss function value does not meet the training stopping condition, adjust the model parameters of the crack recognition model to obtain an updated crack recognition model. Then, use the next training sample to train the updated crack recognition model until the training stopping condition is met, and obtain the trained crack recognition model.

[0044] In actual implementation, considering that the proportion of crack pixels in the whole image is extremely small, in order to avoid the problem of image pixel class imbalance caused by this, during the training of the crack recognition model using each training sample, a composite loss function consisting of the binary cross-entropy loss function and the Desce loss function can be used to constrain the training process of the crack recognition model until the weighted sum of the values ​​of the binary cross-entropy loss function and the Desce loss function satisfies the training stopping condition, and the trained crack recognition model is obtained.

[0045] This application embodiment uses a composite loss function consisting of a binary cross-entropy loss function and a Dessian loss function to constrain the training process of the fracture recognition model during the training of the model using each training sample. This effectively avoids the problem of image pixel class imbalance caused by the extremely small proportion of fracture pixels in the whole image, and can improve the training effect of the fracture recognition model, thereby ensuring the accuracy of identifying well wall fractures using the trained fracture recognition model.

[0046] Optionally, the crack recognition model may include a backbone network with an encoder and a decoder; the encoder and decoder may perform feature fusion through skip connections.

[0047] In one embodiment, step S102 above: inputting the second electrical imaging image of the target wellbore into the trained fracture recognition model to obtain the fracture recognition result of the target wellbore, includes: The second electrical imaging image is input into the trained crack recognition model, and the encoder extracts features from the second electrical imaging image to obtain the first image features. The first image feature is fused with the second image feature through a skip connection to obtain the fused image feature; the second image feature is the image feature extracted by the decoder. The decoder performs feature reconstruction based on the fused image features and outputs the crack identification results of the target well wall.

[0048] Optionally, the encoder may integrate deformable convolutional layers; the decoder may integrate multi-scale attention modules.

[0049] In practical implementation, the second electrical imaging image can be input into the trained crack recognition model. Multiple cascaded convolutional layers in the encoder can sequentially downsample to generate shallow first image features containing high-frequency details. Specifically, some standard convolutional layers in the encoder path are replaced or supplemented with deformable convolutional layers. These deformable convolutional layers can dynamically learn and adjust the sampling point positions of their convolutional kernels according to the specific geometry of the crack, allowing their receptive field to adaptively match the actual direction and width variations of the crack. This results in the extraction of first image features that better characterize complex crack morphologies such as meandering, discontinuous, and varying widths. Especially for meandering cracks, deformable convolutional layers can more effectively capture their complete geometric shape, significantly improving the model's segmentation accuracy for complex crack shapes.

[0050] Each convolutional layer in the decoder can be upsampled sequentially to generate deep second image features containing rich semantics. Then, the model can use skip connections to fuse the first image features containing fine spatial details from the corresponding level of the encoder with the second image features from the current level of the decoder to obtain fused image features.

[0051] During feature fusion, the model can utilize a multi-scale attention module to focus on features more relevant to the current task while suppressing irrelevant or interfering features. For fracture identification, this means the model can automatically focus on key fracture-related features (such as weak fracture responses and typical fracture-formation combinations) while suppressing common interference information in wellbore images, such as noise and bedding boundaries, ultimately outputting a more informative and targeted fused image. This significantly enhances the model's ability to distinguish between ambiguous fracture features and detect minute fractures.

[0052] Next, the model can input the fused image features optimized by the multi-scale attention module into the subsequent upsampling and convolutional layers of the decoder. Based on these optimized fused image features, the decoder performs feature reconstruction step by step to obtain a pixel-level crack prediction mask for the second electrical imaging image. Finally, based on the pixel-level crack prediction mask, the crack identification result of the target well wall can be obtained. For details on the implementation, please refer to the relevant description below, which will not be described here.

[0053] This embodiment of the application improves the accuracy of identifying well wall cracks by inputting a second electrical imaging image into a crack identification model, extracting features from the second electrical imaging image using an encoder to obtain a first image feature, fusing the first image feature with the second image feature through a skip connection to obtain a fused image feature, where the second image feature is the image feature extracted by the decoder, and outputting the crack identification result of the target well wall by performing feature reconstruction based on the fused image feature using the decoder.

[0054] In one embodiment, step S102 above: inputting the second electrical imaging image of the target wellbore into the trained fracture recognition model to obtain the fracture recognition result of the target wellbore, includes: The second electrical imaging image of the target well wall is input into the trained crack recognition model to obtain the pixel-level crack mask of the second electrical imaging image; The crack identification result is determined based on the pixel-level crack mask of the second electrical imaging image.

[0055] Optionally, the pixel-level crack mask of the second electrical imaging image can be represented as a group of associated data of pixel coordinates and crack labels in the second electrical imaging image; or it can be a binary label image of the same size as the second electrical imaging image and labeled pixel by pixel, where white pixels represent cracks and black pixels represent the background, which can be simply referred to as a pixel-level crack mask image.

[0056] In practice, the pixel-level crack mask can be directly used as the crack identification result; alternatively, the crack location and number determined based on the pixel-level crack mask can be used as the crack identification result for the target wellbore. When the pixel-level crack mask is represented as a binary label image, the crack location can be directly marked on the pixel-level crack mask image and the crack number can be counted to obtain the crack identification result for the target wellbore.

[0057] This embodiment of the application improves the accuracy of identifying well wall cracks by inputting a second electrical imaging image of the target well wall into a trained crack recognition model to obtain a pixel-level crack mask of the second electrical imaging image; and by determining the crack recognition result based on the pixel-level crack mask of the second electrical imaging image.

[0058] In one embodiment, determining the crack identification result based on the pixel-level crack mask of the second electrical imaging image includes: A pixel-level crack mask map of the target well wall is generated based on the pixel-level crack mask of the second electrical imaging image. Perform connected component analysis on the crack mask map to identify independent crack regions in the crack mask map; The location and number of fracture areas are marked to obtain the fracture identification results of the target well wall.

[0059] In practical implementation, a pixel-level crack mask map can be generated based on the pixel-level crack mask of the second electrical imaging image, with the same size as the second electrical imaging image and labeled pixel by pixel, thus obtaining the pixel-level crack mask map. Then, connected component analysis can be performed on the pixel-level crack mask map to assign a unique identifier to each unconnected independent crack region, thereby segmenting individual cracks from the pixel-level crack mask map.

[0060] Then, the location of each independent fracture region with a unique identifier in the well depth direction can be marked, and the total number of independent fracture regions with unique identifiers can be counted to obtain the fracture identification result of the target well wall.

[0061] It is worth mentioning that before marking and counting the number of fracture regions, the independent fracture regions obtained through connected component analysis can be filtered based on their size to remove those that are too small and may be false fracture regions caused by tiny artifacts or noise from the model output. Then, the remaining fracture regions after filtering can be marked and counted to obtain the fracture identification results for the target wellbore.

[0062] This application embodiment generates a pixel-level crack mask map of the target well wall based on a pixel-level crack mask of a second electrical imaging image; performs connected component analysis on the crack mask map to determine independent crack regions in the crack mask map; and performs position marking and quantity statistics on the crack regions to obtain the crack identification result of the target well wall, which can improve the accuracy of crack identification in the well wall.

[0063] Optionally, the fracture identification model described above can be a DMSA-Net model using a U-Net network as its backbone network. In a specific embodiment, the fracture fine interpretation method for whole-core CT-enhanced electrical imaging logging provided in this application can be applied to, for example... Figure 2 The illustrated system is a fracture detail interpretation system 100 for whole-core CT-enhanced electrophysiological logging. This system 100 may include a general-purpose computing device 110, such as a high-performance workstation or cloud server.

[0064] The general-purpose computing device 110 may have at least one built-in hardware processor 112 (such as CPU, GPU, etc.) and data storage device 114 (such as memory, hard disk, etc.). The data storage device 114 may store a database 115 of raw wellbore electro-imaging images and core CT scan images, as well as a core software module 116 for realizing the wellbore fracture identification function.

[0065] The core software module 116 may include: a data preprocessing and alignment module 117, which can be used to create a training dataset for the fracture identification model; a DMSA-Net model 119 for identifying wellbore fractures; and a fracture marking and counting module 120 for marking fracture locations and counting fracture numbers from the fracture identification results.

[0066] In actual implementation, the hardware processor 112 can input the image data from the wellbore electrical imaging image and the core CT scan image database 115 into the data preprocessing and alignment module 117 to obtain a training dataset for training the DMSA-Net model 119. Then, the hardware processor 112 can use the training samples in the training dataset to train the DMSA-Net model 119, resulting in a trained DMSA-Net model 119. Next, in response to receiving a new wellbore electrical imaging image uploaded to the system 100 by a logging service company or user via a network interface, the hardware processor 112 can input the new wellbore electrical imaging image into the trained DMSA-Net model 119 to obtain a pixel-level fracture prediction mask for the image. Then, the hardware processor 112 can input the pixel-level fracture prediction mask into the fracture marking and counting module 120 to obtain a structured fracture identification report including fracture location and quantity. Finally, the hardware processor 112 can output the structured fracture identification report via the network interface.

[0067] It is worth mentioning that, when the size of the new wellbore electrical imaging image received is inconsistent with that of the electrical imaging image in the training dataset of DMSA-Net model 119, the hardware processor 112 can call the data preprocessing and alignment module 117 to normalize the new wellbore electrical imaging image before inputting the new wellbore electrical imaging image into the trained DMSA-Net model 119, so as to keep the image size consistent.

[0068] Optionally, the aforementioned electrical imaging images can be images obtained by scanning the target wellbore using a Formation MicroImager (FMI) (also referred to as FMI images). Please see [link to relevant documentation]. Figure 3 The method for fine interpretation of fractures in whole-core CT-enhanced electrical imaging logging provided in this application embodiment may further include the following steps: Acquire FMI images of the target wellbore and CT scan images of the core sample from the same well section; The FMI images of the target wellbore and the CT scan images of the core were sequentially subjected to coarse-grained depth alignment, precise spatial registration, and CT fracture ground truth label mapping to obtain high-fidelity training samples for the DMSA-Net model: FMI images and their corresponding CT fracture ground truth labels. The DMSA-Net model is trained using training samples, and the trained DMSA-Net model is obtained and stored.

[0069] Optionally, the CT crack truth label can be understood as a pixel-level crack truth mask of the CT scan image. The specific implementation process of this embodiment can be found in the description of the above embodiments, and will not be repeated here.

[0070] In practical implementation, the FMI images in the training samples can be segmented into multiple image patches, and the DMSA-Net model can be trained using these multiple image patches to improve the training effect of the DMSA-Net model. In this case, the hardware processor 112 can respond to receiving a new wellbore electrical imaging image by calling the data preprocessing and alignment module 117 to segment the new wellbore electrical imaging image into multiple image patches, and then input the multiple image patches into the trained DMSA-Net model 119 to obtain the pixel-level fracture prediction mask of each image patch; finally, the hardware processor 112 can stitch together the pixel-level fracture prediction masks of each image patch to obtain the pixel-level fracture prediction mask of the entire image.

[0071] The following section, using specific data, explains the technical effectiveness of the fracture fine interpretation method for whole-core CT-enhanced electrical imaging logging provided in this application: Method A: Select a physical core sample from a specific section (e.g., 1 meter deep) of an exploratory well and perform a high-resolution micron-scale CT scan to obtain a three-dimensional CT image of the core sample. Then, by finely segmenting and analyzing the three-dimensional CT image, the absolute number of fractures in that section can be determined.

[0072] Method B: Using the fracture fine interpretation method of whole core CT enhanced electrical imaging logging provided in the embodiments of this application, the FMI image of the well section is input into the trained DMSA-Net model to obtain the number of fractures in the well section.

[0073] Method C: Invite an experienced geological expert to perform routine manual interpretation of the FMI images of the well section, manually pick out and count the number of fractures that the expert believes to exist.

[0074] The number of cracks obtained by methods A, B, and C is shown in the table below: Evaluation indicators Method A Method B Method C Number of cracks 43 articles 24 articles 5 items Comparing the number of fractures in the table above, it is clear that Method C (conventional manual interpretation) identified a very small number of fractures. Compared to the 43 fractures identified using Method A (core CT scan image segmentation), Method C only identified 5, resulting in an identification rate of only 11.6%. This is primarily because the resolution of electrical imaging images is significantly lower than that of CT scan images, and their image quality is easily affected by the downhole environment, leading to a large number of weak, real fractures that cannot be effectively identified by the human eye. Furthermore, the results largely depend on the interpreter's subjective judgment and experience, making it difficult to meet the high-precision requirements for wellbore fracture identification.

[0075] Method B (the fine fracture interpretation method for whole-core CT-enhanced electro-optical imaging logging provided in this application) demonstrated an overwhelming advantage, successfully identifying 24 fractures, nearly five times the number identified manually, and matching the number of fractures identified using Method A with a accuracy exceeding 55%. This strongly proves that the fine fracture interpretation method for whole-core CT-enhanced electro-optical imaging logging provided in this application can effectively identify a large number of more subtle fracture signals that are overlooked by manual analysis, and the generated fracture identification results are significantly improved in both quantity and reliability.

[0076] Please see Figure 4 This application embodiment also provides a fracture fine interpretation device 400 for whole core CT enhanced electro-imaging logging, which may include an acquisition module 401 and an identification module 402.

[0077] The acquisition module 401 can be used to acquire the trained fracture recognition model; the fracture recognition model is trained based on the first electrical imaging image of the target well wall and the CT scan image of the target core; the target core and the target well wall belong to the same well section; The recognition module 402 can be used to input the second electrical imaging image of the target well wall into the trained crack recognition model to obtain the crack recognition result of the target well wall.

[0078] The fracture fine interpretation device for whole-core CT enhanced electrical imaging logging provided in this application embodiment can realize all the steps of the fracture fine interpretation method embodiment of the above whole-core CT enhanced electrical imaging logging, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0079] This application also provides an electronic device, including a processor and a memory. The memory stores a program or instructions that can be executed on the processor. When the program or instructions are executed by the processor, they implement the various steps of the above-described embodiment of the method for fine interpretation of fractures in whole-core CT enhanced electro-imaging logging, and can achieve the same technical effect. To avoid repetition, they will not be described again here.

[0080] Figure 5 To illustrate the hardware structure of the electronic device according to the embodiments of this application, the electronic device includes: The processor 501 can be implemented using a general-purpose central processing unit (CPU), microprocessor, application specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 502 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 502 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 502 and called by the processor 501 to execute the fracture fine interpretation method of whole-core CT enhanced electrical imaging logging according to the embodiments of this application. The input / output interface 503 is used to implement information input and output; The communication interface 504 is used to enable communication and interaction between this electronic device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 505 transmits information between various components of an electronic device (e.g., processor 501, memory 502, input / output interface 503, and communication interface 504); The processor 501, memory 502, input / output interface 503, and communication interface 504 are connected to each other within the electronic device via bus 505.

[0081] The electronic device provided in this application embodiment can realize each step of the above-described embodiment of the method for fine interpretation of fractures by whole-core CT enhanced electrical imaging logging, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0082] This application also provides a computer-readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various steps of the above-described embodiment of the method for fine interpretation of fractures using whole-core CT enhanced electro-imaging logging, and achieve the same technical effect. To avoid repetition, these steps will not be repeated here.

[0083] The processor is the processor in the electronic device described in the above embodiments. The computer-readable storage medium includes computer-readable storage media such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0084] This application also provides a chip, which includes a processor and a communication interface. The communication interface and the processor are coupled. The processor is used to execute programs or instructions to implement the various steps of the above-described embodiment of the method for fine interpretation of fractures in whole-core CT enhanced electrical imaging logging, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0085] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0086] This application provides a computer program product stored in a computer-readable storage medium. When executed by a processor, the program product implements the various steps of the above-described embodiment of the method for fine interpretation of fractures using whole-core CT enhanced electrical imaging logging, and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0087] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not delete other identical elements present in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0088] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods of the various embodiments of this application.

[0089] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A method for fine interpretation of fractures in whole-core CT-enhanced electrical imaging logging, characterized in that, include: Obtain the trained crack recognition model; The fracture identification model is trained based on the first electrical imaging image of the target wellbore and the CT scan image of the target core. The target core and the target wellbore belong to the same well section; The second electrical imaging image of the target well wall is input into the trained crack recognition model to obtain the crack recognition result of the target well wall.

2. The method as described in claim 1, characterized in that, The CT scan image carries a pixel-level crack truth mask; The process of obtaining the trained crack recognition model includes: Image registration is performed between the first electrical imaging image and the CT scan image to obtain a training dataset; the training dataset includes multiple training samples; each training sample includes the first electrical imaging image and a pixel-level crack ground truth mask of the first electrical imaging image. The crack recognition model is trained using the training dataset until the training stopping condition is met, resulting in a fully trained crack recognition model.

3. The method as described in claim 2, characterized in that, The number of the first electro-imaging image and the CT scan image are both multiple; The step of image registration between the first electrical imaging image and the CT scan image to obtain a training dataset includes: For each of the first electrical imaging images, based on the logging depth corresponding to the first electrical imaging image and the core sampling depth corresponding to the plurality of CT scan images, a target CT scan image that matches the first electrical imaging image in depth is determined among the plurality of CT scan images. Based on the matching feature point pairs of the first electro-optical imaging image and the target CT scan image on common geological features, the pixel coordinate transformation matrix of the first electro-optical imaging image and the target CT scan image is determined; Based on the pixel coordinate transformation matrix, the pixel-level crack ground truth mask of the target CT scan image is mapped onto the first electrical imaging image to obtain the pixel-level crack ground truth mask of the first electrical imaging image.

4. The method as described in claim 2, characterized in that, The step of training the crack recognition model using the training dataset until the training stopping condition is met, to obtain the trained crack recognition model, includes: For each training sample, perform the following steps: The training samples are input into the crack recognition model to obtain the pixel-level crack prediction mask of the first electrical imaging image; Based on the pixel-level crack prediction mask and the pixel-level crack ground truth mask of the first electrical imaging image, the target loss function value of the crack recognition model is determined; the target loss function value is a weighted sum of the binary cross-entropy loss function value and the Dessian loss function value. If the target loss function value does not meet the training stopping condition, the model parameters of the crack recognition model are adjusted to obtain an updated crack recognition model. The updated crack recognition model is then trained using the next training sample until the training stopping condition is met, resulting in a fully trained crack recognition model.

5. The method as described in claim 1, characterized in that, The crack recognition model includes a backbone network with an encoder and a decoder; the encoder and the decoder perform feature fusion through skip connections; The step of inputting the second electrical imaging image of the target wellbore into the trained fracture recognition model to obtain the fracture recognition result of the target wellbore includes: The second electrical imaging image is input into the trained crack recognition model, and the encoder is used to extract features from the second electrical imaging image to obtain the first image features; The first image feature is fused with the second image feature through the skip connection to obtain the fused image feature; the second image feature is the image feature extracted by the decoder. The decoder performs feature reconstruction based on the fused image features and outputs the crack identification result of the target well wall.

6. The method as described in claim 1, characterized in that, The step of inputting the second electrical imaging image of the target wellbore into the trained fracture recognition model to obtain the fracture recognition result of the target wellbore includes: The second electrical imaging image of the target wellbore is input into the trained fracture recognition model to obtain the pixel-level fracture mask of the second electrical imaging image; The crack identification result is determined based on the pixel-level crack mask of the second electro-imaging image.

7. The method as described in claim 6, characterized in that, The determination of the crack identification result based on the pixel-level crack mask of the second electrical imaging image includes: A pixel-level crack mask map of the target well wall is generated based on the pixel-level crack mask of the second electrical imaging image. Perform connected component analysis on the crack mask map to determine the independent crack regions in the crack mask map; The location of the fractured areas is marked and the number of fractures is counted to obtain the fracture identification results of the target well wall.

8. A device for fine interpretation of fractures in whole-core CT-enhanced electrical imaging logging, characterized in that, The device includes an acquisition module and an identification module; The acquisition module is used to acquire the trained fracture recognition model; the fracture recognition model is trained based on the first electrical imaging image of the target wellbore and the CT scan image of the target core; the target core and the target wellbore belong to the same well section; The recognition module is used to input the second electrical imaging image of the target well wall into the trained crack recognition model to obtain the crack recognition result of the target well wall.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 7.

Images