A method for constructing a lithology prediction model and automatically drawing a fault geological profile

Abstract

An embodiment of the present invention discloses a method for constructing a lithology prediction model and automatically drawing a fault geological section. Among them, the model construction method includes: obtaining a plurality of original images of the fault geological section and the manual block-level lithology annotation of each original image; extracting geological boundary curves from each original image through edge detection; using each original image as a sample to train an image segmentation model based on deep learning, and the trained model is used to predict the pixel-level lithology distribution of each original image; during the training process, by constraining the lithology prediction results of each pixel in the closed area divided by the geological boundary curve in the original image to tend to be consistent, the lithology prediction results of adjacent closed areas are different, and the number of pixels with inconsistent pixel-level lithology prediction results and block-level lithology annotation is the least, to update the model parameters. This embodiment is based on actual picture data to predict lithology attributes and draw geological sections, and the drawing results are more accurate.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of geological image processing, and in particular to a method for constructing a lithology prediction model and automatically drawing a fault geological profile. Background Art

[0002] A fault is a structure in which the crust is stressed and fractured, and significant relative displacement occurs between rock blocks on both sides of the fracture surface. A geological profile is an actual (or inferred) section showing the geological structure on the surface or at a certain depth along a certain direction. A fault geological profile refers to a geological profile including a fault structure, which can intuitively reveal the geometric characteristics, displacement mode and geological structure background of the fault. Therefore, drawing an accurate fault geological profile is very important for fault and geological structure analysis.

[0003] In the prior art, Patent CN202211729160.0 discloses a method for automatically filling and drawing a geological cross-section, Patent CN202311583117.2 discloses an improved method and system for accurately identifying formation regions in a geological profile, Patent CN202110212860.1 discloses a method for automatically processing the spatial morphology of faults and formations in a geological profile, and Patent CN201611103016.0 discloses a method for dividing a natural grid of a geological profile.

[0004] The methods of the prior art have the following technical problems:

[0005] 1. It can only be based on the data of exploration holes obtained from actual collection (including but not limited to excavation exploration, percussion exploration, drilling, static cone penetration test, rotary cone penetration test, dynamic cone penetration test and engineering geophysical exploration holes), and cannot be based on actual picture data;

[0006] 2. It is only limited to ordinary exploration profiles, and the formation boundary lines obtained by drawing are mostly horizontal, which is quite different from the actual situation;

[0007] 3. It cannot draw the development situation of faults, including fault development characteristics (including fault occurrence, etc.). Summary of the Invention

[0008] The embodiments of the present invention provide a method for constructing a lithology prediction model and automatically drawing a fault geological profile to solve at least one of the above problems.

[0009] In a first aspect, the embodiments of the present invention provide a method for constructing a lithology prediction model of a fault geological profile, including:

[0010] Obtaining a plurality of original images of the fault geological profile, and manual block-level lithology annotations for each original image;

[0011] Extracting geological boundary curves from each original image through edge detection;

[0012] Using each original image as a sample to train a deep learning-based image segmentation model, and the trained model is used to predict the pixel-level lithology distribution of each original image;

[0013] During the training process, by constraining the lithology prediction results of each pixel within the closed area divided by the geological boundary curve in the original image to tend to be consistent, the lithology prediction results of adjacent closed areas are different, and the number of pixels where the pixel-level lithology prediction results are inconsistent with the block-level lithology annotation is minimized to update the model parameters.

[0014] In a second aspect, an embodiment of the present invention provides a method for automatically drawing a fault geological profile, including:

[0015] Obtaining the original image of the fault geological profile to be drawn;

[0016] Inputting the original image into the image segmentation model described in the above embodiment to obtain the pixel-level lithology distribution of the original image;

[0017] Overlaying the boundary line and lithology attributes in the pixel-level lithology distribution on the original image to obtain the final geological profile diagram.

[0018] In a third aspect, an embodiment of the present invention further provides an electronic device, and the electronic device includes:

[0019] One or more processors;

[0020] A memory for storing one or more programs,

[0021] When the one or more programs are executed by the one or more processors, the one or more processors implement the method for constructing a lithology prediction model for a fault geological profile or the method for automatically drawing a fault geological profile described in any embodiment.

[0022] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the method for constructing a lithology prediction model for a fault geological profile or the method for automatically drawing a fault geological profile described in any embodiment.

[0023] An embodiment of the present invention provides a method for constructing a lithology prediction model of a fault geological profile and automatic drawing. Based on data such as actual photographed surface pictures, a geological profile is drawn by computer recognition. The fault development situation and stratigraphic boundary can be obtained in the figure, reflecting the fault development characteristics (including occurrence characteristics such as fault strike and dip). Specifically, this embodiment performs AI deep learning based on the established fault lithology database, outputs the pixel-level lithology distribution of the original image through the fault geological profile lithology prediction model, and determines the block boundaries in the profile based on this lithology distribution, including fault line information; performs stratigraphic pattern filling according to the lithology distribution and block boundaries, and inputs stratigraphic and fault occurrence information for annotation. The drawing result of the fault geological profile by this method is accurate and concise, and can realize automatic filling and drawing of the fault geological profile, greatly improving the efficiency and accuracy of drawing the fault geological profile.

[0024] Particularly, this embodiment takes manual annotation and geological boundary curves as two bases for lithology division respectively, and obtains two lithology block division results respectively. It is found in practical applications that there are certain errors in both lithology block divisions. In some areas, the two division results need to be fused so that the regional boundaries complement each other to obtain a better block division. Therefore, in the model training process of this embodiment, by constraining the lithology prediction results of each pixel in the closed area divided by the geological boundary curve in the original image to tend to be consistent, the lithology prediction results of adjacent closed areas are different, and the number of pixels where the pixel-level lithology prediction result is inconsistent with the block-level lithology annotation is the least, the two division results are mutually compensated, and the manual block-level lithology annotation information is fully utilized to generate a better lithology block division and recognition result. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0026] Figure 1 is a flowchart of a method for constructing a lithology prediction model of a fault geological profile provided by an embodiment of the present invention;

[0027] Figure 2 is a schematic diagram of the original photographed image and the geological profile provided by an embodiment of the present invention. Among them, Figure 2 (a) is a field photographed image of the fault geological profile, Figure 2 (b) is the drawn fault geological profile;

[0028] Figure 3It is a schematic diagram of lithologic block division for the input and output of the lithologic prediction model provided by the embodiment of the present invention;

[0029] Figure 4 It is a schematic diagram of lithologic block division for the input and output of the lithologic prediction model provided by the embodiment of the present invention;

[0030] Figure 5 It is a schematic diagram of the closed block on both sides of the fault provided by the embodiment of the present invention;

[0031] Figure 6 It is a flowchart of a method for automatically drawing a fault geological profile provided by the embodiment of the present invention Figure 7 It is a schematic structural diagram of an electronic device provided by the embodiment of the present invention. Detailed implementation manners

[0032] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below. Apparently, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope protected by the present invention.

[0033] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be construed as a limitation of the present invention. In addition, the terms "first", "second", and "third" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0034] In the description of the present invention, it should also be noted that unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the internal communication of two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0035] Figure 1 It is a flowchart of a method for constructing a lithologic prediction model for a fault geological profile provided by the embodiment of the present invention. This method is executed by an electronic device, as Figure 1 shown, and specifically includes the following steps:

[0036] S110. Obtain a plurality of original images of the fault geological section, and manual block-level lithology annotations for each original image.

[0037] First, take pictures, splice, and produce the fault geological section diagram. In a specific embodiment, the excavated or original fault geological section can be photographed, and the photographing angles can be direct view, upward view, or downward view, and the lighting meets the requirements (the best is cloudy day); then, number and sort the taken pictures, and the photographing accuracy meets: the overlap rate of each photo is 70%-80%. Finally, perform overall splicing according to the picture numbers obtained by photographing to obtain the overall picture of the entire fault geological section, as Figure 2 (a) shown.

[0038] Based on the above overall picture, a fault lithology database can be established. The requirements for the original data include:

[0039] (1) The depth model will use simple tasks and transfer learning techniques, with the lowest data volume requirement (500 - 1000 geological images);

[0040] (2) Original image size: One geological image has a resolution of 1920×1080, in RGB color format, and each image is about 5 - 6MB;

[0041] (3) Use data augmentation techniques to greatly expand the number of effective samples: Common methods include rotation (±10° - 30°), horizontal / vertical flipping, random cropping, lighting adjustment (brightness, contrast, saturation, etc.), and random noise, ensuring that the minimum number of geological images is not less than 500;

[0042] For the above original data, the following preprocessing can be carried out:

[0043] (1) Through manual experience, perform block-level lithology annotation on each original image, including manual division of different lithology blocks in the original image and manual annotation of the lithology of each block. It should be noted that it is not necessary to annotate each pixel, only each block needs to be annotated; after annotation, the lithology attribute annotation values at all pixel positions within the block are the lithology attribute annotation values of the block;

[0044] (2) Import the above image data into the library, and store the images and annotations in different directories respectively for subsequent training and testing;

[0045] (3) Uniformly resize the images and labels to a resolution of 256×256 for easy input into the deep learning model in subsequent operations;

[0046] (4) Normalize the images to the [0, 1] interval for easy optimization of the deep learning model;

[0047] (5) The label remains an integer type and is used to achieve pixel-level classification in subsequent classification tasks.

[0048] After the preprocessing is completed, the dataset is divided based on the processed data, specifically including:

[0049] (1) Divide the dataset into a training set (for training the model), a validation set (for optimizing hyperparameters), and a test set (for evaluating the model performance);

[0050] (2) The default division ratio is: training set:validation set:test set = 64:16:20.

[0051] After the division, visual verification can be performed, such as randomly viewing an image and its corresponding label to ensure that the data loading and processing are correct.

[0052] Finally, data organization is carried out. The processed data is stored in a specified path, a data dictionary is created for easy loading during subsequent training, and it is saved as a JSON file to record the metadata of the dataset division, which is convenient for reading in the training script.

[0053] S120. Extract the geological boundary curves from each original image through edge detection.

[0054] Load the original images that have undergone preprocessing, perform image smoothing and edge detection on the images, extract the edge point coordinates, and fit the main geological boundaries (using polynomial fitting to calculate the x and y coordinates corresponding to the fitted curve). The fitted geological boundaries are used to provide another basis for the division of lithologic blocks.

[0055] In a specific embodiment, first, load the original images, grayscale them and perform Gaussian smoothing to eliminate noise, and use the Canny algorithm to perform edge detection on each original image to obtain the edge points of each original image; then, for the same original image, use the polyfit function to perform quadratic fitting on the edge points corresponding to the same geological boundary in the image to obtain the main geological boundary curve. Combining Figure 2 (b), the main geological boundary curve includes the stratigraphic boundary line that spreads horizontally, and the fault line; multiple geological interface curves and the image boundary divide the entire original image into multiple closed regions, and these closed regions often correspond to different lithologic blocks.

[0056] S130. Train a deep learning-based image segmentation model with each original image as a sample, and the trained model is used to predict the pixel-level lithologic distribution of each original image.

[0057] After edge detection, based on the dataset constructed in S110 and the edge detection results in S120, a deep learning-based image segmentation model is trained. The trained model takes the original image of the fault geological section as the input and the lithology attributes at each pixel position in the original image as the output. Among them, each lithology attribute value is arranged according to the pixel position, jointly constituting the pixel-level lithology distribution map of the original image, and this distribution map is consistent with the size of the input image of the model.

[0058] Optionally, the image segmentation model can adopt the U-Net model, and its specific structure is prior art. Generally, it includes a sampling path (encoder) and an upsampling path (decoder). The input of the U model is an RGB image, and the downsampling path, the bottom bottleneck layer, the upsampling path, and the output layer can be implemented through code. In the model, features are extracted through convolutional layers, max pooling is used for downsampling, and upsampling is used to restore the resolution; and Concatenate is used to fuse features to improve the segmentation accuracy.

[0059] In this embodiment, the U-Net model is trained using the dataset constructed in S110. During the training process, the model parameters can be updated by constraining the lithology prediction results of the pixels in the closed area divided by the geological boundary curve in the original image to tend to be consistent, the lithology prediction results of adjacent closed areas to be different, and the number of pixels where the pixel-level lithology prediction result is inconsistent with the block-level lithology annotation to be the least.

[0060] Furthermore, the following loss function can be constructed:

[0061]

[0062] Among them, k represents the index of the closed area divided by the geological boundary curve in the original image, N represents the total number of closed areas divided by the geological boundary curve in the original image; i represents the pixel index in the closed area with index k, n k represents the total number of pixels in the closed area with index k; μ k represents the mean value of the lithology prediction results of all pixels in the closed area with index k, I p,k,i represents the value of the lithology prediction result of the pixel with index i in the closed area with index k; U(k) represents the set of indices of adjacent closed areas of the closed area with index k, j represents the index of the closed area in the set U(k), μ j represents the mean value of the lithology prediction results of all pixels in the closed area with index j; δ(x) represents the impulse function, when the independent variable x = 0, δ(x) is equal to a preset positive constant, when the independent variable x ≠ 0, δ(x) = 0; num(I b -I p ) represents the pixel-level lithology prediction value I p and the block-level lithology annotation value Ib The number of inconsistent pixels, where α, β, and γ are weight coefficients respectively.

[0063] Specifically, in this embodiment, manual annotation and geological boundary curves in the sample set are respectively used as two bases for lithology division, and two lithology block division results are obtained respectively. For the convenience of distinction and description, the lithology blocks obtained by manual annotation can be called the first lithology blocks, and the lithology blocks obtained by dividing the geological boundary curves can be called the second lithology blocks. It is found in practical applications that there are certain errors in both lithology block divisions. In some areas, the two division results need to be fused so that the regional boundaries can complement each other to obtain a better block division.

[0064] Therefore, this embodiment constructs the above loss function and updates the model parameters by minimizing the loss function. Among them, The minimization of can ensure that the lithology prediction results of each pixel in the closed area divided by the geological boundary curve in the original image tend to be consistent; The minimization of can ensure that the lithology prediction results between adjacent closed areas (indexed as k and j) divided by the geological boundary curve in the original image are different; num(I b -I p ) The minimization of can ensure that the number of pixels with inconsistent pixel-level lithology prediction values and block-level lithology annotation values is minimized, so as to take into account the division results and annotation information of the first lithology blocks on the basis of the second lithology blocks. The three loss function terms together achieve the mutual compensation of the two division results, and make full use of the annotation information of the first lithology blocks to generate better lithology block division and recognition results.

[0065] Exemplarily, if there is a situation where a certain second lithology block in the sample data includes two lithology annotations, such as Figure 3 shown, the black line divides the first lithology block, the red line divides the second lithology block, the lithology attributes of the two first lithology blocks are respectively labeled as "Lithology 1" and "Lithology 2", and the red second lithology block includes both lithology annotations at the same time. In this case, the above loss function can automatically select an optimal lithology attribute (i.e., Lithology 2) for the second lithology block, and adjust the boundary of the second lithology block, taking into account the manual division information. The final division result is as shown in the blue line area in Figure 3 .

[0066] Exemplarily, since the second lithology block is the result obtained by using an image algorithm and is usually more detailed than the manual division result, there may also be a situation in the sample data where a certain second lithology block is ignored in the manual division result. Such as Figure 4As shown, the first lithology blocks are still represented by the black line areas, and the second lithology blocks are represented by the red line areas. There is a smaller second lithology block (Block 3) sandwiched between two larger second lithology blocks (Block 1 and Block 2), and this Block 3 is ignored in the manually divided result shown by the black lines. At this time, the above loss function can enable the two larger second lithology blocks (Block 1 and Block 2) to respectively select the optimal lithology attributes (Lithology 1 and Lithology 2), while retaining the smaller second lithology block (i.e., Block 3, whose Lithology 3 is different from both Lithology 1 and Lithology 2), so that the smaller second lithology block will not be submerged in the final segmentation result, as Figure 4 shown by the blue line area in

[0067] Furthermore, in the actual geological structure, there is a situation where the same lithology is allowed in adjacent closed areas, that is, there is a part of the adjacent boundary along the fault direction on both sides of the fault, as Figure 5 shown. In the calculation from the above loss function, this situation needs to be excluded. Optionally, for the closed area with index k when calculating L, all adjacent closed areas of this area can be extracted first; then, from all the extracted adjacent closed areas, the adjacent closed areas with the adjacent boundary being the fault are removed, and the indices of the remaining adjacent closed areas together form U(k). In this way, the situation shown in Figure 5 can be excluded, so that the strata on both sides of the fault can obtain the correct lithology prediction values according to the lithology of other pixels in their respective closed areas.

[0068] In addition, the Adam optimizer can be used to update the parameters during the training process; the callback function is used to save the best model and terminate in advance to prevent overfitting.

[0069] Based on the above lithology prediction model for the fault geological profile, Figure 6 is a flowchart of an automatic drawing method for a fault geological profile provided by an embodiment of the present invention. As Figure 6 shown, the method specifically includes the following steps:

[0070] S210. Obtain the original image of the fault geological profile to be drawn.

[0071] This original image is a new image outside the above dataset, and it is also an overall picture of the fault geological profile obtained by multi-angle shooting, preference, and stitching in S110.

[0072] S220. Input the original image into the image segmentation model constructed in any of the above embodiments to obtain the pixel-level lithology distribution of the original image.

[0073] S230. Superimpose the boundary lines and lithologic attributes in the pixel-level lithologic distribution on the original image to obtain the final geological cross-section diagram.

[0074] In this step, regions of different classes, including lithologic values and boundary lines, are extracted from the prediction mask (i.e., pixel-level lithologic distribution) generated by the deep learning model, and the extraction results are superimposed on the original image. Different legends and symbols are used to fill the lithologic distribution in the cross-section diagram. Optionally, annotations can also be added to the geological cross-section and the fusion result to mark the main boundaries and lithologic regions, obtaining the final geological cross-section diagram.

[0075] In addition, secondary discrimination and correction can be performed on similar fault information such as joints existing in the figure; and the corresponding cross-section strike, fault occurrence information (strike, dip), fault occurrence information (strike, dip), optically stimulated luminescence, C14 sampling points, Quaternary sediment age, etc. are manually input and marked. The final result is as Figure 2 shown in (b).

[0076] In summary, the embodiment of the present invention provides a method for constructing a lithologic prediction model of a fault geological cross-section and automatic drawing. Based on data such as photographed pictures obtained actually, a geological cross-section diagram is recognized and drawn by a computer. The development of faults and stratigraphic boundaries can be reflected in the figure, and the development characteristics of faults (including occurrence characteristics such as fault strike and dip) can be reflected. Specifically, in this embodiment, AI deep learning is performed based on the established fault lithology database, the pixel-level lithologic distribution of the original image is output through the fault geological cross-section lithologic prediction model, and the block boundaries in the cross-section are determined based on this lithologic distribution, including fault line information; stratigraphic pattern filling is performed according to the lithologic distribution and block boundaries, and stratigraphic and fault occurrence information is input and marked. The drawing result of the fault geological cross-section diagram by this method is accurate and concise, and can realize automatic filling and drawing of the fault geological cross-section, greatly improving the efficiency and accuracy of drawing the fault geological cross-section.

[0077] Figure 7 The following is a schematic structural diagram of an electronic device provided by an embodiment of the present invention. As Figure 7 shown, the device includes a processor 60, a memory 61, an input device 62, and an output device 63; the number of processors 60 in the device can be one or more. Figure 7 Here, one processor 60 is taken as an example; the processor 60, memory 61, input device 62, and output device 63 in the device can be connected through a bus or other means. Figure 7 Here, connection through a bus is taken as an example.

[0078] The memory 61, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the automatic drawing method of the fault geological profile in the embodiments of the present invention. The processor 60 executes various functional applications and data processing of the device by running the software programs, instructions, and modules stored in the memory 61, that is, to implement the above-mentioned automatic drawing method of the fault geological profile.

[0079] The memory 61 may mainly include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created according to the use of the terminal, etc. In addition, the memory 61 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage devices. In some instances, the memory 61 may further include a memory remotely set relative to the processor 60, and these remote memories can be connected to the device through a network. Examples of the above network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0080] The input device 62 can be used to receive input digital or character information, and generate key signal inputs related to the user settings and function controls of the device. The output device 63 may include a display device such as a display screen.

[0081] The embodiments of the present invention also provide a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the automatic drawing method of the fault geological profile in any embodiment.

[0082] The computer storage medium of the embodiments of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the computer-readable storage medium include: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, the computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, apparatus, or device.

[0083] A computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which computer-readable program code is carried. Such a propagated data signal may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. The computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device.

[0084] The program code contained on a computer-readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wire, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0085] The computer program code for performing the operations of the present invention may be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the C language or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (for example, by using an Internet service provider to connect through the Internet).

[0086] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing a lithology prediction model of a fault geological section, characterized in that Including: Obtaining a plurality of original images of the fault geological profile, as well as manual block-level lithology annotations for each original image; Extracting geological boundary curves from each original image through edge detection; Training an image segmentation model based on deep learning with each original image as a sample, and the trained model is used to predict the pixel-level lithology distribution of each original image; During the training process, by constraining the lithology prediction results of each pixel within the closed area divided by the geological boundary curve in the original image to tend to be consistent, the lithology prediction results of adjacent closed areas are different, and the number of pixels where the pixel-level lithology prediction results are inconsistent with the block-level lithology annotations is minimized, to update the model parameters.

2. The method according to claim 1, wherein The obtaining of a plurality of original images of the fault geological profile, as well as manual block-level lithology annotations for each original image, includes: Obtaining a plurality of original images of the fault geological profile; Through manual division, each original image is divided into a plurality of first lithology blocks, and lithology annotations are made for each first lithology block.

3. The method according to claim 1, characterized in that, The extracting of geological boundary curves from each original image through edge detection includes: Using the Canny algorithm to perform edge detection on each original image to obtain the edge points of each original image; Performing quadratic fitting on the edge points in the same original image to obtain the geological boundary curve in the same original image.

4. The method according to claim 1, wherein The training of an image segmentation model based on deep learning with each original image as a sample includes: Training an image segmentation model with a U-Net structure using each original image as a sample, and the trained model is used to output the lithology attributes at each pixel position in each original image.

5. The method according to claim 1, wherein The updating of the model parameters by constraining the lithology prediction results of each pixel within the closed area divided by the geological boundary curve in the original image to tend to be consistent, the lithology prediction results of adjacent closed areas are different, and the number of pixels where the pixel-level lithology prediction results are inconsistent with the block-level lithology annotations is minimized, includes: Constructing the following loss function: where k represents the index of the closed region divided by the geological boundary curve in the original image, N represents the total number of closed regions divided by the geological boundary curve in the original image; i represents the pixel index in the closed region with index k, and n k represents the total number of pixels in the closed region with index k; μ k represents the mean value of the lithology prediction results of all pixels in the closed region with index k, and I p，k，i represents the value of the lithology prediction result of the pixel with index i in the closed region with index k; U(k) represents the set of indices of adjacent closed regions of the closed region with index k, j represents the index of the closed region in the set U(k), and μ j represents the mean value of the lithology prediction results of all pixels in the closed region with index j; δ(x) represents the impulse function, when the independent variable x = 0, δ(x) is a preset positive constant, and when the independent variable x ≠ 0, δ(x) = 0; num(I b -I p ) represents the number of pixels where the pixel-level lithology prediction value I p is inconsistent with the block-level lithology annotation value I b , and α, β, and γ are weight coefficients respectively; Updating the model parameters by minimizing the loss function.

6. The method according to claim 5, characterized in that The updating of the model parameters by minimizing the loss function includes: Extracting all adjacent closed areas of the closed area with index k; Removing the adjacent closed areas with adjacent boundaries being faults, and the indices of the remaining adjacent closed areas together constitute U(k).

7. An automatic drawing method for a fault geological profile, characterized in that, Including: Obtaining the original image of the fault geological profile to be drawn; Inputting the original image into the image segmentation model to obtain the pixel-level lithology distribution of the original image, where the image segmentation model is constructed by using the lithology prediction model construction method for the fault geological profile described in any one of claims 1-6; Overlaying and displaying the boundary line and lithology attributes in the pixel-level lithology distribution on the original image to obtain the final geological profile diagram.

8. The automatic drawing method of the fault geological profile according to claim 7, characterized in that After the overlaying and displaying of the boundary line and lithology attributes in the pixel-level lithology distribution on the original image, it further includes: Adding annotations in the overlay image to mark the main boundaries and lithology areas; Performing secondary discrimination and correction on the fault information in the overlay image, and manually inputting the corresponding profile trend, fault occurrence information, fault occurrence information, optically stimulated luminescence, C14 sampling points, and Quaternary sediment age.

9. An electronic device, characterized in that, Comprising: One or more processors; A memory for storing one or more programs, When the one or more programs are executed by the one or more processors, enabling the one or more processors to implement the method for constructing a lithology prediction model of a fault geological profile according to any one of claims 1-6, or the method for automatically drawing a fault geological profile according to any one of claims 7-8.

10. A computer-readable storage medium, characterized in that, On which a computer program is stored, and when the program is executed by a processor, it implements the method for constructing a lithology prediction model of a fault geological profile according to any one of claims 1-6, or the method for automatically drawing a fault geological profile according to any one of claims 7-8.

Images