A lithology identification method, device, computer equipment and storage medium

Abstract

The present application discloses a lithology identification method, device, computer equipment and storage medium. The lithology identification method comprises the following steps: collecting and organizing formation microresistivity scanning imaging FMI image data and corresponding lithology descriptions; constructing a training data set, a verification data set and a test data set according to a preset ratio, and standardizing the constructed various data sets; establishing a lithology identification model based on a dual-modal little information elimination deep learning model, and training the lithology identification model through the standardized training data set, the verification data set and the test data set; continuously optimizing the lithology identification model using a random gradient descent method until a trained lithology identification model is obtained; testing and identifying the FMI image data in the test data set based on the trained lithology identification model, comparing and analyzing the output prediction results with the standard labels, and obtaining the final lithology identification model.

Description

Technical Field

[0001] The present application relates to the technical field of oil and gas geological research, and specifically to a lithology identification method, device, computer equipment and storage medium. Background Art

[0002] Petrology is a basic concept that describes the properties and characteristics of rocks. The study of lithology can reflect the structure and evolution of the earth. The rocks on the earth are mainly composed of sedimentary rocks, metamorphic rocks and igneous rocks, among which sedimentary rocks are the most important component, accounting for about 70% of the total. The study of sedimentary rocks is not only of academic significance, but also of great economic value. We can determine different types of sedimentary rock surfaces by studying the different combinations and arrangements of lithology in sedimentary rocks, so that we can understand the geological history and environmental evolution process. In addition, different lithology types can also reflect different climatic environments and sedimentary conditions, and the paleoclimate changes at that time can be calculated. In addition to its application in academic research, lithology is one of the key factors affecting the formation and distribution of oil and gas reservoirs in industry. Different lithologies have different physical and chemical characteristics, which will have a significant impact on the formation of oil and gas reservoirs.

[0003] Lithology identification is usually done by sampling the core, and then using a series of measures such as thin sections, scanning electron microscopes, and optical microscopes to observe the core's particles, color, and composition to make the final lithology conclusion. Although reliable core analysis results can be obtained through this method, this manual observation method is highly subjective and requires professional technical skills to make a correct analysis. In addition, this manual method often takes a lot of time and manpower, and the requirements for time and manpower are very high. Therefore, automatic lithology identification technology is crucial to the work of geologists, which can improve their work efficiency and reduce their redundant work.

[0004] In recent years, the development of artificial intelligence and deep learning technology has led to more and more integrated applications of computers with other fields, such as chatting, autonomous driving, medical diagnosis, industrial fault detection, etc. In terms of lithology identification, many current works are usually the study of well logging curves, while the study of FMI (Formation MicroScanner Image) image data is in a stage of improvement that needs further development. Since different lithologies have different characteristics under the action of electricity, FMI image data can also reflect lithology. However, for FMI image data, FMI_STAT (formation microresistivity scanning imaging_static) images and FMI_DYN (formation microresistivity scanning imaging_dynamic) images generally appear in pairs. Therefore, how to obtain two types of feature information by parallel processing of FMI_STAT images and FMI_DYN images, and use these two types of feature information for the final lithology judgment is a technical problem that needs to be solved urgently.

[0005] The content of this background technology description is only for facilitating understanding of the relevant technology in this field and is not regarded as an admission of the prior art. Summary of the invention

[0006] Therefore, the embodiments of the present invention intend to provide a FMI image lithology identification method, device, computer equipment and storage medium based on dual-modality and elimination model, which can not only make full use of FMI image data of two different modes, but also actively learn the information in the FMI image and eliminate the blank band information and redundant information in the FMI image data. This method not only achieves good recognition accuracy in the lithology classification standards of mudstone, algal limestone, limestone and sandstone, but also improves the rationality and stability of the model.

[0007] In a first aspect, an embodiment of the present invention provides a lithology identification method, comprising the following steps:

[0008] S1: Collect and organize formation microresistivity scanning imaging (FMI) image data and corresponding lithology descriptions;

[0009] S2: constructing a training data set, a validation data set, and a test data set according to a preset ratio, and performing standardization processing on the constructed training data set, the validation data set, and the test data set;

[0010] S3: establishing a lithology recognition model based on a dual-modal little information elimination deep learning model, and training the lithology recognition model through the standardized training data set, the verification data set, and the test data set;

[0011] S4: The lithology identification model is continuously optimized by using the stochastic gradient descent method until a trained lithology identification model is obtained;

[0012] S5: Based on the trained lithology recognition model, the FMI image data in the test data set is tested and recognized, and the output prediction results are compared and analyzed with the standard labels to obtain the final lithology recognition model.

[0013] Optionally, the FMI image data includes: static image data and dynamic image data.

[0014] Optionally, the lithology of the FMI image data collected in step S1 includes limestone, sandstone, mudstone and algal limestone.

[0015] Optionally, in step S2, the collected FMI image data is divided into units of a first size, and a portion of the first size units is selected as a test data set, the remaining first size units are further divided into sample data of a second size, and the sample data is divided into a training data set and a verification data set according to a preset ratio.

[0016] Optionally, the standardization process in step S2 is:

[0017]

[0018] Among them, x norm is the standardized value, x is the input image pixel, mean and std are the mean and variance respectively.

[0019] Optionally, the dual-modal little information elimination deep learning model in step S3 includes: a data segmentation module, a block coding and classification header module, a elimination module, an information extraction module and a classification module; the dual-modal little information elimination deep learning model construction process is:

[0020] The data segmentation module segments the standardized data into N image blocks of equal size and non-overlapping, where N is an integer greater than or equal to 1;

[0021] The block coding and classification header module adds position coding and classification headers to the N image blocks;

[0022] The elimination module eliminates image data containing blank band information and redundant information areas;

[0023] The information extraction module extracts data information;

[0024] The classification module classifies and identifies the data and outputs the lithology classification results.

[0025] Optionally, in step S5, the accuracy of the dual-modal little information elimination deep learning model in lithology identification and the effect of the little information elimination module are further evaluated.

[0026] In an embodiment of the present invention, in a second aspect, an embodiment of the present invention provides a lithology identification device, characterized in that it includes:

[0027] Data collection module, used to collect and organize formation microresistivity scanning imaging FMI image data and its corresponding lithology description;

[0028] A data processing module, used to divide the image data into units, select a test data set and a sample data set, divide the sample data set again, and divide the divided sample data into a training data set and a verification data set according to a preset ratio, and perform standardization processing on the constructed training data set, the verification data set and the test data set;

[0029] A lithology identification model construction module is used to establish a lithology identification model based on a dual-modal little information elimination deep learning model, train the lithology identification model through the standardized training data set, the verification data set and the test data set, and continuously optimize the lithology identification model using a stochastic gradient descent method until a trained lithology identification model is obtained;

[0030] The data analysis module performs test recognition on the FMI image data in the test data set based on the trained lithology recognition model, compares and analyzes the output prediction results with the standard labels, and obtains the final lithology recognition model.

[0031] In a third aspect, an embodiment of the present invention provides a computer device, characterized in that the computer device comprises: a memory storing an executable computer program; and a processor, wherein the processor runs the executable computer program in the memory to implement any lithology identification method described in the present invention.

[0032] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements any lithology identification method described in the present invention.

[0033] Other optional features and technical effects of the embodiments of the present invention are partially described below, and partially can be understood by reading this document. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. The elements shown are not limited to the proportions shown in the accompanying drawings. The same or similar reference numerals in the accompanying drawings represent the same or similar elements, wherein:

[0035] Figure 1 A schematic diagram of a process flow of a lithology identification method that can implement an embodiment of the present invention is shown;

[0036] Figure 2 A schematic diagram of a dual-modal architecture of a DDViT deep learning model in a lithology identification method according to an embodiment of the present invention is shown;

[0037] Figure 3 A schematic diagram of a lithology identification device that can implement an embodiment of the present invention is shown;

[0038] Figure 4 A schematic diagram of a computer device that can implement an embodiment of the present invention is shown;

[0039] Figure 5 A schematic diagram showing the result of discarding blank band information and redundant information using a rejection module according to an embodiment of the present invention;

[0040] Figure 6 A schematic diagram showing the identification results of the lithology identification method according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific implementation methods and drawings. Here, the exemplary implementation methods of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0042] As used herein, the term "including" and its variations mean open inclusion, i.e., "including but not limited to". Unless otherwise stated, the term "or" means "and / or". The term "based on" means "based at least in part on". The terms "an example embodiment" and "an embodiment" mean "at least one example embodiment". The term "another embodiment" means "at least one additional embodiment". The terms "first", "second", etc. may refer to different or the same objects. Other explicit and implicit definitions may also be included below.

[0043] In the embodiment of the present invention, a FMI image lithology identification method, device, computer equipment and storage medium based on dual-modality and elimination model are proposed, which can not only make full use of FMI image data of two different modes, but also actively learn the information in the FMI image and eliminate the blank band information and redundant information in the FMI image data. This method not only achieves excellent recognition accuracy in the lithology classification standards of mudstone, algal limestone, limestone and sandstone, but also improves the rationality and stability of the model.

[0044] In some embodiments, the lithology identification device may be combined with the lithology identification method features of any embodiment, and vice versa, which will not be elaborated here.

[0045] In an embodiment of the present invention, a computer device is provided, comprising: a processor and a memory storing a computer program, wherein the processor is configured to execute any lithology identification method of the embodiment of the present invention when running the computer program.

[0046] Figure 1 FIG. 2 shows a schematic flow chart of a lithology identification method that can be implemented in an embodiment of the present invention. Figure 1The lithology identification method shown includes the following steps:

[0047] S1: Collect and organize FMI image data and the corresponding lithology description. The FMI image data may be logging data obtained by Schlumberger's MAXIS-500 logging series, which is not specifically limited in the present embodiment of the application. Figure 1 In the shown embodiment, the drilling depth for collecting FMI data can be the logging section 2780-4480m, and the rock types are mainly limestone, sandstone, mudstone and algal limestone. The embodiment of the present application is not specifically limited. The FMI data includes FMI_STAT (formation microresistivity scanning imaging_static) image data and FMI_DYN (formation microresistivity scanning imaging_dynamic) image data.

[0048] S2: Construct a training data set, a validation data set and a test data set according to a preset ratio, and standardize the constructed training data set, the validation data set and the test data set; in this step, the FMI data collected in step S1 is divided into units of a first size, and the units of the first size can be preset, for example, the units of the first size can be 256x1024, and the present application does not make specific restrictions; divide the FMI data into N units of the first size, N is an integer greater than or equal to 1, select a part of the N units as the test data set, and then divide the remaining units into sample data of a second size again. Similarly, the second size can also be preset, for example, the second size can be 16x1024, and the present application does not make specific restrictions. Then divide the divided sample data into a training data set and a validation data set according to a preset ratio. In the embodiment of the present application, in order to illustrate that the ratio of the training data set to the validation data set is 7:3, the present application does not make specific restrictions. And in this step, the above-mentioned divided training data set, validation data set and test data set are standardized using formula (1);

[0049]

[0050] Among them, x norm is the standardized value, x is the input image pixel in RGB format, and mean and std are the mean and variance calculated on the million-level ImageNet dataset.

[0051] S3: establishing a lithology recognition model based on a dual-modal little information elimination deep learning model, and training the lithology recognition model through the standardized training data set, the verification data set, and the test data set;

[0052] In this step, a deep learning model is first constructed. For example, the deep model can be constructed using the Pytorch framework using the model, and the embodiment of the present application does not make specific restrictions. The embodiment of the present application provides that the deep learning model constructed in this step is a DDViT deep learning model, and the DDViT deep learning model includes a data segmentation module, a block encoding and classification head module, a removal module, a transformer (information extraction) module, and a classification module. The operation flow of the model is as follows:

[0053] First, the standardized data is divided into N image blocks of equal size and non-overlapping, where N can be preset and is an integer greater than or equal to 1, and position coding and classification headers are added to the N image blocks of equal size and non-overlapping, using formula (2) to add the position coding and classification headers;

[0054] X DYN ,X STAT =(Cat(cls DYN ,X DYN ),Cat(cls STAT ,X STAT ))+PE (2)

[0055] Among them, X DYN and X STAT They represent FMI_DYN image and FMI_STAT image data information respectively; Cat function is used to stitch all image blocks together; PE represents position encoding; cls DYN and cls STAT Refers to the classification head, which is responsible for the final lithology classification. Then the obtained X DYN and X STAT The data information is sent to the little information elimination module. In this step, the blank band information and redundant information in the FMI image data are eliminated according to formula (3) and formula (4), and the image data containing the blank band information and redundant information area is discarded.

[0056] X DYN =LD DYN (X DYN )(3)

[0057] X STAT =LD STAT (X STAT )(4)

[0058] Where, LD DYN and LD STAT X DYN and X STAT A data information elimination function is provided, wherein the elimination function discards image data in blank band information and redundant information areas.

[0059] After that, X that has passed through the low-information elimination module DYN and X STAT The data information is sent to each transformer module for information extraction. During the extraction process, information is extracted from data of different modalities (static image data and dynamic image data), as shown in formula (5) and formula (6):

[0060] X DYN =Transformer DYN (X DYN )(5)

[0061] X STAT =Transformer STAT (X STAT )(6)

[0062] Then, the information extracted from each mode is aggregated, and the aggregated information is extracted again, as shown in formula (7), and then the extraction result is input into the classification module for lithology identification, as shown in formula (8).

[0063] X = Transformer(Cat(X DYN ,X STAT ))(7)

[0064] Facies=Class(Mean(cls DYN ,cls STAT ))(8)

[0065] In the formula, X represents the DYN and X STAT The processing result after the information is extracted again by the transformer function after aggregation; Class represents the classification module; Mean represents the average operation of all parameters; Facies represents the final lithology classification result.

[0066] S4: Use the random gradient descent method to continuously optimize the lithology identification model until a trained lithology identification model is obtained. Among them, the trained lithology identification model is a stable, accurate and optimal DDViT deep learning model. In the embodiment of the present application, the optimizer used for the DDViT deep learning model can be AdamW (adaptive gradient optimizer), which is not specifically limited in this application. In this step, the FMI image data is input into the DDViT deep learning model for multiple iterations. For example, the number of iterations can be 100 times, which is not specifically limited in the embodiment of the present application. Each time the predicted value and the true value of the deep learning model are used to calculate the loss, the deep learning model is optimized using the loss, so that the predicted value of the deep learning model continues to approach the true value. After the loss stabilizes, the optimal DDViT lithology identification model that has been trained can be obtained.

[0067] S5: Based on the trained lithology recognition model, the FMI image data in the test data set is tested and recognized, and the output prediction results are compared and analyzed with the standard labels to obtain the final lithology recognition model.

[0068] In this step, the test data set is sent to the trained DDViT deep learning model for testing, and the actual lithology data of the test data set is compared with the lithology data analyzed by the deep learning model, and the accuracy of the deep learning model is calculated to verify the accuracy of the deep learning model and obtain the final lithology identification model.

[0069] In this step, it is necessary not only to analyze the accuracy of the DDViT model in lithology identification, but also to analyze the results of eliminating the little information area by the little information elimination module.

[0070] Figure 2 FIG. 1 shows a schematic diagram of a dual-modal architecture of a DDViT deep learning model in a lithology identification method that can implement an embodiment of the present invention. Figure 2 The dual-modal architecture 20 of the DDViT deep learning model shown in the figure includes: a data segmentation module 201, a block coding and classification header module 202, a culling module 203, an information extraction module 204, and a classification module 205. The information extraction module 204 includes a sub-information extraction module and an aggregate information extraction module. Figure 2 In the embodiment shown, the input data are static image data and dynamic image data. Figure 2 In the embodiment shown, the data segmentation module 201 segments the input FMI_DYN image data and FMI_STAT image data into N image blocks of equal size and non-overlapping, where N can be preset and is an integer greater than or equal to 1. The block coding and classification header module 202 adds position coding and classification headers to the N image blocks of equal size and non-overlapping, and uses formula (2) to add position coding and classification headers.

[0071] The elimination module 203 eliminates the blank band information and redundant information in the FMI image data according to formula (3) and formula (4), and discards the image data containing the blank band information and redundant information area.

[0072] The information extraction module 204 extracts information from the data processed by the elimination module 203. In the extraction process, the information of the data of different modes (static image data and dynamic image data) is extracted respectively, and then the extracted information is aggregated, and further information is extracted from the aggregated result. The information extraction module extracts the X DYN and X STAT The data information is sent to each information extraction module for information extraction, as shown in formula (5) and formula (6).

[0073] Then, the information extracted from each mode is aggregated, and the aggregated information is extracted again, as shown in formula (7), and then the extraction result is input into the classification module 205 for processing, and the final lithology classification result is output, as shown in formula (8).

[0074] Figure 3 FIG. 2 is a schematic diagram showing a lithology identification device that can implement an embodiment of the present invention. Figure 3 The lithology identification device shown includes: a data collection module 301, a data processing module 302, a lithology identification model building module 303 and a data analysis module 304. Figure 3 In the embodiment shown, the data collection module 301 collects and organizes FMI image data and corresponding lithology descriptions, wherein the FMI image data may be obtained by the MAXIS-500 logging series of Schlumberger Company, and the embodiment of the present application does not make specific restrictions; specifically, Figure 3In the embodiment shown, the drilling depth for collecting FMI data can be the logging section 2780-4480m, and the lithology is mainly limestone, sandstone, mudstone and algal limestone, which is not specifically limited in the embodiment of the present application. The FMI data includes FMI_STAT (formation microresistivity scanning imaging_static) image data and FMI_DYN (formation microresistivity scanning imaging_dynamic) image data. The data processing module 302 divides the collected FMI data into units of a first size, which can be preset, for example, the first size can be 256x1024, which is not specifically limited in the present application; the data screening and division module 302 divides the FMI data into N units of the first size, where N is an integer greater than or equal to 1, selects a part of the N units as test data, and then divides the remaining units into sample data of a second size again, and similarly, the second size can also be preset, for example, the second size can be 16x1024, which is not specifically limited in the present application. After the division is completed, the data screening and division module 302 divides the above-mentioned sample data into a training data set and a verification data set according to a preset ratio. In the embodiment of the present application, the ratio of the training data set to the verification data set is 7:3 for illustration, and the present application does not make specific restrictions. And the data processing module 302 uses formula (1) to standardize the data of the divided training data set, verification data set and test data set. The lithology identification model construction module 303 constructs a lithology identification model based on the DDViT deep learning model. The embodiment of the present application can use the Pytorch framework to construct the deep learning model, and the embodiment of the present application does not make specific restrictions. The DDViT deep learning model construction process is similar to the principle of constructing the DDViT deep learning model in the above-mentioned lithology identification method embodiment, and will not be repeated here. After the lithology identification model is constructed, the lithology identification model construction module 303 trains the constructed lithology identification model through the standardized training data set, verification data set and test data set, continuously optimizes the DDViT deep learning model using the stochastic gradient descent method, and iterates until a stable and accurate trained lithology identification model is obtained.

[0075] The data analysis module 304 uses the optimal trained lithology recognition model to recognize the above FMI image data, compares the prediction results of the test data set with the standard label, analyzes the prediction results, and obtains the final lithology recognition model. The final lithology recognition model can also be used to predict the well logging data to obtain the final lithology recognition result.

[0076] Figure 4 A schematic diagram of a computer device that can implement an embodiment of the present invention is shown. Figure 4The computer device shown includes: a processor 401, a memory 402, and a computer program 403 stored in the memory 402. When the computer program is executed by the processor, the following steps are performed:

[0077] Collect and organize formation microresistivity scanning imaging (FMI) image data and corresponding lithology descriptions;

[0078] Constructing a training data set, a validation data set, and a test data set according to a preset ratio, and performing standardization processing on the constructed training data set, the validation data set, and the test data set;

[0079] Establishing a lithology recognition model based on a dual-modal little information elimination deep learning model, and training the lithology recognition model through the standardized training data set, the verification data set and the test data set;

[0080] The stochastic gradient descent method is used to continuously optimize the lithology identification model until a trained lithology identification model is obtained;

[0081] Based on the trained lithology recognition model, the FMI image data in the test data set is tested and recognized, and the output prediction results are compared and analyzed with the standard labels to obtain the final lithology recognition model.

[0082] Figure 5 A schematic diagram showing the result of discarding blank band information and redundant information using the elimination module of an embodiment of the present invention. Figure 5 The results of some FMI image data after the elimination module in the deep learning model are shown, such as Figure 5 The elimination module provided in the embodiment of the present application can well discard the blank band information and redundant information. The method provided in the embodiment of the present application can be applied to the lithology recognition of the FMI image data of the exploration well, further improving the automatic lithology recognition capability of the FMI image data.

[0083] Figure 6 A schematic diagram showing the identification result of the lithology identification method according to an embodiment of the present invention; Figure 6 The lithology recognition effect of some deep learning models in the target well is shown, such as Figure 6 As shown, 601 represents mudstone, 602 represents limestone, 603 represents algal limestone, and 604 represents sandstone. Figure 6 It can be seen that the deep learning model provided in the embodiment of the present application can achieve an accuracy of 90.8% in the test data set after multiple iterations.

[0084] The embodiment of the present application sends the two different characteristics of image information, FMI_DYN image and FMI_STAT image, into the dual-modal architecture of the present application, so that the information of the two modes can be fully raised, and the accuracy of lithology classification can be further improved. By using the elimination module provided in the embodiment of the present application, the blank band information and redundant information in the FMI image data can be eliminated, which can make the model more stable and reasonable on the one hand, and reduce the black box nature of the deep learning network model on the other hand.

[0085] The devices, modules or units described in the above embodiments may be implemented by a computer or its associated components. The computer may be, for example, a mobile terminal, a smart phone, a personal computer, a laptop computer, a vehicle-mounted human-computer interaction device, a personal digital assistant, a media player, a navigation device, a game console, a tablet computer, a wearable device, a smart TV, an Internet of Things system, a smart home, an industrial computer, a server or a combination thereof.

[0086] Although not shown, in an embodiment of the present invention, a storage medium is provided, the storage medium storing a computer program, the computer program being configured to execute any file difference-based compilation method of the embodiments of the present invention when executed.

[0087] Storage media in embodiments of the present invention include permanent and non-permanent, removable and non-removable items that can be used to store information by any method or technology. Examples of storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0088] The methods, programs, systems, devices, etc. of the embodiments of the present invention may be executed or implemented in a single or multiple networked computers, or may be practiced in a distributed computing environment. In the embodiments of this specification, in these distributed computing environments, tasks may be performed by remote processing devices connected via a communication network.

[0089] Those skilled in the art should understand that the embodiments of the present specification may be provided as methods, devices or computer program products. Therefore, those skilled in the art may imagine that the implementation of the functional modules / units or controllers and related method steps described in the above embodiments may be implemented in software, hardware or a combination of software / hardware.

[0090] Unless explicitly stated, the actions or steps of the methods, programs, and embodiments of the present invention do not have to be performed in a specific order and can still achieve the desired results. In some implementations, multitasking and parallel processing are also possible or may be advantageous.

[0091] In this article, multiple embodiments of the present invention are described, but for the sake of brevity, the description of each embodiment is not exhaustive, and the same or similar features or parts between the embodiments may be omitted. In this article, "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" are meant to be applicable to at least one embodiment or example according to the present invention, but not all embodiments. The above terms do not necessarily mean to refer to the same embodiment or example. In the absence of contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples.

[0092] The exemplary method and apparatus of the present invention have been specifically shown and described with reference to the above embodiments, which are only examples of the best modes for implementing the present method and apparatus. It will be appreciated by those skilled in the art that various changes may be made to the embodiments of the apparatus and method described herein when implementing the present apparatus and / or method without departing from the spirit and scope of the present invention as defined in the appended claims.

Claims

1. A lithology identification method, characterized in that: The steps include: S1: Collect and organize formation microresistivity scanning imaging (FMI) image data and corresponding lithology descriptions; S2: constructing a training data set, a validation data set, and a test data set according to a preset ratio, and performing standardization processing on the constructed training data set, the validation data set, and the test data set; S3: A lithology recognition model is established based on a dual-modal little information elimination deep learning model, and the lithology recognition model is trained by the standardized training data set, the verification data set and the test data set, wherein the dual-modal little information elimination deep learning model includes: a data segmentation module, a block encoding and classification header module, a elimination module, an information extraction module and a classification module, and the dual-modal little information elimination deep learning model construction process is: The data segmentation module segments the standardized data into N image blocks of equal size and non-overlapping, where N is an integer greater than or equal to 1. The block coding and classification header module adds position coding and classification header to the N image blocks. The elimination module eliminates image data containing blank band information and redundant information areas. The information extraction module extracts data information, The classification module classifies and identifies the data and outputs the lithology classification result; S4: The lithology identification model is continuously optimized by using the stochastic gradient descent method until a trained lithology identification model is obtained; S5: Based on the trained lithology recognition model, the FMI image data in the test data set is tested and recognized, and the output prediction results are compared and analyzed with the standard labels to obtain the final lithology recognition model.

2. The lithology identification method according to claim 1, characterized in that: The FMI image data includes: static image data and dynamic image data.

3. The lithology identification method according to claim 1, characterized in that: The lithology of the FMI image data collected in step S1 includes limestone, sandstone, mudstone and algal limestone.

4. The lithology identification method according to claim 1, characterized in that: In the step S2, the collected FMI image data is divided into units of a first size, and a part of the first size units is selected as a test data set, the remaining first size units are further divided into sample data of a second size, and the sample data is divided into a training data set and a verification data set according to a preset ratio.

5. The lithology identification method according to claim 1, characterized in that: The standardization process in step S2 is as follows: Among them, x norm is the standardized value, x is the input image pixel, mean and std are the mean and variance respectively.

6. The lithology identification method according to claim 1, characterized in that: In step S5, the accuracy of the dual-modal little information elimination deep learning model in lithology identification and the effect of the little information elimination module are further evaluated.

7. A lithology identification device, characterized in that: include: Data collection module, used to collect and organize formation microresistivity scanning imaging FMI image data and its corresponding lithology description; A data processing module is used to divide the image data into units, select a test data set and a sample data set, divide the sample data set again, and divide the divided sample data into a training data set and a verification data set according to a preset ratio, and perform standardization processing on the constructed training data set, the verification data set and the test data set; The lithology identification model construction module is used to establish a lithology identification model based on a dual-modal little information elimination deep learning model, train the lithology identification model through the standardized training data set, the verification data set and the test data set, and continuously optimize the lithology identification model using a stochastic gradient descent method until a trained lithology identification model is obtained, wherein the dual-modal little information elimination deep learning model includes: a data segmentation module, a block encoding and classification header module, a elimination module, an information extraction module and a classification module, and the dual-modal little information elimination deep learning model construction process is: The data segmentation module segments the standardized data into N image blocks of equal size and non-overlapping, where N is an integer greater than or equal to 1. The block coding and classification header module adds position coding and classification header to the N image blocks. The elimination module eliminates image data containing blank band information and redundant information areas. The information extraction module extracts data information, The classification module classifies and identifies the data and outputs the lithology classification result; The data analysis module performs test recognition on the FMI image data in the test data set based on the trained lithology recognition model, compares and analyzes the output prediction results with the standard labels, and obtains the final lithology recognition model.

8. A computer device, characterized in that: The computer device comprises: a memory storing an executable computer program; A processor, wherein the processor runs the executable computer program in the memory to implement the lithology identification method described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the lithology identification method according to any one of claims 1 to 6 is implemented.

Images