Debris-generating dolomite sedimentary microfacies identification method based on algorithm

Abstract

The invention discloses an algorithm-based detritus dolomite sedimentary microfacies identification method, and relates to the technical field of rock sedimentary microfacies identification, and the method comprises the following steps: collecting a detritus dolomite sample, and carrying out analysis to obtain the structure and characteristics of the detritus dolomite sample; extracting features in the structure and the characteristics; selecting a machine learning algorithm; using known detritus dolomite sedimentary microfacies data as a training set, and training a machine learning algorithm to obtain a microfacies recognition model; and then the microfacies identification model is applied to actual detritus dolomite sedimentary microfacies identification, the method can obviously reduce the labor force required by detritus dolomite sedimentary microfacies identification, so that the laboratory analysis is time-saving and labor-saving, and when the microfacies identification model is used for identifying detritus dolomite, the identification result is more objective, effective, accurate and reliable; and the microscopic characteristics or microstructure of the rock stratum can also be captured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of rock sedimentary microfacies identification, and particularly relates to a method for identifying bioclastic dolomite sedimentary microfacies based on an algorithm. BACKGROUND

[0002] Bioclastic dolomite is a unique sedimentary rock that contains abundant bioclastic fragments within its structure. These fragments provide researchers with a wealth of information about ancient environments, ecosystems, and depositional conditions. Therefore, the analysis of bioclastic dolomite is of great significance in geology, oil and gas exploration, and environmental science.

[0003] Existing analysis techniques for bioclastic dolomite rely on macroscopic petrographic observations and a series of geophysical and geochemical tests. However, these methods are often labor-intensive, resulting in laboratory analysis typically requiring a significant amount of time and resources. Additionally, the analysis results may vary depending on different interpreters, especially in complex rock structures. Furthermore, traditional analysis methods may not capture certain microscopic features or microstructures.

[0004] Therefore, there is an urgent need to innovate on the basis of existing technology to find a method for identifying bioclastic dolomite sedimentary microfacies that is time-saving and labor-saving, objective and effective, accurate and reliable, and can capture microscopic features or microstructures. SUMMARY

[0005] The present application aims to provide a method for identifying bioclastic dolomite sedimentary microfacies based on an algorithm, which solves the problem of existing bioclastic dolomite analysis techniques relying on macroscopic petrographic observations and a series of geophysical and geochemical tests, requiring significant manpower and resources, and laboratory analysis being time-consuming and labor-intensive. Additionally, the identification results of bioclastic dolomite using traditional techniques have significant differences due to different interpreters, and the identification results are not objective and accurate enough.

[0006] The present application is achieved by the following technical solutions: A method for identifying bioclastic dolomite sedimentary microfacies based on an algorithm, comprising the following steps: S1, collecting bioclastic dolomite samples for analysis to obtain the structure and characteristics of the bioclastic dolomite samples; S2, extracting features from the structure and characteristics in step S1; S3, selecting a machine learning algorithm based on the features obtained in step S2; S4, using known bioclastic dolomite sedimentary microfacies data as a training set to further train the machine learning algorithm and obtain a microfacies identification model; S5, applying the microfacies identification model in step S4 to actual bioclastic dolomite sedimentary microfacies identification.

[0007] Furthermore, in step S1, the methods for analyzing the collected bioclastic dolomite samples include petrographic, geophysical, and geochemical analysis.

[0008] Furthermore, in step S1, the structure of the bioclastic dolomite sample includes biotype, grain structure, mineral content, grain content, bioclastic type, cement characteristics, and bioclastic content.

[0009] Furthermore, in step S1, the characteristics of the bioclastic dolomite sample include mineral content, cementation form, geophysical properties and geochemical properties, as well as time series characteristics, frequency domain characteristics, statistical characteristics and texture and shape characteristics.

[0010] Furthermore, in step S2, the features include particle content, bioclastic content, geophysical properties, and geochemical properties.

[0011] Furthermore, in step S3, the machine learning algorithm includes decision tree, random forest, support vector machine, and neural network algorithm.

[0012] Furthermore, in step S4, when the machine learning algorithm is a decision tree learning algorithm, the following steps are included: S4.1. Multiple known bioclastic dolomite sedimentary micro-correspondences correspond to multiple bioclastic dolomites. For each bioclastic dolomite, extract its texture and shape features, time series features, and frequency domain information. S4.2 Next, each bioclastic dolomite is judged by the decision tree algorithm. The judgment result is compared with the corresponding bioclastic dolomite sedimentary microfacies to determine the deviation between the judgment result and the corresponding bioclastic dolomite sedimentary microfacies. The decision tree algorithm is then adjusted according to the deviation. S4.3 Then, according to the adjusted decision tree algorithm, each bioclastic dolomite is compared with the bioclastic dolomite sedimentary microfacies again until the comparison results are within the preset range. The adjusted decision tree algorithm is then used as the microfacies identification model.

[0013] Furthermore, in step S4, when the machine learning algorithm is the random forest learning algorithm, the following steps are included: S4.1 For each bioclastic dolomite, construct multiple decision trees, and then use multiple known bioclastic dolomite sedimentary microfacies data as training sets. Multiple known bioclastic dolomite sedimentary microfacies correspond to multiple bioclastic dolomites. For each bioclastic dolomite, extract the texture and shape features, time series features and frequency domain information of the bioclastic dolomite. S4.2 Next, each bioclastic dolomite is analyzed using multiple decision trees. The results of these decision trees are then aggregated and averaged. The average result is then compared with the corresponding bioclastic dolomite sedimentary microfacies to determine the deviation between the average result and the corresponding bioclastic dolomite sedimentary microfacies. Based on this deviation, the multiple decision trees are adjusted. S4.3. Then, according to the adjusted decision trees, each bioclastic dolomite is compared with the bioclastic dolomite sedimentary microfacies until the comparison results are within the preset range. The adjusted decision trees are then sorted out to form a random forest, which is used as the microfacies identification model.

[0014] Furthermore, in step S4, when the machine learning algorithm is a support vector machine learning algorithm, the following steps are included: S4.1. Multiple known bioclastic dolomite sedimentary microstructures correspond to multiple bioclastic dolomites. For each bioclastic dolomite, the time series features of that bioclastic dolomite are extracted, and then multiple time series features are obtained. These multiple time series features are then combined to form a hyperplane. S4.2. Then, based on the distance between the hyperplane and the time series features, the sedimentary microfacies of the bioclastic dolomite are determined. If the deviation between the determined sedimentary microfacies of the bioclastic dolomite and the actual sedimentary microfacies of the bioclastic dolomite is within the set deviation range, it indicates that the determined result is relatively accurate, and the entire algorithm can be used as a microfacies identification model.

[0015] Furthermore, in step S4, when the machine learning algorithm is a neural network algorithm learning algorithm, the following steps are included: S4.1 First, the texture and shape features of several known bioclastic dolomite are input into the neural network model, and the model is trained based on the texture and shape features. S4.2 Then, the neural network model outputs the identified sedimentary microfacies of the bioclastic dolomite. The output sedimentary microfacies of the bioclastic dolomite are compared with the actual sedimentary microfacies of the bioclastic dolomite. If the difference between the output sedimentary microfacies of the bioclastic dolomite and the actual sedimentary microfacies of the bioclastic dolomite is within the preset difference range, it indicates that the identification result of the neural network model is accurate, and the neural network model can be used as the microfacies identification model; if the difference between the output sedimentary microfacies of the bioclastic dolomite and the actual sedimentary microfacies of the bioclastic dolomite is not within the preset difference range, the neural network model is adjusted. S4.3 Input the texture and shape features of multiple known bioclastic dolomite rocks into the adjusted neural network model. The adjusted neural network model will continue to output the identified sedimentary microfacies of bioclastic dolomite. Then, the output sedimentary microfacies of bioclastic dolomite are compared with the actual sedimentary microfacies of bioclastic dolomite. If the difference between the output sedimentary microfacies of bioclastic dolomite and the actual sedimentary microfacies of bioclastic dolomite is within the preset difference range, the adjusted neural network model is used as the microfacies identification model. If the difference between the output sedimentary microfacies of bioclastic dolomite and the actual sedimentary microfacies of bioclastic dolomite is not within the preset difference range, the neural network model is adjusted until the difference between the output sedimentary microfacies of bioclastic dolomite and the actual sedimentary microfacies of bioclastic dolomite is within the preset difference range.

[0016] Compared with the prior art, the present invention has the following advantages and beneficial effects: I. This invention employs an algorithm-based method for identifying microfacies in bioclastic dolomite sediments. This method includes collecting bioclastic dolomite samples, analyzing them to obtain their structure and characteristics; extracting features from the structure and characteristics; selecting a machine learning algorithm based on these features; using known bioclastic dolomite sedimentary microfacies data as a training set to train the machine learning algorithm and obtain a microfacies identification model; and applying the microfacies identification model to actual bioclastic dolomite sedimentary microfacies identification. Compared to existing technologies, this method significantly reduces the labor required for bioclastic dolomite sedimentary microfacies identification, making laboratory analysis more time- and labor-saving.

[0017] Second, in this invention, a microfacies identification model is used to identify bioclastic dolomite, which avoids the difference in identification results due to different interpreters, and the identification results are more objective, effective, accurate and reliable.

[0018] Third, this invention can also capture the microscopic features or microstructure of rock strata, which facilitates related research. Attached Figure Description

[0019] Figure 1 This is a flowchart of the bioclastic dolomite sedimentary microfacies identification method of the present invention.

[0020] Figure 2 This is a schematic diagram of a decision tree in Example 1. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0022] Example 1 This embodiment provides an algorithm-based method for identifying sedimentary microfacies in bioclastic dolomite. It is applicable to the identification of open-pitted, limestone-dominated bioclastic dolomite sedimentary microfacies. Specifically, it can be applied to marine bioclastic limestone interlayers with low dolomitization, where the dolomitization is minimal and difficult to predict. It can also be applied to ordinary bioclastic dolomite. (Reference) Figure 1 Specifically, it includes the following steps: Step 1: Collect bioclastic dolomite samples, analyze them, and obtain the structure and characteristics of the bioclastic dolomite samples.

[0023] Bioclastic dolomite is a type of carbonate rock composed of bioclastic material and crystalline dolomite. Bioclastic dolomite typically forms in shallow, warm-water environments, particularly in areas with abundant biological activity and clastic deposition. The presence of bioclastic dolomite often reflects the biological activity and marine environment of its formation period, helping geologists reconstruct ancient ecosystems and climatic conditions. In these environments, clastic material is deposited on the seabed and combines with dolomite crystals to form rock strata.

[0024] Generally, when collecting bioclastic dolomite samples, representative bioclastic dolomite is selected for analysis to obtain the structure and characteristics of representative bioclastic dolomite samples.

[0025] The analytical methods used for bioclastic dolomite samples include petrographic, geophysical, and geochemical analyses. The following section will provide a detailed introduction to petrographic, geophysical, and geochemical analyses: Petrology is the study of the formation, evolution, and classification of rocks, and it is an important branch of geology. Based on the type of rocks studied, petrology can be divided into three main parts: igneous petrology, sedimentary petrology, and metamorphic petrology. Sedimentary petrology studies the formation, properties, distribution, and depositional environments of sedimentary rocks. Sedimentary rocks are formed from materials on the Earth's surface through erosion, transport, deposition, and compaction, and include clastic rocks (such as sandstone and shale), chemical sedimentary rocks (such as limestone and salt rocks), and organic sedimentary rocks (such as coal).

[0026] Geophysical analysis is the study of the Earth's physical properties and processes. It involves using physical methods and techniques to investigate the Earth's internal structure, dynamic processes, material distribution, and physical characteristics. This discipline is crucial for understanding the Earth's formation, evolution, resource exploration, and environmental issues. Geophysical analysis has wide applications in oil, natural gas, mineral, and groundwater exploration. Furthermore, it provides an important scientific foundation and technical support for understanding the Earth's deep structure, earthquake prediction, environmental monitoring, and underground engineering.

[0027] Geochemical analysis uses chemical methods to study the composition and chemical changes of the Earth and its various parts (including the atmosphere, water bodies, crust, and mantle). Geochemistry provides invaluable information about how Earth's materials interact, how they form, and how they evolve. Geochemical analysis offers in-depth understanding and methodologies for geological exploration, environmental science, Earth system history, and the Earth's internal structure. With advancements in modern instruments and technologies, geochemistry has developed into a highly complex and multidisciplinary field, capable of providing answers to a wide range of scientific and applied questions.

[0028] In this step, petrographic, geophysical, and geochemical analysis methods are used to obtain the structural and tectonic characteristics of the bioclastic dolomite samples, including the bioclastic content, biotype, cementation characteristics, grain structure, geophysical characteristics, and geochemical characteristics of the bioclastic dolomite.

[0029] Step 2: Extract the features from the structure and properties in Step 1.

[0030] In this step, since the structure of the bioclastic dolomite sample is relatively clear, features such as structure and morphology, as well as time series features, frequency domain features, and statistical features, are selected as the basis for the next step of model construction.

[0031] This involves using image analysis software to extract features from scanning electron microscope images, including the shape, size, and distribution of bioclastic grains; calculating the relative mineral content from X-ray diffraction data; and extracting information about rock structure, including bedding, fractures, and contact relationships, based on thin section analysis. The extracted structural and property features include mineral content, cementation forms, bioclastic content, geophysical properties, and geochemical properties.

[0032] Furthermore, based on the above petrographic, geophysical, and geochemical analyses, the structure (including cement content, bioclastic assemblage type, and bioclastic content) and properties (including geophysical and geochemical properties) of the bioclastic dolomite samples were obtained, and features were extracted from their structure and properties. Feature extraction is a crucial step in machine learning and data analysis; it involves extracting information from raw data that is meaningful or relevant to a specific task (such as classification, regression, or clustering). Selecting or constructing the correct features can significantly improve model performance.

[0033] The following are some common features of feature extraction: Statistical characteristics: Calculate statistical values ​​based on the distribution of data, including mean, standard deviation, maximum value, minimum value, median, etc.

[0034] Time series features: For time series data, features such as trends, seasonality, and periodicity can be extracted.

[0035] Frequency domain features: By transforming data from the time domain to the frequency domain (e.g., using Fourier transform), the frequency features of a signal can be extracted.

[0036] Texture and shape features: For image data, features such as texture, shape, edges, and key points can be extracted.

[0037] The feature extraction method in this step can simplify the model input while maximizing relevant information. The extracted features are meaningful and non-redundant, and help improve the model's generalization ability.

[0038] In this approach, when extracting features, it is necessary to consider the correlation between features, the interpretability of features, and the relationship between features and the target variable.

[0039] Step 3: Select a machine learning algorithm based on the features obtained in Step 2.

[0040] In this step, the machine learning algorithm is a decision tree machine learning algorithm. A decision tree is a classification and regression algorithm that progressively splits features, making a "decision" for each split, thereby ultimately achieving its classification or prediction goal. A major advantage of decision trees is their high interpretability; they can be intuitively represented as a "tree-like" structure, making them very easy to understand.

[0041] The specific operating steps are as follows: First, an initial assessment is made of the bioclastic dolomite based on its texture and shape features. If the texture and shape features of the bioclastic dolomite meet the requirements, the time series features are then compared with those of the bioclastic dolomite. If the time series of the bioclastic dolomite meets the requirements, the frequency domain features are then compared with the frequency domain information extracted from the bioclastic dolomite. If the frequency domain information meets the requirements, the decision tree-based learning algorithm is then constructed and can be trained subsequently.

[0042] refer to Figure 2 , can be according to Figure 2 The learning algorithm for constructing decision trees is described in the above method.

[0043] In this embodiment, a decision tree machine learning algorithm is used, which can intuitively reflect the relationship between the characteristics of bioclastic dolomite; it is highly interpretable and easy to understand for bioclastic dolomite.

[0044] Step 4: Use known bioclastic dolomite sedimentary microfacies data as a training set to train a machine learning algorithm and obtain a microfacies recognition model.

[0045] The training set is a subset of data used to train models in machine learning and deep learning projects. The main goal of the training set is to fine-tune the model's parameters using this data, enabling the model to learn the basic patterns and relationships within the data.

[0046] The specific operating steps are as follows: I. Multiple known bioclastic dolomite sedimentary microstructures correspond to multiple bioclastic dolomites, thereby extracting the texture and shape features, time series features, and frequency domain information of each bioclastic dolomite; II. Then, each bioclastic dolomite is judged by the decision tree algorithm. The judgment result is compared with the corresponding bioclastic dolomite sedimentary microfacies to determine the deviation between the judgment result and the corresponding bioclastic dolomite sedimentary microfacies. The decision tree algorithm is adjusted according to the deviation. III. Then, using the adjusted decision tree algorithm, each bioclastic dolomite is compared with the bioclastic dolomite sedimentary microfacies again until the comparison results are within the preset range. The adjusted decision tree algorithm is then used as the microfacies identification model.

[0047] Step 5: Apply the microfacies identification model to the actual identification of microfacies in bioclastic dolomite deposits.

[0048] Among them, the obtained microfacies identification model can be applied to the microfacies of bioclastic dolomite sediments with relatively complex geological structures, thereby enabling accurate identification of the sedimentary microfacies of bioclastic dolomite.

[0049] The algorithm-based sedimentary microfacies identification method for bioclastic dolomite, as described in this embodiment, can solve the problem of identifying sedimentary microfacies in bioclastic dolomite under complex geological environments. Compared with traditional manual methods for identifying sedimentary microfacies, this method provides more accurate and reliable identification results for sedimentary microfacies in bioclastic dolomite. Based on the texture and shape characteristics of bioclastic dolomite, it can objectively and accurately determine sedimentary microfacies, significantly reducing the workload of analysts.

[0050] Furthermore, the algorithm-based microfacies identification method for bioclastic dolomite sediments in this embodiment can also effectively predict bioclastic dolomite with low dolomitization, especially for thin-layered Permian dolomite in marine environments. Compared with the traditional method, which relies on manual identification when predicting and identifying thin-layered Permian dolomite in marine environments, the identification method of this application has fewer problems with differences in identification results due to different identification personnel.

[0051] Furthermore, the identification method of this embodiment can be repeated an unlimited number of times, that is, the identification method of this application is repeatable, and the identification result obtained each time is the same, avoiding the problem of different results obtained by different people in the traditional method, thereby effectively improving the accuracy of identification of bioclastic dolomite.

[0052] Example 2 This embodiment provides an algorithm-based method for identifying sedimentary microfacies in bioclastic dolomite, applicable to the identification of sedimentary microfacies in bioclastic dolomite with relatively complex geological structures. The method specifically includes the following steps: Step 1: Collect bioclastic dolomite samples, analyze them, and obtain the structure and characteristics of the bioclastic dolomite samples.

[0053] In this step, petrographic, geophysical, and geochemical analysis methods are used to obtain the structure and characteristics of the bioclastic dolomite samples, including the bioclastic content, bioclastic type, cement characteristics, geophysical characteristics, and geochemical characteristics.

[0054] Step 2: Extract the features from the structure and properties in Step 1.

[0055] In this step, since the texture and structural features of the bioclastic dolomite sample are relatively clear, structural and morphological features, as well as time series features, frequency domain features, and statistical features, were selected as the basis for the next step of model construction.

[0056] This involves using image analysis software to extract features from scanning electron microscope images, including grain shape, size, and distribution; calculating the relative mineral content from X-ray diffraction data; and extracting information about rock structure, such as bedding, fractures, and contact relationships, based on thin section analysis. The extracted structural and property features include mineral content, grain content, bioclastic content, geophysical properties, and geochemical properties.

[0057] Step 3: Select a machine learning algorithm based on the features obtained in Step 2.

[0058] In this step, the machine learning algorithm is a random forest algorithm. Random forests are an ensemble algorithm that constructs multiple decision trees and uses the average or majority vote of their outputs to make the final decision. This method can effectively reduce overfitting of decision trees and generally provides better prediction accuracy.

[0059] The specific operating steps are as follows: I. Multiple known bioclastic dolomite sedimentary microfacies correspond to multiple bioclastic dolomites. For each bioclastic dolomite, its texture and shape features, time series features, and frequency domain information are extracted. Then, each bioclastic dolomite is judged using a decision tree algorithm. The judgment result is compared with the corresponding bioclastic dolomite sedimentary microfacies to determine the deviation between the judgment result and the corresponding bioclastic dolomite sedimentary microfacies. The decision tree algorithm is adjusted according to the deviation. Then, each bioclastic dolomite is compared with the bioclastic dolomite sedimentary microfacies again according to the adjusted decision tree algorithm until the comparison result is within the preset range. The adjusted decision tree algorithm is then used as a decision tree.

[0060] II. Multiple known bioclastic dolomite sedimentary microfacies correspond to multiple bioclastic dolomites. For each bioclastic dolomite, its texture and shape features are extracted. Then, each bioclastic dolomite is judged using a decision tree algorithm. The judgment result is compared with the corresponding bioclastic dolomite sedimentary microfacies to determine the deviation between the judgment result and the corresponding bioclastic dolomite sedimentary microfacies. Based on the deviation, the decision tree algorithm is adjusted. Then, each bioclastic dolomite is compared with the bioclastic dolomite sedimentary microfacies again according to the adjusted decision tree algorithm until the comparison result is within the preset range. The adjusted decision tree algorithm is then used as a decision tree.

[0061] III. Multiple known bioclastic dolomite sedimentary microfacies correspond to multiple bioclastic dolomites. For each bioclastic dolomite, the time series characteristics and frequency domain information of the bioclastic dolomite are extracted. Then, each bioclastic dolomite is judged using a decision tree algorithm. The judgment result is compared with the corresponding bioclastic dolomite sedimentary microfacies to determine the deviation between the judgment result and the corresponding bioclastic dolomite sedimentary microfacies. According to the deviation, the decision tree algorithm is adjusted. Then, each bioclastic dolomite is compared with the bioclastic dolomite sedimentary microfacies again according to the adjusted decision tree algorithm until the comparison result is within the preset range. The adjusted decision tree algorithm is then used as a decision tree.

[0062] In this embodiment, a random forest machine learning algorithm is used to construct multiple decision trees for bioclastic dolomite, and the final decision is made by averaging their outputs or by majority voting. This method can effectively reduce overfitting of decision trees and generally provides better prediction accuracy.

[0063] Step 4: Use known bioclastic dolomite sedimentary microfacies data as a training set to train a machine learning algorithm and obtain a microfacies recognition model.

[0064] The training set is a subset of data used to train models in machine learning and deep learning projects. The main goal of the training set is to fine-tune the model's parameters using this data, enabling the model to learn the basic patterns and relationships within the data.

[0065] The specific operating method includes the following steps: S4.1 For each bioclastic dolomite, construct multiple decision trees, and then use multiple known bioclastic dolomite sedimentary microfacies data as training sets. Multiple known bioclastic dolomite sedimentary microfacies correspond to multiple bioclastic dolomites, thereby extracting the texture and shape features, time series features and frequency domain information of each bioclastic dolomite. S4.2 Then, each bioclastic dolomite is analyzed using multiple decision trees. The results of the multiple decision trees are summarized and averaged. The average result is then compared with the corresponding bioclastic dolomite sedimentary microfacies to determine the deviation between the average result and the corresponding bioclastic dolomite sedimentary microfacies. Based on this deviation, the multiple decision trees are adjusted. S4.3 Then, according to the adjusted decision trees, each bioclastic dolomite is compared with the bioclastic dolomite sedimentary microfacies until the comparison results are within the preset range. The adjusted decision trees are then sorted out to form a random forest, which is used as the microfacies identification model.

[0066] Step 5: Apply the microfacies identification model to the actual identification of microfacies in bioclastic dolomite deposits.

[0067] Among them, the obtained microfacies identification model can be applied to the microfacies of bioclastic dolomite sediments with relatively complex geological structures, thereby enabling accurate identification of the sedimentary microfacies of bioclastic dolomite.

[0068] The algorithm-based sedimentary microfacies identification method for bioclastic dolomite in this embodiment can solve the problem of identifying sedimentary microfacies in bioclastic dolomite under complex geological environments. Compared with traditional manual methods for identifying sedimentary microfacies, this method provides more accurate and reliable results for identifying sedimentary microfacies in bioclastic dolomite. Based on the texture and shape characteristics of bioclastic dolomite, sedimentary microfacies can be objectively and accurately determined, significantly reducing the workload of analysts.

[0069] Furthermore, the algorithm-based microfacies identification method for bioclastic dolomite sediments in this embodiment can effectively predict bioclastic dolomite with low dolomitization, especially for thin-layered Permian dolomite in marine environments. Compared to traditional methods, which rely on manual identification when predicting and identifying thin-layered Permian dolomite in marine environments, this method has fewer issues with differences in identification results due to different personnel. Moreover, this method can be repeated an unlimited number of times, meaning it is repeatable and yields the same results each time, avoiding the problem of different results obtained by different personnel in traditional methods. This effectively improves the accuracy of bioclastic dolomite identification.

[0070] Specifically, for dolomite with a low degree of dolomitization in the Permian system, existing methods rely on manual identification. This method is not only labor-intensive but also prone to inconsistencies in identification results among different personnel. The identification method proposed in this application, however, allows for repeatable identification with consistent results each time, and can completely replace manual identification. This enables more accurate identification of dolomite with a low degree of dolomitization and effectively reduces the labor intensity.

[0071] Example 3 This embodiment provides an algorithm-based method for identifying sedimentary microfacies in bioclastic dolomite, applicable to the identification of sedimentary microfacies in bioclastic dolomite with relatively complex geological structures. The method specifically includes the following steps: Step 1: Collect bioclastic dolomite samples, analyze them, and obtain the structure and characteristics of the bioclastic dolomite samples.

[0072] In this step, petrographic, geophysical, and geochemical analysis methods are used to obtain the structure and characteristics of the bioclastic dolomite samples, including the properties, distribution, sedimentary environment, internal structure, geophysical characteristics, geochemical characteristics, grain content, bioclastic content, and mineral content of the bioclastic dolomite.

[0073] Step 2: Extract the features from the structure and properties in Step 1.

[0074] In this step, since the time series characteristics of bioclastic dolomite are relatively clear and easy to determine, time series and other features, as well as texture and shape features, frequency domain features, and statistical features, were selected as the basis for the next step of model construction.

[0075] This involves using image analysis software to extract features such as grain shape, size, and distribution from scanning electron microscope images; calculating the relative mineral content from X-ray diffraction data; and extracting information about rock structure, such as bedding, fractures, and contact relationships, based on thin section analysis. The extracted structural and property features include, but are not limited to, mineral content, grain content, bioclastic content, geophysical properties, and geochemical properties.

[0076] Step 3: Select a machine learning algorithm based on the features obtained in Step 2.

[0077] In this step, the machine learning algorithm is a Support Vector Machine (SVM) algorithm. Support Vector Machine (SVM) is a classification and regression algorithm. For classification problems, the goal of SVM is to find a hyperplane that maximizes the separation of data points between the two classes. SVM is particularly suitable for high-dimensional data, and in some cases, when the data is linearly inseparable, SVM can handle non-linear data by using the so-called kernel trick.

[0078] The specific operating steps are as follows: S3.1 First, determine whether the time series characteristics of bioclastic dolomite are linearly separable. If the time series characteristics of bioclastic dolomite are linearly separable, then extend the time series of bioclastic dolomite from two dimensions to a multidimensional space to form a hyperplane of the time series. S3.2 Then, the temporal characteristics of multiple bioclastic dolomite rocks are separated by a hyperplane; S3.3 Then, by measuring the distance from the hyperplane to the temporal characteristics of each bioclastic dolomite, the sedimentary microfacies of the bioclastic dolomite are determined.

[0079] Among them, there is a corresponding relationship between the sedimentary microfacies of dolomite and the distance between time characteristics and the hyperplane.

[0080] Step 4: Use known bioclastic dolomite sedimentary microfacies data as a training set to train a machine learning algorithm and obtain a microfacies recognition model.

[0081] The training set is a subset of data used to train models in machine learning and deep learning projects. The main goal of the training set is to fine-tune the model's parameters using this data, enabling the model to learn the basic patterns and relationships within the data.

[0082] Specifically, multiple known bioclastic dolomite sedimentary microfacies correspond to multiple bioclastic dolomites. For each bioclastic dolomite, time-series features are extracted, resulting in multiple time-series features. These features are then combined to form a hyperplane. The sedimentary microfacies of the bioclastic dolomite are determined based on the distance between the hyperplane and the time-series features. If the deviation between the determined sedimentary microfacies and the actual sedimentary microfacies of the bioclastic dolomite is within a set deviation range, the determination result is considered relatively accurate, and the entire algorithm can be used as a microfacies identification model.

[0083] If the deviation between the determined sedimentary microfacies of the bioclastic dolomite and the actual sedimentary microfacies of the bioclastic dolomite is not within the set deviation range, then the hyperplane is adjusted, that is, the hyperplane is re-determined, and then the distance between the hyperplane and the time series features is re-determined to determine the sedimentary microfacies of the bioclastic dolomite, until the deviation between the determined sedimentary microfacies of the bioclastic dolomite and the actual sedimentary microfacies of the bioclastic dolomite is within the set deviation range.

[0084] Step 5: Apply the microfacies identification model to the actual identification of microfacies in bioclastic dolomite deposits.

[0085] Among them, the obtained microfacies identification model can be applied to the microfacies of bioclastic dolomite sediments with relatively complex geological structures, thereby enabling accurate identification of the sedimentary microfacies of bioclastic dolomite.

[0086] The algorithm-based sedimentary microfacies identification method for bioclastic dolomite in this embodiment can solve the problem of identifying sedimentary microfacies in bioclastic dolomite under complex geological environments. Compared with traditional manual methods for identifying sedimentary microfacies, this method provides more accurate and reliable results for identifying sedimentary microfacies in bioclastic dolomite. Based on the texture and shape characteristics of bioclastic dolomite, sedimentary microfacies can be objectively and accurately determined, significantly reducing the workload of analysts.

[0087] Furthermore, the algorithm-based microfacies identification method for bioclastic dolomite in this embodiment can effectively predict bioclastic dolomite with low dolomitization, especially for thin-layered Permian dolomite in marine environments. Compared to traditional methods, which rely on manual identification when predicting and identifying thin-layered Permian dolomite in marine environments, the identification method of this application rarely suffers from differences in identification results due to different personnel. Moreover, the identification method of this application can be repeated an unlimited number of times, meaning it is repeatable and yields the same identification result each time. This avoids the problem of different personnel obtaining different results in traditional methods, thereby effectively improving the accuracy of bioclastic dolomite identification.

[0088] In addition, the identification method provided in this application can also be used to identify dolomite with a high degree of dolomitization, thereby improving the accuracy of dolomite identification and reducing the labor intensity of manual labor.

[0089] Example 4 This embodiment provides an algorithm-based method for identifying sedimentary microfacies in bioclastic dolomite, applicable to the identification of sedimentary microfacies in bioclastic dolomite with relatively complex geological structures. The method specifically includes the following steps: Step 1: Collect bioclastic dolomite samples, analyze them, and obtain the structure and characteristics of the bioclastic dolomite samples.

[0090] In this step, petrographic, geophysical, and geochemical analysis methods are used to obtain the structure and characteristics of the bioclastic dolomite samples, including the properties, distribution, sedimentary environment, internal structure, geophysical characteristics, geochemical characteristics, grain content, bioclastic content, and mineral content of the bioclastic dolomite.

[0091] Step 2: Extract the features from the structure and properties in Step 1.

[0092] In this step, the texture of the bioclastic dolomite sample is relatively clear, and at least texture and shape features are selected.

[0093] This involves using image analysis software to extract features such as grain shape, size, and distribution from scanning electron microscope images; calculating the relative mineral content from X-ray diffraction data; and extracting information about rock structure, such as bedding, fractures, and contact relationships, based on thin section analysis. The extracted structural and property features include, but are not limited to, mineral content, grain content, bioclastic content, geophysical properties, and geochemical properties.

[0094] Step 3: Select a machine learning algorithm based on the features obtained in Step 2.

[0095] In this step, the machine learning algorithm is a neural network machine learning algorithm. A neural network is an algorithm that mimics the working mechanism of the human brain. It consists of many simple units (called "neurons"), which are organized in layers and connected to each other. The neural networks used in deep learning have many layers and are called deep neural networks. Neural networks and deep learning have shown excellent performance in areas such as image and speech recognition and natural language processing, but they can also be applied to other types of data.

[0096] Step 4: Use known bioclastic dolomite sedimentary microfacies data as a training set to train a machine learning algorithm and obtain a microfacies recognition model.

[0097] The training set is a subset of data used to train models in machine learning and deep learning projects. The main goal of the training set is to fine-tune the model's parameters using this data, enabling the model to learn the basic patterns and relationships within the data.

[0098] The specific operating method includes the following steps: S4.1 First, the bioclastic dolomite is trained based on texture and shape features, that is, the texture and shape features of multiple known bioclastic dolomite are input into the neural network model; S4.2 Then, the sedimentary microfacies of the bioclastic dolomite identified by the neural network model are output and compared with the actual sedimentary microfacies of the bioclastic dolomite. If the difference between the output sedimentary microfacies and the actual sedimentary microfacies of the bioclastic dolomite is within the preset difference range, it indicates that the result of the neural network model is accurate and the neural network model can be used as the microfacies identification model. If the difference between the output sedimentary microfacies and the actual sedimentary microfacies of the bioclastic dolomite is not within the preset difference range, the neural network model is adjusted, for example, by resetting the values ​​of the convolutional layers of the neural network model.

[0099] S4.3 Input the texture and shape features of multiple known bioclastic dolomite rocks into the adjusted neural network model. The adjusted neural network model will continue to output the identified sedimentary microfacies of bioclastic dolomite. Then, the output sedimentary microfacies of bioclastic dolomite are compared with the actual sedimentary microfacies of bioclastic dolomite. If the difference between the output sedimentary microfacies of bioclastic dolomite and the actual sedimentary microfacies of bioclastic dolomite is within the preset difference range, the adjusted neural network model is used as the microfacies identification model. If the difference between the output sedimentary microfacies of bioclastic dolomite and the actual sedimentary microfacies of bioclastic dolomite is not within the preset difference range, the neural network model is adjusted until the difference between the output sedimentary microfacies of bioclastic dolomite and the actual sedimentary microfacies of bioclastic dolomite is within the preset difference range.

[0100] In this embodiment, a neural network machine learning algorithm is used to quickly determine the sedimentary microfacies of bioclastic dolomite, thereby improving the efficiency of determining the sedimentary microfacies of bioclastic dolomite.

[0101] Step 5: Apply the microfacies identification model to the actual identification of microfacies in bioclastic dolomite deposits.

[0102] Among them, the obtained microfacies identification model can be applied to the microfacies of bioclastic dolomite sediments with relatively complex geological structures, so as to accurately identify the sedimentary microfacies of bioclastic dolomite, and also to make rapid identification.

[0103] The algorithm-based sedimentary microfacies identification method for bioclastic dolomite in this embodiment can solve the problem of identifying sedimentary microfacies in bioclastic dolomite under complex geological environments. Compared with traditional manual methods for identifying sedimentary microfacies, this method provides more accurate and reliable identification results for sedimentary microfacies in bioclastic dolomite, and has higher identification efficiency. Based on the texture and shape characteristics of bioclastic dolomite, sedimentary microfacies can be objectively and accurately determined, significantly reducing the workload of analysts.

[0104] Furthermore, the algorithm-based microfacies identification method for bioclastic dolomite in this embodiment can effectively predict bioclastic dolomite with low dolomitization, especially for thin-layered Permian dolomite in marine environments. Compared to traditional methods, which rely on manual identification for predicting and identifying thin-layered Permian dolomite in marine environments, this method has fewer issues with inconsistent identification results due to different personnel. Moreover, this method can be repeated an unlimited number of times, ensuring repeatability and consistent results each time. This avoids the problem of different results obtained by different personnel in traditional methods, effectively improving the accuracy of bioclastic dolomite identification and consequently enhancing the accuracy of subsequent geological predictions based on the identification results.

[0105] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications or equivalent changes made to the above embodiments based on the technical essence of the present invention shall fall within the protection scope of the present invention.

Claims

1. An algorithm-based method for identifying sedimentary microfacies in bioclastic dolomite, characterized in that, Includes the following steps: S1. Collect bioclastic dolomite samples, analyze them, and obtain the structure and characteristics of the bioclastic dolomite samples; S2. Extract the features from the structure and properties in step S1; S3. Select a machine learning algorithm based on the features obtained in step S2; S4. Using known bioclastic dolomite sedimentary microfacies data as a training set, a machine learning algorithm is trained to obtain a microfacies recognition model. S5. Apply the microfacies identification model from step S4 to the actual identification of microfacies in bioclastic dolomite deposits.

2. The algorithm-based method for identifying sedimentary microfacies in bioclastic dolomite according to claim 1, characterized in that: In step S1, the methods for analyzing the collected bioclastic dolomite samples include petrographic, geophysical, and geochemical analysis.

3. The algorithm-based method for identifying sedimentary microfacies in bioclastic dolomite according to claim 1, characterized in that: In step S1, the structure of the bioclastic dolomite sample includes biotype, grain structure, mineral content, grain content, bioclastic type, cement characteristics, and bioclastic content.

4. The algorithm-based method for identifying sedimentary microfacies in bioclastic dolomite according to claim 1, characterized in that: In step S1, the characteristics of the bioclastic dolomite sample include mineral content, cementation form, geophysical properties and geochemical properties, as well as time series characteristics, frequency domain characteristics, statistical characteristics and texture and shape characteristics.

5. The algorithm-based method for identifying sedimentary microfacies in bioclastic dolomite according to claim 1, characterized in that: In step S2, the features include particle content, bioclastic content, geophysical properties, and geochemical properties.

6. The algorithm-based method for identifying sedimentary microfacies in bioclastic dolomite according to claim 1, characterized in that: In step S3, the machine learning algorithms include decision trees, random forests, support vector machines, and neural network algorithms.

7. The algorithm-based method for identifying sedimentary microfacies in bioclastic dolomite according to claim 6, characterized in that: In step S4, when the machine learning algorithm is a decision tree learning algorithm, the following steps are included: S4.

1. Multiple known bioclastic dolomite sedimentary micro-correspondences correspond to multiple bioclastic dolomites. For each bioclastic dolomite, extract its texture and shape features, time series features, and frequency domain information. S4.2 Next, each bioclastic dolomite is judged by the decision tree algorithm. The judgment result is compared with the corresponding bioclastic dolomite sedimentary microfacies to determine the deviation between the judgment result and the corresponding bioclastic dolomite sedimentary microfacies. The decision tree algorithm is then adjusted according to the deviation. S4.3 Then, according to the adjusted decision tree algorithm, each bioclastic dolomite is compared with the bioclastic dolomite sedimentary microfacies again until the comparison results are within the preset range. The adjusted decision tree algorithm is then used as the microfacies identification model.

8. The algorithm-based method for identifying sedimentary microfacies in bioclastic dolomite according to claim 4, characterized in that: In step S4, when the machine learning algorithm is the random forest learning algorithm, the following steps are included: S4.1 For each bioclastic dolomite, construct multiple decision trees, and then use multiple known bioclastic dolomite sedimentary microfacies data as training sets. Multiple known bioclastic dolomite sedimentary microfacies correspond to multiple bioclastic dolomites. For each bioclastic dolomite, extract the texture and shape features, time series features and frequency domain information of the bioclastic dolomite. S4.2 Next, each bioclastic dolomite is analyzed using multiple decision trees. The results of these decision trees are then aggregated and averaged. The average result is then compared with the corresponding bioclastic dolomite sedimentary microfacies to determine the deviation between the average result and the corresponding bioclastic dolomite sedimentary microfacies. Based on this deviation, the multiple decision trees are adjusted. S4.

3. Then, according to the adjusted decision trees, each bioclastic dolomite is compared with the bioclastic dolomite sedimentary microfacies until the comparison results are within the preset range. The adjusted decision trees are then sorted out to form a random forest, which is used as the microfacies identification model.

9. The algorithm-based method for identifying sedimentary microfacies in bioclastic dolomite according to claim 4, characterized in that: In step S4, when the machine learning algorithm is a support vector machine learning algorithm, the following steps are included: S4.

1. Multiple known bioclastic dolomite sedimentary microstructures correspond to multiple bioclastic dolomites. For each bioclastic dolomite, the time series features of that bioclastic dolomite are extracted, and then multiple time series features are obtained. These multiple time series features are then combined to form a hyperplane. S4.

2. Then, based on the distance between the hyperplane and the time series features, the sedimentary microfacies of the bioclastic dolomite are determined. If the deviation between the determined sedimentary microfacies of the bioclastic dolomite and the actual sedimentary microfacies of the bioclastic dolomite is within the set deviation range, it indicates that the determined result is relatively accurate, and the entire algorithm can be used as a microfacies identification model.

10. The algorithm-based method for identifying sedimentary microfacies in bioclastic dolomite according to claim 4, characterized in that: In step S4, when the machine learning algorithm is a neural network algorithm learning algorithm, the following steps are included: S4.1 First, the texture and shape features of several known bioclastic dolomite are input into the neural network model, and the model is trained based on the texture and shape features. S4.2 Then, the sedimentary microfacies of the bioclastic dolomite identified by the neural network model are output and compared with the actual sedimentary microfacies of the bioclastic dolomite. If the difference between the output sedimentary microfacies of the bioclastic dolomite and the actual sedimentary microfacies of the bioclastic dolomite is within the preset difference range, it indicates that the identification result of the neural network model is accurate and the neural network model can be used as the microfacies identification model; otherwise, the neural network model needs to be adjusted. S4.3 Then, input the texture and shape features of multiple known bioclastic dolomite into the adjusted neural network model, and repeat step S4.2 until the difference between the sedimentary microfacies of bioclastic dolomite output by the adjusted neural network model and the actual sedimentary microfacies of bioclastic dolomite is within the preset difference range. The resulting neural network model is used as the microfacies recognition model.

Images