System and method for x-ray diffraction mineral composition analysis based on machine learning using domain knowledge

Abstract

Disclosed are a system and method for X-ray diffraction mineral composition analysis based on machine learning using domain knowledge. The method for X-ray diffraction mineral composition analysis based on machine learning using domain knowledge comprises the steps of: receiving X-ray diffraction data of a mineral as input; normalizing and preprocessing each piece of the input X-ray diffraction data; generating a clustering model using the preprocessed X-ray diffraction data; receiving as input and preprocessing new X-ray diffraction data, and then classifying the preprocessed new X-ray diffraction data as general composition data or specific composition data by using the clustering model; and estimating the mineral composition by analyzing the new X-ray diffraction data classified as general composition data by using a mineral composition analysis model.

Description

Machine learning-based X-ray diffraction mineral composition analysis system and method utilizing domain knowledge

[0001] The present invention relates to a machine learning-based X-ray diffraction mineral composition analysis system and method utilizing domain knowledge.

[0002]

[0003] Due to concerns about global warming and energy security, the development of gas hydrates, which are widely distributed around the world and possess an estimated 3,000 trillion cubic meters of reserves, is becoming increasingly important. In particular, South Korea's net energy import dependence was 88.1% as of 2010, with most of its energy imported from abroad. Consequently, the importance of gas hydrates in South Korea's surrounding areas is also significant. Gas hydrates, primarily composed of methane, emit less carbon dioxide than traditional resources like coal and oil, making them environmentally friendly. However, due to their unique characteristics compared to traditional resources, commercial production of gas hydrates requires extensive research on the geological and fluid properties of the deposits.

[0004] Mineralogical composition data provides information on the formation environment and origin of sediments. Furthermore, analyzing the composition of constituent minerals is crucial for characterizing petroleum reservoirs. Such mineral composition data can be obtained through X-ray diffraction (X-ray diffraction) experiments. X-ray diffraction experiments involve scanning a powdered sample (10 μm or less) with X-rays and measuring the intensity of the diffracted X-rays. The resulting X-ray diffraction intensities are then analyzed by experts using analysis software to determine the mineral composition. While individual minerals exhibit characteristic intensities depending on the X-ray incidence angle, sediments, composed of multiple minerals, exhibit complex patterns, making interpretation highly dependent on experts. Furthermore, when analyzing large quantities of samples, the time consumed by repetitive tasks poses significant limitations.

[0005]

[0006] The present invention provides a machine learning-based X-ray diffraction mineral composition analysis system and method utilizing domain knowledge, which classifies X-ray diffraction data of sediments into general composition data or specific composition data using a clustering model, estimates the composition value using a mineral composition analysis model in the case of general composition data, and allows an expert to analyze the specific composition data.

[0007]

[0008] According to one aspect of the present invention, a machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge performed by an X-ray diffraction mineral composition analysis system is disclosed.

[0009] A machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge according to an embodiment of the present invention includes the steps of receiving X-ray diffraction data of a mineral, preprocessing the input X-ray diffraction data by normalizing it for each data, creating a clustering model using the preprocessed X-ray diffraction data, creating a mineral composition analysis model using the preprocessed X-ray diffraction data, receiving new X-ray diffraction data and preprocessing it, and then classifying the preprocessed new X-ray diffraction data into general composition data or specific composition data using the clustering model, and estimating a mineral composition by analyzing the new X-ray diffraction data classified as general composition data using the mineral composition analysis model.

[0010] The above preprocessing step preprocesses the intensity values ​​included in the X-ray diffraction data for each data using min-max scaling.

[0011] The step of generating the clustering model includes the steps of: separating the preprocessed X-ray diffraction data into training and verification data and test data; learning the clustering model with the training and verification data; selecting a specific composition criterion for performance evaluation of the clustering model using the training and verification data; classifying test data into general composition data and specific composition data according to the selected specific composition criterion using the learned clustering model; generating a confusion matrix using the classification result; selecting precision among confusion matrix performance evaluation indices to determine the optimal number of clusters; and determining the number of clusters with the highest precision of the generated confusion matrix as the optimal number of clusters to generate an optimized clustering model.

[0012] The step of generating the above mineral composition analysis model includes the step of dividing the preprocessed X-ray diffraction data into training data, verification data, and test data, the step of performing deep learning with the training data and the verification data, and the step of optimizing hyper parameters of the deep learning model generated through the deep learning using the verification data and the test data to generate an optimized mineral composition analysis model.

[0013] The new X-ray diffraction data classified as the above-mentioned special composition data are analyzed by experts to estimate the mineral composition.

[0014] The above X-ray diffraction data includes intensity values ​​and mineral composition values.

[0015] According to another aspect of the present invention, a machine learning-based X-ray diffraction mineral composition analysis system utilizing domain knowledge is disclosed.

[0016] A machine learning-based X-ray diffraction mineral composition analysis system utilizing domain knowledge according to an embodiment of the present invention includes a memory for storing commands and a processor for executing the commands, wherein the commands perform an X-ray diffraction composition analysis method including the steps of: receiving X-ray diffraction data of a mineral; preprocessing the input X-ray diffraction data by normalizing them by data; generating a clustering model using the preprocessed X-ray diffraction data; generating a mineral composition analysis model using the preprocessed X-ray diffraction data; receiving new X-ray diffraction data and preprocessing it, and then classifying the preprocessed new X-ray diffraction data into general composition data or specific composition data using the clustering model; and estimating a mineral composition by analyzing the new X-ray diffraction data classified as general composition data using the mineral composition analysis model.

[0017]

[0018] A machine learning-based X-ray diffraction mineral composition analysis system and method utilizing domain knowledge according to an embodiment of the present invention classifies X-ray diffraction data of sediments into general composition data or specific composition data using a clustering model, estimates a composition value using a mineral composition analysis model in the case of general composition data, and allows an expert to analyze the specific composition data.

[0019]

[0020] FIG. 1 is a flowchart schematically illustrating a machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge performed by an X-ray diffraction mineral composition analysis system according to an embodiment of the present invention.

[0021] Figure 2 is a flowchart schematically illustrating detailed steps of step S300 of Figure 1.

[0022] Figure 3 is a flowchart schematically illustrating detailed steps of step S400 of Figure 1.

[0023] FIGS. 4 to 11 are diagrams for explaining a machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge according to the embodiment of the present invention of FIG. 1.

[0024] FIG. 12 is a diagram schematically illustrating the configuration of a machine learning-based X-ray diffraction mineral composition analysis system utilizing domain knowledge according to an embodiment of the present invention.

[0025]

[0026] As used herein, singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, terms such as "consist of" or "include" should not be construed to necessarily include all components or steps described in the specification, and should be construed to mean that some of the components or steps may not be included, or that additional components or steps may be included. In addition, terms such as "part" and "module" described in the specification mean a unit that processes at least one function or operation, which may be implemented by hardware or software, or by a combination of hardware and software.

[0027]

[0028] Hereinafter, various embodiments of the present invention will be described in detail with reference to the attached drawings.

[0029] FIG. 1 is a flowchart schematically illustrating a machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge performed by an X-ray diffraction mineral composition analysis system according to an embodiment of the present invention, FIG. 2 is a flowchart schematically illustrating detailed steps of step S300 of FIG. 1, FIG. 3 is a flowchart schematically illustrating detailed steps of step S400 of FIG. 1, and FIGS. 4 to 11 are diagrams for explaining a machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge according to an embodiment of the present invention of FIG. 1. Hereinafter, a machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge according to an embodiment of the present invention will be described with reference to FIG. 1, with reference to FIGS. 2 to 11.

[0030] At step S100, the X-ray diffraction mineral composition analysis system receives X-ray diffraction data of the mineral.

[0031] Figure 4 shows the locations of eight boreholes from the UBGH-1 and UBGH-2 projects conducted in the Ulleung Basin, East Sea. In the present invention, X-ray diffraction data of 488 data points obtained through the UBGH-1 and UBGH-2 large-scale drilling exploration projects conducted in the Ulleung Basin, East Sea in 2007 and 2010 were used, and these data points were provided by the Gas Hytrate Business Group. These data points are sediment data obtained from a total of eight boreholes, and the borehole locations are as shown in Figure 4, and the number of data points per borehole is as shown in the table below.

[0032]

[0033] Of the total 488 data, 439 (90%) were used for model training, and the remaining 49 (10%) were used for testing.

[0034] X-ray diffraction data consist of 3,100 intensity values ​​(input layer) organized at 0.02° intervals in the incident angle range of 3.01° to 64.99° and 12 mineral composition values ​​(output layer) analyzed from these. Here, the 12 minerals are quartz, albite, opal-A, calcite, muscovite, dolomite, chlorite, kaolinite, illite, pyrite, NaCl, and K-feldspar.

[0035] For example, the intensity values ​​according to the angle of incidence of data No. 1 are as shown in Fig. 5, and the mineral composition values ​​are as shown in the table below.

[0036]

[0037] In step S200, the X-ray diffraction mineral composition analysis system preprocesses the input X-ray diffraction data by normalizing it for each data.

[0038] For example, the intensity values ​​of the input X-ray diffraction data can be preprocessed using the min-max scaling mathematical formula below.

[0039]

[0040] Here, X ms represents the normalized intensity value through min-max scaling, and X max,j and X min,j represent the maximum and minimum intensity values ​​in the jth data, respectively.

[0041] That is, the X-ray diffraction composition analysis system according to an embodiment of the present invention preprocesses the input X-ray diffraction data for each data. This is because the absolute size and exact position of the peak are not important for interpretation, and differences occur due to the mineralogical characteristics of the data and experimental errors, and the ratio and tendency of the intensity values ​​according to the incident angle within a data set are important. Therefore, in the present invention, the X-ray diffraction data is normalized between 0 and 1 to match the scale.

[0042] Figure 6 shows the X-ray diffraction data format according to the preprocessing method. In general, when preprocessing by factor used in machine learning, it can be confirmed that the raw data format of (a) is completely damaged into the format of (b), and when the min-max scaler for each data is applied, it can be seen that the format of the raw data is maintained and the scale is also corrected to 0~1, as in (c). Therefore, it is judged that applying min-max scaling for each data is appropriate only for X-ray diffraction data.

[0043] At step S300, the X-ray diffraction mineral composition analysis system creates a clustering model using preprocessed X-ray diffraction data.

[0044] Below, with reference to Fig. 2, the detailed steps of step S300 will be described.

[0045] In step S310, the X-ray diffraction mineral composition analysis system separates the preprocessed X-ray diffraction data into training and verification data and test data.

[0046] At step S320, the X-ray diffraction mineral composition analysis system trains a clustering model with training and validation data.

[0047] Figure 7 illustrates the process of the k-means clustering algorithm, a type of clustering algorithm. Referring to Figure 7, the k-means clustering algorithm predetermines the number of clusters, k, and then assigns data to each cluster based on the initial center. It then repeatedly updates the cluster average until the center no longer fluctuates, thereby classifying the data closest to the center into a single cluster. Because the results vary depending on the number of clusters, it is important to select the optimal number of clusters for the intended purpose.

[0048] In step S330, the X-ray diffraction mineral composition analysis system selects specific composition criteria for performance evaluation of the clustering model using training and verification data.

[0049] For example, Figure 8 shows a boxplot of the compositional values ​​of 12 minerals and the range of compositional values ​​for general composition data determined using training and validation data. Any value outside this range is classified as an unusual composition.

[0050] In step S340, the X-ray diffraction mineral composition analysis system classifies test data into general composition data and specific composition data using the learned clustering model.

[0051] At step S350, the X-ray diffraction mineral composition analysis system generates a confusion matrix using the classification results.

[0052] At step S360, the X-ray diffraction mineral composition analysis system selects precision among the confusion matrix performance evaluation indices to determine the optimal number of clusters.

[0053] As indicators for quantitatively analyzing clustering results, recall and precision of the confusion matrix are used and can be expressed by the following mathematical formula.

[0054]

[0055] That is, recall is the ratio of the number of data accurately predicted as general composition data among the actual general composition data, and precision is the ratio of the number of data predicted as general composition data among the number of data that are actual general composition data.

[0056] Considering that the purpose of the present invention is to determine whether or not there is specific composition data within a general composition cluster, it is appropriate to determine the number of clusters based on the precision indicating the proportion of actual general composition data among the classified data.

[0057] In step S370, the X-ray diffraction mineral composition analysis system determines the number of clusters with the highest precision of the generated confusion matrix as the optimal number of clusters to generate an optimized clustering model.

[0058] Figure 9 shows the confusion matrix for 49 Ulleung Basin test data. Referring to Figure 9, recall was 100% when the number of clusters was 3 and 92.7% when the number of clusters was 5, and precision was 83.7% when the number of clusters was 3 and 100% when the number of clusters was 5. Considering recall, the optimal number of clusters should be 3, but since precision is a suitable indicator for the present invention, the optimal number of clusters may be 5.

[0059] In step S400, the X-ray diffraction mineral composition analysis system creates a mineral composition analysis model using preprocessed X-ray diffraction data.

[0060] Below, with reference to Fig. 3, the detailed steps of step S400 will be described.

[0061] In step S410, the X-ray diffraction mineral composition analysis system separates the preprocessed X-ray diffraction data into training data, verification data, and test data.

[0062] At step S420, the X-ray diffraction mineral composition analysis system performs deep learning with training data and verification data.

[0063] The table below shows the average and deviation of the coefficient of determination (COD) and mean absolute error (MAE) according to preprocessing methods using a convolutional neural network (CNN), a deep learning algorithm. Feature_CNN and Sample_CNN, respectively, performed factor-wise and data-wise normalization. The COD for data-wise normalization was 0.757, compared to factor-wise normalization, confirming the reasonableness of data-wise normalization.

[0064]

[0065] In step S430, the X-ray diffraction mineral composition analysis system optimizes hyper parameters of a deep learning model generated through deep learning using verification and test data to generate a mineral composition analysis model.

[0066] Again, referring to Fig. 1, steps S500 to S700 will be described.

[0067] At step S500, the X-ray diffraction mineral composition analysis system receives new X-ray diffraction data, preprocesses it, and then classifies the preprocessed new X-ray diffraction data into general composition data or specific composition data using the generated clustering model.

[0068] At step S600, the X-ray diffraction mineral composition analysis system estimates the mineral composition using a mineral composition analysis model generated from new X-ray diffraction data classified as general composition data.

[0069] At step S700, new X-ray diffraction data classified as unusual mineral composition data are analyzed by experts to estimate the mineral composition.

[0070] Figure 10 shows the locations of the UBGH-1 and UBGH-2 projects and the Korea Land boreholes, and Figure 11 shows the confusion matrix results for the additional verification data from Korea Land. Using 54 data points obtained from Korea Land, located in the northern periphery of the Ulleung Basin as shown in Figure 10, we examined the scalability of the clustering model to additional data from nearby areas. As shown in Figure 11, consistent with the previous results, when the number of clusters was 5, the accuracy was 100% compared to 3, confirming that there were no unusual compositional data within the general compositional clusters.

[0071]

[0072] FIG. 12 is a drawing schematically illustrating the configuration of a machine learning-based X-ray diffraction mineral composition analysis system utilizing domain knowledge according to an embodiment of the present invention.

[0073] Referring to FIG. 12, a machine learning-based X-ray diffraction mineral composition analysis system utilizing domain knowledge according to an embodiment of the present invention includes a processor (10), a memory (20), a communication unit (30), and an interface unit (40).

[0074] The processor (10) may be a CPU or semiconductor device that executes processing instructions stored in memory (20).

[0075] The memory (20) may include various types of volatile or non-volatile memory media. For example, the memory (20) may include ROM, RAM, etc.

[0076] For example, the memory (20) can store commands for performing a machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge according to an embodiment of the present invention.

[0077] The communication unit (30) is a means for transmitting and receiving data with other devices through a communication network.

[0078] The interface unit (40) may include a network interface and a user interface for connecting to a network.

[0079]

[0080] Meanwhile, the components of the aforementioned embodiments can be easily understood from a process perspective. That is, each component can be understood as a separate process. Furthermore, the processes of the aforementioned embodiments can be easily understood from the perspective of the device components.

[0081] In addition, the technical contents described above may be implemented in the form of program commands that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program commands, data files, data structures, etc., alone or in combination. The program commands recorded on the medium may be those specially designed and configured for the embodiments or may be known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, and hardware devices specially configured to store and execute program commands, such as ROMs, RAMs, and flash memories. Examples of program commands include not only machine language codes generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter, etc. The hardware devices may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

[0082]

[0083] The above-described embodiments of the present invention are disclosed for the purpose of illustration, and those skilled in the art with common knowledge of the present invention will be able to make various modifications, changes, and additions within the spirit and scope of the present invention, and such modifications, changes, and additions should be considered to fall within the scope of the following patent claims.

Claims

1. In a machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge performed by an X-ray diffraction composition analysis system, Step of inputting X-ray diffraction data of a mineral; A step of preprocessing the input X-ray diffraction data by normalizing it for each data; A step of creating a clustering model using the above preprocessed X-ray diffraction data; A step of creating a mineral composition analysis model using the above preprocessed X-ray diffraction data; A step of inputting new X-ray diffraction data, preprocessing the data, and then classifying the preprocessed new X-ray diffraction data into general composition data or specific composition data using the clustering model; and A machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge, comprising a step of estimating mineral composition by analyzing new X-ray diffraction data classified as the above general composition data using the above mineral composition analysis model.

2. In paragraph 1, The above preprocessing step is, A machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge, characterized in that the intensity values ​​included in the above X-ray diffraction data are preprocessed for each data using min-max scaling.

3. In paragraph 1, The steps for creating the above clustering model are: A step of separating the above preprocessed X-ray diffraction data into training and validation data and test data; A step of training a clustering model using the above training and validation data; A step of selecting a specific composition criterion for performance evaluation of a clustering model using the above training and verification data; A step of classifying test data into general composition data and specific composition data according to the selected specific composition criteria using the learned clustering model; A step of generating a confusion matrix using the above classification results; In order to determine the optimal number of clusters, a step of selecting precision among the confusion matrix performance evaluation indicators; and A machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge, characterized in that it includes a step of generating an optimized clustering model by determining the number of clusters with the highest precision of the generated confusion matrix as the optimal number of clusters.

4. In paragraph 1, The step of creating the above mineral composition analysis model is: A step of separating the above preprocessed X-ray diffraction data into training data, verification data, and test data; A step of performing deep learning with the training data and the verification data; and A machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge, characterized in that it comprises a step of generating an optimized mineral composition analysis model by optimizing hyper parameters of a deep learning model generated through the deep learning performance using the verification data and the test data.

5. In paragraph 1, A machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge, characterized in that new X-ray diffraction data classified as the above-mentioned special composition data is analyzed by an expert to estimate the mineral composition.

6. In paragraph 1, A machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge, characterized in that the above X-ray diffraction data includes intensity values ​​and mineral composition values.

7. In a machine learning-based X-ray diffraction mineral composition analysis system utilizing domain knowledge, memory for storing instructions; and Including a processor that executes the above instructions, The above command is, Step of inputting X-ray diffraction data of a mineral; A step of preprocessing the input X-ray diffraction data by normalizing it for each data; A step of creating a clustering model using the above preprocessed X-ray diffraction data; A step of creating a mineral composition analysis model using the above preprocessed X-ray diffraction data; A step of inputting new X-ray diffraction data, preprocessing the data, and then classifying the preprocessed new X-ray diffraction data into general composition data or specific composition data using the clustering model; and A machine learning-based X-ray diffraction mineral composition analysis system utilizing domain knowledge, characterized in that it performs a machine learning-based X-ray diffraction mineral composition analysis method utilizing domain knowledge, including a step of estimating a mineral composition by analyzing new X-ray diffraction data classified as the general composition data using the mineral composition analysis model.

Images