Fluid inclusion analysis method based on big data structure

Abstract

The invention discloses a fluid inclusion analysis method based on a big data structure. The method comprises the following steps: step 1, collecting a sample; step 2, data cleaning and arrangement; step 3, data processing and analysis; 4, identifying the position and morphological characteristics of the inclusion in the image by using YOLOv10; step 5, oil and gas filling period and paleo-pressure evolution analysis: step 5-1, data analysis and feature extraction; 5-2, the oil and gas charging period is recognized; 5-3, palaeopressure change curves of different oil and gas charging periods are drawn, the trend of palaeopressure change is captured, and the future pressure evolution trend is predicted; 5-4, according to the pressure evolution trend and the oil and gas filling period, an oil and gas favorable gathering area is deduced; and step 6, oil and gas accumulation time calibration. According to the method, a large amount of fluid inclusion data can be processed and analyzed, the data can be rapidly processed and analyzed, and the efficiency is improved on the premise that the analysis accuracy and reliability are guaranteed.

Description

Technical Field

[0001] The invention belongs to the field of oil and gas accumulation, and particularly relates to a fluid inclusion analysis method based on a large data structure. Background Art

[0002] In the field of oil and gas accumulation, studying the phase, homogenization temperature, and freezing point temperature of fluid inclusions is of great significance for understanding and predicting the accumulation and migration of oil and gas. The current fluid inclusion analysis technology is mainly based on traditional data statistical methods, and its main analysis process is:

[0003] 1. Data collection: Use a high-power microscope to observe fluid inclusions in a specific layer in a specific area, and use a hot and cold stage to collect data on the uniform temperature and freezing point temperature of inclusions in host minerals of different periods.

[0004] 2. Data preprocessing: Preprocess the collected data, including eliminating invalid data, pressure calculation, etc., to ensure the accuracy and reliability of the data.

[0005] 3. Data storage: The preprocessed data is manually stored in Excel for subsequent analysis and processing.

[0006] 4. Data analysis: Use Excel to analyze the stored data, extract the key features of fluid inclusions, analyze the oil and gas filling periods, etc.

[0007] 5. Result display: The analysis results are displayed in a visual form to facilitate user observation and analysis.

[0008] This analysis method generally has the following problems:

[0009] 1. Low data processing efficiency: Traditional fluid inclusion analysis techniques often rely on manual operations and limited data processing capabilities, resulting in slow processing of large amounts of sample data and unable to meet the needs of rapid exploration.

[0010] 2. Insufficient accuracy of analysis results: Due to the limitations of data processing methods, traditional analysis techniques often have errors in determining oil and gas filling periods, reservoir characteristics, and oil and gas accumulation time, affecting the accuracy and reliability of exploration results.

[0011] 3. Insufficient data mining: Traditional fluid inclusion analysis is often limited to specific regions and specific layers. It lacks data collection, processing and analysis of multiple regions and multiple layers. It has defects in the three-dimensional exploration of multiple layers at the basin level and cannot meet the needs of later oil and gas exploration.

[0012] Therefore, a new technology is needed to effectively analyze the stages, homogenization temperature, and freezing point temperature of fluid inclusions, and then analyze the stages of oil and gas accumulation and the migration paths of oil and gas. Summary of the invention

[0013] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a fluid inclusion analysis method based on a large data structure that can process and analyze a large amount of fluid inclusion data, quickly process and analyze the data, and improve efficiency while ensuring the accuracy and reliability of the analysis.

[0014] The object of the present invention is achieved through the following technical solution: a fluid inclusion analysis method based on big data structure, comprising the following steps:

[0015] Step 1, sample collection: Use a microscope to carefully observe the host minerals of the fluid inclusions, analyze the morphological characteristics of the inclusions, and identify the fluid inclusions of the host minerals of different periods; and collect the uniform temperature, freezing point temperature, inclusion morphology, inclusion size, gas-liquid ratio, paleopressure and chemical composition content of the fluid inclusions of the host minerals of different periods;

[0016] Step 2: Data cleaning and organization, including the following sub-steps:

[0017] Step 2-1, manually input the collected raw data, select the inclusions of the host mineral period with a gas-liquid ratio of less than 20%, and then select the homogenization temperature, freezing point temperature, paleopressure and chemical composition content of the inclusions of different host mineral periods as valid data;

[0018] Step 2-2: Mark the data exceeding the 94% confidence interval as outliers and remove them;

[0019] Step 2-3, eliminate data points with the same gas-liquid ratio, uniform temperature and freezing point temperature;

[0020] Step 2-4: Standardize the cleaned and organized data using the same labels;

[0021] Step 2-5: Use distributed database technology to establish an original database using the processed data;

[0022] Step 3: Data processing and analysis, including the following sub-steps:

[0023] Step 3-1: Preprocessing of semi-structured data in the database: Use data cleaning technology to correct errors in the data, delete invalid values ​​and missing values ​​in the database and perform average processing;

[0024] Step 3-2: Normalize the pre-processed uniform temperature, freezing point temperature, paleopressure and chemical composition content data respectively. Data normalization Min-Max normalization refers to scaling the data so that it falls into a small specific interval [0, 1]. The specific formula is:

[0025]

[0026] Where: X′ represents the normalized data, X represents the original data, and X max , X min are the maximum and minimum values ​​in the original data respectively;

[0027] Step 3-3, extract features from the normalized homogeneous temperature, freezing point temperature, paleopressure and chemical composition content:

[0028]

[0029] Where: R i,j In the feature extraction process, it represents the feature weight of the jth extraction of the i-th type of data. According to the sparse distribution of the features in space, the weights of the four different features of uniform temperature, freezing point temperature, paleopressure and chemical composition content are determined; R represents the original data; W i,j Indicates the data simplicity attribute, and uses principal component analysis for feature selection; χ(k) indicates the degree of data spatial deviation; k indicates the degree of data outlier;

[0030] Step 4: Identify the location and morphological features of fluid inclusions in the image: Collect a microscopic image of the inclusion to be identified, and use the supervised learning model YOLOv10 to identify the location and morphological features of the fluid inclusions in the image;

[0031] Step 5: Analysis of oil and gas filling stages and paleopressure evolution, including the following sub-steps:

[0032] Step 5-1, data analysis and feature extraction: Compare the morphological characteristics of the inclusions output by YOLOv10 with the morphological characteristics of the inclusions of different host mineral periods stored in the database to determine the mineral period of the inclusions; at the same time, find the homogenization temperature, freezing point temperature, chemical composition content and paleopressure of the inclusions of the mineral period in the database;

[0033] Step 5-2, oil and gas charging period identification;

[0034] The homogenization temperature data of the inclusions are corrected using a regression model based on a neural network: the homogenization temperature, as well as the four input features of freezing point temperature, paleopressure, and chemical composition content are input into a multilayer perceptron, which automatically adjusts the weights of the input features according to the algorithm in step 3-3 and outputs the corrected formation temperature of the inclusions;

[0035] The K-means clustering algorithm is used to cluster the output inclusion formation temperature and identify different oil and gas filling periods;

[0036] Step 5-3, paleopressure evolution analysis: Based on the classification results obtained by clustering algorithm, the ARIMA model, a time series analysis method, is used to draw paleopressure change curves for different oil and gas filling periods, capture the trend of paleopressure changes, and predict the future pressure evolution trend;

[0037] Step 5-4, analysis of favorable oil and gas accumulation areas: Based on the pressure evolution trend and oil and gas filling period, use the spatial analysis tool based on geographic information system to infer favorable oil and gas accumulation areas;

[0038] Step 6, calibration of oil and gas accumulation time, including the following sub-steps:

[0039] Step 6-1, data preparation: import the collected burial history and thermal evolution history data of geological structures into the database; use ETL tools to standardize the data;

[0040] Step 6-2, temperature matching: Use the DTW algorithm to match the homogenization temperature of the inclusion with the formation temperature in the burial history and thermal evolution history; during the matching process, the DTW algorithm will automatically detect the similarity between the formation temperature and the homogenization temperature of the inclusion, and find the best matching point by dynamically adjusting the path, so as to obtain the formation time of the inclusion;

[0041] Step 6-3, determination of oil and gas accumulation time: conduct a comprehensive analysis of the formation time of the inclusions and other geological evidence of oil and gas accumulation, and use the Bayesian inference method to finally determine the oil and gas accumulation time;

[0042] Step 6-4: Result visualization: Use Matplotlib or Plotly to draw temperature-depth graphs and time-temperature graphs to display the calibration results.

[0043] The beneficial effects of the present invention are: being able to process and analyze a large amount of fluid inclusion data, quickly process and analyze the data, and improve efficiency while ensuring the accuracy and reliability of the analysis. The analysis results are displayed in a visual manner, making the results more intuitive and easy to understand. Moreover, the technology is not only applicable to geological research in energy fields such as oil and natural gas, but can also be applied to resource exploration and development in the fields of metal minerals, non-metallic minerals, etc. DETAILED DESCRIPTION

[0044] Definitions of Abbreviations and Key Terms

[0045] 1. Big data structure: Big data structure refers to the algorithms and technologies for processing and storing big data.

[0046] 2. Fluid inclusions: Fluid inclusions are fluid samples of different sizes and shapes that are captured during the formation of minerals, enclosed by mineral crystals, represent the characteristics of the original fluid when the minerals are formed, exist in the lattice defects or voids of the main minerals, and are composed of liquid, gas, solid or a combination of these.

[0047] The technical solution of the present invention is further described below.

[0048] A fluid inclusion analysis method based on big data structure comprises the following steps:

[0049] Step 1, sample collection: Use a microscope to carefully observe the host minerals of the fluid inclusions, identify the type of minerals and analyze their occurrence. Observe the type of inclusions, analyze the morphological characteristics of the inclusions, such as shape, size, and distribution, and identify the fluid inclusions of different periods of host minerals; and collect the homogenization temperature, freezing point temperature, inclusion shape, inclusion size, gas-liquid ratio, paleopressure, and chemical composition content of the fluid inclusions of different periods of host minerals;

[0050] Step 2: Data cleaning and organization, including the following sub-steps:

[0051] Step 2-1, manually input the collected raw data (including host mineral period, homogenization temperature, freezing point temperature, inclusion morphology, inclusion size, gas-liquid ratio, paleopressure and chemical composition content), select the inclusions of the host mineral period with a gas-liquid ratio less than 20%, and then select the homogenization temperature, freezing point temperature, paleopressure and chemical composition content of the inclusions of different host mineral periods as valid data;

[0052] Step 2-2: Use statistical methods (such as box plots, standard deviation method, etc.) to mark data exceeding the 94% confidence interval as outliers and eliminate them;

[0053] Step 2-3, remove data points with the same gas-liquid ratio, uniform temperature and freezing point temperature to ensure the uniqueness of the data;

[0054] Step 2-4: Standardize the cleaned and sorted data using the same labels to ensure data consistency and comparability;

[0055] Step 2-5: Use distributed database technology to establish the original database for the processed data; based on the data type, the data is stored in different nodes in a distributed manner. Multiple independent computing units (nodes) form a system, and the nodes are connected through a network and work together to complete common tasks. Compared with traditional database technology, this technology performs shard storage based on the geographical location and layer of the data source. At the same time, distributed database technology supports efficient data storage, retrieval and management functions, meeting the needs of subsequent analysis.

[0056] Step 3: Data processing and analysis, including the following sub-steps:

[0057] Step 3-1. Preprocessing of semi-structured data in the database: Use data cleaning technology in big data to correct errors in the data, and based on the problems found, delete and average the invalid values ​​and missing values ​​in the database (delete invalid values ​​and fill missing values ​​with the mean).

[0058] Step 3-2: Normalize the pre-processed uniform temperature, freezing point temperature, paleopressure and chemical composition content data, including data normalization, integration of data of the same type, classification of data with the same attributes, etc.; Data normalization Min-Max normalization refers to scaling the data so that it falls into a small specific interval [0, 1]. The specific formula is:

[0059]

[0060] Where: X′ represents the normalized data, X represents the original data, and X max , X min They are the maximum and minimum values ​​in the original data respectively; then data integration is performed: similar data from different sources are merged into a consistent data storage area.

[0061] Step 3-3, extract features from the normalized homogeneous temperature, freezing point temperature, paleopressure and chemical composition content:

[0062]

[0063] Where: R i,j In the feature extraction process, it represents the feature weight of the jth extraction of the i-th type of data. According to the sparse distribution of the features in space, the weights of the four different features of uniform temperature, freezing point temperature, paleopressure and chemical composition content are determined. R represents the original data; W i,j It represents the data parsimony attribute, which refers to the number of attributes retained after some form of simplification or dimensionality reduction. Principal component analysis (PCA) is used for feature selection. χ(k) represents the degree of data space deviation, which refers to the change in the distribution of data in the feature space. The degree of data deviation is evaluated by calculating the distance between data points and using clustering algorithms. k represents the degree of data outlier, which refers to the number of outliers or outliers in the data set that are significantly different from most data points. The statistical method Z-Score is used to identify and calculate outliers.

[0064] Step 4: Identify the location and morphological features of fluid inclusions in the image: Collect microscopic images of the inclusions to be identified, and perform corresponding preprocessing on the microscopic image data based on the Apache Spark data processing framework, including resizing and normalization. Then, use the supervised learning model YOLOv10 to identify the location and morphological features of the fluid inclusions in the image.

[0065] Step 5: Analysis of oil and gas filling stages and paleopressure evolution, including the following sub-steps:

[0066] Step 5-1, data analysis and feature extraction: Compare the morphological characteristics of the inclusions output by YOLOv10 with the morphological characteristics of the inclusions of different host mineral periods stored in the database to determine the mineral period of the inclusions; at the same time, find the homogenization temperature, freezing point temperature, chemical composition content and paleopressure of the inclusions of the mineral period in the database;

[0067] Step 5-2, oil and gas filling period identification: The homogenization temperature data of the inclusions are corrected using a regression model based on a neural network; the homogenization temperature, as well as the four input features of freezing point temperature, paleopressure, and chemical composition content are input into a multilayer perceptron (MLP), which automatically adjusts the weights of the input features according to the algorithm in step 3-3 and outputs the corrected formation temperature of the inclusions. The correction process can obtain a more accurate inclusion formation temperature by eliminating the influence of pressure and considering correction factors of different mineralization environments.

[0068] The K-means clustering algorithm is used to cluster the output inclusion formation temperature and identify different oil and gas charging periods. The MLP-corrected data is automatically divided into several classes corresponding to different oil and gas charging periods using the K-means clustering algorithm. The algorithm automatically classifies the data by calculating the distance from each data point to the center of each class, and continuously optimizes the classification results during the iteration process.

[0069] Step 5-3, paleopressure evolution analysis: Based on the classification results obtained by the aforementioned clustering algorithm, the paleopressure change curves of different oil and gas filling periods are drawn using the ARIMA model, a time series analysis method, to capture the trend of paleopressure changes, predict the future pressure evolution trend, and provide a basis for oil and gas reservoir analysis; cluster analysis directly processes temperature data, but the results are indirectly related to mineral stages and paleopressure. The temperature and paleopressure of inclusions in different mineral stages at the time of formation are key factors reflecting geological history. Cluster analysis helps to infer the formation conditions of these inclusions in different mineral stages by identifying the pattern of temperature data. Using the results of K-means clustering (i.e., different oil and gas filling periods), combine it with the corresponding paleopressure data, and apply time series analysis methods (such as ARIMA models) to generate paleopressure change curves of different oil and gas filling periods. Capture the trend of paleopressure changes, analyze the pressure changes during oil and gas filling in different periods, and provide a basis for further oil and gas reservoir analysis.

[0070] In order to fully understand the oil and gas accumulation process, mineral stages, temperature data and paleopressure data need to be integrated and analyzed. For example, the oil and gas charging stages obtained through cluster analysis can be matched with the mineral stages, and further combined with paleopressure change data to analyze how the geological environment changes under different mineral stages and charging stages.

[0071] Step 5-4, analysis of favorable oil and gas accumulation areas: Based on the pressure evolution trend and oil and gas filling periods, use the spatial analysis tool based on the geographic information system to infer favorable oil and gas accumulation areas; this tool combines the paleopressure data of different periods with the geographical location information to generate a spatial distribution map of oil and gas accumulation, and predicts favorable oil and gas accumulation areas based on the spatial analysis results;

[0072] Step 6, calibration of oil and gas accumulation time, including the following sub-steps:

[0073] Step 6-1, data preparation: import the burial history and thermal evolution history data of the collected geological structures (including time, stratigraphic information, burial depth, geothermal heat flow value, and stratigraphic temperature) into the database; use ETL tools to standardize the data;

[0074] Step 6-2, temperature matching: Use the DTW algorithm to match the homogenization temperature of the fluid inclusion with the formation temperature in the burial history and thermal evolution history; during the matching process, the DTW algorithm automatically detects the similarity between the formation temperature and the homogenization temperature of the inclusion, and dynamically adjusts the path to find the best matching point, thereby obtaining the formation time of the inclusion;

[0075] The Dynamic Time Warping (DTW) algorithm is used to compare the similarity of two time series, even if the two series are not completely aligned on the time axis, they can be effectively compared. In geological analysis, DTW can be used to match the uniform temperature series of fluid inclusions and the formation temperature series to find the best matching path.

[0076] (1) Prepare time series data:

[0077] Homogenized temperature time series: The homogenized temperature data of fluid inclusions is used as a time series. It includes the homogenized temperatures of different inclusions and reflects the geological conditions of different formation time periods.

[0078] Formation temperature time series: The formation temperature history data extracted from geological data forms a time series. It shows the temperature changes of the formation at different time points in the geological history.

[0079] (2) Apply the DTW algorithm:

[0080] Calculate the distance matrix: First, calculate the Euclidean distance matrix between the uniform temperature time series and the formation temperature time series. Each element of the matrix represents the distance between two temperature points.

[0081] Finding the optimal matching path: The DTW algorithm uses dynamic programming to find the path that minimizes the overall distance, reflecting the best match between two time series.

[0082] (3) Matching result analysis:

[0083] Time inference: Through the best matching path found by DTW, the position of the formation temperature corresponding to the uniform temperature on the geological time axis is inferred, and then the formation time of the fluid inclusion is determined.

[0084] Determination of the time of oil and gas accumulation: Based on the correspondence between the formation time of inclusions and related geological events (such as tectonic movements, temperature changes, etc.), the time of oil and gas accumulation can be further inferred.

[0085] Step 6-3, determination of oil and gas accumulation time: conduct a comprehensive analysis of the formation time of the inclusions and other geological evidence of oil and gas accumulation (such as the initial filling time of oil and gas, tectonic movement time, etc.), and use the Bayesian inference method to finally determine the oil and gas accumulation time;

[0086] Step 6-4, Result Visualization: Use Matplotlib or Plotly to draw temperature-depth and time-temperature graphs to display the calibration results. Through interactive visualization tools, users can dynamically adjust the displayed content of the graph to better understand the time process and related parameters of oil and gas accumulation. And the visualization results can be interpreted and evaluated in combination with the knowledge and experience of geological experts. Compare and analyze the characteristics of fluid inclusions in different regions, evaluate the potential and exploration prospects of oil and gas resources, and provide decision support for oil and gas exploration and development.

[0087] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A fluid inclusion analysis method based on big data structure, characterized in that: The following steps are involved: Step 1: Sample collection: Use a microscope to carefully observe the host minerals of the fluid inclusions, analyze the morphological characteristics of the inclusions, and identify the fluid inclusions of the host minerals of different periods; and collect the uniform temperature, freezing point temperature, inclusion morphology, inclusion size, gas-liquid ratio, paleopressure and chemical composition content of the fluid inclusions of the host minerals of different periods; Step 2: Data cleaning and organization, including the following sub-steps: Step 2-1, manually input the collected raw data, select the inclusions of the host mineral period with a gas-liquid ratio of less than 20%, and then select the homogenization temperature, freezing point temperature, paleopressure and chemical composition content of the inclusions of different host mineral periods as valid data; Step 2-2: Mark the data exceeding the 94% confidence interval as outliers and remove them; Step 2-3, eliminate data points with the same gas-liquid ratio, uniform temperature and freezing point temperature; Step 2-4: Standardize the cleaned and organized data using the same labels; Step 2-5: Use distributed database technology to establish an original database using the processed data; Step 3: Data processing and analysis, including the following sub-steps: Step 3-1: Preprocessing of semi-structured data in the database: Use data cleaning technology to correct errors in the data, delete invalid values ​​and missing values ​​in the database and perform average processing; Step 3-2: Normalize the pre-processed uniform temperature, freezing point temperature, paleopressure and chemical composition content data respectively. Data normalization Min-Max normalization refers to scaling the data so that it falls into a small specific interval [0, 1]. The specific formula is: Where: X′ represents the normalized data, X represents the original data, and X max , X min are the maximum and minimum values ​​in the original data respectively; Step 3-3, extract features from the normalized homogeneous temperature, freezing point temperature, paleopressure and chemical composition content: Where: R i,j In the feature extraction process, it represents the feature weight of the jth extraction of the i-th type of data. According to the sparse distribution of the features in space, the weights of the four different features of uniform temperature, freezing point temperature, paleopressure and chemical composition content are determined; R represents the original data; W i,j Indicates the data simplicity attribute, and uses principal component analysis for feature selection; χ(k) indicates the degree of data spatial deviation; k indicates the degree of data outlier; Step 4: Identify the location and morphological features of fluid inclusions in the image: Collect a microscopic image of the inclusion to be identified, and use the supervised learning model YOLOv10 to identify the location and morphological features of the fluid inclusions in the image; Step 5: Analysis of oil and gas filling stages and paleopressure evolution, including the following sub-steps: Step 5-1, data analysis and feature extraction: Compare the morphological characteristics of the inclusions output by YOLOv10 with the morphological characteristics of the inclusions of different host mineral periods stored in the database to determine the mineral period of the inclusions; at the same time, find the homogenization temperature, freezing point temperature, chemical composition content and paleopressure of the inclusions of the mineral period in the database; Step 5-2, oil and gas charging period identification; The homogenization temperature data of the inclusions are corrected using a regression model based on a neural network: the homogenization temperature, as well as the four input features of freezing point temperature, paleopressure, and chemical composition content are input into a multilayer perceptron, which automatically adjusts the weights of the input features according to the algorithm in step 3-3 and outputs the corrected formation temperature of the inclusions; The K-means clustering algorithm is used to cluster the output inclusion formation temperature and identify different oil and gas filling periods; Step 5-3, paleopressure evolution analysis: Based on the classification results obtained by clustering algorithm, the ARIMA model, a time series analysis method, is used to draw paleopressure change curves for different oil and gas filling periods, capture the trend of paleopressure changes, and predict the future pressure evolution trend; Step 5-4, analysis of favorable oil and gas accumulation areas: Based on the pressure evolution trend and oil and gas filling period, use the spatial analysis tool based on geographic information system to infer favorable oil and gas accumulation areas; Step 6, calibration of oil and gas accumulation time, including the following sub-steps: Step 6-1, data preparation: import the collected burial history and thermal evolution history data of geological structures into the database; use ETL tools to standardize the data; Step 6-2, temperature matching: Use the DTW algorithm to match the homogenization temperature of the inclusion with the formation temperature in the burial history and thermal evolution history; during the matching process, the DTW algorithm will automatically detect the similarity between the formation temperature and the homogenization temperature of the inclusion, and find the best matching point by dynamically adjusting the path, so as to obtain the formation time of the inclusion; Step 6-3, determination of oil and gas accumulation time: conduct a comprehensive analysis of the formation time of the inclusions and other geological evidence of oil and gas accumulation, and use the Bayesian inference method to finally determine the oil and gas accumulation time; Step 6-4: Result visualization: Use Matplotlib or Plotly to draw temperature-depth graphs and time-temperature graphs to display the calibration results.

Images