Multi-parameter integrated oil exploration data analysis method

Abstract

The invention relates to a multi-parameter integrated oil exploration data analysis method, which comprises the steps of collecting original data in an oil exploration process, the original data comprising original seismic data, original logging data and original geological data; cleaning, denoising, aligning and normalizing the original data to obtain a multi-parameter data set; carrying out dimensionality reduction on the multi-parameter data set to obtain a dimensionality-reduced feature data set; the feature data set after dimension reduction is input into a support vector machine, an oil and gas reservoir feature classification result is obtained, and the support vector machine passes through a historical multi-parameter data set and a corresponding oil and gas reservoir category; and according to the oil and gas reservoir feature classification result and the feature data set after dimension reduction, constructing a three-dimensional geologic model and analyzing reservoir distribution and oil and gas reservoir features to obtain a reserve prediction result. According to the method, earthquake, logging and geological data are effectively integrated, and the accuracy of oil and gas reservoir feature recognition and the reliability of reserve prediction are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of oil exploration, and particularly to a multi-parameter integrated oil exploration data analysis method. Background Art

[0002] During the process of oil exploration, the collected raw data usually includes various types, such as seismic data, logging data, geological data, etc. These data are easily affected by factors such as environmental noise and equipment errors during the collection process, resulting in uneven data quality. However, in the data preprocessing stage, how to effectively remove noise without losing key information, and how to accurately correct time deviation and amplitude difference, is a technical difficulty.

[0003] And due to the complexity and uncertainty of geological conditions, how to accurately predict reserves and production capacity is still a technical problem. Especially in multi-parameter data sets, the interaction and influence mechanism between different parameters have not been fully clarified, and how to establish an accurate mathematical model to describe these relationships is the focus and difficulty of current research. Summary of the Invention

[0004] The purpose of the present invention is to provide a multi-parameter integrated oil exploration data analysis method, which effectively integrates seismic, logging and geological data, and improves the accuracy of hydrocarbon reservoir feature identification and the reliability of reserve prediction.

[0005] To achieve the above purpose, the present invention provides the following solution:

[0006] A multi-parameter integrated oil exploration data analysis method, comprising:

[0007] Collecting raw data during the process of oil exploration, wherein the raw data includes raw seismic data, raw logging data and raw geological data;

[0008] Cleaning, denoising, aligning and normalizing the raw data to obtain a multi-parameter data set;

[0009] Reducing the dimension of the multi-parameter data set to obtain a dimension-reduced feature data set;

[0010] Inputting the dimension-reduced feature data set into a support vector machine to obtain a hydrocarbon reservoir feature classification result, wherein the support vector machine is trained by a historical multi-parameter data set and the corresponding hydrocarbon reservoir categories;

[0011] Constructing a three-dimensional geological model and analyzing reservoir distribution according to the hydrocarbon reservoir feature classification result and the dimension-reduced feature data set to obtain a reserve prediction result.

[0012] Optionally, cleaning the raw data includes:

[0013] If the number of data points in the original data exceeds a preset threshold range, it is marked as an outlier;

[0014] According to the marked outliers, the corresponding data points are removed from the original data;

[0015] The mean interpolation method is used to fill in the missing data points after removing the outliers, and the cleaned original data is obtained.

[0016] Optionally, denoising the original data includes: decomposing the cleaned original data by wavelet transform method, identifying high-frequency noise components, filtering out noise interference, and obtaining a denoised data set.

[0017] Optionally, aligning the original data includes: calculating the time deviation using the cross-correlation algorithm, correcting the time axis of the denoised seismic data, and obtaining time-aligned seismic data and logging data;

[0018] According to the time-aligned seismic data, logging data, and geological data, a time-aligned data set is obtained.

[0019] Optionally, normalizing the original data includes:

[0020] Using the min-max normalization method to calculate the minimum and maximum values of the amplitudes for the time-aligned seismic data and logging data;

[0021] According to the minimum and maximum values of the amplitudes, the amplitude values are mapped to a preset unified interval to obtain an amplitude-normalized data set;

[0022] Using an interpolation algorithm to convert the depth sequence of the logging data in the amplitude-normalized data set into a time sequence to obtain the multi-parameter data set.

[0023] Optionally, dimensionality reduction of the multi-parameter data set to obtain a dimensionality-reduced feature data set includes:

[0024] Using the principal component analysis method to extract data features and determine the eigenvectors;

[0025] According to the eigenvectors, calculate the covariance matrix, and solve for the eigenvalues and eigenvectors;

[0026] Sort according to the eigenvalue magnitudes and select the top K eigenvectors to form a dimensionality reduction matrix;

[0027] Multiply the dimensionality reduction matrix and the multi-parameter data set to obtain the dimensionality-reduced feature data set.

[0028] Optionally, constructing a three-dimensional geological model includes: based on the classification result of the reservoir characteristics and the dimensionality-reduced feature dataset, constructing the three-dimensional geological model by establishing a stratigraphic grid model, wherein during the modeling process, a stochastic simulation method is used to simulate the spatial distribution of reservoir physical properties parameters including porosity and permeability.

[0029] Optionally, analyzing the reservoir distribution and obtaining the reserve prediction result includes:

[0030] Analyzing the reservoir distribution characteristics according to the three-dimensional geological model to obtain the reservoir distribution analysis result;

[0031] Dividing the reserve calculation units according to the reservoir distribution analysis result;

[0032] Obtaining reserve-related data according to drilling, well testing and logging data, and the classification result of the reservoir characteristics, wherein the reserve-related data includes the oil layer division standard, the effective thickness of a single well, the crude oil density, the effective porosity, the oil saturation, and the oil-bearing area;

[0033] Obtaining the reserve prediction result according to the reserve-related data.

[0034] The beneficial effects of the present invention are as follows: The present invention discloses a multi-parameter integrated oil exploration data analysis method. This method first performs outlier screening and noise filtering on the original seismic, logging and geological data, and then unifies the multi-source data into the same structure through time alignment and amplitude normalization processing. Then, principal component analysis is used to reduce the data dimension and extract key features. Finally, a support vector machine algorithm is used to train a classification model to identify non-linear relationships, and a three-dimensional geological model is constructed in combination with the geological modeling method to achieve reserve prediction. Through multi-step data processing and machine learning algorithms, the present invention effectively integrates seismic, logging and geological data, improves the accuracy of reservoir characteristics identification and the reliability of reserve prediction, and provides an important basis for oil and gas exploration and development decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0036] Figure 1 It is a flowchart of a multi-parameter integrated oil exploration data analysis method according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0038] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0039] In oil exploration, geophysical exploration (abbreviated as geophysical prospecting) is the most effective, scientific, and economical exploration method. With the continuous improvement of the exploration and development degree of mining areas, the exploration geological targets have changed from mainly looking for large-scale structural oil and gas reservoirs to mainly focusing on fine structural description and lithologic oil and gas reservoir prediction. The exploration targets are becoming more and more refined, and the exploration conditions are becoming more and more complex. This requires the acquisition accuracy of geophysical prospecting technology to be continuously improved (i.e., "high-precision geophysical prospecting") to meet the new exploration accuracy requirements.

[0040] As Figure 1 shown, this embodiment provides a multi-parameter integrated oil exploration data analysis method, including:

[0041] Collecting the original data in the process of oil exploration, where the original data includes original seismic data, original logging data, and original geological data;

[0042] Cleaning, denoising, aligning, and normalizing the original data to obtain a multi-parameter data set;

[0043] Reducing the dimension of the multi-parameter data set to obtain a dimension-reduced feature data set;

[0044] Inputting the dimension-reduced feature data set into a support vector machine to obtain an oil and gas reservoir feature classification result, where the support vector machine is trained through historical multi-parameter data sets and corresponding oil and gas reservoir categories;

[0045] According to the oil and gas reservoir feature classification result and the dimension-reduced feature data set, constructing a three-dimensional geological model and analyzing the reservoir distribution and oil and gas reservoir features to obtain a reserve prediction result.

[0046] Specifically, seismic data, logging data, and geological data are easily affected by factors such as environmental noise and equipment errors during the acquisition process, resulting in uneven data quality. First, it is necessary to clean the original data, remove outliers and noise interference, and ensure the integrity and reliability of the data. At the same time, since there may be differences in the time base and amplitude range of different data sources, it is necessary to perform time correction and amplitude normalization on the data to ensure the consistency of the data in the time axis and numerical range.

[0047] In the multi-parameter integration stage, different types of data have different physical meanings and data structures. How to effectively fuse these heterogeneous data to form a unified multi-parameter data set is another technical challenge. For example, seismic data usually exists in the form of time series, while logging data is a depth sequence. In this embodiment, data of different dimensions are aligned and matched, and valuable information is extracted from the massive data.

[0048] Furthermore, cleaning the original data includes:

[0049] If the data points in the original data exceed the preset threshold range, they are marked as outliers;

[0050] According to the marked outliers, the corresponding data points are removed from the original data;

[0051] The mean interpolation method is used to fill in the missing data points after removing the outliers, and the cleaned original data is obtained.

[0052] Specifically, the acquisition of seismic data, logging data, and geological data involves different acquisition methods. Seismic data is obtained by artificial seismic sources to generate seismic waves and record the reflection information of the underground medium on the seismic waves; logging data is measured by downhole detection tools for the physical properties of rocks; geological data includes lithology, structure, and other characteristics. For example, the seismic data of a certain oilfield includes information such as reflection time and amplitude, logging data records parameters such as acoustic travel time and density, and geological data describes the lithology distribution of the formation. Data merging needs to consider the sampling intervals and coordinate systems of different data. For example, in a certain logging curve, if the measured value exceeds this range, it is marked as an anomaly. For example, in acoustic travel time data, the normal range is 50 - 100 microseconds / foot, and the data points outside this range need to be marked and removed. Mean interpolation is an effective method for dealing with missing data. For the removed outliers, the average value of its adjacent normal data points can be used to fill them. For example, if the density value of a certain section of the logging curve is abnormal, it can be replaced by the average value of the adjacent normal density values above and below. Outlier detection ensures data quality, interpolation and standardization achieve data unity, and ultimately provide a reliable basis for oil and gas exploration and development.

[0053] Furthermore, denoising the original data includes: decomposing the cleaned original data by wavelet transform method, identifying high-frequency noise components, filtering out noise interference, and obtaining the denoised data set.

[0054] Furthermore, aligning the original data includes: calculating the time deviation using the cross-correlation algorithm, correcting the time axis of the denoised seismic data, and obtaining the time-aligned seismic data and logging data;

[0055] According to the time-aligned seismic data, logging data, and geological data, a time-aligned data set is obtained.

[0056] Specifically, the cross-correlation algorithm determines the time deviation by calculating the correlation degree between two data sequences. There will be a time reference difference during the acquisition of seismic data and logging data. For example, a certain seismic profile shows that the carbonate rock layer is at a depth of one thousand meters at 0.5 seconds, while the logging instrument measures the actual depth of this layer at 0.6 seconds, indicating a time deviation of 0.1 seconds. This deviation can be accurately identified through cross-correlation calculation, and the seismic data time axis can be corrected accordingly.

[0057] Furthermore, the normalization of the original data includes:

[0058] Using the min-max normalization method, calculate the minimum and maximum values of the amplitudes for the time-aligned seismic data and logging data.

[0059] According to the minimum and maximum values of the amplitudes, map the amplitude values to a preset unified interval to obtain a dataset with normalized amplitudes.

[0060] Use the interpolation algorithm to convert the depth sequence of the logging data in the amplitude-normalized dataset into a time sequence to obtain a multi-parameter dataset.

[0061] Specifically, when performing numerical mapping on seismic data and logging data, for example, the amplitude range of seismic data is from -1000 to +2000, and the range of logging data is from 50 to 150. Through normalization, the two types of data can be uniformly mapped to the interval from 0 to 1, making the data comparable. The normalized data can better reflect the relative relationship between different types of data. The interpolation algorithm is the key technology for realizing the conversion of the depth sequence of logging data to the time sequence. Cubic spline interpolation is commonly used to maintain the smoothness of the data, so that the converted logging data is aligned with the seismic data in the time domain. For example, the sampling interval of logging data in a certain oilfield is 0.2 meters, while the sampling interval of seismic data is 2 milliseconds. Through interpolation, the logging data can be resampled to the same time points for subsequent analysis.

[0062] Furthermore, dimensionality reduction of the multi-parameter dataset to obtain the reduced-dimensional feature dataset includes:

[0063] Use the principal component analysis method to extract data features and determine the eigenvectors.

[0064] According to the eigenvectors, calculate the covariance matrix and solve for the eigenvalues and eigenvectors.

[0065] Sort according to the eigenvalue magnitudes and select the top K eigenvectors to form a dimensionality reduction matrix.

[0066] Multiply the dimensionality reduction matrix by the multi-parameter dataset to obtain the reduced-dimensional feature dataset.

[0067] Specifically, principal component analysis retains the main features of data through dimensionality reduction. Assuming that the normalized data contains features in twenty dimensions, it can be found through calculating the covariance matrix that the eigenvalues corresponding to the first five eigenvectors account for 95% of the total variance, indicating that these five eigenvectors already contain the main information of the data. Therefore, these five eigenvectors can be selected to form a dimensionality reduction matrix, greatly reducing the data dimension. The calculation of the covariance matrix is an effective tool for analyzing the correlation of multi-parameter data. Taking petroleum exploration as an example, if there are parameters such as reservoir porosity, permeability, and resistivity, a covariance matrix can be constructed to reveal the correlations between the parameters. For example, porosity and permeability are usually positively correlated, and there is also a corresponding relationship between oil saturation and resistivity logging curves. By calculating the eigenvalues and eigenvectors, the dominant factors can be identified.

[0068] Furthermore, constructing a three-dimensional geological model includes: according to the classification results of hydrocarbon reservoir characteristics and the dimensionality-reduced feature dataset, a three-dimensional geological model is constructed by establishing a stratigraphic grid model, where during the modeling process, spatial distribution simulation of reservoir physical properties parameters including porosity and permeability is carried out through stochastic simulation methods.

[0069] Furthermore, analyzing the reservoir distribution and hydrocarbon reservoir characteristics to obtain the reserve prediction result includes:

[0070] Analyzing the reservoir distribution characteristics according to the three-dimensional geological model to obtain the reservoir distribution analysis result;

[0071] Dividing the reserve calculation units according to the reservoir distribution analysis result;

[0072] Obtaining reserve-related data according to drilling, well testing, logging data, and hydrocarbon reservoir characteristics classification results, and the reserve-related data includes oil layer division criteria, effective thickness of a single well, crude oil density, effective porosity, oil saturation, and oil-bearing area;

[0073] Obtaining the reserve prediction result according to the reserve-related data.

[0074] Specifically, the reserve calculation units are divided from bottom to top vertically.

[0075] Obtaining the reserve prediction result according to the reserve-related data includes:

[0076]

[0077] Nz = Nρ0

[0078] Where, N is the geological reserve of crude oil; Nz is the geological reserve of crude oil; A o is the oil-bearing area; h is the effective thickness; is the effective porosity; S oi is the original oil saturation; B 0i is the formation volume factor of crude oil; ρ0 is the crude oil density.

[0079] The acquisition of the oil-bearing area includes:

[0080] For the divided calculation units, relying on the drilling, well testing and logging of the same horizon in adjacent areas, data is collected.

[0081] Based on the collected data, horizon calibration, full 3D stereoscopic fine interpretation and well-seismic combination are used to determine the trap structure.

[0082] According to the classification results of the characteristics of the oil and gas reservoir, the oil-bearing area is delineated according to the isobath of the trap structure.

[0083] The acquisition of reserve-related data includes: for each calculation unit, the effective thickness isolines are drawn based on the effective thickness of a single well and the distribution of the oil layer, and each unit determines the effective thickness by using the effective thickness isoline weighing method. The porosity interpretation model is selected in combination with core analysis data. The effective porosity of a single well is weighted by thickness, and the porosity of the same horizon on the plane is determined by the average of well points. The oil saturation of a single well is determined by using the volume weighing method.

[0084] The embodiments described above are only descriptions of the preferred embodiments of the present invention, and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A multi-parameter integrated oil exploration data analysis method, characterized in that, Including: Collecting original data during the petroleum exploration process, where the original data includes original seismic data, original logging data, and original geological data; Cleaning, denoising, aligning, and normalizing the original data to obtain a multi-parameter data set; Reducing the dimension of the multi-parameter data set to obtain a feature data set after dimension reduction; Inputting the feature data set after dimension reduction into a support vector machine to obtain an oil and gas reservoir feature classification result, where the support vector machine is trained through historical multi-parameter data sets and corresponding oil and gas reservoir categories; Constructing a three-dimensional geological model based on the oil and gas reservoir feature classification result and the feature data set after dimension reduction, and analyzing the reservoir distribution to obtain a reserve prediction result.

2. The multi-parameter integrated oil exploration data analysis method according to claim 1, wherein Cleaning the original data includes: If the data points in the original data exceed the preset threshold range, they are marked as outliers; Removing the corresponding data points from the original data according to the marked outliers; Using the mean interpolation method to fill the missing data points after removing the outliers to obtain the cleaned original data.

3. The multi-parameter integrated oil exploration data analysis method according to claim 2, characterized in that Denoising the original data includes: decomposing the cleaned original data through the wavelet transform method, identifying high-frequency noise components, and filtering out noise interference to obtain a denoised data set.

4. The multi-parameter integrated oil exploration data analysis method according to claim 3, wherein Aligning the original data includes: calculating the time deviation using the cross-correlation algorithm and correcting the time axis of the denoised seismic data to obtain time-aligned seismic data and logging data; Obtaining a time-aligned data set based on the time-aligned seismic data, logging data, and geological data.

5. The multi-parameter integrated oil exploration data analysis method according to claim 4, characterized in that Normalizing the original data includes: Using the min-max normalization method to calculate the minimum and maximum values of the amplitudes for the time-aligned seismic data and logging data; Mapping the amplitude values to a preset unified interval according to the minimum and maximum values of the amplitudes to obtain an amplitude-normalized data set; Using an interpolation algorithm to convert the depth sequence of the logging data in the amplitude-normalized data set into a time sequence to obtain the multi-parameter data set.

6. The multi-parameter integrated oil exploration data analysis method according to claim 1, wherein Reducing the dimension of the multi-parameter data set to obtain a feature data set after dimension reduction includes: Using the principal component analysis method to extract data features and determine the eigenvectors; Calculating the covariance matrix according to the eigenvectors, and solving for the eigenvalues and eigenvectors; Sorting according to the eigenvalue magnitudes and selecting the top K eigenvectors to form a dimension reduction matrix; Multiplying the dimension reduction matrix and the multi-parameter data set to obtain the feature data set after dimension reduction.

7. The multi-parameter integrated oil exploration data analysis method according to claim 1, characterized in that, Constructing a three-dimensional geological model includes: constructing the three-dimensional geological model by establishing a stratigraphic grid model based on the oil and gas reservoir feature classification result and the feature data set after dimension reduction. During the modeling process, the spatial distribution of reservoir physical properties parameters including porosity and permeability is simulated through a stochastic simulation method.

8. The multi-parameter integrated oil exploration data analysis method according to claim 1, wherein, Analyzing the reservoir distribution to obtain a reserve prediction result includes: Analyzing the reservoir distribution characteristics based on the three-dimensional geological model to obtain a reservoir distribution analysis result; Dividing reserve calculation units according to the reservoir distribution analysis result; According to the drilling, well testing and logging data and the classification results of the oil and gas reservoir characteristics, obtain the data related to reserves, where the data related to reserves includes the oil layer division standard, the effective thickness of a single well, the crude oil density, the effective porosity, the oil saturation, and the oil-bearing area; According to the data related to reserves, obtain the reserve prediction result.

Images