CN114817356B

CN114817356B - A method for fusing oilfield core data and well logging data

Info

Publication number: CN114817356B
Application number: CN202110116109.1A
Authority: CN
Inventors: 李春雷; 张林凤; 靳彩霞; 许立伟; 赵蕾; 刘建涛; 姜兴兴; 马青; 杨河山; 张明安
Original assignee: China Petroleum and Chemical Corp; Exploration and Development Research Institute of Sinopec Shengli Oilfield Co
Current assignee: China Petroleum and Chemical Corp; Exploration and Development Research Institute of Sinopec Shengli Oilfield Co
Priority date: 2021-01-28
Filing date: 2021-01-28
Publication date: 2024-12-24
Anticipated expiration: 2041-01-28
Also published as: CN114817356A

Abstract

The invention relates to an oilfield exploration data processing method, in particular to an oilfield core data and logging data fusion method. The method utilizes a density-based DBSCAN algorithm to perform respective clustering analysis on core data and logging data and obtain a clustering result, finds a common well set, uses each coring well in the common well set as a label class, adopts a supervised learning K-nearest neighbor algorithm to subdivide the corresponding logging class until all logging data are classified, ensures that each logging data can be classified as the closest coring well, and realizes fusion application of logging data and core data with two scales and different densities. The method can be popularized and applied to the research of oil reservoir engineering and geologic modeling, and effectively improves the accuracy and efficiency of geologic research.

Description

Oil field core data and logging data fusion method

Technical Field

The invention relates to an oilfield exploration data processing method, in particular to an oilfield core data and logging data fusion method.

Background

Oilfield research belongs to interdisciplinary and multi-field research, and relates to links of geophysical exploration, geological research, drilling, logging, development testing and the like, and data acquisition means and technologies have obvious differences, so that data expression forms are different. Meanwhile, in geological research, different layers of the reservoir, such as different levels of geological units, are divided to form data with different scales. The oil field core data and the logging data are two types of data with different expression forms, dimensions, spaces and existence densities. The core data are various attribute value expressions of the geological object core, the longitudinal data expression is in a centimeter level range, the well logging data are various attribute expressions of the geological object well layer, the longitudinal data expression is in a kilometer range, and the distribution of the core data and the well logging data on an oilfield plane is basically distributed according to the density ratio of 1:15 of the well number. But both types of data are the most direct reflection of the evidence reservoir conditions, and both types of data are the most fundamental requirements for oilfield fusion applications. The core work of the fusion is core homing.

Core homing refers to unifying the drilling depth of a core taken by drilling to a certain standard depth, and the standard depth used in scientific research work at present generally refers to logging depth.

Chinese patent application CN108227036A discloses a method for resetting a fine-grained sedimentary rock core, which comprises the following steps of 1) data collection and data preparation, 2) establishing a lithology logging plate by using a sensitive logging curve, 3) selecting and calibrating a typical lithology section and a lithology combination section by using logging lithology, carrying out overall preliminary resetting of the core by comparing analysis with a core centimeter level fine description or a core scanning image, and 4) carrying out local fine resetting of the preliminary resetting core by using analysis test data and the core centimeter level fine description. The invention aims to provide a method for homing a fine-grained sedimentary rock core based on multi-parameter combination of test data-core description-core image-lithology combination-logging curve, which provides a feasible technical system for reducing the exploration risk of unconventional oil gas in a fine-grained sedimentary rock layer and improving the exploration success rate of high-yield and high-efficiency well positions.

The Chinese patent application CN110134918A discloses an automatic core homing method based on a sliding window method, which sequentially comprises a preliminary depth matching step, a sliding window constructing step, a correlation coefficient calculating step and a screening step, wherein the sliding window method is adopted, and the preliminary depth matching of logging data and experimental data measured by the core and the correlation coefficient calculation of porosity parameters when the porosity parameters slide along a logging curve are combined to perform intelligent calculation processing of the automatic core homing, so that the system error between the cable depth and the drill rod depth is reduced to the maximum extent, and the calculation efficiency and accuracy of the core homing are improved.

At present, rock cores are unified on a logging curve by adopting the corresponding relation between lithology and electrical property, and the method is complicated, is greatly influenced by human factors and has relatively poor accuracy. Later, the core ground gamma test technology is developed, and is a test technology developed aiming at the reuse and deep research of core data. The natural gamma homing of the rock core ground is to perform a natural gamma test of the rock core ground, and the measured curve and the logging natural gamma curve are used for comparing and homing, so that the natural gamma homing is less interfered by human factors, the accuracy is greatly improved, and the method is convenient and quick. However, a great disadvantage is that the number of wells for performing the surface core gamma scan is very small, and the ratio of the number of wells to the coring number is 1:80. Therefore, the core homing work is completed in time in a large proportion, two kinds of data are fused, and the meaning is not great, because the proportion of core gamma data and logging data fusion is 1:1200. Geological research is a global effort that requires obtaining sufficient reservoir property data reflected by the core, and combining the logging data for research.

Therefore, a method must be found to generalize the core data to all logging data, so as to achieve comprehensive fusion of reservoir physical properties reflected by the core and reservoir strata reflected by the logging data.

Disclosure of Invention

The invention mainly aims to provide a novel method for fusing oil field core data and logging data. The method comprises the steps of fully utilizing the existing full-quantity logging data and coring data of the oil field, optimally applying a density-based method DBSCAN algorithm and a supervised learning K-nearest neighbor algorithm, establishing a logging data and coring data fusion application classification model based on business condition constraints corresponding to multiple attributes of reservoir depth, and forming a whole set of coring data obtained by all logging single wells of the oil field, so that the whole set becomes a necessary path for obtaining reservoir physical property data in geological research. The method has comprehensive research data and can accurately reflect the current situation of the oil reservoir and the reservoir.

In order to achieve the above purpose, the present invention adopts the following technical scheme:

The invention provides a fusion method of oilfield core data and logging data, which comprises the following steps:

1) Establishing a machine learning data training set, collecting full-quantity logging data and core data, and respectively establishing a logging data training set for cluster analysis and a core data training set for cluster analysis;

2) Based on the data training set, performing cluster analysis by using a DBSCAN algorithm to respectively obtain a logging data cluster analysis result set and a core data cluster analysis result set;

3) Determining a common cluster well set;

4) Labeling each well in a common cluster well set;

5) Further classifying by adopting a supervised K-nearest neighbor algorithm;

6) Completing supervised learning classification of all logging data;

7) And obtaining core data with the closest reservoir physical properties.

Further, in step 1), the full-scale logging data is collected, the types of the common logging curves are analyzed, the data units are unified, data cleaning is completed, the data are standardized and normalized, and a logging data training set for cluster analysis is established.

Further, in the step 1), the core data comprise hole, seepage, saturation, carbon conventional experimental data, oil-water phase seepage data, wettability data, sensitivity analysis data and mercury-pressing data, and the data are cleaned and normalized to establish a core data training set for cluster analysis.

Further, in step 2), based on the logging data training set, further performing feature engineering analysis, performing cluster analysis by using a DBSCAN algorithm to obtain a logging data cluster analysis result set A= { a ₁,a₂,……a_m }, and based on the core data training set, further performing feature engineering analysis, performing cluster analysis by using the DBSCAN algorithm to obtain a core data cluster analysis result set B= { B ₁,b₂,……b_n }.

Further, in step 3), logging data a _i is taken, coring data b _j with common well numbers is automatically found, and a common well set C _i＝{c₁,c₂,……c_k is established, wherein k < n, and k < m.

Further, in step 4), each well in the well set C _i＝{c₁,c₂,……c_k is taken as a class and used as a label for further subdivision of the next step of logging data according to the core data, and a label set is established.

Further, the coordinate data of each well is used as constraint conditions, the wells in the well logging a _i are further classified according to the C _i label set by adopting a supervised K-nearest neighbor algorithm, the classification of all the wells in the well logging a _i is completed, and iterative optimization is continuously carried out to reach a target value.

Further, steps 3) through 5) are repeated until each log finds the closest coring well.

Compared with the prior art, the invention has the following advantages:

(1) The method can realize the fusion of two types of data with different scales and different distribution densities, namely logging data and core data, in real time. The method is not limited by test conditions, oil development stages and the like, is not limited by traditional empirical formulas, realizes the fusion of two types of data in real time, and realizes the on-demand integration requirement of basic data of oilfield geological research business.

(2) The physical property data complete set of all wells of the oil field can be formed by adopting the method. The method comprises the steps of performing classification learning twice, performing unsupervised learning for the first time to obtain primary classification, performing supervised learning for the second time to obtain more accurate classification, adding professional constraint to enable all wells of the oil field to obtain the closest physical parameter value, enabling physical data of the oil field to be expanded from point to surface, and being capable of being used as new data resources of the oil field. The method solves the most important problem of serious data loss of reservoir physical properties in the research process of intelligent oil fields and intelligent oil reservoirs.

(3) The method can be popularized and applied to the research of oil reservoir engineering and geologic modeling, and effectively improves the accuracy and efficiency of geologic research.

Drawings

The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the invention.

Fig. 1 is a flowchart of a method for fusing oilfield core data and logging data according to an embodiment of the present invention.

Detailed Description

It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the invention. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the present invention. As used herein, the singular forms also are intended to include the plural forms unless the context clearly indicates otherwise, and furthermore, it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, and/or combinations thereof.

In order to enable those skilled in the art to more clearly understand the technical scheme of the present invention, the technical scheme of the present invention will be described in detail with reference to specific embodiments.

Example 1

As shown in fig. 1, the method for fusing oilfield core data and logging data comprises the following steps:

Step 110, a machine learning data training set is established. The method comprises the steps of collecting full-quantity logging data, analyzing common logging curve types including SP, R4, RIID/RIIS, CAL, AC, GR, POR and the like, unifying data units, completing data cleaning, standardizing and normalizing the data, establishing a logging data training set for cluster analysis, simultaneously analyzing core data, mainly various physical property data of the core including hole, seepage, saturation, carbon conventional experimental data, oil-water seepage data, wettability data, sensitivity analysis data, mercury-pressing data and the like, cleaning and normalizing the data, and establishing a core data training set for cluster analysis.

And 120, performing cluster analysis by using a DBSCAN algorithm. Based on the core data training set, further performing feature engineering analysis, performing cluster analysis by using a DBSCAN algorithm to obtain a cluster analysis result set A= { a ₁,a₂,……a_m }, and performing feature engineering analysis, performing cluster analysis by using the DBSCAN algorithm to obtain a cluster analysis result set B= { B ₁,b₂,……b_n }.

Step 130. Determining a common cluster well set. An automatic corresponding program is required to be written, logging data a _i are acquired, coring data b _j with common well numbers are automatically found, and a common well set C _i＝{c₁,c₂,……c_k is established, wherein k is smaller than n and k is smaller than m.

Step 140. Labeling each well in the common cluster well set. Taking each well in the well set C _i＝{c₁,c₂,……c_k as a class, taking the class as a label of the next step of well logging data further subdivided according to the core data, and establishing a label set.

And 150, further classifying by adopting a supervised K-neighbor algorithm. The coordinate data of each well is used as constraint conditions, the wells in the well logging a _i are further classified according to the C _i label set by adopting a supervised K-nearest neighbor algorithm, and the classification of all the wells in the well logging a _i is completed, namely the corresponding specific core wells. And carrying out algorithm evaluation according to the algorithm evaluation indexes such as EVS, MAE, MSE, R and the like, and continuously carrying out iterative optimization to reach the target value.

Step 160. All log data completes the supervised learning classification. Repeating the third to fifth steps until all log data are classified as N-T (where 0< = T < N), i.e. each log is found to be the closest cored well.

Step 170. Each well with logging data obtains core data with closest reservoir properties. Logging utilizes core data of the coring wells with the same classification to realize fusion of two types of data with different scales and different distribution densities.

The two types of data fusion processes in the method realize automation, the traditional method is basically in a manual state, the traditional method generally comprises the steps of borrowing the drawing material of the core logging, picking up or lowering the value character record from the core picking tool, manually comparing the difference value between the logging depth and the coring depth on the core logging map to obtain core homing values with different depths if the character record is not available, manually homing each sample depth of the core, and finally corresponding the sample homing depth to the logging depth to realize fusion of the core logging drawing material and the core logging drawing material. Compared with the analysis result of the traditional method, the method has the advantages that the prediction accuracy of the model of the method reaches 92% of that of the traditional method, but the application efficiency of data fusion is improved by 8-9 times.

The above examples are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above examples, and any other changes, modifications, substitutions, combinations, and simplifications that do not depart from the spirit and principle of the present invention should be made in the equivalent manner, and the embodiments are included in the protection scope of the present invention.

Claims

1. A method for fusing oilfield core data and well logging data, characterized in that it comprises the following steps:

1) Establish a machine learning data training set, collect all the well logging data and core data, and establish a well logging data training set for cluster analysis and a core data training set for cluster analysis respectively;

2) Based on the data training set, cluster analysis is performed using the DBSCAN algorithm to obtain a set of well logging data cluster analysis results and a set of core data cluster analysis results;

3) determine the common clustering wells;

4) Label each well in the common cluster well set;

5) Use supervised K-nearest neighbor algorithm for further classification;

6) Complete supervised learning classification of all logging data;

7) Obtain core data that is closest to reservoir properties;

In step 2), based on the well logging data training set, feature engineering analysis is further performed, and cluster analysis is performed using the DBSCAN algorithm to obtain a well logging data cluster analysis result set A = {a ₁ , a ₂ , ... a _m }; based on the core data training set, feature engineering analysis is further performed, and cluster analysis is performed using the DBSCAN algorithm to obtain a core data cluster analysis result set B = {b ₁ , b ₂ , ... b _n };

In step 3), determine the common clustered well set; write an automatic correspondence program, take the logging data a _i class, automatically find the coring data b _j class with the common well number, and establish a common well set C _i ={c ₁ ,c ₂ ,...c _k }, k<n, k<m;

In step 5, the supervised K-nearest neighbor algorithm is used for further classification; the coordinate data of each well is used as a constraint condition, and the wells in the logging a _i class are further classified according to the _Ci label set using the supervised K-nearest neighbor algorithm to complete the classification of all wells in the logging a _i class, that is, the corresponding specific core wells, and the algorithm is evaluated according to the algorithm evaluation indicators such as EVS, MAE, MSE, R2, and iterative optimization is continuously performed to reach the target value;

In step 6, supervised learning classification is performed on all well logging data; steps 3) to 5) are repeated until the closest coring well is found for each well logging.

2. The method according to claim 1 is characterized in that, in step 1), all logging data are collected, common logging curve types are analyzed, data units are unified, data cleaning is completed, and the data is standardized and normalized to establish a logging data training set for cluster analysis.

3. The method according to claim 1 is characterized in that, in step 1), the core data includes porosity, permeability, saturation, carbon conventional experimental data, oil-water phase permeability data, wettability data, sensitivity analysis data, and mercury injection data; the data is cleaned and normalized to establish a core data training set for cluster analysis.

4. The method according to claim 1, characterized in that in step 4), each well in the well set _Ci = { _c1 , _c2 , ..., _ck } is taken as a class, and used as a label for further subdividing the logging data according to the core data in the next step to establish a label set.