Methods, computing devices, and storage media for adding labels to well logs

Abstract

The application provides a method, a computing device and a storage medium for adding labels to well logging curves. The method comprises: analyzing a plurality of well logging curves of a target reservoir, and selecting at least one curve meeting a preset condition from the plurality of well logging curves as a key well logging curve according to an analysis result; based on geological knowledge of lithofacies types of the target reservoir at different depths, using a preset classification algorithm, determining, for at least one key well logging curve of the target reservoir at each depth, a probability that each key well logging curve at the depth points to each lithofacies type in the geological knowledge, establishing an association relationship between a lithofacies type corresponding to a maximum probability and the corresponding key well logging curve, and then adding a lithofacies type label to the at least one key well logging curve at different depths. The method is beneficial to improving the speed and accuracy of adding labels to well logging curves, greatly reducing the workload and labor cost of manually adding labels, and is beneficial to the smooth progress of exploration and development.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of reservoir prediction, and particularly relates to a method for adding labels to well logging curves, a computing device and a storage medium. BACKGROUND

[0002] In recent years, some new methods have appeared for clastic reservoir prediction, such as using deep learning methods to predict. In the process of using deep learning methods to carry out reservoir prediction, a large amount of labeled data is needed to train the deep learning model, and these labeled data are generally labeled well logging curves. A relatively new method is to manually mark well logging curves for deep learning training according to relatively clear lithofacies.

[0003] However, for a region that is in the process of exploration and development, if the labels of the well logging curves are manually added, there are mainly two problems: (1) heavy workload, generally there are many wells in the study area, the sampling rate of the well logging curves is high, and the accuracy requirement of adding labels is high, the above three factors directly determine the huge workload of well logging curve marking, and further the progress of well logging curve lithofacies marking directly affects the progress of exploration and development; (2) strong multi-solution, since multiple well logging curves need to be referred to in the process of lithofacies marking, for example, the reservoir of clastic rock is often not completely identified by only one curve, and the well is affected by various factors such as engineering construction and mud invasion, the marking of the well logging curve is difficult, and the marking result is greatly affected by the subjective factors of the technical personnel. Moreover, the marking error of the well logging curve will greatly affect the result of reservoir prediction, resulting in poor prediction effect.

[0004] Therefore, a method for quickly, objectively and effectively adding labels to well logging curves is needed. SUMMARY

[0005] The main purpose of the present application is to provide a method for adding labels to well logging curves, a computing device and a storage medium, so as to improve the speed and intelligence of adding labels to well logging curves.

[0006] In a first aspect, the present application provides a method for adding labels to well logging curves, comprising: analyzing a plurality of well logging curves of a target reservoir, and selecting at least one curve meeting a preset condition from the plurality of well logging curves as a key well logging curve according to an analysis result; based on geological knowledge of lithofacies types of the target reservoir at different depths, using a preset classification algorithm, determining, for at least one key well logging curve of the target reservoir at each depth, a probability that each key well logging curve at the depth points to each lithofacies type in the geological knowledge, establishing an association relationship between a lithofacies type corresponding to a maximum probability and the corresponding key well logging curve, and then adding a lithofacies type label to the at least one key well logging curve at different depths respectively.

[0007] In one embodiment, the method of analyzing the plurality of well logging curves of the target reservoir, and screening at least one curve meeting the preset condition from the plurality of well logging curves as the key well logging curve comprises: performing principal component analysis on the plurality of well logging curves of the target reservoir, determining the information amount contained in each well logging curve, and screening at least one curve whose information amount meets the preset condition from the plurality of well logging curves as the key well logging curve.

[0008] In one embodiment, the method of screening at least one well logging curve whose information amount meets the preset condition from the plurality of well logging curves as the key well logging curve comprises: sorting the plurality of well logging curves according to their respective information amounts from large to small, determining the cumulative proportion of the cumulative information amount of each well logging curve in the sorting to the total information amount of all well logging curves, and taking each well logging curve corresponding to the minimum cumulative proportion reaching the preset proportion threshold as the key well logging curve.

[0009] In one embodiment, the preset classification algorithm comprises a K-Mean unsupervised classification algorithm.

[0010] In one embodiment, after determining the probability of each key well logging curve at the depth pointing to each lithofacies type in the geological understanding, and before establishing the association between the lithofacies type corresponding to the maximum probability and the corresponding key well logging curve, the method further comprises a cross-plot analysis step, which comprises: taking the lithofacies type corresponding to the maximum probability as the lithofacies type of the target reservoir at the depth, thereby obtaining the lithofacies types of the target reservoir at different depths; performing cross-plot analysis on the lithofacies types of the target reservoir at different depths by using at least two well logging curves other than the key well logging curves, determining whether the determined lithofacies types of the target reservoir at different depths are correct according to the results of the cross-plot analysis, and determining the correctness of the determined lithofacies types of the target reservoir at different depths, and when the correctness reaches a preset correctness threshold, establishing the association between the lithofacies type corresponding to the maximum probability and the corresponding key well logging curve for each depth of the target reservoir again.

[0011] In one embodiment, the at least two well logging curves other than the key well logging curves comprise a shear wave velocity curve, a compressional wave velocity curve and a density curve; the cross-plot analysis on the lithofacies types of the target reservoir at different depths by using the at least two well logging curves other than the key well logging curves comprises: determining a Lame constant curve and a Poisson's ratio curve according to the shear wave velocity curve, the compressional wave velocity curve and the density curve, and determining a product curve of the Lame constant curve and the density curve; and performing cross-plot between the Poisson's ratio curve and the product curve for the lithofacies types of the target reservoir at different depths.

[0012] In one embodiment, when the accuracy does not reach the preset accuracy threshold, the classification parameters of the preset classification algorithm are adjusted, the probability of each lithofacies type of each key logging curve of the target reservoir at each depth in the geological understanding is re-determined by using the adjusted preset classification algorithm, and the crossplot analysis step is re-executed.

[0013] In a second aspect, the present application provides a training method of a reservoir lithofacies type prediction model, comprising: adding labels to logging curves by using the method for adding labels to logging curves as described above to generate training samples; training a preset reservoir lithofacies type deep learning model by using a training sample set composed of multiple training samples to obtain a trained reservoir lithofacies type prediction model.

[0014] In a third aspect, the present application provides a method for predicting reservoir lithofacies types, comprising: obtaining multiple logging curves of a target reservoir; and determining the lithofacies types of the target reservoir at different depths based on the multiple logging curves by using the trained reservoir lithofacies type prediction model obtained by using the method as described above.

[0015] In a fourth aspect, the present application provides a computing device, comprising: a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the steps of the method for adding labels to logging curves as described above or the steps of the training method of a reservoir lithofacies type prediction model as described above or the steps of the method for predicting reservoir lithofacies types as described above are executed.

[0016] In a fifth aspect, the present application provides a storage medium storing a computer program, and when the computer program is executed by a processor, the steps of the method for adding labels to logging curves as described above or the steps of the training method of a reservoir lithofacies type prediction model as described above or the steps of the method for predicting reservoir lithofacies types as described above are executed.

[0017] By the method of the present application, the logging curves are classified based on the geological understanding of the reservoir lithofacies types, and the logging curves are labeled with the lithofacies types according to the classification results, which replaces the method for manually labeling the logging curves in the prior art, is conducive to improving the speed and accuracy of labeling the logging curves, has a high level of intelligence, greatly reduces the workload and labor cost of manual labeling, and is conducive to the smooth progress of exploration and development. BRIEF DESCRIPTION OF DRAWINGS

[0018] The drawings accompanying the specification of the present application form a part thereof, serve to provide further understanding of the present application, and together with the description of the exemplary embodiments of the present application and the explanation thereof serve to explain the present application, and do not constitute improper limitations on the present application. In the drawings:

[0019] Figure 1 Flow chart of a method for adding labels to well logs according to an exemplary embodiment of the present application;

[0020] Figure 2 Flow chart of a method for adding labels to well logs according to a specific embodiment of the present application;

[0021] Figure 3A Schematic diagram of results of each component obtained by performing principal component analysis on a plurality of well logs according to a specific embodiment of the present application;

[0022] Figure 3B Schematic diagram of cumulative information amount of each component obtained by performing principal component analysis on a plurality of well logs according to a specific embodiment of the present application;

[0023] Figure 3C Schematic diagram of correlation analysis results of each component obtained by performing principal component analysis on a plurality of well logs according to a specific embodiment of the present application;

[0024] Figure 4 Schematic diagram of analysis results of crossplot analysis of facies types at different depths of a target reservoir using water saturation-mud content according to a specific embodiment of the present application;

[0025] Figure 5 Schematic diagram of analysis results of crossplot analysis of facies types at different depths of a target reservoir using density-mud content according to a specific embodiment of the present application;

[0026] Figure 6 Schematic diagram of analysis results of crossplot analysis of facies types at different depths of a target reservoir using water saturation-density according to a specific embodiment of the present application;

[0027] Figure 7 Schematic diagram of analysis results of crossplot analysis of facies types at different depths of a target reservoir using GR-density according to a specific embodiment of the present application;

[0028] Figure 8 Schematic diagram of analysis results of crossplot analysis of facies types at different depths of a target reservoir using Lame constant* density-Poisson ratio according to a specific embodiment of the present application;

[0029] Figure 9 Schematic diagram of analysis results of crossplot analysis of facies types at different depths of a target reservoir using Lame constant* density-Poisson ratio-facies according to a specific embodiment of the present application;

[0030] Figure 10A schematic diagram of an analysis result of crossplot analysis of a longitudinal wave impedance-Poisson's ratio-lithofacies on a target reservoir at different depths according to an embodiment of the present application;

[0031] Figure 11 A schematic diagram of a result of predicting lithofacies according to an embodiment of the present application;

[0032] Figure 12 A schematic diagram of a flow of a method of adding labels to real-time logging curves according to an embodiment of the present application. DETAILED DESCRIPTION

[0033] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments.

[0034] Deep learning method is used to carry out reservoir prediction, which needs to use a large amount of labeled data to train the deep learning model. At present, the work of adding labels to logging curves is completed manually on the logging. However, the manual labeling work is large in workload and strong in multi-solution. The labeling error of the logging curve will have a great influence on the result of reservoir prediction, resulting in poor prediction effect.

[0035] In view of the problems of large workload and strong multi-solution in the related art, a method of adding labels to logging curves, a computing device and a storage medium are provided, which automatically add lithofacies labels to key logging curves by using principal component analysis and unsupervised classification method, effectively improve the generation quality and efficiency of the electrical (lithological) phase label, so as to train the deep learning model for lithofacies prediction by using the logging curve after adding the label, so as to facilitate the prediction of reservoir lithofacies type.

[0036] Embodiment one

[0037] The embodiment provides a method for adding labels to logging curves, Figure 1 A flowchart of a method for adding labels to logging curves according to an exemplary embodiment of the present application. As shown in Figure 1 The method of the embodiment can include:

[0038] S100: analyzing a plurality of logging curves of a target reservoir, and selecting at least one curve meeting a preset condition from the plurality of logging curves as a key logging curve according to an analysis result.

[0039] S200: based on the geological understanding of the lithofacies types of the target reservoir at different depths, using a preset classification algorithm, for at least one key logging curve of the target reservoir at each depth, determining the probability of each key logging curve at the depth pointing to each lithofacies type in the geological understanding, establishing the association between the lithofacies type corresponding to the maximum probability and the corresponding key logging curve, and then adding a lithofacies type label to at least one key logging curve at different depths.

[0040] Through the above steps, the key logging curve of the target reservoir is determined, and the key logging curve is classified based on the geological understanding of the lithofacies types of the target reservoir, so as to add a lithofacies type label to the key logging curve. The key logging curve is labeled efficiently and quickly, which saves labor cost and facilitates the smooth progress of exploration and development.

[0041] The preset condition can be set as needed, for example, the fluctuation degree of the logging curve can be large enough, or the difference between the maximum value and the minimum value of the logging curve can be large enough, or a logging curve sensitive to the lithofacies type can be selected as the key logging curve. The preset classification algorithm includes K-Mean unsupervised classification algorithm, of course, other classification algorithms can also be selected as needed, which is not limited herein.

[0042] In one example, a plurality of logging curves of the target reservoir are analyzed, and at least one curve meeting the preset condition is selected from the plurality of logging curves as the key logging curve according to the analysis result, which can include: performing principal component analysis on the plurality of logging curves of the target reservoir, determining the information amount contained in each logging curve, and selecting at least one curve whose information amount meets the preset condition from the plurality of logging curves as the key logging curve.

[0043] The at least one logging curve whose information amount meets the preset condition is selected from the plurality of logging curves as the key logging curve, which can include: sorting the plurality of logging curves according to the information amount contained in each logging curve from large to small, determining the cumulative proportion of the cumulative information amount corresponding to each logging curve in the sorting to the total information amount of all logging curves, and taking each logging curve corresponding to the minimum cumulative proportion reaching the preset proportion threshold as the key logging curve.

[0044] In another example, after determining the probability of each key logging curve pointing to each lithofacies type in the geological understanding at the depth, and before establishing the correlation between the lithofacies type corresponding to the maximum probability and the corresponding key logging curve, the method can further include a crossplot analysis step, including: taking the lithofacies type corresponding to the maximum probability as the lithofacies type of the target reservoir at the depth, thereby obtaining the lithofacies type of the target reservoir at each different depth; using at least two logging curves other than the key logging curve to perform crossplot analysis on the lithofacies type of the target reservoir at each different depth, determining whether the determined lithofacies type of the target reservoir at each different depth is correct according to the results of the crossplot analysis, and determining the correctness of the determined lithofacies type of the target reservoir at each different depth, and when the correctness reaches a preset correctness threshold, establishing the correlation between the lithofacies type corresponding to the maximum probability and the corresponding key logging curve for each depth of the target reservoir again.

[0045] In the crossplot analysis, the at least two logging curves other than the key logging curve can include a shear wave velocity curve, a compressional wave velocity curve, and a density curve; the crossplot analysis on the lithofacies type of the target reservoir at each different depth using the at least two logging curves other than the key logging curve can include: determining a Lame constant curve and a Poisson's ratio curve according to the shear wave velocity curve, the compressional wave velocity curve, and the density curve, and determining a product curve of the Lame constant curve and the density curve; and performing crossplot between the Poisson's ratio curve and the product curve for the lithofacies type of the target reservoir at each different depth.

[0046] When the correctness does not reach the preset correctness threshold, the classification parameters of the preset classification algorithm are adjusted, the probability of each key logging curve pointing to each lithofacies type in the geological understanding at each depth of the target reservoir is determined again using the adjusted preset classification algorithm, and the crossplot analysis step is re-executed.

[0047] By repeatedly verifying the lithofacies type label added to the key logging curve through cyclic execution of the step of classifying the key logging curve and the crossplot analysis step, the accuracy of the label added to the logging curve is improved.

[0048] Through the method of the present application, the logging curve is classified based on the geological understanding of the lithofacies type of the reservoir, and the lithofacies type label is added to the logging curve according to the classification result, which replaces the method of adding the label to the logging curve by manual operation in the prior art, and is beneficial to improving the speed and accuracy of adding the label to the logging curve, has a high intelligent level, greatly reduces the workload and labor cost of manual label addition, and is beneficial to the smooth progress of exploration and development.

[0049] Embodiment Two

[0050] This embodiment provides a specific embodiment of a method for adding labels to well logging curves. As shown in the figure, the method of this embodiment can include the following steps: Figure 2

[0051] (1) Input all well logging data in the specified block range, perform PCA principal component analysis on the target reservoir's multiple well logging curve data, determine the key well logging curves by matrix singular value decomposition statistical analysis of the well logging curve parameters, eliminate redundant information, and provide the data basis for clustering analysis;

[0052] (2) According to the geological understanding, determine the lithofacies types contained in the target reservoir, correct the abnormal points of multiple key well logging curves, etc. Preprocessing, establishing an initial probability model;

[0053] (3) Use the K-Mean unsupervised classification method to perform unsupervised classification on the key well logging curves, and get the classification results. Mainly use the average distance between the mean vector of multiple key curves and the preset parameters to judge the lithofacies type represented by each key well logging curve;

[0054] (4) Use rock physics analysis to perform multi-parameter crossplot analysis on the classification results. Multi-parameters such as shale content-water saturation crossplot, shale content-density crossplot, Lame constant-density and Poisson's ratio-lithofacies crossplot, etc. Through iterative analysis, the accuracy of the lithofacies type represented by the key well logging curves can be adjusted by adjusting the parameters of the clustering analysis, and more accurate lithofacies type labels can be added to the key well logging curves;

[0055] This method solves the problem of workload and human error caused by manual hooking in the previous analysis process, can generate a large number of accurate and effective labels, effectively improves the generation quality and efficiency of the lithofacies type labels of well logging curves, and provides a good data set for the application of subsequent deep learning and other intelligent algorithms.

[0056] The method steps of this specific embodiment are as follows:

[0057] Step 1: Collect the actual well logging data of a certain work area and perform PCA principal component analysis:

[0058] (1) According to the previous experience of this work area, select neutron porosity, density, resistivity, shale content and water saturation curves for well logging electrical facies division (considering that these curves are the basis for dividing lithofacies results).

[0059] (2) First, reduce the curve dimension through principal component analysis, and prepare for subsequent unsupervised electrical facies classification.

[0060] ​Firstly, the corresponding relationship between the characteristic vectors PC1-PC5 of the matrix composed of neutron porosity, density, resistivity, shale content and water saturation and the characteristic vectors PC1-PC5 is determined through the quantitative analysis of PCA. PC1-PC5 are sorted according to their eigenvalues from large to small, the cumulative eigenvalues are calculated according to the sorting, a plurality of characteristic vectors corresponding to the minimum number of eigenvalues reaching the preset cumulative eigenvalue threshold are selected, the information amount of each logging curve is calculated according to the selected characteristic vectors, and a plurality of logging curves with the largest information amount are taken as key logging curves.

[0061] As shown in Figures 3A-3C According to the results of principal component analysis, the characteristic vector PC1 accounts for 56.8% of the information amount of all curves, among which, the remaining four curves contribute except density; the characteristic vector PC2 accounts for 24.3% of the total information, mainly contributed by density; the characteristic vector PC3 accounts for 9.0% of the total information, mainly contributed by water saturation and shale content; the first three groups account for 90.2% of the total information. Therefore, the first three groups of PC1-PC5 can be used to represent the matrix information of the entire logging parameter, and the information amount contained in each logging curve is calculated to determine the key logging curve, thereby effectively reducing the data dimension of the logging curve.

[0062] (1) PC1 is positively correlated with neutron porosity, shale content and water saturation, and negatively correlated with resistivity;

[0063] (2) PC2 is negatively correlated with RHOB (density) curve, and has low correlation with other curves;

[0064] (3) PC3 is positively correlated with water saturation, and negatively correlated with shale content;

[0065] Since PC4 and PC5 only contain less than 10% of the effective information, when doing logging electrical phase analysis, PC4 and PC5 curves (most of which are redundant information) are removed to improve the reliability of the results of logging electrical phase separation.

[0066] The main principle of principal component analysis is as follows:

[0067] Principal component analysis (PCA) is a multivariate statistical method for studying the correlation between multiple variables, which studies how to reveal the internal structure between multiple variables through a few principal components, that is, to derive a few principal components from the original variables, so that they can retain as much information as possible of the original variables and are mutually independent. The analysis of principal components can effectively extract the information of the data, and reduce the dimension of the data without loss of information, and the principal components become another way to display the data.

[0068] The essence of PCA is to determine an orthogonal transformation of a coordinate system, to transform a set of variables that can have correlations into a set of linearly uncorrelated variables, and to maximize the variance of the transformed data points along the new coordinate axes in the new coordinate system. The transformed set of variables is called principal components (or principal components). The PCA processing procedure is briefly described as follows:

[0069] (1) Suppose we have p well logging response curves, each curve has n measurement samples, and the sample matrix is

[0070]

[0071] First, in order to eliminate the influence of the actual data dimension, the sample matrix is standardized as follows,

[0072]

[0073] Wherein, The standardized matrix Z is obtained.

[0074] (2) The correlation coefficient matrix R is obtained,

[0075]

[0076] Wherein,

[0077] (3) The characteristic equation |R-λI p | is solved, p eigenvalues λ1>λ2>…>λ p >0 in descending order and corresponding p eigenvectors A i =(a 1i ,a 2i ,…,a pi ) are obtained, (i=1,2,…,p)

[0078] (4) The standardized data sample variables are transformed into components, that is,

[0079] F=AX, or

[0080] F1 is called the first component, F2 is called the second component, …, F p is called the pth component.

[0081] (5) According to the given threshold value (q=0.85), when the sum of the eigenvalues of the current m principal components is greater than the given threshold value q, the value of m is determined by the cumulative contribution of the eigenvalue as follows,

[0082]

[0083] At this time, m < p, the first m components are selected to represent the original sample, and key logging curves in the logging curves are determined according to the m components, thereby playing a role of dimension reduction.

[0084] In the second step, it is determined that there should be four facies in the region according to geological understanding, it is clear that the four facies correspond to mudstone, gas-bearing water layer, gas layer and dry layer respectively, and the initial logging data is corrected, and the classification vector of each facies type is set.

[0085] In the third step, the K-Mean analysis method of unsupervised clustering analysis driven by pure data is used to determine the facies type pointed by each key logging curve, the facies type is added as the label of the corresponding key logging curve, and the cross-plot analysis is performed on the classification result to improve the accuracy of the label.

[0086] On this basis, through the feature analysis of each key logging curve, four sets of electric (rock) facies are finally identified by unsupervised classification, as shown in the following table.

[0087] Table 1- electric (rock) facies feature division

[0088] Main features Explanation Electrical facies 1 High natural gamma, high neutron porosity, low resistivity, no hydrocarbons Shale Electrical facies 2 Medium to low resistivity, high water saturation, low shale content (Gas) water layer Electrical facies 3 High resistivity, low water saturation, low shale content Gas layer Electrical facies 4 High density, low natural gamma, no hydrocarbons Dry layer

[0089] The resistivity is high, indicating high gas content, and through model application analysis of specific well curves, the four sets of electric (rock) facies can be well matched on the logging curves.

[0090] Figure 4 Fig. 1 is a schematic diagram of the analysis result of cross-plot analysis of the facies type of the target reservoir at different depths according to the water saturation-mud content classification result according to an embodiment of the present application, Figure 5 Fig. 2 is a schematic diagram of the analysis result of cross-plot analysis of the facies type of the target reservoir at different depths according to the density-mud content classification result according to an embodiment of the present application, Figure 6 Fig. 3 is a schematic diagram of the analysis result of cross-plot analysis of the facies type of the target reservoir at different depths according to the water saturation-density classification result according to an embodiment of the present application, Figure 7 Fig. 4 is a schematic diagram of the analysis result of cross-plot analysis of the facies type of the target reservoir at different depths according to the GR-density classification result according to an embodiment of the present application, Figure 8 Fig. 5 is a schematic diagram of the analysis result of cross-plot analysis of the facies type of the target reservoir at different depths according to the Lame constant * density-Poisson ratio classification result according to an embodiment of the present application, Figure 9 Fig. 6 is a schematic diagram of the analysis result of cross-plot analysis of the facies type of the target reservoir at different depths according to the Lame constant * density-Poisson ratio-facies classification result according to an embodiment of the present application, Figure 10Fig. 1 is a schematic diagram of an analysis result of crossplot analysis of the P-wave impedance-Poisson's ratio-lithofacies of a target reservoir at different depths according to an embodiment of the present application. The method of the present embodiment can be used to predict the lithofacies types contained in a target area according to well logging curves, for example, to supplement the lithofacies types obtained by using the method of the present application. Figure 11 In this way, the lithofacies description of the well logging curves of each well at each depth point can be obtained. From the perspective of each well, taking well A1 as an example, the curve marked with the electrical facies column in the figure is the lithofacies type curve of the target reservoir of the well, which is actually one-dimensional data such as 2.0, 1.0, 1.0, 3.0, 4.0, and the like, similar to a square wave signal.

[0091] The present application automatically obtains the electrical (rock) facies curve according to multiple well logging curves, solves the problems of workload and errors caused by manual hooking in the previous analysis process, and effectively improves the quality and efficiency of the generation of the electrical (rock) facies label.

[0092] Embodiment three

[0093] The present embodiment provides a training method of a reservoir lithofacies type prediction model, which comprises: adding labels to well logging curves by using the method of adding labels to well logging curves as described above to generate training samples; and training a preset reservoir lithofacies type deep learning model by using a training sample set composed of multiple training samples to obtain a trained reservoir lithofacies type prediction model. Figure 12 Fig. 2 is a schematic diagram of the flow of a method of adding labels to real-time well logging curves according to an embodiment of the present application.

[0094] Embodiment four

[0095] The present embodiment provides a method of predicting reservoir lithofacies types, which comprises: obtaining multiple well logging curves of a target reservoir; and determining the lithofacies types of the target reservoir at different depths based on the multiple well logging curves by using a trained reservoir lithofacies type prediction model obtained by using the method as described above.

[0096] Embodiment five

[0097] The present embodiment provides a computing device, which comprises: a memory and a processor, and the memory stores a computer program, when the computer program is executed by the processor, the steps of the method of adding labels to well logging curves as described above or the steps of the training method of a reservoir lithofacies type prediction model as described above or the steps of the method of predicting reservoir lithofacies types as described above are executed.

[0098] In one embodiment, the computing device can comprise one or more processors (CPUs), input / output interfaces, network interfaces, and memories.

[0099] Memory can include non-persistent memory and / or volatile memory, random access memory (RAM), and / or non-volatile memory, e.g., read only memory (ROM) or flash memory (FLASH RAM), among others, in a computer readable medium. Memory is an example of computer readable media.

[0100] Embodiment six

[0101] The embodiments provide a storage medium storing a computer program, which, when executed by a processor, performs the steps of the method of adding labels to well logs as described above or the steps of the method of training a reservoir facies type prediction model as described above or the steps of the method of predicting reservoir facies types as described above.

[0102] The computer program can be in any combination of one or more storage media. The storage media can be a readable signal medium or a readable storage medium.

[0103] The readable storage medium, for example, can include an electric, a magnetic, a optical, an electromagnetic, an infrared, or a semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium can include an electrical connection having one or more wires, a portable disc, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0104] The readable signal medium can include a propagated data signal with readable computer program code, which is embodied in a baseband or as part of a carrier wave. Examples of a propagated signal can take many forms, e.g., an electromagnetic signal, an optical signal, or any suitable combination thereof. The readable signal medium can also be any storage medium that can be used to carry or store program code in any manner, which can be read and executed or otherwise used by an instruction execution system, apparatus, or device.

[0105] The computer program code embodied on the storage medium can be transmitted by any programmed medium, including wired, wireless, optical, RF, or any suitable combination of the foregoing.

[0106] Computer programs for performing operations of the present application can be written in any combination of one or more programming languages. Programming languages can include a procedural programming language, such as the "C" programming language and / or other programming languages such as Java, C++, etc. The computer programs can execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's device and partly on a remote computing device or entirely on the remote computing device or server. In the latter scenario, the remote computing device can be connected to the user's computing device through any type of network, including a local area network ("LAN") or a wide area network ("WAN"), or the connection can be made to an external computing device, such as through the Internet using an Internet Service Provider ("ISP").

[0107] It is to be understood that the terms so far as the terminology used herein is concerned are for the purpose of describing particular embodiments and are not intended to be limiting in any way. This description and claims appended hereto should be read to include all equivalents as allowed by the applicable Indus.

[0108] It should be noted that the terms "first", "second", and the like, herein do not denote any ordinal, sequential, or chronologic significance, but are used for purposes of nomenclature only. It is also to be understood that the terminology used herein is for the purpose of describing particular embodiments only, and is not intended to be limiting.

[0109] It should be understood that the exemplary embodiments described herein can be implemented in various forms of hardware, software, or a combination thereof, and can be implemented in several manners, all of which have been contemplated to be within the scope of the present disclosure. Additionally or alternatively, certain steps can be omitted, combined into a single step, or divided into multiple steps. These embodiments are provided so that the present disclosure will be thorough and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art, and should not be interpreted to limit the present application.

[0110] While the principles of the application have been described above in connection with specific embodiments, it is to be understood that this description is made only by way of example and not as a limitation of the scope of the application. The teachings of the present application are intended to be broadly applied to the various aspects of the application. The various aspects of the present application are intended to be encompassed by the appended claims.

Claims

1. A method of tagging a well log, characterized by, The method comprises the following steps: analyzing a plurality of logging curves of a target reservoir, and selecting at least one curve meeting preset conditions from the plurality of logging curves as a key logging curve according to an analysis result; based on geological knowledge of lithofacies types of the target reservoir at different depths, using a preset classification algorithm, for at least one key logging curve of the target reservoir at each depth, determining a probability of each key logging curve at the depth pointing to each lithofacies type in the geological knowledge, establishing an association relationship between a lithofacies type corresponding to a maximum probability and a corresponding key logging curve, and then adding a lithofacies type label to the at least one key logging curve at different depths respectively; after determining the probability of each key logging curve at the depth pointing to each lithofacies type in the geological knowledge, and before establishing the association relationship between the lithofacies type corresponding to the maximum probability and the corresponding key logging curve, the method further comprises the following steps: an intersection analysis step, comprising: taking the lithofacies type corresponding to the maximum probability as a lithofacies type of the target reservoir at the depth, and then obtaining lithofacies types of the target reservoir at different depths; for the lithofacies types of the target reservoir at different depths, using at least two logging curves other than the key logging curve for intersection analysis, judging whether the determined lithofacies types of the target reservoir at different depths are correct according to an intersection analysis result, and determining a correctness rate of the determined lithofacies types of the target reservoir at different depths, when the correctness rate reaches a preset correctness threshold, and then establishing the association relationship between the lithofacies type corresponding to the maximum probability and the corresponding key logging curve for each depth of the target reservoir; the at least two logging curves other than the key logging curve comprise: a shear wave velocity curve, a longitudinal wave velocity curve and a density curve; for the lithofacies types of the target reservoir at different depths, using at least two logging curves other than the key logging curve for intersection analysis, comprising: determining a Lame constant curve and a Poisson's ratio curve according to the shear wave velocity curve, the longitudinal wave velocity curve and the density curve, and determining a product curve of the Lame constant curve and the density curve; for the lithofacies types of the target reservoir at different depths, performing intersection between the Poisson's ratio curve and the product curve.

2. The method of tagging a well log of claim 1, wherein, The method comprises the following steps: performing principal component analysis on a plurality of logging curves of a target reservoir, determining an information amount contained in each logging curve, and selecting at least one curve meeting a preset condition as a key logging curve from the plurality of logging curves.

3. The method of tagging a well log of claim 2, wherein, The method comprises the following steps: sorting the plurality of logging curves according to their respective information amounts from large to small, determining a cumulative proportion of a cumulative information amount corresponding to each logging curve in a total sum of information amounts of all logging curves, and taking each logging curve corresponding to a minimum cumulative proportion reaching a preset proportion threshold as a key logging curve.

4. The method of tagging a well log of claim 1, wherein, The preset classification algorithm comprises a K-Mean unsupervised classification algorithm.

5. The method of tagging a well log of claim 1, wherein, When the accuracy does not reach the preset accuracy threshold, a classification parameter of the preset classification algorithm is adjusted, the probability of each lithofacies type of each key logging curve of the target reservoir at each depth in the geological understanding is re-determined by using the adjusted preset classification algorithm, and the cross-plot analysis step is re-executed.

6. A training method for a reservoir facies type prediction model, characterized in that, The method comprises: The logging curve is labeled by using the method for labeling logging curves according to any one of claims 1 to 5 to generate training samples; A preset reservoir lithofacies type deep learning model is trained by using a training sample set composed of multiple training samples to obtain a trained reservoir lithofacies type prediction model.

7. A method of predicting reservoir lithofacies types, characterized by, The method comprises: A plurality of logging curves of a target reservoir are obtained; Based on the trained reservoir lithofacies type prediction model obtained by using the method according to claim 6, the lithofacies types of the target reservoir at different depths are determined according to the plurality of logging curves.

8. A computing device, comprising: The method comprises: A memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the steps of the method for labeling logging curves according to any one of claims 1 to 5 or the steps of the training method of the reservoir lithofacies type prediction model according to claim 6 or the steps of the method for predicting the reservoir lithofacies type according to claim 7 are executed.

9. A storage medium storing a computer program, characterized by The computer program is executed by the processor to execute the steps of the method for labeling logging curves according to any one of claims 1 to 5 or the steps of the training method of the reservoir lithofacies type prediction model according to claim 6 or the steps of the method for predicting the reservoir lithofacies type according to claim 7.

Images