Intelligent identification method for logging lithofacies of unbalanced sample, related method and device

Abstract

The invention discloses a logging lithofacies intelligent identification method of an unbalanced sample, and a related method and device. The method comprises the following steps: establishing a lithofacies sample data set according to pre-obtained logging attributes corresponding to various rock phases of an actual well in a research area, and sampling to obtain a plurality of first sample subsets; for each first sample subset, calculating a mean value of each logging attribute corresponding to all the sample data of each type of lithofacies to obtain each attribute feature of each type of lithofacies, further determining the weight of each attribute feature, and further calculating a weighted feature parameter of each type of lithofacies to obtain a second sample subset; inputting each second sample subset into a base classifier trained based on the resampled lithofacies sample data set for identification, and determining a logging lithofacies identification probability result in each base classifier; and determining lithofacies types of different depths of the actual well in the research area according to the predetermined combined weight of each base classifier and the corresponding logging lithofacies identification probability result.

Description

Technical Field

[0001] The present application relates to a method for intelligently identifying well logging lithofacies of non-equilibrium samples, and related methods and devices. Background Art

[0002] Lithofacies refers to the characteristics of rocks formed in a certain sedimentary environment, including rock mineral composition, structure, texture and color. Accurate identification and classification of lithofacies can provide important basis for reservoir parameter calculation, development plan design, fracturing reconstruction construction, reservoir description and basin estimation, and is an important part of reservoir evaluation. Well logging data is widely used in lithofacies identification and evaluation due to its high vertical resolution and good continuity. However, with the continuous deepening of exploration and development, reservoirs are becoming increasingly complex and highly heterogeneous. In complex lithology reservoirs, especially those with thin interlayers or thin interlayers, the lithology samples obtained from core or logging data generally have more main lithology and fewer minority lithology, and the number of samples of different lithology is unevenly distributed. The accuracy of lithology identification using traditional core calibration logging or deep learning methods is often not high. Summary of the invention

[0003] In order to achieve higher-precision logging lithofacies identification of non-balanced samples, the embodiments of the present application provide a method for intelligently identifying logging lithofacies of non-balanced samples, and related methods and devices.

[0004] In a first aspect, an embodiment of the present application provides a method for intelligently identifying well logging lithofacies of a non-equilibrium sample, the method comprising:

[0005] According to the well logging attributes corresponding to various types of rocks in actual wells in the study area obtained in advance, a lithofacies sample data set is established, and a plurality of first sample subsets are obtained by sampling;

[0006] For each of the first sample subsets, calculating the mean of each well logging attribute corresponding to all sample data of each type of lithofacies, and obtaining each attribute feature of each type of lithofacies;

[0007] Based on the attribute characteristics of each type of lithofacies, determining the weight of each attribute characteristic;

[0008] According to the weights of the attribute characteristics, weighted characteristic parameters of each type of lithofacies are calculated to obtain a second sample subset;

[0009] Input each of the second sample subsets into a base classifier trained based on the resampled lithofacies sample data set for identification, and determine the well logging lithofacies identification probability results in each of the base classifiers;

[0010] The lithofacies types of actual wells at different depths in the study area are determined according to the predetermined combined weight of each base classifier and the corresponding logging lithofacies identification probability results.

[0011] In one or some optional implementations of the embodiments of the present application, the trained base classifiers are obtained by the following method:

[0012] Establish a historical lithofacies sample data set based on the well logging attributes corresponding to each type of rock obtained in advance;

[0013] Resampling the historical lithofacies sample data set according to a preset non-equilibrium index to obtain a resampled historical sample data set, and sampling to obtain a plurality of first historical sample subsets;

[0014] For each of the first historical sample subsets, the mean values ​​of the well logging attributes corresponding to all sample data of each type of lithofacies are calculated to obtain the attribute characteristics of each type of lithofacies;

[0015] Based on the attribute characteristics of each type of lithofacies, determining the weight of each attribute characteristic;

[0016] According to the weights of the attribute characteristics, weighted characteristic parameters of each type of lithofacies are calculated to obtain a second historical sample subset;

[0017] Input each of the second historical sample subsets into a base classifier for identification, and determine the historical well logging lithofacies identification probability results in each base classifier;

[0018] Compare the historical logging lithofacies recognition probability results in each base classifier with the real results of the logging lithofacies obtained in advance to determine whether the prediction accuracy of each base classifier reaches a preset threshold;

[0019] If so, the trained base classifiers are obtained.

[0020] In one or some optional implementations of the embodiment of the present application, the historical lithofacies sample data set is resampled according to the preset non-equilibrium index to obtain a resampled historical sample data set, and a plurality of first historical sample subsets are sampled, including:

[0021] Determine the majority lithofacies and the minority lithofacies according to the number of samples of different lithofacies types in the historical lithofacies sample data set;

[0022] According to the preset non-equilibrium index, most types of lithofacies are undersampled and minority types of lithofacies are oversampled until the number of samples of different lithofacies types in the historical lithofacies sample data set is balanced, a resampled historical sample data set is obtained, and multiple first historical sample subsets are sampled.

[0023] In one or some optional implementations of the embodiment of the present application, for each of the first sample subsets, the mean of each logging attribute corresponding to all sample data of each type of lithofacies is calculated to obtain each attribute feature of each type of lithofacies, including:

[0024] For each of the first sample subsets, the mean values ​​of the well logging attributes corresponding to all sample data of each type of lithofacies are calculated based on the following formula 1:

[0025]

[0026] In the formula, r ij is the mean value of the jth well logging attribute of the lth training sample of the i-th lithofacies in the first sample subset; ijl is the jth logging attribute value of the lth sample in the i-th lithofacies in the first sample subset; n c is the number of samples in the first sample subset;

[0027] The mean of each logging attribute corresponding to all sample data of each type of lithofacies is taken as the attribute characteristics of each type of lithofacies.

[0028] In one or some optional implementations of the embodiments of the present application, determining the weight of each attribute feature based on each attribute feature of each type of lithofacies includes:

[0029] For the first sample subset, based on the attribute features of each type of lithofacies, calculating the deviation of each attribute feature from the same attribute features in other types of lithofacies;

[0030] According to the deviation, the total deviation of each attribute characteristic of each type of lithofacies is calculated;

[0031] The weight of each attribute feature is determined according to the total deviation.

[0032] In one or some optional implementations of the embodiment of the present application, the weighted characteristic parameters of each type of lithofacies are calculated according to the weights of the attribute characteristics to obtain the second sample subset, including:

[0033] According to the weights of the attribute characteristics, the weighted characteristic parameters of each type of lithofacies are calculated based on the following formula 2:

[0034]

[0035] In the formula, rt i is the weighted characteristic parameter of the i-th type of lithofacies in the second sample subset; r ij is the jth weighted characteristic parameter of the i-th type of lithofacies in the first sample subset; The weight of the jth weighted feature parameter corresponding to the maximum value of the total deviation of all the features;

[0036] For each of the first sample subsets, a second sample subset is obtained according to the weighted characteristic parameters of each type of lithofacies.

[0037] In one or some optional implementations of the embodiments of the present application, after obtaining the trained base classifiers, the method further includes:

[0038] Determine the precision and recall of each base classifier according to the historical logging lithofacies recognition probability results in each trained base classifier and the real results of the logging lithofacies obtained in advance;

[0039] According to the precision and recall of each base classifier, the well logging lithofacies identification performance index of each base classifier is determined based on the following formula 3:

[0040]

[0041] Where, F is the logging lithofacies recognition performance index of a base classifier; P p is the precision of a base classifier; P r is the recall rate of a base classifier;

[0042] According to the well logging lithofacies identification performance index of each base classifier, the combination weight of each base classifier is determined based on the following formula 4:

[0043]

[0044] In the formula, w t is the combined weight of the tth base classifier; F t is the well logging lithofacies recognition performance index of the t-th base classifier; F min The minimum value of the well logging lithofacies identification performance index of all base classifiers; F max The maximum value of the well logging lithofacies identification performance index of all base classifiers.

[0045] In one or some optional implementations of the embodiment of the present application, after establishing the lithofacies sample data set according to the well logging attributes corresponding to various types of rocks in the actual wells in the study area obtained in advance, the method further includes:

[0046] All the logging curves in the lithofacies sample data set are normalized.

[0047] In one or some optional implementations of the embodiment of the present application, the method of determining the lithofacies types at different depths of actual wells in the study area according to the predetermined combined weight of each base classifier and the corresponding logging lithofacies identification probability result includes:

[0048] For each well logging sampling depth of the actual wells in the study area, the probability of each type of lithofacies is determined based on the following formula 5 according to the pre-determined combined weight of each base classifier and the corresponding well logging lithofacies identification probability result:

[0049]

[0050] Where p is the probability of each type of lithofacies; w i is the weight of the i-th classifier; p i is the probability of each lithofacies in the i-th classifier; T is the number of the second sample subset;

[0051] For each logging sampling depth of the actual wells in the study area, according to the corresponding probability of each type of lithofacies, the lithofacies with the highest probability is taken as the corresponding lithofacies type to determine the lithofacies types of different depths of the actual wells in the study area.

[0052] In a second aspect, an embodiment of the present application provides a method for training a well logging lithofacies intelligent recognition model for non-balanced samples, the method comprising:

[0053] Establish a historical lithofacies sample data set based on the well logging attributes corresponding to each type of rock obtained in advance;

[0054] Resampling the historical lithofacies sample data set according to a preset non-equilibrium index to obtain a resampled historical sample data set, and sampling to obtain a plurality of first historical sample subsets;

[0055] For each of the first historical sample subsets, the mean values ​​of the well logging attributes corresponding to all sample data of each type of lithofacies are calculated to obtain the attribute characteristics of each type of lithofacies;

[0056] Based on the attribute characteristics of each type of lithofacies, determining the weight of each attribute characteristic;

[0057] According to the weights of the attribute characteristics, weighted characteristic parameters of each type of lithofacies are calculated to obtain a second historical sample subset;

[0058] Input each of the second historical sample subsets into a base classifier for identification, and determine the historical well logging lithofacies identification probability results in each base classifier;

[0059] Compare the historical logging lithofacies recognition probability results in each base classifier with the real results of the logging lithofacies obtained in advance to determine whether the prediction accuracy of each base classifier reaches a preset threshold;

[0060] If so, the trained base classifiers are obtained.

[0061] In a third aspect, an embodiment of the present application provides a well logging lithofacies intelligent identification device for a non-equilibrium sample, the device comprising:

[0062] A first sample establishment module is used to establish a lithofacies sample data set according to the well logging attributes corresponding to various types of rocks in actual wells in the study area obtained in advance, and to sample and obtain a plurality of first sample subsets;

[0063] A first determination module is used to calculate the mean of each well logging attribute corresponding to all sample data of each type of lithofacies for each of the first sample subsets, so as to obtain each attribute feature of each type of lithofacies;

[0064] A second determination module is used to determine the weight of each attribute feature based on each attribute feature of each type of lithofacies;

[0065] A second sample establishment module is used to calculate the weighted characteristic parameters of each type of lithofacies according to the weights of the attribute characteristics to obtain a second sample subset;

[0066] A third determination module is used to input each of the second sample subsets into a base classifier trained based on the resampled lithofacies sample data set for identification, and determine the well logging lithofacies identification probability results in each of the base classifiers;

[0067] The fourth determination module is used to determine the lithofacies types of different depths of actual wells in the study area according to the predetermined combined weight of each base classifier and the corresponding logging lithofacies identification probability result.

[0068] In a fourth aspect, an embodiment of the present application provides a well logging lithofacies intelligent identification model training device for non-balanced samples, the device comprising:

[0069] The third sample establishment module is used to establish a historical lithofacies sample data set according to the well logging attributes corresponding to various types of rocks obtained in advance;

[0070] A sampling module, used for resampling the historical lithofacies sample data set according to a preset non-equilibrium index to obtain a resampled historical sample data set, and sampling to obtain a plurality of first historical sample subsets;

[0071] A fifth determination module, configured to calculate, for each of the first historical sample subsets, the mean values ​​of the well logging attributes corresponding to all sample data of each type of lithofacies, and obtain the attribute characteristics of each type of lithofacies;

[0072] A sixth determination module, used to determine the weight of each attribute feature based on each attribute feature of each type of lithofacies;

[0073] A seventh determination module is used to calculate the weighted characteristic parameters of each type of lithofacies according to the weights of the attribute characteristics to obtain a second historical sample subset;

[0074] An eighth determination module is used to input each of the second historical sample subsets into a base classifier for identification, and determine the historical well logging lithofacies identification probability results in each base classifier;

[0075] A judgment module, used to compare the historical logging lithofacies recognition probability results in each base classifier with the real results of the logging lithofacies obtained in advance, and judge whether the prediction accuracy of each base classifier reaches a preset threshold;

[0076] The ninth determination module is used to obtain the trained base classifiers if the prediction accuracy of the base classifiers determined by the judgment module reaches a preset threshold.

[0077] In a fifth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for intelligent identification of well logging lithofacies of unbalanced samples as described above, and / or the method for training a model for intelligent identification of well logging lithofacies of unbalanced samples.

[0078] In a sixth aspect, an embodiment of the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for intelligent identification of well logging lithofacies of unbalanced samples as described above is implemented, and / or the method for training a model for intelligent identification of well logging lithofacies of unbalanced samples is implemented.

[0079] In a seventh aspect, an embodiment of the present application provides a computer program product comprising instructions. When the computer program product is run on a computer device, the computer device executes the method for intelligent identification of well logging lithofacies of unbalanced samples as described above, and / or the method for training a model for intelligent identification of well logging lithofacies of unbalanced samples.

[0080] In an eighth aspect, an embodiment of the present application provides a chip, the chip including a processor and a communication interface, the communication interface and the processor are coupled, and the processor is used to run a computer program or instruction to implement the above-mentioned method for intelligent identification of well logging lithofacies of unbalanced samples, and / or the method for training a model for intelligent identification of well logging lithofacies of unbalanced samples.

[0081] The beneficial effects of the above technical solution provided by the embodiment of the present application include at least:

[0082] The method for intelligent identification of well logging lithofacies of unbalanced samples provided in the embodiment of the present application is to weight the lithofacies characteristics of multiple first sample subsets constructed based on the well logging attributes of actual wells in the study area to obtain weighted characteristic parameters, thereby establishing multiple second sample subsets, and inputting the multiple second sample subsets into multiple base classifiers trained based on the lithofacies sample data set after resampling the unbalanced samples for identification, thereby determining the lithofacies types of different depths of actual wells in the study area. The method adopts an integrated learning algorithm, constructs a feature weight matrix to weight the lithofacies characteristic parameter values ​​of the first sample subset to establish a second sample subset, and the base classifiers trained based on the sample data set trained to achieve a balanced number of lithofacies samples of different types by resampling the unbalanced samples, and the well logging lithofacies identification performance of each base classifier determines the combination mechanism of the integrated learning algorithm, which can effectively identify the lithofacies types of different depths of actual wells in the study area, has a high overall identification accuracy, greatly improves the lithofacies identification effect of unbalanced samples, is simple and easy to implement, has a high actual processing efficiency, and is easy to promote and apply.

[0083] Other features and advantages of the present application will be described in the subsequent description, and partly become apparent from the description, or be understood by practicing the present application. The purpose and other advantages of the present application can be realized and obtained by the structures specifically pointed out in the written description, claims, and drawings.

[0084] The technical solution of the present application is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] The accompanying drawings are used to provide a further understanding of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings:

[0086] Figure 1 A schematic diagram of the steps of a method for intelligent identification of well logging lithofacies of a non-equilibrium sample provided in an embodiment of the present application;

[0087] Figure 2 A comparison diagram of a result section of lithofacies identification performed by a well logging lithofacies intelligent identification method of a non-balanced sample in an actual well in the study area provided in an embodiment of the present application and a core result section;

[0088] Figure 3 A schematic diagram of the steps of a method for training a well logging lithofacies intelligent identification model for non-balanced samples provided in an embodiment of the present application;

[0089] Figure 4 A schematic diagram of the structure of a well logging lithofacies intelligent identification device for non-equilibrium samples provided in an embodiment of the present application;

[0090] Figure 5 A schematic diagram of the structure of a well logging lithofacies intelligent identification model training device for non-balanced samples provided in an embodiment of the present application. DETAILED DESCRIPTION

[0091] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0092] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0093] It should also be understood that the term “and / or” used in the specification and appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0094] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

[0095] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0096] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0097] It should be understood that the size of the serial numbers of the steps in the following embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0098] In order to illustrate the technical solution of the present application, a specific embodiment is provided below for illustration.

[0099] The inventors found that in the prior art, with the continuous deepening of exploration and development, the reservoirs are becoming increasingly complex and highly heterogeneous. In complex lithology reservoirs, especially those containing thin interlayers or thin interlayers, the lithology samples obtained from core or logging data generally have more main lithology and fewer minority lithology, and the number of samples of different lithology is unevenly distributed. However, the accuracy of lithology identification using traditional core calibration logging or deep learning methods is often not high, which does not meet the inventors' expectations. Based on this, the inventors have made this application after further research and development, providing a logging lithofacies intelligent identification method, related methods and devices for non-balanced samples.

[0100] Embodiment 1

[0101] The present application embodiment provides a method for intelligently identifying well logging lithofacies of non-equilibrium samples, referring to Figure 1 As shown, the method includes:

[0102] S101: According to the well logging attributes corresponding to various types of rocks in actual wells in the study area obtained in advance, a lithofacies sample data set is established, and a plurality of first sample subsets are obtained by sampling.

[0103] In the embodiment of the present application, core data of actual wells in the study area are collected in advance, the lithofacies corresponding to the cores are identified, and the main lithofacies types are determined, assuming a total of N types. According to the logging response characteristics of the actual wells in the study area, the logging curves required for lithofacies identification are selected. Generally, the selection is made based on the sensitivity of the logging curves to the lithofacies. For example, the values ​​of the gamma, density, and resistivity logging curves of different lithofacies are different. Then, gamma, density, and resistivity logging curves can be selected as input curves for lithofacies identification. It is assumed that a total of m logging curves are selected. On this basis, a lithofacies sample data set is established, assuming that it contains M samples. Among them, the lithofacies data of a depth and its corresponding logging curve value are taken as a sample data as a whole. The logging curve value is the input data of the sample, and the lithofacies data is the label of the sample.

[0104] In the embodiment of the present application, after establishing the lithofacies sample data set according to the well logging attributes corresponding to various types of rocks in the actual wells in the study area obtained in advance, the following is further included:

[0105] All the logging curves in the lithofacies sample data set are normalized.

[0106] In the embodiment of the present application, all the original logging curve data in the lithofacies sample data set are normalized based on the following formula 6:

[0107]

[0108] In the formula, R i is the logging response value of the i-th logging curve, R i (min) is the minimum value of the i-th logging curve in the entire depth section, R i (max) is the maximum value of the ith logging curve in the entire depth section.

[0109] The normalized lithofacies sample data set is sampled to form T first sample subsets, denoted as N t (t=1~T).

[0110] S102: For each of the first sample subsets, calculate the mean of each well logging attribute corresponding to all sample data of each type of lithofacies to obtain each attribute feature of each type of lithofacies.

[0111] In the above step S102, for each of the first sample subsets, the mean of each well logging attribute corresponding to all sample data of each type of lithofacies is calculated to obtain each attribute feature of each type of lithofacies, including:

[0112] For each of the first sample subsets, the mean values ​​of the well logging attributes corresponding to all sample data of each type of lithofacies are calculated based on the following formula 1:

[0113]

[0114] In the formula, r ij is the first sample subset N t The mean value of the jth well logging attribute of the lth training sample of the i-th lithofacies in f ijl is the first sample subset N t The jth well logging attribute value of the lth training sample in the i-th lithofacies; n c is the number of samples in the first sample subset;

[0115] The mean of each logging attribute corresponding to all sample data of each type of lithofacies is taken as the attribute characteristics of each type of lithofacies.

[0116] In the embodiment of the present application, for a first sample subset N t , assuming that there are c types of lithofacies, with a total of n c Based on the above formula 1, the mean of all logging attributes corresponding to all sample data of each type of lithofacies in the first sample subset can be calculated, and the mean of each logging attribute corresponding to all sample data of each type of lithofacies is used as each attribute feature of each type of lithofacies. All attribute features in the first sample subset are constructed into an attribute feature matrix, i.e., a lithofacies feature parameter matrix, denoted as R. The lithofacies feature parameter matrix R is shown in the following formula 7:

[0117]

[0118] In the formula, r cm is the first sample subset N t The mth attribute characteristic of the cth type of lithofacies.

[0119] S103: Based on the attribute characteristics of each type of lithofacies, determine the weight of each attribute characteristic.

[0120] In the above step S103, the weight of each attribute feature is determined based on each attribute feature of each type of lithofacies, specifically including:

[0121] For the first sample subset, based on the attribute features of each type of lithofacies, calculating the deviation of each attribute feature from the same attribute features in other types of lithofacies;

[0122] According to the deviation, the total deviation of each attribute characteristic of each type of lithofacies is calculated;

[0123] The weight of each attribute feature is determined according to the total deviation.

[0124] In the embodiment of the present application, in a first sample subset, the sum of the differences between a certain characteristic parameter of a certain lithofacies and the characteristic parameters of other lithofacies is called the deviation. Therefore, for each first sample subset, based on each attribute feature of each type of lithofacies, i.e., the characteristic parameter, the deviation of each attribute feature from the same attribute feature in other types of lithofacies can be calculated based on the following formula 8:

[0125]

[0126] In the formula, f ij is the first sample subset N t The deviation of the jth attribute characteristic of the i-th lithofacies in r ij is the first sample subset N t The jth attribute characteristic of the i-th lithofacies; kj is the first sample subset N t The jth attribute feature of the kth lithofacies in; c is the first sample subset N t Total number of medium lithofacies types.

[0127] After obtaining the deviation of each attribute feature of each type of lithofacies in the above manner, for each first sample subset, the total deviation of each attribute feature of each type of lithofacies is calculated based on the following formula 9:

[0128]

[0129] Where f is the total deviation of each attribute characteristic of each lithofacies; G j is the weight of the jth attribute feature; m is the total number of attribute features; c is the first sample subset N t Total number of medium lithofacies types.

[0130] By using the above formula 9, the weight of each attribute corresponding to the maximum value of the total deviation f is calculated, and the weight of each attribute feature is determined.

[0131] S104: Calculate the weighted characteristic parameters of each type of lithofacies according to the weights of the attribute characteristics to obtain a second sample subset.

[0132] In the above step S104, the weighted characteristic parameters of each type of lithofacies are calculated according to the weights of the attribute characteristics to obtain the second sample subset, which specifically includes:

[0133] According to the weights of the attribute characteristics, the weighted characteristic parameters of each type of lithofacies are calculated based on the following formula 2:

[0134]

[0135] In the formula, rt i The second sample subset Mt The weighted characteristic parameter of the i-th lithofacies; r ij is the first sample subset N t The jth weighted characteristic parameter of the i-th lithofacies; The weight of the jth weighted feature parameter corresponding to the maximum value of the total deviation of all the features;

[0136] For each of the first sample subsets, a second sample subset is obtained according to the weighted characteristic parameters of each type of lithofacies.

[0137] In the embodiment of the present application, according to the weights of the attribute features of each type of lithofacies in each first sample subset obtained in step S103, based on the above formula 2, the weighted characteristic parameters of each type of lithofacies are calculated, thereby replacing the attribute features of each type of lithofacies in the first sample subset with the weighted characteristic parameters of each type of lithofacies to obtain the second sample subset. For each first sample subset, the above S102 to S104 are executed to obtain multiple second sample subsets, which are recorded as M t .

[0138] S105: Input each of the second sample subsets into a base classifier trained based on the resampled lithofacies sample data set for identification, and determine the well logging lithofacies identification probability results in each of the base classifiers.

[0139] In an embodiment of the present application, the obtained multiple second sample subsets are input into multiple base classifiers for well logging lithofacies identification according to the correspondence between one second sample subset and one trained base classifier, and the corresponding well logging lithofacies identification probability results are obtained in each base classifier.

[0140] In the embodiment of the present application, the trained base classifiers are obtained specifically by the following method:

[0141] Establish a historical lithofacies sample data set based on the well logging attributes corresponding to each type of rock obtained in advance;

[0142] Resampling the historical lithofacies sample data set according to a preset non-equilibrium index to obtain a resampled historical sample data set, and sampling to obtain a plurality of first historical sample subsets;

[0143] For each of the first historical sample subsets, the mean values ​​of the well logging attributes corresponding to all sample data of each type of lithofacies are calculated to obtain the attribute characteristics of each type of lithofacies;

[0144] Based on the attribute characteristics of each type of lithofacies, determining the weight of each attribute characteristic;

[0145] According to the weights of the attribute characteristics, weighted characteristic parameters of each type of lithofacies are calculated to obtain a second historical sample subset;

[0146] Input each of the second historical sample subsets into a base classifier for identification, and determine the historical well logging lithofacies identification probability results in each base classifier;

[0147] Compare the historical logging lithofacies recognition probability results in each base classifier with the real results of the logging lithofacies obtained in advance to determine whether the prediction accuracy of each base classifier reaches a preset threshold;

[0148] If so, the trained base classifiers are obtained.

[0149] In the embodiment of the present application, during the training process of each base classifier, historical core data of a certain historical research area is collected in advance, the lithofacies corresponding to the core are identified, and the main lithofacies types are determined. According to the logging response characteristics of the historical research area, the logging curve required for lithofacies identification is selected, and a lithofacies sample data set is established on this basis. Among them, the lithofacies data of a depth and its corresponding logging curve value are taken as a sample data.

[0150] After establishing the historical lithofacies sample data set and normalizing the logging curves, it is also necessary to resample the non-balanced historical lithofacies sample data set, combining oversampling and undersampling methods to balance the number of different types of lithofacies samples.

[0151] In the embodiment of the present application, the historical lithofacies sample data set is resampled according to the preset non-equilibrium index to obtain a resampled historical sample data set, and a plurality of first historical sample subsets are sampled, including:

[0152] Determine the majority lithofacies and the minority lithofacies according to the number of samples of different lithofacies types in the historical lithofacies sample data set;

[0153] According to the preset non-equilibrium index, most types of lithofacies are undersampled and minority types of lithofacies are oversampled until the number of samples of different lithofacies types in the historical lithofacies sample data set is balanced, a resampled historical sample data set is obtained, and multiple first historical sample subsets are sampled.

[0154] In the embodiment of the present application, the majority lithofacies and minority lithofacies, as well as the number of samples corresponding to the majority lithofacies and minority lithofacies, are determined based on the number of samples of different lithofacies types in the historical lithofacies sample data set obtained by statistics. The majority lithofacies is denoted as P, and its number of samples is denoted as N. P ; The minority lithofacies is recorded as N, and the number of samples is recorded as N N The sample number imbalance index is set to the preset imbalance index ε. When the ratio of the number of samples of the majority lithofacies to the number of samples of the minority lithofacies is greater than ε, that is, N P / N NWhen ε > ε, the majority lithofacies P is undersampled, while the minority lithofacies N is oversampled. The specific implementation method is as follows:

[0155] In the entire historical lithofacies sample data set, each sample in the minority lithofacies N is recorded as N i , for each sample in the minority lithofacies N, calculate each sample N i The number of samples of non-minority lithofacies N in the n nearest neighbor samples is recorded as n1. If n1 = n, then the corresponding sample N i Marked as noise data; if n / 2≤n1<n, then the corresponding sample N i Marked as boundary data, if 0≤n1<n / 2, then the corresponding sample N i Marked as safe data.

[0156] In the entire historical lithofacies sample data set, each sample in the majority lithofacies P is recorded as P i , for each sample in the majority lithofacies P, calculate each sample P i The number of samples of non-majority lithofacies P among the n nearest neighbor samples is recorded as n1. If n1 = n, then the corresponding sample P i Marked as noise data; if n / 2≤n1<n, then the corresponding sample P i Marked as boundary data; if 0≤n1<n / 2, the corresponding sample P i Marked as safe data.

[0157] The samples marked as noise data are removed from the sample data set of the majority lithofacies P to complete the undersampling of the majority lithofacies samples. At this time, the number of samples of lithofacies P is N. P1 .

[0158] After the above undersampling steps, the minority lithofacies N are oversampled. New data are generated only for the boundary data of the minority lithofacies N and its neighbors. i From the nearest neighbor samples, select k nearest neighbor samples belonging to the minority lithofacies N, and then randomly select j nearest neighbor samples from the k nearest neighbor samples to calculate the sample N of the boundary data i The distance s to each neighboring sample j (j represents N i The jth nearest neighbor sample of , the value range of j is 1~k), and based on the following formula 10, the random number r j (0<r j <1)Generate(N P1 -N N ) / ε new sample data:

[0159] N j =N i+r j ×s j Formula 10;

[0160] Where N j is the jth sample of the synthetic minority lithofacies N; r j is a random number, 0<r j <1;s j is the sample N of the boundary data i The distance to each neighboring sample.

[0161] Repeat the above steps of generating new data for the boundary data of the minority lithofacies N and its neighbors until the number of samples of different lithofacies types in the historical lithofacies sample data set is balanced, and obtain the resampled sample data set. Sampling is performed on the resampled sample data set to form multiple first historical sample subsets.

[0162] After obtaining multiple first historical sample subsets in the above manner, the attribute features of each first historical sample subset after resampling are weighted by constructing a feature weight matrix to establish a second historical sample set. The specific implementation method refers to steps S102 to S104 and will not be repeated here.

[0163] In an embodiment of the present application, a decision tree is used as a base classifier, and the number of base classifiers is selected according to actual needs. The obtained multiple second historical sample subsets are input into a base classifier according to the corresponding relationship of one second historical sample subset, and the multiple second historical sample subsets are input into multiple base classifiers for well logging lithofacies identification model training, and the corresponding well logging lithofacies identification probability results are obtained in each base classifier.

[0164] Compare the corresponding historical logging lithofacies identification probability results obtained by training in each base classifier with the real results of the logging lithofacies obtained in advance, determine the prediction accuracy of each base classifier for logging lithofacies identification, and judge whether the prediction accuracy reaches the preset threshold. If so, the trained base classifiers are obtained. If not, adjust the parameters of each base classifier and repeat the above model training process, that is, repeatedly input multiple second historical sample subsets into multiple base classifiers for training, obtain the corresponding historical logging lithofacies identification probability results and compare them with the real results of the logging lithofacies obtained in advance. Among them, the preset threshold is set according to the actual situation. For example, if the accuracy reaches 80%, it is considered that the base classifier is trained.

[0165] In the embodiment of the present application, after obtaining the trained base classifiers, the method further includes:

[0166] Determine the precision and recall of each base classifier according to the historical logging lithofacies recognition probability results in each trained base classifier and the real results of the logging lithofacies obtained in advance;

[0167] According to the precision and recall of each base classifier, the well logging lithofacies identification performance index of each base classifier is determined based on the following formula 3:

[0168]

[0169] Where, F is the logging lithofacies recognition performance index of a base classifier; P p is the precision of a base classifier; P r is the recall rate of a base classifier;

[0170] According to the well logging lithofacies identification performance index of each base classifier, the combination weight of each base classifier is determined based on the following formula 4:

[0171]

[0172] In the formula, w t is the combined weight of the tth base classifier; F t is the well logging lithofacies recognition performance index of the t-th base classifier; F min The minimum value of the well logging lithofacies identification performance index of all base classifiers; F max The maximum value of the well logging lithofacies identification performance index of all base classifiers.

[0173] In the embodiment of the present application, after the trained base classifiers are obtained, the accuracy rate P of each base classifier for well logging lithofacies identification is calculated based on the following formula 11 according to the historical well logging lithofacies identification probability results in each trained base classifier and the real results of the well logging lithofacies obtained in advance. p and recall P r :

[0174]

[0175] In the formula, the accuracy P p The recall rate P is the ratio of the number of samples correctly identified as a certain type of lithofacies to the number of samples identified as that type of lithofacies. r It is the ratio of the number of correctly identified lithofacies samples to the actual number of samples of that lithofacies.

[0176] After obtaining the precision and recall of each base classifier in the above manner, the well logging lithofacies recognition performance index of each base classifier is determined based on the above formula 3, and the combination weight of each base classifier in the result combination (i.e., the combination of prediction results obtained in each base classifier) ​​is determined based on the above formula 4 according to the well logging lithofacies recognition performance index of each base classifier. After obtaining the combination weight of each base classifier, the probability that the sample obtained by each base classifier belongs to each lithofacies can be weighted to obtain the probability that the sample belongs to each lithofacies, and then the lithofacies corresponding to the maximum probability is used as the classification result of the entire model. Based on the lithofacies recognition performance of the base classifier, the combination mechanism of the base classifier in the ensemble learning algorithm is adjusted using the weighted voting method, which can significantly improve the lithofacies recognition ability of the model.

[0177] S106: Determine the lithofacies types of actual wells at different depths in the study area according to the predetermined combined weight of each base classifier and the corresponding logging lithofacies identification probability result.

[0178] In the above step S106, the method of determining the lithofacies types of different depths of actual wells in the study area according to the predetermined combined weight of each base classifier and the corresponding logging lithofacies identification probability result specifically includes:

[0179] For each well logging sampling depth of the actual wells in the study area, the probability of each type of lithofacies is determined based on the following formula 5 according to the pre-determined combined weight of each base classifier and the corresponding well logging lithofacies identification probability result:

[0180]

[0181] Where p is the probability of each type of lithofacies; w i is the weight of the i-th classifier; p i is the probability of each lithofacies in the i-th classifier; T is the number of the second sample subset;

[0182] For each logging sampling depth of the actual wells in the study area, according to the corresponding probability of each type of lithofacies, the lithofacies with the highest probability is taken as the corresponding lithofacies type to determine the lithofacies types of different depths of the actual wells in the study area.

[0183] In the embodiment of the present application, at a certain logging sampling depth point of the actual well in the study area, the logging lithofacies identification probability result obtained by logging lithofacies identification in each base classifier at the depth point is combined with the combined weights of each base classifier, and based on the above formula 5, the final probability of the depth point belonging to each lithofacies is determined, and the lithofacies type corresponding to the logging lithofacies identification result with the largest final probability is selected as the lithofacies type of the depth point. The above process of determining the lithofacies type at the depth according to the combined weight of each base classifier and the corresponding logging lithofacies identification probability result is repeated at different logging sampling depths of the actual well in the study area, until the logging lithofacies identification of the entire depth interval of the actual well in the study area is completed, and the lithofacies types of different depths of the actual well in the study area are determined.

[0184] In the embodiment of the present application, in a certain research area, the well logging lithofacies intelligent identification method of non-balanced samples is used to perform well logging lithofacies identification, and the identification results obtained are as follows: Figure 2 As shown in the figure, by comparing the lithofacies profile obtained from core or logging with the lithofacies profile identified using the ensemble learning algorithm, it can be seen that the accuracy of lithofacies identification using the trained ensemble learning algorithm is high, which greatly improves the lithofacies identification effect.

[0185] The method for intelligent identification of well logging lithofacies of unbalanced samples provided in the embodiment of the present application is to weight the lithofacies characteristics of multiple first sample subsets constructed based on the well logging attributes of actual wells in the study area to obtain weighted characteristic parameters, thereby establishing multiple second sample subsets, and inputting the multiple second sample subsets into multiple base classifiers trained based on the lithofacies sample data set after resampling the unbalanced samples for identification, thereby determining the lithofacies types of different depths of actual wells in the study area. The method adopts an integrated learning algorithm, constructs a feature weight matrix to weight the lithofacies characteristic parameter values ​​of the first sample subset to establish a second sample subset, and the base classifiers trained based on the sample data set trained to achieve a balanced number of lithofacies samples of different types by resampling the unbalanced samples, and the well logging lithofacies identification performance of each base classifier determines the combination mechanism of the integrated learning algorithm, which can effectively identify the lithofacies types of different depths of actual wells in the study area, has a high overall identification accuracy, greatly improves the lithofacies identification effect of unbalanced samples, is simple and easy to implement, has a high actual processing efficiency, and is easy to promote and apply.

[0186] Embodiment 2

[0187] Based on the same inventive concept, the present application also provides a method for training a well logging lithofacies intelligent recognition model for non-balanced samples, referring to Figure 3 As shown, the method includes:

[0188] S201: Establish a historical lithofacies sample data set based on the well logging attributes corresponding to each type of rock obtained in advance.

[0189] S202: resampling the historical lithofacies sample data set according to a preset non-equilibrium index to obtain a resampled historical sample data set, and sampling to obtain a plurality of first historical sample subsets.

[0190] S203: For each of the first historical sample subsets, calculate the mean of each well logging attribute corresponding to all sample data of each type of lithofacies to obtain each attribute feature of each type of lithofacies.

[0191] S204: Determine the weight of each attribute feature based on each attribute feature of each type of lithofacies.

[0192] S205: Calculate the weighted characteristic parameters of each type of lithofacies according to the weights of the attribute characteristics to obtain a second historical sample subset.

[0193] S206: Input each of the second historical sample subsets into a base classifier for identification, and determine the historical well logging lithofacies identification probability results in each base classifier.

[0194] S207: Compare the historical logging lithofacies recognition probability results in each base classifier with the actual logging lithofacies results obtained in advance to determine whether the prediction accuracy of each base classifier reaches a preset threshold. If so, execute S208; if not, execute S209.

[0195] S208: Obtain the trained base classifiers.

[0196] S209: Adjust the parameters of each base classifier and repeat the above model training process of S206 to S207.

[0197] Embodiment 3

[0198] Based on the same inventive concept, the present application also provides a well logging lithofacies intelligent identification device for non-equilibrium samples, referring to Figure 4 As shown, the device comprises:

[0199] The first sample establishment module 101 is used to establish a lithofacies sample data set according to the well logging attributes corresponding to various types of rocks in actual wells in the study area obtained in advance, and to sample and obtain a plurality of first sample subsets;

[0200] A first determination module 102 is used to calculate the mean of each well logging attribute corresponding to all sample data of each type of lithofacies for each of the first sample subsets, so as to obtain each attribute feature of each type of lithofacies;

[0201] The second determination module 103 is used to determine the weight of each attribute feature based on each attribute feature of each type of lithofacies;

[0202] The second sample establishment module 104 is used to calculate the weighted characteristic parameters of each type of lithofacies according to the weights of the attribute characteristics to obtain a second sample subset;

[0203] The third determination module 105 is used to input each of the second sample subsets into a base classifier trained based on the resampled lithofacies sample data set for identification, and determine the well logging lithofacies identification probability results in each of the base classifiers;

[0204] The fourth determination module 106 is used to determine the lithofacies types at different depths of actual wells in the study area according to the predetermined combined weight of each base classifier and the corresponding logging lithofacies identification probability result.

[0205] Embodiment 4

[0206] Based on the same inventive concept, the present application also provides a well logging lithofacies intelligent identification model training device for non-balanced samples, referring to Figure 5 As shown, the device comprises:

[0207] The third sample establishment module 201 is used to establish a historical lithofacies sample data set according to the well logging attributes corresponding to various types of rocks obtained in advance;

[0208] A sampling module 202 is used to resample the historical lithofacies sample data set according to a preset non-equilibrium index to obtain a resampled historical sample data set, and to sample to obtain a plurality of first historical sample subsets;

[0209] The fifth determination module 203 is used to calculate the mean of each well logging attribute corresponding to all sample data of each type of lithofacies for each of the first historical sample subsets, so as to obtain each attribute feature of each type of lithofacies;

[0210] The sixth determination module 204 is used to determine the weight of each attribute feature based on each attribute feature of each type of lithofacies;

[0211] The seventh determination module 205 is used to calculate the weighted characteristic parameters of each type of lithofacies according to the weights of the attribute characteristics to obtain a second historical sample subset;

[0212] An eighth determination module 206 is used to input each of the second historical sample subsets into a base classifier for identification, and determine the historical well logging lithofacies identification probability results in each base classifier;

[0213] A judgment module 207 is used to compare the historical logging lithofacies recognition probability results in each base classifier with the real results of the logging lithofacies obtained in advance, and judge whether the prediction accuracy of each base classifier reaches a preset threshold;

[0214] The ninth determination module 208 is used to obtain the trained base classifiers if the prediction accuracy of each base classifier determined by the determination module reaches a preset threshold.

[0215] Embodiment 5

[0216] Based on the same inventive concept, an embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, it implements the method for intelligent identification of well logging lithofacies of unbalanced samples as described in the above embodiment 1, and / or the method for training a model for intelligent identification of well logging lithofacies of unbalanced samples as described in the above embodiment 2.

[0217] Embodiment 6

[0218] Based on the same inventive concept, an embodiment of the present application also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for intelligent identification of well logging lithofacies of unbalanced samples as described in the above-mentioned embodiment 1 and / or the method for training a model for intelligent identification of well logging lithofacies of unbalanced samples as described in the above-mentioned embodiment 2 are implemented.

[0219] Embodiment 7

[0220] Based on the same inventive concept, an embodiment of the present application also provides a computer program product comprising instructions. When the computer program product is run on a computer device, the computer device executes the method for intelligent identification of well logging lithofacies of unbalanced samples as described in the above embodiment 1, and / or the method for training a model for intelligent identification of well logging lithofacies of unbalanced samples as described in the above embodiment 2.

[0221] Embodiment 8

[0222] Based on the same inventive concept, an embodiment of the present application further provides a chip, the chip comprising a processor and a communication interface, the communication interface and the processor are coupled, and the processor is used to run a computer program or instruction to implement the method for intelligent identification of well logging lithofacies of unbalanced samples as described in the above embodiment 1, and / or the method for training a model for intelligent identification of well logging lithofacies of unbalanced samples as described in the above embodiment 2.

[0223] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) that contain computer-usable program code.

[0224] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0225] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0226] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0227] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for intelligent identification of well logging lithofacies of non-equilibrium samples, characterized in that: include: According to the well logging attributes corresponding to various types of rocks in actual wells in the study area obtained in advance, a lithofacies sample data set is established, and a plurality of first sample subsets are obtained by sampling; For each of the first sample subsets, calculating the mean of each well logging attribute corresponding to all sample data of each type of lithofacies, and obtaining each attribute feature of each type of lithofacies; Based on the attribute characteristics of each type of lithofacies, determining the weight of each attribute characteristic; According to the weights of the attribute characteristics, weighted characteristic parameters of each type of lithofacies are calculated to obtain a second sample subset; Input each of the second sample subsets into a base classifier trained based on the resampled lithofacies sample data set for identification, and determine the well logging lithofacies identification probability results in each of the base classifiers; The lithofacies types of actual wells at different depths in the study area are determined according to the predetermined combined weight of each base classifier and the corresponding logging lithofacies identification probability results.

2. The method according to claim 1, characterized in that The trained base classifiers are obtained in the following way: Establish a historical lithofacies sample data set based on the well logging attributes corresponding to each type of rock obtained in advance; Resampling the historical lithofacies sample data set according to a preset non-equilibrium index to obtain a resampled historical sample data set, and sampling to obtain a plurality of first historical sample subsets; For each of the first historical sample subsets, the mean values ​​of the well logging attributes corresponding to all sample data of each type of lithofacies are calculated to obtain the attribute characteristics of each type of lithofacies; Based on the attribute characteristics of each type of lithofacies, determining the weight of each attribute characteristic; According to the weights of the attribute characteristics, weighted characteristic parameters of each type of lithofacies are calculated to obtain a second historical sample subset; Input each of the second historical sample subsets into a base classifier for identification, and determine the historical well logging lithofacies identification probability results in each base classifier; Comparing the historical logging lithofacies recognition probability results in each base classifier with the actual logging lithofacies results obtained in advance, and judging whether the prediction accuracy of each base classifier reaches a preset threshold; If so, the trained base classifiers are obtained.

3. The method according to claim 2, characterized in that The historical lithofacies sample data set is resampled according to the preset non-equilibrium index to obtain a resampled historical sample data set, and a plurality of first historical sample subsets are sampled, including: Determine the majority lithofacies and the minority lithofacies according to the number of samples of different lithofacies types in the historical lithofacies sample data set; According to the preset non-equilibrium index, most types of lithofacies are undersampled and minority types of lithofacies are oversampled until the number of samples of different lithofacies types in the historical lithofacies sample data set is balanced, a resampled historical sample data set is obtained, and multiple first historical sample subsets are sampled.

4. The method according to claim 1, characterized in that For each of the first sample subsets, the mean of each well logging attribute corresponding to all sample data of each type of lithofacies is calculated to obtain each attribute feature of each type of lithofacies, including: For each of the first sample subsets, the mean values ​​of the well logging attributes corresponding to all sample data of each type of lithofacies are calculated based on the following formula 1: In the formula, r ij is the mean value of the jth well logging attribute of the lth training sample of the i-th lithofacies in the first sample subset; ijl is the jth logging attribute value of the lth sample in the i-th lithofacies in the first sample subset; n c is the number of samples in the first sample subset; The mean of each logging attribute corresponding to all sample data of each type of lithofacies is taken as the attribute characteristics of each type of lithofacies.

5. The method according to claim 1, characterized in that Determining the weight of each attribute feature based on each attribute feature of each type of lithofacies includes: For the first sample subset, based on the attribute features of each type of lithofacies, calculating the deviation of each attribute feature from the same attribute features in other types of lithofacies; According to the deviation, the total deviation of each attribute characteristic of each type of lithofacies is calculated; The weight of each attribute feature is determined according to the total deviation.

6. The method according to claim 1, characterized in that The weighted characteristic parameters of each type of lithofacies are calculated according to the weights of the attribute characteristics to obtain a second sample subset, including: According to the weights of the attribute characteristics, the weighted characteristic parameters of each type of lithofacies are calculated based on the following formula 2: In the formula, rt i is the weighted characteristic parameter of the i-th type of lithofacies in the second sample subset; r ij is the jth weighted characteristic parameter of the i-th type of lithofacies in the first sample subset; The weight of the jth weighted feature parameter corresponding to the maximum value of the total deviation of all the features; For each of the first sample subsets, a second sample subset is obtained according to the weighted characteristic parameters of each type of lithofacies.

7. The method according to claim 2, characterized in that After the trained base classifiers are obtained, the method further includes: Determine the precision and recall of each base classifier according to the historical logging lithofacies recognition probability results in each trained base classifier and the real results of the logging lithofacies obtained in advance; According to the precision and recall of each base classifier, the well logging lithofacies identification performance index of each base classifier is determined based on the following formula 3: Where, F is the well logging lithofacies recognition performance index of a base classifier; P p is the precision of a base classifier; P r is the recall rate of a base classifier; According to the well logging lithofacies identification performance index of each base classifier, the combination weight of each base classifier is determined based on the following formula 4: In the formula, w t is the combined weight of the tth base classifier; F t is the well logging lithofacies recognition performance index of the t-th base classifier; F min The minimum value of the well logging lithofacies identification performance index of all base classifiers; F max The maximum value of the well logging lithofacies identification performance index of all base classifiers.

8. The method according to claim 1, characterized in that After establishing the lithofacies sample data set based on the well logging attributes corresponding to various types of rocks in the actual wells in the study area obtained in advance, the following is also included: All the logging curves in the lithofacies sample data set are normalized.

9. The method according to any one of claims 1 to 8, characterized in that The method of determining the lithofacies types of different depths of actual wells in the study area according to the predetermined combined weight of each base classifier and the corresponding logging lithofacies identification probability result includes: For each well logging sampling depth of the actual wells in the study area, the probability of each type of lithofacies is determined based on the following formula 5 according to the pre-determined combined weight of each base classifier and the corresponding logging lithofacies identification probability result: Where p is the probability of each type of lithofacies; w i is the weight of the i-th classifier; p i is the probability of each lithofacies in the i-th classifier; T is the number of the second sample subset; For each logging sampling depth of the actual wells in the study area, according to the corresponding probability of each type of lithofacies, the lithofacies with the highest probability is taken as the corresponding lithofacies type to determine the lithofacies types of different depths of the actual wells in the study area.

10. A method for training a well logging lithofacies intelligent identification model for non-balanced samples, characterized in that: include: Establish a historical lithofacies sample data set based on the well logging attributes corresponding to each type of rock obtained in advance; Resampling the historical lithofacies sample data set according to a preset non-equilibrium index to obtain a resampled historical sample data set, and sampling to obtain a plurality of first historical sample subsets; For each of the first historical sample subsets, the mean values ​​of the well logging attributes corresponding to all sample data of each type of lithofacies are calculated to obtain the attribute characteristics of each type of lithofacies; Based on the attribute characteristics of each type of lithofacies, determining the weight of each attribute characteristic; According to the weights of the attribute characteristics, weighted characteristic parameters of each type of lithofacies are calculated to obtain a second historical sample subset; Input each of the second historical sample subsets into a base classifier for identification, and determine the historical well logging lithofacies identification probability results in each base classifier; Comparing the historical logging lithofacies recognition probability results in each base classifier with the actual logging lithofacies results obtained in advance, and judging whether the prediction accuracy of each base classifier reaches a preset threshold; If so, the trained base classifiers are obtained.

11. An intelligent identification device for well logging lithofacies of non-equilibrium samples, characterized in that: include: A first sample establishment module is used to establish a lithofacies sample data set according to the well logging attributes corresponding to various types of rocks in actual wells in the study area obtained in advance, and to sample and obtain a plurality of first sample subsets; A first determination module is used to calculate the mean of each well logging attribute corresponding to all sample data of each type of lithofacies for each of the first sample subsets, so as to obtain each attribute feature of each type of lithofacies; A second determination module is used to determine the weight of each attribute feature based on each attribute feature of each type of lithofacies; A second sample establishment module is used to calculate the weighted characteristic parameters of each type of lithofacies according to the weights of the attribute characteristics to obtain a second sample subset; A third determination module is used to input each of the second sample subsets into a base classifier trained based on the resampled lithofacies sample data set for identification, and determine the well logging lithofacies identification probability results in each of the base classifiers; The fourth determination module is used to determine the lithofacies types of different depths of actual wells in the study area according to the predetermined combined weight of each base classifier and the corresponding logging lithofacies identification probability result.

12. A well logging lithofacies intelligent identification model training device for non-balanced samples, characterized in that: include: The third sample establishment module is used to establish a historical lithofacies sample data set according to the well logging attributes corresponding to various types of rocks obtained in advance; A sampling module, used for resampling the historical lithofacies sample data set according to a preset non-equilibrium index to obtain a resampled historical sample data set, and sampling to obtain a plurality of first historical sample subsets; A fifth determination module, configured to calculate, for each of the first historical sample subsets, the mean values ​​of the well logging attributes corresponding to all sample data of each type of lithofacies, and obtain the attribute characteristics of each type of lithofacies; A sixth determination module, used to determine the weight of each attribute feature based on each attribute feature of each type of lithofacies; A seventh determination module is used to calculate the weighted characteristic parameters of each type of lithofacies according to the weights of the attribute characteristics to obtain a second historical sample subset; An eighth determination module is used to input each of the second historical sample subsets into a base classifier for identification, and determine the historical well logging lithofacies identification probability results in each base classifier; A judgment module, used to compare the historical logging lithofacies recognition probability results in each base classifier with the real results of the logging lithofacies obtained in advance, and judge whether the prediction accuracy of each base classifier reaches a preset threshold; The ninth determination module is used to obtain the trained base classifiers if the prediction accuracy of the base classifiers determined by the judgment module reaches a preset threshold.

13. A computer-readable storage medium, wherein instructions are stored in the computer-readable storage medium. When the instructions are executed on a terminal, the terminal executes the method for intelligent identification of well logging lithofacies of unbalanced samples as described in any one of claims 1 to 9, and / or the method for training a model for intelligent identification of well logging lithofacies of unbalanced samples as described in claim 10.

14. A computer device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for intelligent identification of well logging lithofacies of unbalanced samples according to any one of claims 1 to 9 and / or the method for training a model for intelligent identification of well logging lithofacies of unbalanced samples according to claim 10 is implemented.

15. A computer program product comprising instructions, which, when executed on a computer device, enables the computer device to execute the method for intelligent identification of well logging lithofacies of unbalanced samples as claimed in any one of claims 1 to 9, and / or the method for training a model for intelligent identification of well logging lithofacies of unbalanced samples as claimed in claim 10.

16. A chip, comprising a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run a computer program or instruction to implement the method for intelligent identification of well logging lithofacies of unbalanced samples according to any one of claims 1 to 9, and / or the method for training a model for intelligent identification of well logging lithofacies of unbalanced samples according to claim 10.

Images