Reservoir fluid type identification method, device, equipment and medium

Abstract

The embodiment of the invention provides a reservoir fluid type identification method and device, equipment and a medium. The method comprises the steps that logging data and spatial information obtained by logging a target position of a reservoir are obtained, then the logging data and the spatial information are input into a pre-obtained fluid type recognition model for recognition processing, and the fluid type of the target position is obtained, the fluid type identification model is obtained by training based on a semi-supervised learning method and an AdaBoost-neural network algorithm. According to the method, the dependence on physical sampling and manual interpretation is reduced, and an intelligent reservoir fluid type identification method capable of rapidly and accurately identifying reservoir fluid types by utilizing limited well logging data of known fluid types and a large number of well logging data of unknown fluid types is developed.

Description

Technical Field

[0001] The present application relates to the field of oil and gas exploration and production, and in particular to a method, device, equipment and medium for identifying reservoir fluid types. Background Art

[0002] Natural gas hydrates and oil are both important energy resources, and their development and utilization have attracted widespread attention from energy scientists and industry around the world. Through seismic exploration, logging data analysis and other technologies, the types of fluids such as oil, gas, and water in the reservoir can be identified, providing key information for the discovery and development of oil and gas reservoirs. In the exploration and development of natural gas hydrates and oil, accurately identifying the type of fluid in the reservoir is a key step in evaluating its economic value and development feasibility.

[0003] Traditional reservoir fluid identification technology requires physical sampling and manual interpretation of new wells to identify the reservoir fluid type of the well. The existing technology lacks an intelligent identification method that can use limited logging data of known fluid types and a large amount of logging data of unknown fluid types to quickly and accurately identify reservoir fluid types. Summary of the invention

[0004] The embodiments of the present application provide a method, apparatus, device and medium for identifying reservoir fluid types, so as to achieve the effect of using limited historical logging data of known fluid types and a large amount of historical logging data of unknown fluid types to train a model, and using the model to intelligently, quickly and accurately identify reservoir fluid types, thereby reducing the reliance on physical sampling and manual interpretation during the identification process.

[0005] In a first aspect, an embodiment of the present application provides a method for identifying a reservoir fluid type, comprising:

[0006] Acquire logging data and spatial information obtained by logging a target location of a reservoir;

[0007] Inputting the logging data and the spatial information into a pre-acquired fluid type recognition model for recognition processing to obtain the fluid type at the target location;

[0008] Wherein, the fluid type recognition model is obtained based on semi-supervised learning method and AdaBoost-neural network algorithm training.

[0009] In a possible implementation manner, the spatial information includes well logging location information and well logging depth for logging the target location; the well logging location information includes the longitude and latitude of the well logging location.

[0010] In a possible implementation manner, before inputting the well logging data and the spatial information into a pre-acquired fluid type recognition model for recognition processing, the method further includes:

[0011] Acquire a historical well logging data set and a historical fluid type set, wherein the historical well logging data set includes well logging data obtained by multiple well loggings at different locations of different reservoirs and spatial information of each well logging, and the historical fluid type set includes fluid types analyzed by each well logging at different locations corresponding to the historical well logging data set;

[0012] Based on the historical well logging data set and the historical fluid type set, a sample set is obtained, wherein the sample set includes a first part of samples and a second part of samples, wherein the first part of samples includes well logging data, spatial information corresponding to the well logging data, and fluid types, and the second part of samples includes well logging data and spatial information corresponding to the well logging data;

[0013] Constructing an initial fluid type recognition model based on the AdaBoost-neural network algorithm;

[0014] Based on the sample set, the initial fluid type recognition model is trained to obtain the fluid type recognition model.

[0015] In a possible implementation manner, the performing model training on the initial fluid type recognition model based on the sample set to obtain the fluid type recognition model includes:

[0016] According to the first part of samples, the initial fluid type recognition model is trained, and model parameters are adjusted to obtain a trained fluid type recognition model;

[0017] Using a semi-supervised learning method, the trained fluid type recognition model is used to predict the second part of samples respectively, so as to obtain a plurality of pseudo labels corresponding to the second part of samples;

[0018] A new sample set is constructed according to the first part of samples, the second part of samples and the multiple pseudo labels, and the initial fluid type recognition model is retrained by the new sample set to obtain the fluid type recognition model.

[0019] In a possible implementation manner, the acquiring of a historical well logging data set and a historical fluid type set includes:

[0020] Based on the exploration data of the drilled layer section or the drilled single well, multiple well logging data and the spatial information of each well logging are obtained;

[0021] According to the test production data and coring analysis results of each test production at different locations, the fluid type corresponding to each well logging is determined to obtain the historical fluid type set;

[0022] Preprocessing the plurality of well logging data to obtain a plurality of processed well logging data;

[0023] The processed multiple well logging data and the spatial information of each well logging are standardized to obtain the historical well logging data set.

[0024] In one possible implementation, the preprocessing includes the following processing steps:

[0025] Cleaning erroneous data and placeholders in the plurality of well logging data;

[0026] removing abnormal data from the plurality of well logging data;

[0027] The missing values ​​in the plurality of well logging data are supplemented by other well logging data within a preset depth range.

[0028] In a possible implementation, the method further includes:

[0029] Based on the logging data and spatial information of the target location, and the fluid type of the target location, the fluid type identification model is optimized to obtain a new fluid type identification model.

[0030] In a second aspect, an embodiment of the present application provides a device for identifying a reservoir fluid type, comprising:

[0031] The first acquisition module: acquires the logging data and spatial information obtained by logging the target position of the reservoir;

[0032] Identification module: input the logging data and the spatial information into a pre-acquired fluid type identification model for identification processing to obtain the fluid type of the target location; wherein the fluid type identification model is obtained by training based on the AdaBoost-neural network algorithm.

[0033] In a third aspect, an embodiment of the present application provides a device for identifying a reservoir fluid type, including: a memory, a processor;

[0034] The memory stores computer-executable instructions;

[0035] The processor executes the computer-executable instructions stored in the memory, so that the processor executes the above first aspect and / or various possible implementations of the first aspect.

[0036] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the first aspect above and / or various possible implementations of the first aspect.

[0037] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the above first aspect and / or various possible implementation methods of the first aspect.

[0038] The reservoir fluid type identification method, device, equipment and medium provided in the embodiments of the present application obtain the logging data and spatial information obtained by logging the target position of the reservoir in advance, and input the logging data and the spatial information into the pre-acquired fluid type identification model trained based on the AdaBoost-neural network algorithm for identification processing, so as to obtain the technical means of the fluid type of the target position, so as to achieve the effect of using limited historical logging data of known fluid types and a large amount of historical logging data of unknown fluid types to train the model, and use the model to intelligently, quickly and accurately identify the reservoir fluid type, thereby reducing the dependence on physical sampling and manual interpretation in the identification process. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0040] Figure 1 A flow chart of Embodiment 1 of the method for identifying reservoir fluid type provided in the present application;

[0041] Figure 2 A flow chart of Embodiment 2 of the method for identifying reservoir fluid types provided in this application;

[0042] Figure 3 A flow chart of Embodiment 3 of the method for identifying reservoir fluid type provided in the present application;

[0043] Figure 4 A flowchart of Embodiment 4 of the method for identifying reservoir fluid types provided in the present application;

[0044] Figure 5 The comparison result of the actual and predicted layer labels after the model verification in the fifth embodiment of the reservoir fluid type identification method provided in the present application;

[0045] Figure 6 This is the classification result of the confusion matrix comparison after model verification in Example 5 of the reservoir fluid type identification method provided in this application.

[0046] Figure 7 A schematic diagram of the structure of a device for identifying reservoir fluid types provided in this application;

[0047] Figure 8 A schematic diagram of the structure of the reservoir fluid type identification device provided in this application.

[0048] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0049] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0050] First, the terms involved in this application are explained:

[0051] Cross-Validation: Cross-Validation (CV) is a commonly used method for building machine learning models and verifying model parameters. It is generally used to evaluate the performance of a machine learning model and can also be used for hyperparameter tuning. The basic idea of ​​this method is to divide the original data set into smaller subsets, which are used as training sets and test sets to train and test the model, thereby evaluating the performance of the model.

[0052] Well logging data: Well logging data is an important data generated during oil and gas exploration and development, which records the physical properties of rocks and fluids in the wellbore. Well logging data is obtained through well logging technology, which uses different physical methods (such as electrical, acoustic, nuclear, chemical, etc.) to measure various parameters in the wellbore, such as resistivity, well diameter, natural gamma, acoustic wave time difference, density, neutron, natural potential, etc. These data are crucial for understanding the properties of underground rocks, identifying oil and gas reservoirs, evaluating oil and gas reserves, and formulating subsequent development plans.

[0053] In order to clearly understand the technical solution of the present application, the solution of the prior art is introduced in detail below.

[0054] Natural gas hydrate is a potential energy source existing in marine sediments and permafrost areas on land. At the same time, petroleum is one of the most important energy and chemical raw materials in the world today. The development and utilization of both has attracted extensive attention from energy scientists and industry around the world. In the exploration and development of natural gas hydrates, accurately identifying the fluid type in the reservoir is a key step in evaluating its economic value and development feasibility.

[0055] Traditional fluid identification technologies mostly rely on drilling sampling and field testing. These methods are not only costly, but also have certain risks in technical implementation, such as the possibility of causing damage to the reservoir structure or changing the original fluid distribution. Existing logging interpretation methods such as acoustic logging and resistivity logging often rely on experienced geologists for data analysis and interpretation, which is not only time-consuming, but also its accuracy depends largely on personal experience and subjective judgment. Under complex geological conditions such as high-salinity environments and areas with mixed fluids, the accuracy of traditional logging technology and data interpretation methods will drop significantly. In addition, these methods often lack sufficient flexibility and adaptability when dealing with rapidly changing geological environments.

[0056] Based on the description of the existing reservoir fluid type identification method in the above background technology, the previous reservoir fluid type identification method lacks the ability to effectively integrate multiple logging data to quickly and effectively identify fluids in today's rapidly developing artificial intelligence technology, which limits its application in rapid decision-making scenarios. In addition, the prior art lacks a limited number of well logging data of known fluid types and a large number of well logging data of unknown fluid types to train a model, and uses the model to intelligently, quickly and accurately identify reservoir fluid types.

[0057] The method for identifying reservoir fluid types provided in the present application is to input the logging data and spatial information of the target reservoir position into the fluid type identification model obtained by training based on the semi-supervised learning method and the AdaBoost-neural network algorithm to obtain the fluid type of the target position. The present application uses spatial information as input data, so that the constructed model can not only perform efficient identification, but also has the ability to determine the fluid type in combination with spatial information. Because the large amount of high-quality acquisition of fluid types from historical logging data is often time-consuming and labor-intensive, the present application adopts a semi-supervised learning training method in the process of fluid identification, so that the model can automatically extract meaningful feature patterns from the logging data of a large amount of unknown fluid types, reducing the reliance on physical sampling and manual interpretation in the identification process. In addition, the present application applies the machine learning algorithm of the AdaBoost-neural network, in which the deep neural network has powerful feature learning and nonlinear modeling capabilities. By constructing a multi-layer neural network structure, the model can automatically learn the hierarchical feature representation in the data, capture the complex nonlinear relationship between geological parameters, and improve the recognition accuracy; and the integrated learning strategy of AdaBoost further enhances the adaptability and stability of the model under variable geological conditions. Therefore, the method provided in the present application solves the technical problems in the prior art that the reservoir fluid identification process is time-consuming and labor-intensive, is greatly influenced by subjective factors, and lacks an intelligent identification method that can use limited logging data of known fluid types and a large amount of logging data of unknown fluid types to quickly and accurately identify reservoir fluid types.

[0058] The reservoir fluid type identification method provided in this application can be applied to fluid exploration fields such as natural gas hydrate exploration and oil exploration. Especially in the middle and late stages of exploration, in order to more accurately grasp the fluid type and reserves of the reservoir, multiple wells need to be logged. Based on a small amount of historical logging data of known fluid types and a large amount of historical logging data of unknown fluid types, the reservoir fluid type identification method provided in this application can be used to identify the logging data of new wells.

[0059] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0060] Figure 1 This is a flow chart of Embodiment 1 of the method for identifying reservoir fluid types provided in this application, as shown in FIG. Figure 1 As shown, the method includes:

[0061] S101: Acquire logging data and spatial information obtained by logging a target position of a reservoir.

[0062] In this solution, it should be understood that in actual situations, if the fluid type of a certain reservoir is to be identified, the explorer needs to log the reservoir, wherein the number of wells and the logging locations are determined according to information such as geological structure and exploration targets during the logging process. When logging a well, logging can be performed at different depths of the well to obtain logging data and spatial information of multiple target locations corresponding to the well. During the identification process, the logging data and spatial information of multiple target locations can be input into a pre-trained model for reservoir fluid identification, and finally, multiple fluid types corresponding to the above multiple target locations are obtained.

[0063] In this step, if the fluid type at the target location in the reservoir is to be identified, the target location in the reservoir needs to be logged in advance. By logging the well corresponding to the target location, the logging data of the target location and the spatial information corresponding to the logging data can be obtained.

[0064] Optionally, the logging method used can be resistivity logging, caliper logging, natural gamma logging, acoustic time difference logging, density logging, neutron logging, natural potential logging, etc., and the acquired logging data may also correspond to resistivity, caliper, natural gamma ray, acoustic time difference, density, neutron ray, natural potential, etc.

[0065] In the specific application of this solution, it should be understood that logging data is a set of curves of measured data changing with well depth. Exemplarily, natural potential logging is the process of measuring the natural potential changing with well depth in a well using a natural potential logging instrument, and the measured curve is called a natural potential curve. Therefore, a position in the reservoir corresponds to a set of logging data, and the logging data is the values ​​of resistivity, well diameter, natural gamma ray, acoustic wave time difference, density, neutron ray, natural potential, etc. corresponding to the position.

[0066] Optionally, the spatial information obtained by logging the target position of the reservoir includes the logging position information and the logging depth of the target position, wherein the logging position information includes the longitude and latitude of the logging position.

[0067] In the specific application implementation of this solution, it should be understood that in order to better capture the spatial distribution characteristics of the reservoir, it is necessary to obtain the logging location information of the target location. When logging operations are performed on the well corresponding to the target location, the location of the well is the logging location information of the target location, including the longitude and latitude of the logging location. In addition, in order to explore the sedimentary law of the reservoir at the target location that changes with depth, it is necessary to obtain the logging depth of the target location, where the logging depth is the well depth corresponding to the location.

[0068] S102: input the logging data and spatial information into a pre-acquired fluid type recognition model for recognition processing to obtain the fluid type at the target location, wherein the fluid type recognition model is trained based on a semi-supervised learning method and an AdaBoost-neural network algorithm.

[0069] In this step, the previously acquired logging data and spatial information corresponding to the target position are used as input data, and this input data is input into the fluid type recognition model trained based on the semi-supervised learning method and the AdaBoost-neural network algorithm. After the fluid type recognition model recognizes the input data, the fluid type of the target position of the reservoir can be automatically obtained.

[0070] Optionally, the fluid type that can be identified by the fluid type identification model may include: oil, water, gas, hydrate or other fluids. The fluid type can be determined according to actual exploration conditions, and this solution does not impose any limitation on this.

[0071] In the specific application of this solution, it should be understood that the data input into the fluid type identification model is generally the logging data of a new well corresponding to the reservoir where the target location is located. By inputting the logging data of the new drilling well, the model can identify the fluid types at multiple locations in the reservoir corresponding to the well in real time, which can help geological engineers make more accurate resource assessments and mining decisions.

[0072] In the specific application of this scheme, it should be understood that the semi-supervised learning method and AdaBoost-neural network algorithm used in this scheme effectively improve the generalization ability and prediction accuracy of the model. The semi-supervised learning training method can enable the model to automatically extract meaningful feature patterns from the logging data of massive unknown fluid types; the deep neural network has a strong feature learning and nonlinear modeling ability. By constructing a multi-layer neural network structure, the model can automatically learn the hierarchical feature representation in the data, capture the complex nonlinear relationship between geological parameters, and improve the recognition accuracy; and the integrated learning strategy of AdaBoost further enhances the adaptability and stability of the model under variable geological conditions. Therefore, the fluid type recognition model trained by this algorithm can quickly and effectively respond to complex and variable geological conditions, and has flexibility and stability. Furthermore, the fluid type recognition technology of this scheme can support geological engineers in making accurate resource assessments and mining decisions, which will greatly improve the economic benefits and efficiency of gas hydrate and oil reservoir exploration and development, and is conducive to the sustainable development of resources.

[0073] In the reservoir fluid type identification method provided in this embodiment, the well logging data and spatial information corresponding to the pre-acquired target position are used as input to the fluid type identification model trained based on the semi-supervised learning method and the AdaBoost-neural network algorithm, and the fluid type at the target position is obtained through rapid identification by the model. This solution uses a semi-supervised learning method to train a fluid type identification model based on the AdaBoost-neural network algorithm, and introduces the spatial information of the well logging into the model, so that the constructed model can not only perform efficient identification, but also has the ability to determine the fluid type in combination with spatial information, realizing intelligent, fast and accurate identification of fluids under various complex geological conditions, and optimizing resource assessment and mining decision-making processes.

[0074] Figure 2 This is a flow chart of Embodiment 2 of the method for identifying reservoir fluid types provided in this application, as shown in FIG. Figure 2 As shown, before the well logging data and spatial information are input into the pre-acquired fluid type identification model for identification processing, the method further includes:

[0075] S201: Acquire a historical well logging data set and a historical fluid type set.

[0076] In this solution, before the well logging data and spatial information of the target location are input into the pre-acquired fluid type recognition model for recognition processing, the fluid type recognition model needs to be trained to obtain the fluid type recognition model. The data used for training includes a historical well logging data set and a historical fluid type set. After the sample set is confirmed based on the historical well logging data set and the historical fluid type set, the model can be trained using the sample data to obtain the fluid type recognition model.

[0077] In this step, the historical logging data set obtained includes logging data obtained by logging at different locations of different reservoirs for multiple times and spatial information of each logging, wherein the spatial information includes: longitude, latitude and logging depth information of each logging location. The historical fluid type set obtained includes the fluid type obtained by logging analysis at different locations each time corresponding to the historical logging data set.

[0078] S202: Based on the historical well logging data set and the historical fluid type set, a sample set is obtained, wherein the sample set includes a first part of samples and a second part of samples, the first part of samples includes well logging data, spatial information corresponding to the well logging data, and fluid type, and the second part of samples includes well logging data and spatial information corresponding to the well logging data.

[0079] In this step, it is necessary to collect data from all wells, including the historical logging data set and historical fluid type set collected in step S201, and obtain a sample set for model training based on the data of these wells, wherein the first part of the sample data comes from the historical logging data of known fluid types, therefore, the first part of the samples includes logging data, spatial information corresponding to the logging data, and fluid type; the second part of the sample data comes from the historical logging data of unknown fluid types, therefore, the second part of the samples includes logging data and spatial information corresponding to the logging data.

[0080] S203: Constructing an initial fluid type recognition model based on the AdaBoost-neural network algorithm.

[0081] In this step, a fluid type recognition model is initially constructed based on the AdaBoost-neural network algorithm. It should be understood that the AdaBoost algorithm is an integrated learning method that aims to create a strong learner by combining multiple weak learners. In this solution, the performance of the fluid recognition model is enhanced by combining AdaBoost with a neural network.

[0082] S204: According to the first part of samples, the initial fluid type recognition model is trained, and model parameters are adjusted to obtain a trained fluid type recognition model.

[0083] In this step, before training the initial fluid type recognition model based on the first part of samples, a training set for model training and a validation set for model validation need to be confirmed based on the first part of samples.

[0084] Specifically, the first part of the samples in the sample set is divided into a training set and a validation set according to a preset ratio. The preset ratio can be determined according to the actual training situation, and this solution does not limit this. For example, the preset ratio is set to 7:3, then 70% of the sample set is randomly used as a training set and 30% of the sample is used as a validation set, which can be expressed by the following formula:

[0085] W train , W test =split(W,train_size=0.7)

[0086] In the above formula, W is the first part of the sample, W train is the training set, W test is the validation set.

[0087] In a specific application, training the initial fluid type recognition model according to the first part of samples may include:

[0088] (1) Initialize the sample weights. According to the number of samples in the training set determined by the first part of the samples, assign the same weight to each sample, which can be expressed as:

[0089]

[0090] In the formula, w i represents the weight of the i-th sample, N represents the total number of samples in the training set, and the samples in the training set are represented as (X1, Y1)(X2, Y2), ..., (X N ,Y N ).

[0091] (2) Start iterative training, the total number of iterations is M.

[0092] A. According to the sample weight w i (m) Train the weak learner, where the weak learner is a neural network classifier. The sample weights in each iteration are expressed as w i (m) Indicates, where m represents the number of iterations, m=1,2,......,M.

[0093] B. After one iteration of training, the weighted error rate of all weak learners is calculated. Specifically, according to the accuracy of the prediction output of a weak learner for all samples, the weighted error rate of the weak learner after this round of training is determined, and the weighted error rate of the tth weak learner ε tThe calculation formula is:

[0094]

[0095] In the formula, h t (X i ) represents the predicted output of the t-th weak learner for the i-th sample; is an indicator function, such as the weak learner's prediction error for the sample Output 1, otherwise Output 0.

[0096] C. Calculate the weights of all weak learners. Specifically, according to the weighted error rate of the t-th weak learner, determine the weight of the t-th weak learner. The weight α of the t-th weak learner t The calculation formula is as follows:

[0097]

[0098] D. Update the weight of the sample in the training set corresponding to the weak learner according to the current weight of the weak learner, the actual output of the sample, and the predicted output of the weak learner for the sample. The new weight of the i-th sample The calculation formula is as follows:

[0099]

[0100] After calculating the new weights of all samples in the training set corresponding to the t-th weak learner through the above formula, the sample weights are normalized to ensure that the sum of the sample weights in the training set is 1, and the update of the sample weights in the training set corresponding to the t-th weak learner is completed. According to this method, the update of the sample weights in the training set corresponding to all weak learners is updated.

[0101] (3) After M rounds of iterative training, the prediction models of all weak learners are combined according to the final weights of each weak learner to obtain the strong learner model H(x). The basic form of this model can be expressed as:

[0102]

[0103] In the formula, h t (x) is the prediction of the tth weak learner for the input data x, where the weak learner is a neural network classifier, and the sign function is used to determine the final predicted class label, which determines the final output fluid type by calculating the weighted sum of all weak learners.

[0104] Through the above steps, the preliminary training of the initial fluid type recognition model can be completed.

[0105] Optionally, in the process of adjusting the model hyperparameters, the optimal hyperparameter combination can be found through methods such as cross-validation, grid search, and Bayesian optimization to optimize the prediction accuracy and generalization ability of the model. The number and type of hyperparameters are determined by the hyperparameters involved in the model, and this case does not limit this.

[0106] According to the optimal hyperparameters found above, after training on the training set, the trained fluid type recognition model H can be obtained. * (x).

[0107] In this step, the initial fluid type recognition model is first trained according to the training set to obtain the fluid recognition model H(x), and then the hyperparameters in the model are automatically adjusted by cross-validation and other methods to obtain the trained fluid type recognition model H * (x).

[0108] S205: using a semi-supervised learning method, respectively predicting the second part of samples through the trained fluid type recognition model to obtain multiple pseudo labels corresponding to the second part of samples;

[0109] In this step, by predicting the logging data and spatial information in the second part of samples without fluid type information, the predicted fluid type corresponding to the logging data can be obtained, and this fluid type is the pseudo label corresponding to the logging data.

[0110] S206 constructs a new sample set according to the first part of samples, the second part of samples and multiple pseudo labels, and retrains the initial fluid type recognition model through the new sample set to obtain a fluid type recognition model.

[0111] In this step, it is necessary to merge the first part of samples with fluid type information, the second part of samples without fluid type information, and the pseudo labels corresponding to the logging data in the second part of samples into a sample set, obtain a new sample set, and retrain the initial fluid type recognition model through the new sample set to obtain a fluid type recognition model. The newly obtained fluid type recognition model captures the new features or patterns in the second part of samples.

[0112] In the specific application implementation of this solution, it should be understood that after retraining the initial fluid type recognition model, the recognition accuracy of this model needs to be evaluated and verified, specifically:

[0113] In step S204, based on the validation set confirmed by the first part of samples, the retrained fluid type recognition model is validated and the model performance is evaluated, which can be specifically expressed by the following formula:

[0114]

[0115] In the above formula, N correct Represents the correct number of validations on the validation set after validation, N test is the total number of validation sets.

[0116] Optionally, before performing model verification on the retrained fluid type recognition model, the performance of the AdaBoost-neural network model may be evaluated using evaluation methods such as cross-validation to ensure that the model has high accuracy and stability.

[0117] In the reservoir fluid type identification method provided in this embodiment, the first part of samples and the second part of samples are obtained by obtaining a set of historical well logging data and a set of historical fluid types, and a semi-supervised learning algorithm is used to train the fluid type identification model based on the AdaBoost-neural network model algorithm to obtain pseudo labels, and finally the final fluid type identification model is retrained based on all samples and pseudo labels. In this solution, the semi-supervised algorithm enables the model to learn the intrinsic structure and characteristics of the data even when well logging data of known fluid types are scarce. The method of combining AdaBoost with neural networks effectively improves the generalization ability and prediction accuracy of the model, and effectively improves the scope of application of the model under complex geological conditions.

[0118] Figure 3 This is a flow chart of Embodiment 3 of the method for identifying reservoir fluid types provided in this application, as shown in FIG. Figure 2 As shown, a historical logging data set and a historical fluid type set are obtained, including:

[0119] S301: Acquire multiple well logging data and spatial information of each well logging according to the exploration data of the drilled layer section or the drilled single well;

[0120] In this step, the exploration data of the drilled layer section or the drilled single well include but are not limited to resistivity, acoustic wave time difference, density, neutron and other logging data, test production data and coring analysis results. Based on the acquired exploration data, multiple logging data and spatial information of each logging can be obtained.

[0121] S302: Determine the fluid type corresponding to each well logging according to the test production data and coring analysis results of each test production at different locations, and obtain a historical fluid type set;

[0122] In this step, based on the test production data of each test production at different locations in the exploration data obtained and the core sampling analysis results, after laboratory processing and expert experience analysis, the fluid type corresponding to each well logging can be obtained, and then a historical fluid type set can be obtained.

[0123] S303: Preprocessing the multiple well logging data to obtain multiple processed well logging data.

[0124] Optionally, the step of preprocessing the plurality of well logging data may include: data cleaning, removing outliers, and correcting missing values. In a specific implementation, the following steps may be used for preprocessing:

[0125] A. Cleaning erroneous data and placeholders in the plurality of well logging data.

[0126] Optionally, the error data and placeholders may include: -999, -9999, -99999, -9999, -999.25, 0. In a specific application, the error data and placeholders in a plurality of well logging data may be replaced with null values ​​to represent missing data, so as to clean the well logging data, which may be expressed by the following formula:

[0127] X i =Nan,X i ∈{-999,-9999,-99999,-999.25,0}

[0128] Where, X i Represents multiple logging data, including resistivity, wellbore diameter, natural gamma, acoustic wave time difference, density, neutron, natural potential, etc.

[0129] B. removing abnormal data from the plurality of well logging data;

[0130] Optionally, the abnormal data includes: negative values ​​in the logging data except the natural potential data and extreme values ​​in all the logging data. In a specific application, for all negative values ​​except the natural potential data, the abnormal data can be removed by replacing the negative values ​​with null values, which can be expressed by the following formula:

[0131] X i =Nan,(X i <0),(X i <0)

[0132] Among them, X i The logging data cannot be a natural potential curve.

[0133] For the extreme values ​​in all logging data, the standard deviation method is used to eliminate outliers. Specifically, the mean value μ and standard deviation δ of the corresponding curve of each logging data are calculated first, and then the conditional judgment is used to mark the values ​​exceeding the mean value ± 2 times the standard deviation as null values, which can be expressed by the following formula:

[0134] X i =Nan,(X i -μ>2δ)

[0135] C. The missing values ​​in multiple well logging data are supplemented by other well logging data within a preset depth range.

[0136] For example, the average value of other logging data within 10 m above and below the missing value depth can be used to supplement and correct the missing value to complete the data processing, which can be expressed by the following formula:

[0137] X i =average(X i [Depth i -10, Depth i +10]),(X i =Nan)

[0138] Where, Depth i The well logging depths for missing values ​​in multiple well logging data.

[0139] In the specific application of this solution, it should be understood that after preprocessing multiple well logging data, the well logging depth information can be introduced by sorting the well logging data used in the historical well logging data set according to the depth sequence, so that the subsequent training retains certain sedimentary regularity information that changes with depth. After the well logging depth information is introduced by this method, the spatial information of each well logging only includes the longitude and latitude of each well logging position. The longitude and latitude of each well logging position can be represented by spatial coordinates (x, y).

[0140] S304: Standardizing the processed multiple well logging data and the spatial information of each well logging to obtain the historical well logging data set.

[0141] Optionally, the Z-score method can be used to standardize the preprocessed multiple logging data and the spatial information of each logging. The specific standardization formula is as follows:

[0142]

[0143] Where X is multiple well logging data or the spatial coordinates corresponding to the well logging data, X standardized is the standardized logging data or the spatial coordinates corresponding to the logging data, μ is the average value of the curve corresponding to the logging data, and σ is the standard deviation of the curve corresponding to the logging data.

[0144] In the specific application of this solution, it should be understood that by obtaining the spatial information of each well logging, it is possible to have the ability to determine the fluid type in combination with the spatial information, and to better capture the spatial distribution characteristics of the hydrate reservoir.

[0145] In the reservoir fluid type identification method provided in this embodiment, by obtaining the exploration data of the drilled layer section or the drilled single well, multiple well logging data and the spatial information of each well logging are obtained and the fluid type corresponding to each well logging is determined to obtain a historical fluid type set, and then the multiple well logging data are pre-processed, and the processed multiple well logging data and the spatial information of each well logging are standardized to obtain a historical well logging data set. After pre-processing the multiple well logging data, this solution can improve the accuracy of the well logging data in the historical well logging data set, thereby improving the accuracy of the model; and through the above-mentioned standardization processing, the dimension effect is eliminated and the algorithm performance is improved, thereby ensuring that the data quality meets the requirements of model training.

[0146] Figure 4 This is a flow chart of Embodiment 4 of the method for identifying reservoir fluid types provided in this application, as shown in FIG. Figure 4 As shown, after the well logging data and spatial information of the target position of the reservoir are input into the pre-acquired fluid type identification model for identification processing to obtain the fluid type of the target position, it also includes:

[0147] Based on the well logging data and spatial information of the target location, as well as the fluid type of the target location, the fluid type identification model can be optimized to obtain a new fluid type identification model, which specifically includes the following steps:

[0148] S401: acquiring new historical logging data based on the logging data and spatial information of the target location, and acquiring new historical fluid types according to the fluid type of the target location.

[0149] In this scheme, the newly acquired logging data and spatial information of the target location, as well as the fluid type of the target location, can be used to update and optimize the AdaBoost-neural network model through an incremental learning method, so that the model can continuously learn and adapt to new data.

[0150] In this step, the newly acquired logging data and spatial information of the target location, as well as the fluid type of the target location, are also subjected to preprocessing and standardization as described in the third embodiment to obtain new historical logging data and new historical fluid types.

[0151] S402: Acquire a new sample set based on new historical logging data and new historical fluid types.

[0152] In the specific application of this solution, it should be understood that the target object of logging using the logging method is the well corresponding to a certain reservoir, rather than a certain position in the reservoir. Therefore, the logging data and spatial information input into the model are the logging data and spatial information of multiple target positions. Therefore, the new sample set may include the logging data and spatial information of multiple target positions, as well as the fluid type of the target position, so that the sample set may include multiple samples, each of which includes a set of logging data, spatial information corresponding to the logging data, and fluid type, wherein the new sample set is represented by W new express.

[0153] S403: Based on the new sample set, the fluid type recognition model is trained to obtain a new fluid type recognition model.

[0154] In this step, the new sample set needs to be divided into a new training set W according to the preset ratio. new,train And the new validation set W new,test The step of training the fluid type recognition model to obtain a new fluid type recognition model includes the following steps:

[0155] A. Train the fluid type recognition model based on the new training set. Update the weight of the existing weak classifier according to the impact of the new training set on the error rate. It can be expressed by the following formula:

[0156]

[0157] In the formula, α t is the original weight of the existing t-th weak classifier, α' t is the new weight of the existing t-th weak classifier, and γ is an adjustment factor that can dynamically adjust the weight according to the error rate of the new sample.

[0158] B. Generally speaking, the new training set has new features or patterns that the previous model has not captured well. Therefore, this step requires adding a new classifier, which is expressed by the following formula:

[0159] h new = RandomForest(W new )

[0160] α new =calculate_weight(h new ,W new )

[0161] In the above formula, h new is the new weak trainer, α new is the weight of the new weak classifier.

[0162] C. Add the new weak trainer and its weight to the existing fluid type recognition model, which can be expressed by the following formula:

[0163]

[0164] In the above formula, M'(x) represents the updated fluid type identification model

[0165] D. Perform model verification on the updated fluid type recognition model to obtain a new fluid type recognition model. Specifically, it is necessary to use sample data on the new verification set to verify the updated fluid type recognition model and evaluate the model performance, which can be expressed by the following formula:

[0166]

[0167] In the above formula, N correct_new Represents the correct number of validations on the new validation set after validation, N test_new is the total number of new validation sets.

[0168] Optionally, the validation set used for validation may also be the sum of the original validation set and the new validation set.

[0169] In the reservoir fluid type identification method provided in this embodiment, based on the well logging data and spatial information of the target location, as well as the fluid type of the target location, the fluid type identification model is optimized, thereby obtaining a new fluid type identification model. This solution adopts an incremental learning method, which enables the model to continuously learn and adapt to the new data environment by updating the model optimization method when receiving new data, and the entire optimization process does not need to be trained from scratch. Furthermore, this feature enables the model to be continuously optimized over time, maintaining its prediction accuracy and adaptability under new geological conditions, and effectively improving the application scope and sustained performance of the model under complex geological conditions.

[0170] The following is a specific fifth embodiment of the present invention, in which the reservoir fluid type identification method of the present invention is used to identify several typical hydrate drilling wells in Shenhu Sea Area of ​​my country.

[0171] After the fluid type recognition model is obtained based on semi-supervised learning and AdaBoost-neural network model training, the retrained fluid type recognition model is verified based on the validation set. Figure 5 is the comparison result between the actual and predicted layer labels after model verification, where the layer label is the fluid type at the target location described in other embodiments. Figure 5 As shown in the figure, the actual classification results are represented by solid points, and the recognition prediction results are represented by the symbol '×'. It can be seen that the actual classification results are basically consistent with the prediction results, and the prediction accuracy is above 0.985. In addition, Figure 6The confusion matrix after model validation compares the classification results, such as Figure 6 As shown in the figure, the confusion matrix shows that the classification results are accurate, and new well prediction can be carried out based on this training model.

[0172] Through the specific embodiments provided above, it can be seen that the reservoir fluid identification method provided in this application can accurately identify and predict the type of natural gas hydrate reservoir fluid, has stronger applicability and practicality than previous methods, and has certain industrial application value.

[0173] Figure 7 A schematic diagram of the structure of the reservoir fluid type identification device provided in this application, such as Figure 7 As shown, the reservoir fluid type identification device 50 provided in this embodiment includes:

[0174] The first acquisition module 501 is used to acquire the logging data and spatial information obtained by logging the target position of the reservoir;

[0175] The identification module 502 is used to input the logging data and the spatial information into a pre-acquired fluid type identification model for identification processing to obtain the fluid type of the target location, wherein the fluid type identification model is trained based on a semi-supervised learning method and an AdaBoost-neural network algorithm.

[0176] The reservoir fluid type identification device provided in this embodiment is used to execute the technical solution in any of the aforementioned method embodiments. Its implementation principle and technical effects are similar and will not be described in detail here.

[0177] In a possible implementation, the spatial information includes well logging location information and well logging depth for logging the target location; the well logging location information includes the longitude and latitude of the well logging location.

[0178] Based on the above-mentioned reservoir fluid type identification device 50, in a specific implementation manner, before inputting the well logging data and the spatial information into the pre-acquired fluid type identification model for identification processing, the reservoir fluid type identification device 50 may further include:

[0179] The second acquisition module 503 is used to acquire a historical logging data set and a historical fluid type set, wherein the historical logging data set includes logging data obtained by logging at different locations of different reservoirs for multiple times and spatial information of each logging, and the historical fluid type set includes fluid types analyzed by logging at different locations each time corresponding to the historical logging data set;

[0180] A processing module 504 is used to obtain a sample set based on the historical well logging data set and the historical fluid type set, wherein the sample set includes a first part of samples and a second part of samples, wherein the first part of samples includes well logging data, spatial information corresponding to the well logging data, and fluid types, and the second part of samples includes well logging data and spatial information corresponding to the well logging data;

[0181] A construction module 505 is used to construct an initial fluid type recognition model based on the AdaBoost-neural network algorithm;

[0182] The training module 506 is used to perform model training on the initial fluid type recognition model based on the sample set to obtain the fluid type recognition model.

[0183] In a specific implementation, in order to implement model training for the initial fluid type recognition model based on the sample set to obtain the fluid type recognition model, the training module 506 is specifically used to:

[0184] According to the first part of samples, the initial fluid type recognition model is trained, and model parameters are adjusted to obtain a trained fluid type recognition model;

[0185] Using a semi-supervised learning method, the trained fluid type recognition model is used to predict the second part of samples respectively, so as to obtain a plurality of pseudo labels corresponding to the second part of samples;

[0186] A new sample set is constructed according to the first part of samples, the second part of samples and the multiple pseudo labels, and the initial fluid type recognition model is retrained by the new sample set to obtain the fluid type recognition model.

[0187] In a specific implementation, in order to obtain a historical well logging data set and a historical fluid type set, the second acquisition module 503 is specifically used to:

[0188] Based on the exploration data of the drilled layer section or the drilled single well, multiple well logging data and the spatial information of each well logging are obtained;

[0189] According to the test production data and coring analysis results of each test production at different locations, the fluid type corresponding to each well logging is determined to obtain the historical fluid type set;

[0190] Preprocessing the plurality of well logging data to obtain a plurality of processed well logging data, wherein the preprocessing includes the following processing steps:

[0191] Cleaning erroneous data and placeholders in the plurality of well logging data;

[0192] removing abnormal data from the plurality of well logging data;

[0193] The missing values ​​in the plurality of well logging data are supplemented by other well logging data within a preset depth range.

[0194] The processed multiple well logging data and the spatial information of each well logging are standardized to obtain the historical well logging data set.

[0195] Based on the above-mentioned reservoir fluid type identification device 50, in a specific embodiment, the reservoir fluid type identification device 50 may further include:

[0196] The updating module 507 is used to optimize the fluid type identification model based on the logging data and spatial information of the target location and the fluid type of the target location to obtain a new fluid type identification model.

[0197] The reservoir fluid type identification device 50 provided in any of the aforementioned embodiments can execute the method provided in the aforementioned method embodiment, and its implementation principle and technical effect are similar, which will not be described in detail in this embodiment.

[0198] Figure 8 This is a schematic diagram of the structure of the reservoir fluid type identification device provided in this application. Figure 8 As shown, the electronic device 60 provided in this embodiment includes: at least one processor 601 and a memory 602. Optionally, the device 60 also includes a communication component 603. The processor 601, the memory 602 and the communication component 603 are connected via a bus 604.

[0199] In a specific implementation process, at least one processor 601 executes the computer execution instructions stored in the memory 602, so that at least one processor 601 executes the above method.

[0200] The specific implementation process of the processor 601 can be found in the above method embodiment, and its implementation principle and technical effect are similar, so this embodiment will not be repeated here.

[0201] In the above embodiments, it should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the invention may be directly implemented as being executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.

[0202] The memory may include a high-speed memory (Random Access Memory, RAM), and may also include a non-volatile memory (Non-volatile Memory, NVM), such as at least one disk memory.

[0203] The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the bus in the drawings of the present application is not limited to only one bus or one type of bus.

[0204] The present application also provides a computer program product, including a computer program, which implements the above method when executed by a processor.

[0205] The present application also provides a computer-readable storage medium, in which computer-executable instructions are stored. When a processor executes the computer-executable instructions, the above method is implemented.

[0206] The above-mentioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general or special-purpose computer.

[0207] An exemplary readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (Application Specific Integrated Circuits, referred to as: ASIC). Of course, the processor and the readable storage medium can also exist in the device as discrete components.

[0208] The division of units is only a logical function division, and there may be other divisions in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0209] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0210] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0211] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0212] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk and other media that can store program codes.

[0213] Finally, it should be noted that those skilled in the art will readily conceive of other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses or adaptations of the present invention, which follow the general principles of the present invention and include common knowledge or customary technical means in the art not disclosed by the present invention, are not limited to the precise structure described above and shown in the drawings, and may be modified and changed in various ways without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A method for identifying reservoir fluid type, characterized in that: include: Acquire logging data and spatial information obtained by logging a target location of a reservoir; Inputting the logging data and the spatial information into a pre-acquired fluid type recognition model for recognition processing to obtain the fluid type at the target location; Wherein, the fluid type recognition model is obtained based on semi-supervised learning method and AdaBoost-neural network algorithm training.

2. The method according to claim 1, characterized in that: The spatial information includes the well logging location information and the well logging depth for logging the target location; the well logging location information includes the longitude and latitude of the well logging location.

3. The method according to claim 1 or 2, characterized in that: Before inputting the logging data and the spatial information into a pre-acquired fluid type recognition model for recognition processing, the method further includes: Acquire a historical well logging data set and a historical fluid type set, wherein the historical well logging data set includes well logging data obtained by multiple well loggings at different locations of different reservoirs and spatial information of each well logging, and the historical fluid type set includes fluid types analyzed by each well logging at different locations corresponding to the historical well logging data set; Based on the historical well logging data set and the historical fluid type set, a sample set is obtained, wherein the sample set includes a first part of samples and a second part of samples, wherein the first part of samples includes well logging data, spatial information corresponding to the well logging data, and fluid types, and the second part of samples includes well logging data and spatial information corresponding to the well logging data; Constructing an initial fluid type recognition model based on the AdaBoost-neural network algorithm; Based on the sample set, the initial fluid type recognition model is trained to obtain the fluid type recognition model.

4. The method according to claim 3, characterized in that The step of training the initial fluid type recognition model based on the sample set to obtain the fluid type recognition model includes: According to the first part of samples, the initial fluid type recognition model is trained, and model parameters are adjusted to obtain a trained fluid type recognition model; Using a semi-supervised learning method, the trained fluid type recognition model is used to predict the second part of samples respectively, so as to obtain a plurality of pseudo labels corresponding to the second part of samples; A new sample set is constructed according to the first part of samples, the second part of samples and the multiple pseudo labels, and the initial fluid type recognition model is retrained by the new sample set to obtain the fluid type recognition model.

5. The method according to claim 3, characterized in that: The obtaining of a historical well logging data set and a historical fluid type set includes: Based on the exploration data of the drilled layer section or the drilled single well, multiple well logging data and the spatial information of each well logging are obtained; Based on the test production data and coring analysis results of multiple test productions at different locations, determine the fluid type corresponding to each well logging, and obtain the historical fluid type set; Preprocessing the plurality of well logging data to obtain a plurality of processed well logging data; The processed multiple well logging data and the spatial information of each well logging are standardized to obtain the historical well logging data set.

6. The method according to claim 5, characterized in that Preprocessing includes the following processing steps: Cleaning erroneous data and placeholders in the plurality of well logging data; removing abnormal data from the plurality of well logging data; The missing values ​​in the plurality of well logging data are supplemented by other well logging data within a preset depth range.

7. The method according to claim 1 or 2, characterized in that: The method further comprises: Based on the logging data and spatial information of the target location, and the fluid type of the target location, the fluid type identification model is optimized to obtain a new fluid type identification model.

8. A device for identifying reservoir fluid type, characterized in that: include: Acquisition module: used to acquire logging data and spatial information obtained by logging the target position of the reservoir; Identification module: used to input the logging data and the spatial information into a pre-acquired fluid type identification model for identification processing to obtain the fluid type of the target location; wherein the fluid type identification model is obtained by training based on the AdaBoost-neural network algorithm.

9. A device for identifying reservoir fluid type, characterized in that: include: Memory, processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor performs the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor.

11. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images