Method and system for evaluating permeability of carbonate reservoir based on nuclear magnetic resonance

Abstract

The invention discloses a method and system for evaluating the permeability of a carbonate reservoir based on nuclear magnetic resonance, and the method comprises the steps: obtaining nuclear magnetic resonance logging T2 spectrum test data and permeability data of the reservoir, and dividing a corresponding training set and a test set; performing data standardization processing on input and output of the training set to eliminate differences of units and orders of magnitude among different input vectors; inputting the standardized training set into a preset deep learning model for training to obtain a permeability calculation model; and testing the performance of the permeability calculation model by using the test set. The calculation precision of nuclear magnetic resonance logging information for evaluating the permeability of the carbonate reservoir is improved, and effective support is provided for reservoir parameter fine evaluation and later oil reservoir development.

Description

Technical Field

[0001] The present invention relates to the technical field of oil and gas well exploration, and in particular to a method and system for evaluating the permeability of carbonate reservoirs based on nuclear magnetic resonance. Background Art

[0002] Evaluation of reservoir permeability is a very important part of reservoir description and quantitative description of remaining oil, which directly determines the determination of reservoir production capacity and the design of development plan. Reasonable reservoir evaluation must be based on accurate permeability evaluation. Among them, the size of permeability is determined by multiple factors such as rock porosity, pore geometry, wettability, and mud content. Among them, the nuclear magnetic resonance T2 spectrum can describe the pore throat distribution characteristics of the rock, and the pore throat distribution directly determines the size of the permeability. Therefore, the permeability calculated by nuclear magnetic resonance is often considered to be the most accurate method. There are two classic models of permeability calculation based on nuclear magnetic resonance, SDR and Coates. By establishing the relationship between the transverse relaxation time of nuclear magnetic resonance logging and permeability, the transverse relaxation time can be quickly and intuitively expressed as permeability. Carbonate reservoirs have different properties from conventional sandstone reservoirs. Their sedimentation is controlled by multiple diagenesis and there are many types of reservoir space, resulting in complex pore structure and severe heterogeneity. The permeability of carbonate reservoirs calculated based on the two classic models of SDR and Coates using the nuclear magnetic resonance logging T2 spectrum is not accurate. Therefore, the conventional NMR logging permeability calculation model has low calculation accuracy in carbonate reservoirs and cannot meet the requirements of fine reservoir characterization. Summary of the invention

[0003] The embodiments of the present invention provide a method and system for evaluating the permeability of carbonate reservoirs based on nuclear magnetic resonance, thereby solving the technical problem of low accuracy in permeability calculation in carbonate reservoirs in the prior art and achieving the technical effect of more accurate permeability calculation.

[0004] In the first aspect, to solve the above technical problems, the embodiments of the present invention provide the following technical solutions:

[0005] A method for evaluating the permeability of a carbonate reservoir based on nuclear magnetic resonance comprises: obtaining nuclear magnetic resonance logging T2 spectrum test data and permeability data of the reservoir, and dividing the data into corresponding training sets and test sets; performing data standardization processing on the input and output of the training set to eliminate the differences in units and orders of magnitude between different input vectors; inputting the standardized training set into a preset deep learning model for training to obtain a permeability calculation model; and using the test set to test the performance of the permeability calculation model.

[0006] Optionally, the step of performing data standardization processing on the input and output of the training set specifically includes: calculating the original input vector of the input and output data and its average value and standard deviation; comparing the difference between the original input vector and the average value with the standard deviation to obtain a standardized input vector.

[0007] Optionally, the step of inputting the standardized training set into a preset deep learning model for training specifically includes: establishing a mapping function f(x) of the nuclear magnetic resonance logging T2 spectrum test data and the permeability data, and inputting the mapping function f(x) into the training set; constructing a prediction model h(x i )=y i -f(x i ), where h(x i ) is the predicted value, y i is the true value; the mapping function is continuously iterated through the deep training model, so that f(x i ) approaches y i , until the residual of the corresponding loss function is reduced to the preset range, and the permeability calculation model is output.

[0008] Optionally, the step of using the test set to test the performance of the permeability calculation model specifically includes: inputting the test set into the permeability calculation model to obtain first permeability data; calculating a correlation coefficient and a mean absolute error based on the first permeability data and the permeability data obtained by core analysis in the test set; and obtaining an evaluation result by comparing the range interpretation corresponding to the correlation coefficient and the mean absolute error.

[0009] Optionally, the step of obtaining the nuclear magnetic resonance logging T2 spectrum test data of the reservoir specifically includes: inverting and obtaining the nuclear magnetic resonance logging T2 spectrum based on the nuclear magnetic resonance echo data of the reservoir.

[0010] Optionally, the above step of dividing the corresponding training set and test set specifically includes: dividing the corresponding training set and test set based on a random principle.

[0011] Optionally, the above method further includes:

[0012] Construct a deep learning model based on one or more of the convolutional neural network algorithm, the recurrent neural network algorithm, and the transformer algorithm.

[0013] In a second aspect, a system for evaluating the permeability of carbonate reservoirs based on nuclear magnetic resonance is provided, the system comprising:

[0014] A data acquisition module is used to acquire the nuclear magnetic resonance logging T2 spectrum test data and permeability data of the reservoir, and divide the corresponding training set and test set;

[0015] A standardization processing module is used to perform data standardization processing on the input and output of the above training set to eliminate the differences in units and orders of magnitude between different input vectors;

[0016] A training module is used to input the standardized training set into a preset deep learning model for training to obtain a permeability calculation model;

[0017] The performance testing module is used to test the performance of the permeability calculation model using the test set.

[0018] In a third aspect, an electronic device is provided, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps corresponding to the method of the first aspect when executing the computer program.

[0019] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, the steps corresponding to the above method in the first aspect are implemented.

[0020] One or more technical solutions provided in the embodiments of the present invention have at least the following technical effects or advantages:

[0021] The present invention divides the reservoir's nuclear magnetic resonance logging T2 spectrum test data and the permeability data set obtained from core analysis into a training set and a test set, and uses deep learning to continuously iterate the permeability calculation model until it is close to the true value. This improves the calculation accuracy of carbonate reservoir permeability evaluated by nuclear magnetic resonance logging data, and provides effective support for the fine evaluation of reservoir parameters and later reservoir development. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 A flow chart of a method for evaluating carbonate reservoir permeability based on nuclear magnetic resonance provided by the present invention;

[0024] Figure 2 The absolute error frequency distribution of the carbonate reservoir permeability evaluated based on nuclear magnetic resonance in the present invention on the test set;

[0025] Figure 3 Absolute error frequency distribution of the traditional NMR Coates model provided by the present invention on the test set;

[0026] Figure 4A schematic diagram of the structure of a system for evaluating carbonate reservoir permeability based on nuclear magnetic resonance provided by the present invention;

[0027] Figure 5 The present invention provides a schematic structural diagram of an electronic device. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0029] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. The terms "first", "second", "third", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0031] It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solutions of the present application, rather than limitations of the technical solutions of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments may be combined with each other.

[0032] The technical solution of the embodiment of the present invention is to solve the above technical problems, and the overall idea is as follows:

[0033] In order to effectively use NMR logging data to evaluate the permeability of carbonate reservoirs, it is necessary to fully consider the reservoir information contained in the NMR logging T2 spectrum. The permeability calculation model is continuously iterated using a deep learning model until it is close to the true value, thereby obtaining a more accurate permeability calculation value.

[0034] In an embodiment of the present invention, there is provided Figure 1 A method for evaluating carbonate reservoir permeability based on nuclear magnetic resonance is shown, the method comprising steps S101 to S104:

[0035] Step S101, obtaining nuclear magnetic resonance logging T2 spectrum test data and permeability data of the reservoir, and dividing the data into corresponding training sets and test sets;

[0036] It should be noted that the nuclear magnetic resonance logging T2 spectrum can be obtained by adjusting the parameters of the nuclear magnetic resonance logging instrument, such as magnetic field strength, pulse sequence, etc., combined with other properties of the reservoir for evaluation; it can also be obtained by inverting the nuclear magnetic resonance echo data of the target reservoir. Specifically, the inversion of the nuclear magnetic resonance logging T2 spectrum specifically includes: collecting echo data, including spin echo signals and geomagnetic field signals in the formation. Preprocessing the collected echo data, including denoising, filtering, etc., to improve the signal-to-noise ratio and resolution of the signal. The signal-to-noise ratio of the nuclear magnetic resonance echo train can be set to be greater than 10dB to ensure the accuracy of the test. The preprocessed echo data is Fourier transformed to obtain a frequency domain signal. The frequency domain signal is inverted to obtain the nuclear magnetic resonance logging T2 reflecting the reservoir. In addition, for the division of the training set and the test set, the random principle is adopted, 80% of the samples are used as training data, and 20% of the samples are used as test data. Among them, the random principle adopted ensures that both the training set and the test set can represent the overall data from the original data set. Because of random partitioning, no matter what the characteristics of the data are, it is possible to be divided into the training set or the test set, which can avoid inaccurate model performance evaluation due to data selection bias. At the same time, random partitioning ensures that the amount of data in the training set and the test set is equal, so that the model has enough data for learning and testing during the training and testing process, ensuring the fairness of the model evaluation.

[0037] Step S102, performing data standardization processing on the input and output of the training set to eliminate the differences in units and magnitudes between different input vectors;

[0038] It should be noted that eliminating the differences in units and magnitudes between different input vectors is intended to facilitate comparison and weighting of indicators of different units or levels. Specifically: First, calculate the original input vector of the input and output data and its average value and standard deviation. Then, compare the difference between the original input vector and the average value with the standard deviation to obtain the standardized input vector. Specifically:

[0039] To standardize the input variables of the training data, first calculate the average value of the original input vector The standard deviation of the original input vector is calculated as follows:

[0040]

[0041]

[0042] According to formulas (1) and (2), the normalized input vector is obtained

[0043]

[0044] For the output variable (permeability) of the training data, the calculation formula is as follows:

[0045]

[0046] Among them, Y Nor is the normalized output vector, Y Ori is the original output vector, is the average value of the original output vector, is the standard deviation of the original output vector.

[0047] Step S103, inputting the standardized training set into a preset deep learning model for training to obtain a permeability calculation model;

[0048] It should be noted that this embodiment uses a deep learning method to extract the mapping relationship between various input parameters, and calculates the permeability of the target reservoir based on the mapping relationship. The parameter is the nuclear magnetic resonance T2 spectrum. Taking the model constructed based on the convolutional neural network (CNN) algorithm as an example, the pre-constructed model is trained using the training set, and the trained model is used as the permeability calculation model.

[0049] Specifically, a mapping function f(x) between the NMR logging T2 spectrum test data and the permeability data is established and input into the training set; that is, the NMR T2 spectrum after data standardization is used as x in the training set. The core analysis permeability after data standardization is used as y in the training set, and a mapping function f(x) from x to y is established. i ).

[0050] Initialize the parameters, get an initial mapping function f0(x), and build a prediction model h(x i )=y i -f(x i ), where h(x i ) is the predicted value, y i is the true value; constantly changing f(x i ), so that f(x i ) gradually approaches y i Through the deep training model, the mapping function is continuously iterated to make the loss function smaller and realize negative gradient fitting until the residual of the corresponding loss function is reduced to the preset range, and the permeability calculation model is output. Among them, the corresponding loss function is:

[0051] L(y,f t (x))=L(y,f t-1 (x)+h t (x)) (5)

[0052] Among them, h t (x) represents the residual, which is equivalent to the value of the negative gradient of the loss function in the current model. The smaller the residual, the better the fitting effect. t (x) represents the strong learner finally obtained in the tth iteration. The strong learning function can be expressed as:

[0053] f k (x) = f k-1 (x)+vh k (x) (6)

[0054] Among them, k represents the number of iterations and v represents the step size.

[0055] Finally, the training effect can be determined based on the size of the residual. If the training effect does not meet the requirements, the training can be continued, and the number of iterations and step size can be adjusted to obtain the target model; if the training result meets the requirements, the trained model is used as the model for calculating the permeability.

[0056] Step S104, using the test set to test the performance of the permeability calculation model. This is used to evaluate the prediction accuracy. Specifically, the remaining 20% ​​of the data set is used as the test data set to evaluate the model performance, and the test set is input into the permeability calculation model to obtain the first permeability data. Based on the first permeability data and the permeability data obtained from the core analysis in the test set, the correlation coefficient γ and the mean absolute error MAE are calculated; the specific formula is as follows:

[0057]

[0058]

[0059] The evaluation results are obtained by comparing the range interpretations of the correlation coefficient and the mean absolute error. Among them, γ is calculated by dividing the sample covariance of the two lists by the product of the sample variance, which is the correlation coefficient of the two lists. Between -1 and 1, close to 1 means that the two lists are highly correlated and the prediction accuracy is good; close to -1 means that the two lists are less correlated and the prediction accuracy is poor; close to 0 means that the two lists are not correlated and the prediction accuracy is neither good nor bad. The smaller the mean absolute error MAE, the closer the model prediction value is to the true value, and the better the prediction performance of the model.

[0060] Furthermore, the method also includes: constructing a deep learning model based on one or more of a convolutional neural network algorithm, a recurrent neural network algorithm, and a converter algorithm.

[0061] It should be noted that the convolutional neural network algorithm is shown in the above embodiment. The selection of the convolutional neural network algorithm, the recurrent neural network algorithm and the converter algorithm is optimized according to the needs. When using multiple algorithms to generate a model, different algorithms are used to iterate the same data respectively, and then the trained models are compared based on the same validation set for accuracy, and the one with the highest accuracy is selected, thereby further improving the calculation accuracy of the model.

[0062] In order to more clearly illustrate the beneficial effects of the embodiments of the present invention, the carbonate reservoir permeability evaluation method provided by the embodiment of the present invention is compared with some conventional methods for evaluating the test data set. Figure 2 and Figure 3 Provide explanation.

[0063] Figure 2 The absolute error frequency distribution of the evaluation method provided by the embodiment of the present invention on the test set, Figure 3 It is the absolute error frequency distribution of the traditional nuclear magnetic resonance Coates model on the test set. It can be observed that the method provided by the embodiment of the present invention has a more concentrated calculation result error. The correlation coefficients of the two methods (the former is the present invention, and the latter is the traditional method) on the test set are 0.853 and 0.75 respectively. The MAE is 0.567 and 0.856 respectively. It can be seen that the evaluation method provided by the embodiment of the present invention has a calculation result that is significantly better than the traditional method.

[0064] Based on the same inventive concept, an embodiment of the present invention provides a system for evaluating carbonate reservoir permeability based on nuclear magnetic resonance, such as Figure 4 As shown, including:

[0065] The data acquisition module 201 is used to acquire the nuclear magnetic resonance logging T2 spectrum test data and permeability data of the reservoir, and divide the corresponding training set and test set;

[0066] A standardization processing module 202 is used to perform data standardization processing on the input and output of the training set to eliminate the differences in units and magnitudes between different input vectors;

[0067] A training module 203 is used to input the standardized training set into a preset deep learning model for training to obtain a permeability calculation model;

[0068] The performance testing module 204 is used to test the performance of the permeability calculation model using a test set.

[0069] Based on the same inventive concept, an embodiment of the present invention provides an electronic device, such as Figure 5As shown, it includes: a memory 302, a processor 301, and a computer program stored in the memory 302 and executable on the processor 301. When the processor 301 executes the computer program, a method for evaluating the permeability of a carbonate reservoir based on nuclear magnetic resonance is implemented.

[0070] Based on the same inventive concept, this embodiment provides a computer-readable storage medium having a computer program stored thereon, characterized in that when the program is executed by a processor, a method for evaluating the permeability of a carbonate reservoir based on nuclear magnetic resonance is implemented.

[0071] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks 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 produce 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.

[0072] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0073] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for evaluating the permeability of carbonate reservoirs based on nuclear magnetic resonance, characterized in that: The method comprises: Obtain the nuclear magnetic resonance logging T2 spectrum test data and permeability data of the reservoir, and divide the corresponding training set and test set; Performing data normalization on the input and output of the training set to eliminate differences in units and magnitudes between different input vectors; The standardized training set is input into a preset deep learning model for training to obtain a permeability calculation model; The performance of the permeability calculation model is tested using the test set.

2. The method according to claim 1, characterized in that The step of performing data standardization processing on the input and output of the training set specifically includes: Calculate the original input vector of the input and output data and its average value and standard deviation; The difference between the original input vector and the average value is compared with the standard deviation to obtain a standardized input vector.

3. The method according to claim 1, characterized in that The step of inputting the standardized training set into a preset deep learning model for training specifically includes: Establishing a mapping function f(x) between the nuclear magnetic resonance logging T2 spectrum test data and the permeability data, and inputting the mapping function f(x) into the training set; Constructing the prediction model h(x i )=y i -f(x i ), where h(x i ) is the predicted value, y i is the true value; Through the deep training model, the mapping function is continuously iterated so that f(x i ) approaches y i , until the residual of the corresponding loss function is reduced to the preset range, and the permeability calculation model is output.

4. The method according to claim 1, characterized in that The step of using the test set to test the performance of the permeability calculation model specifically includes: Inputting the test set into the permeability calculation model to obtain first permeability data; Calculating a correlation coefficient and a mean absolute error based on the first permeability data and the permeability data obtained by core analysis in the test set; The evaluation results are obtained by comparing the range interpretation corresponding to the correlation coefficient and the mean absolute error.

5. The method according to claim 1, characterized in that The step of obtaining the nuclear magnetic resonance logging T2 spectrum test data of the reservoir specifically includes: The nuclear magnetic resonance logging T2 spectrum is obtained by inversion based on the nuclear magnetic resonance echo data of the reservoir.

6. The method according to claim 1, characterized in that The step of dividing the corresponding training set and test set specifically includes: Based on the random principle, the corresponding training set and test set are divided.

7. The method according to claim 1, characterized in that The method further comprises: Construct a deep learning model based on one or more of the convolutional neural network algorithm, the recurrent neural network algorithm, and the transformer algorithm.

8. A system for evaluating carbonate reservoir permeability based on nuclear magnetic resonance, characterized in that: The system comprises: A data acquisition module is used to acquire the nuclear magnetic resonance logging T2 spectrum test data and permeability data of the reservoir, and divide the corresponding training set and test set; A standardization processing module, used for performing data standardization processing on the input and output of the training set to eliminate the differences in units and magnitudes between different input vectors; A training module is used to input the standardized training set into a preset deep learning model for training to obtain a permeability calculation model; A performance testing module is used to test the performance of the permeability calculation model using the test set.

9. An electronic device, characterized in that: The electronic device comprises: a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the method steps described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps corresponding to the method according to any one of claims 1 to 7 are implemented.

Images