Method, device, equipment and medium for predicting velocity in parallel bedding direction of shale reservoir
By combining neural network algorithms with core test data and well logging interpretation curves, the problems of high modeling difficulty and low prediction accuracy in traditional methods have been solved, achieving efficient and low-cost velocity prediction in the parallel bedding direction of shale reservoirs.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA PETROLEUM & CHEMICAL CORP
- Filing Date
- 2024-12-12
- Publication Date
- 2026-06-12
AI Technical Summary
Traditional seismic reservoir prediction methods suffer from problems such as high modeling difficulty, low prediction accuracy, high equipment requirements, and high personnel requirements when predicting velocities in shale reservoirs along the bedding direction. In addition, the large number of input parameters leads to low efficiency.
A neural network algorithm is used to predict the velocity parallel to bedding direction by establishing a training model based on core test data and well logging interpretation curves, thereby reducing input parameters and improving prediction accuracy.
It simplifies the prediction process, reduces time and labor costs, while ensuring prediction accuracy and improving the prediction efficiency of shale reservoir velocities parallel to bedding direction.
Smart Images
Figure CN122194259A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of geophysical technology, and in particular to a method, apparatus, equipment and medium for predicting velocities in the parallel bedding direction of shale reservoirs. Background Technology
[0002] Shale gas reservoir development has promising prospects and potential. Unlike conventional reservoirs, shale reservoirs exhibit strong anisotropy, which limits the effectiveness of traditional seismic reservoir prediction methods. The anisotropy of shale reservoirs refers to the significant bedding structure developed in shale due to sedimentary, diagenetic, and tectonic processes, resulting in different physical and chemical properties of the same shale in different directions. Velocities perpendicular to bedding are relatively easy to obtain, while velocities parallel to bedding generally can only be obtained through rock physics experiments.
[0003] Currently, most methods estimate velocities along the parallel bedding direction by establishing shale petrophysical models. However, establishing reliable shale petrophysical models requires not only substantial data support but also the selection of appropriate computational models based on specific needs. Some computational models are highly complex and require powerful computing capabilities. Furthermore, using petrophysical models to predict velocities along the parallel bedding direction in shale reservoirs faces numerous challenges, including high modeling difficulty, low prediction accuracy, high equipment requirements, and demanding personnel requirements. Summary of the Invention
[0004] The purpose of this application is to provide at least one method, apparatus, equipment and medium for predicting velocities in the parallel bedding direction of shale reservoirs, which can at least reduce the number of input parameters while ensuring prediction accuracy.
[0005] To address the aforementioned technical problems, at least one embodiment of this application provides a method for predicting velocities in shale reservoirs along the bedding direction, comprising:
[0006] Collect core test data and well logging interpretation curves of the target shale reservoir; the core test data includes at least: mineral content test data, TOC content test data, and velocity test data parallel to the bedding direction; correspondingly, the well logging interpretation curves include at least: mineral content curves and TOC content curves;
[0007] Establish the calibration relationship between the core test data and the well logging interpretation curve;
[0008] Based on the core test data, a training model is obtained by training a neural network algorithm. In the training model, the mineral content test data and the TOC content test data are used as the input layer dataset, and the velocity test data in the parallel bedding direction is used as the output layer dataset.
[0009] The training model is applied to the well logging interpretation curve to obtain the velocity prediction curve of the well logging interpretation curve for the parallel bedding direction.
[0010] In some embodiments, the mineral content test data includes at least one of the following: clay content test data, silica content test data, and calcium content test data;
[0011] Accordingly, the mineral content curve includes at least one of the following: clay content curve, silica content curve, and calcium content curve.
[0012] In some embodiments, the velocity test data in the parallel layering direction includes at least one of the following: longitudinal wave velocity test data in the parallel layering direction and transverse wave velocity test data in the parallel layering direction.
[0013] In some embodiments, applying the training model to the well logging interpretation curve to obtain the velocity prediction curve of the well logging interpretation curve for the parallel bedding direction includes:
[0014] Using the mineral content curve and the TOC content curve in the well logging interpretation curve as input signals to the training model, the predicted P-wave velocity and / or S-wave velocity of the well logging interpretation curve for the parallel bedding direction are obtained.
[0015] In some embodiments, the training model obtained by training a neural network algorithm based on the core test data includes:
[0016] The core test data is divided into a training set and a test set according to a preset ratio;
[0017] The mineral content test data and TOC content test data in the training set are used as input signals to the neural network algorithm, and the velocity test data in the parallel bedding direction is used as the output signal to train the neural network algorithm to obtain the training model.
[0018] In some embodiments, the training model obtained by training a neural network algorithm based on the core test data further includes:
[0019] The training model is tested using the test set to determine whether it meets preset conditions. When the training model meets the preset conditions, the training model is obtained.
[0020] In some embodiments, testing whether the trained model meets preset conditions using the test set includes:
[0021] The trained model is evaluated using the test set to obtain performance evaluation metrics.
[0022] The training model is obtained when the performance evaluation index meets the preset conditions; the performance evaluation index includes the training set correlation coefficient, the training set validation correlation coefficient, and the test set correlation coefficient.
[0023] In some embodiments, the preset condition includes: the training set correlation coefficient, the training set validation correlation coefficient, and the test set correlation coefficient are all greater than a preset value, wherein the preset value is greater than or equal to 0.9.
[0024] In some embodiments, dividing the core test data into a training set and a test set according to a preset ratio includes:
[0025] The core test data are divided into sets with a ratio of 8:2 between the number of samples in the training set and the number of samples in the test set.
[0026] At least one embodiment of this application also provides a device for predicting the velocity of shale reservoirs parallel to bedding direction, comprising:
[0027] The collection module is used to collect core test data and well logging interpretation curves of the target shale reservoir; the core test data includes at least: mineral content test data, TOC content test data, and velocity test data parallel to the bedding direction; the well logging interpretation curves include at least: mineral content curves and TOC content curves.
[0028] A module is established to establish the scale relationship between the core test data and the well logging interpretation curve;
[0029] The training module is used to train a training model based on the core test data using a neural network algorithm. The training model uses the mineral content test data and the TOC content test data as the input layer dataset and the velocity test data in the parallel bedding direction as the output layer dataset.
[0030] An application module is used to apply the training model to the well logging interpretation curve to obtain the velocity prediction curve of the well logging interpretation curve for the parallel bedding direction.
[0031] At least one embodiment of this application also provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the above-described method for predicting the velocity in the parallel bedding direction of shale reservoirs.
[0032] At least one embodiment of this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, performs the above-described prediction of the velocity in the shale reservoir parallel bedding direction.
[0033] The embodiments of this application provide a method, apparatus, equipment, and medium for predicting the velocity in the parallel bedding direction of shale reservoirs. Based on core test data and well logging interpretation curves, and using a neural network algorithm, this method predicts the velocity in the parallel bedding direction of the target shale reservoir, reducing the number of input parameters while ensuring prediction accuracy. This prediction method is simple and easy to implement, effectively saving time and labor costs. Attached Figure Description
[0034] One or more embodiments are illustrated by way of example with reference to the accompanying drawings, and these illustrative descriptions do not constitute a limitation on the embodiments.
[0035] Figure 1 This is a flowchart of a method for predicting shale reservoir velocity along the parallel bedding direction, provided in one embodiment of this application;
[0036] Figures 2-3 This is a schematic diagram of shale core sample preparation and velocity measurement provided in one embodiment of this application;
[0037] Figure 4 This is a schematic diagram of core test data projected onto the corresponding depth position on the corresponding well logging interpretation curve according to an embodiment of this application;
[0038] Figure 5 This is a schematic diagram of the basic structure of an artificial neural network provided in one embodiment of this application;
[0039] Figure 6 This is a schematic diagram illustrating the algorithm principle of an artificial neural network provided in one embodiment of this application;
[0040] Figure 7 This is a prediction result diagram of the predicted shear wave velocity and longitudinal wave velocity obtained by using a trained model, provided in one embodiment of this application;
[0041] Figure 8 This is a schematic diagram of a velocity prediction curve in the parallel bedding direction obtained from a training model, provided in an embodiment of this application.
[0042] Figure 9 This is a flowchart of a method for predicting shale reservoir velocity along the bedding direction provided in another embodiment of this application;
[0043] Figure 10 This is a flowchart of a method for predicting shale reservoir velocity along the bedding direction provided in another embodiment of this application;
[0044] Figure 11 This is a block diagram of a shale reservoir velocity prediction device in the direction of parallel bedding provided in one embodiment of this application;
[0045] Figure 12 This is a schematic diagram of the structure of an electronic device provided in one embodiment of this application. Detailed Implementation
[0046] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the various embodiments of this application will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details have been provided in the various embodiments of this application to help readers better understand this application. However, the technical solutions claimed in this application can be implemented even without these technical details and various changes and modifications based on the following embodiments. The division of the various embodiments below is for the convenience of description and should not constitute any limitation on the specific implementation of this application. The various embodiments can be combined with and referenced by each other without contradiction.
[0047] To facilitate understanding of the embodiments of this application, relevant content in the field of geophysics will be introduced first.
[0048] In geophysics, the main focus is on the anisotropy of shale velocity parameters, specifically the difference between velocities perpendicular to and parallel to bedding planes. Common geophysical methods for velocity testing include rock physics experiments, geophysical logging, and seismic exploration. Velocities perpendicular to bedding planes are relatively easy to obtain, while velocities parallel to bedding planes are generally only obtainable through rock physics experiments.
[0049] Currently, most methods estimate velocities along the parallel bedding direction by establishing shale petrophysical models. However, establishing reliable shale petrophysical models requires not only extensive data support, such as rock and mineral data, microstructural characteristic data, and elastic parameter data, but also the selection of appropriate computational models based on the physical contact patterns between different parts. Some computational models are very complex and require powerful computing capabilities. Furthermore, due to the complex pore structure and organic matter distribution of shale reservoirs, using petrophysical models to predict velocities along the parallel bedding direction in shale reservoirs presents challenges such as high modeling difficulty, low prediction accuracy, high equipment requirements, and high personnel demands.
[0050] To address the technical problems of numerous required parameters and low prediction accuracy, this application proposes a method for predicting shale reservoir velocities in the parallel bedding direction. The implementation details of the shale reservoir velocities in the parallel bedding direction prediction method in this embodiment are described below. The following implementation details are provided for ease of understanding and are not essential for implementing this solution.
[0051] Example 1:
[0052] The method for predicting shale reservoir velocities along the bedding direction in this embodiment can be applied to electronic devices with communication, computing, and data storage capabilities. The specific process can be as follows: Figure 1 As shown, it includes:
[0053] Step 101: Collect core test data and well logging interpretation curves of the target shale reservoir; the core test data shall include at least: mineral content test data, TOC content test data, and velocity test data in the direction of parallel bedding; correspondingly, the well logging interpretation curves shall include at least: mineral content curves and TOC content curves.
[0054] Specifically, step 101 is the data preparation stage, during which core test data and well logging interpretation curves of the target shale reservoir are collected. The core test data of the target shale reservoir can be obtained from core analysis experiments, and includes mineral content test data and TOC content test data from sampling points at different depths. Sampling points are the locations where core analysis experiments are conducted on the cored sections. Core test data is generally a set of discrete data with unequal intervals. Well logging interpretation curves are an important curve in petroleum geological exploration; they reflect not only the mineral composition of the target shale reservoir but also its structure and porosity. Mineral content test data can be obtained by testing the core samples collected from the well. These minerals include clay, silica, and calcareous materials. Total organic carbon (TOC) is the most commonly used indicator of organic matter content.
[0055] In one embodiment, the mineral content test data includes at least one of the following: clay content test data, silica content test data, and calcium content test data; correspondingly, the mineral content curve includes at least one of the following: clay content curve, silica content curve, and calcium content curve.
[0056] Velocity test data parallel to the bedding plane includes one of the following: longitudinal wave velocity test data parallel to the bedding plane and transverse wave velocity test data parallel to the bedding plane. Specifically, sonic logging instruments can be used to obtain velocity test data parallel to the bedding plane. Sonic logging instruments perform sonic measurements parallel to the wellbore. When the wellbore trajectory is perpendicular to the formation bedding plane, the wave velocity obtained in sonic logging is perpendicular to the formation; when the wellbore trajectory is parallel to the formation bedding plane, the wave velocity obtained in sonic logging is parallel to the formation. For example, refer to... Figures 2-3The diagram illustrates, for example, the preparation and velocity measurement of shale core samples. Velocity measurements can be performed by drilling shale core samples at different bedding angles. The bedding angle is the angle between the axial direction of the shale core sample and the bedding plane; 0° is taken for the direction parallel to the bedding plane, and 90° is taken for the direction perpendicular to the bedding plane. When performing velocity testing in the direction parallel to the bedding plane, the core sample preparation and velocity measurement can be performed as follows: Figure 2 As shown, sampling and testing are carried out along the bedding direction of the shale reservoir core to obtain P-wave velocity and S-wave velocity test data parallel to the bedding direction.
[0057] Step 102: Establish the scale relationship between core test data and well logging interpretation curves.
[0058] In this embodiment, step 102 can be the data quality control stage. In this step, the core test data can be projected onto the corresponding depth of the well logging interpretation curve to perform depth repositioning of the core test data, analyze the correlation between the core test data and the well logging interpretation curve, and check the consistency between the core test data and the well logging interpretation curve to ensure good consistency between the core test data and the well logging data, so as to establish the scale relationship between the core test data and the well logging interpretation curve.
[0059] In order to control the quality of core test data, such as Figure 4 As shown, core test data is projected onto the corresponding depth of the well logging interpretation curve to ensure good consistency between the core test data and the well logging interpretation curve. In this embodiment, the core test data is projected onto the corresponding depth position on the well logging interpretation curve. When the core test data and the well logging interpretation curve basically overlap, it indicates that the difference between the two is small, and the consistency between the core test data and the well logging interpretation curve is considered good. Therefore, the core test data can be used as a training dataset. Figure 4 As shown, mathematical statistics methods can be used to establish the scale relationship between core test data and well logging interpretation curves based on the depth of the shale rock experimental sample.
[0060] There are two methods for extracting data from well logging interpretation curves. Method 1: Manually read the data directly from the well logging interpretation curves using graph paper. Method 2: Use computer technology to digitize the graphs corresponding to the well logging interpretation curves, complete the data conversion, and then extract the information. After extracting the data from the well logging interpretation curves, the data can be cleaned to remove outliers.
[0061] Step 103: Based on the core test data, a training model is obtained by training a neural network algorithm. In the training model, mineral content test data and TOC content test data are used as the input layer dataset, and velocity test data in the parallel bedding direction are used as the output layer dataset.
[0062] In this embodiment, step 103 can be the model training stage, where an artificial neural network can be used to establish a training model between mineral content, TOC content and velocity in the parallel bedding direction, thereby obtaining a training model between shale reservoir lithological composition and velocity in the parallel bedding direction.
[0063] Artificial neural networks are function models based on the structure and function of the human brain, such as... Figure 5 and Figure 6 As shown, this example illustrates the basic structure and algorithm of an artificial neural network. It includes an input layer, hidden layers, and an output layer. The hidden layer can be one or more layers. Neurons between layers are unidirectionally connected, and neurons within a layer are independent of each other. The output layer can take any value. The backpropagation error is calculated based on the network's expected output, and the weights are adjusted accordingly. The network's learning and training process is essentially a weight adjustment process, which ends when the output error is less than a given value. Artificial neural network models can approximate any nonlinear mapping with any precision, possess a certain degree of fault tolerance, and are suitable for handling complex problems.
[0064] like Figure 6 As shown, this example illustrates the computational principle of a neural network model containing one hidden layer, assuming the number of samples is n, x i (1≤i≤n) is the input value, w i (1≤i≤n) represents the input weights, y i (1≤i≤n) represents the output value. Each input neuron in the neural network has a corresponding weight, thus obtaining the input signal u. i =∑ n w i x i -θ i , where θ i Let f(t) represent the threshold for the i-th sample. Assume the system transfer function is f(t) = 1 / (1+e^(-t / t)). t Then the system output signal
[0065] In this embodiment, for example, the initial weight is given as 1, that is, the weight of each sample is equal, and the number of hidden neuron layers is 10.
[0066] In this embodiment, mineral content test data and TOC content test data from core test data are used as the input layer dataset, and velocity test data in the parallel bedding direction are used as the output layer dataset. The initial parameters of the neural network are set to obtain the training model.
[0067] Step 104: Apply the trained model to the well logging interpretation curve to obtain the velocity prediction curve of the well logging interpretation curve for the parallel bedding direction.
[0068] In this embodiment, step 104 is the model application stage. Since the well logging interpretation curve does not include the velocity curve in the parallel bedding direction, the trained model is used to predict the obtained velocity curve in the parallel bedding direction.
[0069] In one example, the trained model is applied to well logging interpretation curves to obtain velocity prediction curves for the parallel bedding direction, including:
[0070] Using the mineral content curve and TOC content curve in the well logging interpretation curve as the input signal for training the model, the predicted P-wave velocity and / or S-wave velocity curves of the well logging interpretation curve for the parallel bedding direction are obtained.
[0071] like Figure 8 As shown, the clay content curve, silica content curve, calcium content curve, and TOC content curve in the well logging interpretation curve are used as input signals for training the model. This yields the predicted P-wave velocity and S-wave velocity curves for the parallel bedding direction. Furthermore, the accuracy of the obtained predicted P-wave velocity and S-wave velocity curves for the parallel bedding direction can be determined by comparing the velocity test data at different depths of the core test data with the predicted velocity curves, thus ensuring prediction accuracy.
[0072] The method for predicting shale reservoir velocities along the bedding plane provided in this application uses an artificial neural network to predict the P-wave and S-wave velocities along the bedding plane of the target shale reservoir. This method reduces the number of input parameters while maintaining prediction accuracy. The method is simple and easy to implement, effectively saving time and labor costs.
[0073] Example 2:
[0074] The method for predicting shale reservoir velocities along the bedding direction in this embodiment can be applied to electronic devices with communication, computing, and data storage capabilities. The specific process can be as follows: Figure 9 As shown, it includes:
[0075] Step 201: Collect core test data and well logging interpretation curves of the target shale reservoir; the core test data shall include at least: mineral content test data, TOC content test data, and velocity test data in the direction of parallel bedding; correspondingly, the well logging interpretation curves shall include at least: mineral content curves and TOC content curves.
[0076] Specifically, step 201 is the data preparation stage, during which core test data and well logging interpretation curves of the target shale reservoir are collected. The core test data of the target shale reservoir can be obtained from core analysis experiments, and includes mineral content test data and TOC content test data from sampling points at different depths. Sampling points are the locations where core analysis experiments are conducted on the cored sections. Core test data is generally a set of discrete data with unequal intervals. Well logging interpretation curves are an important curve in petroleum geological exploration; they reflect not only the mineral composition of the target shale reservoir but also its structure and porosity. Mineral content test data can be obtained by testing the core samples collected from the well. These minerals include clay, silica, and calcareous materials. Total organic carbon (TOC) is the most commonly used indicator of organic matter content.
[0077] In one embodiment, the mineral content test data includes at least one of the following: clay content test data, silica content test data, and calcium content test data; correspondingly, the mineral content curve includes at least one of the following: clay content curve, silica content curve, and calcium content curve.
[0078] Velocity test data parallel to the bedding plane includes one of the following: longitudinal wave velocity test data parallel to the bedding plane and transverse wave velocity test data parallel to the bedding plane. Specifically, sonic logging instruments can be used to obtain velocity test data parallel to the bedding plane. Sonic logging instruments perform sonic measurements parallel to the wellbore. When the wellbore trajectory is perpendicular to the formation bedding plane, the wave velocity obtained in sonic logging is perpendicular to the formation; when the wellbore trajectory is parallel to the formation bedding plane, the wave velocity obtained in sonic logging is parallel to the formation. For example, refer to... Figures 2-3 The diagram illustrates, for example, the preparation and velocity measurement of shale core samples. Velocity measurements can be performed by drilling shale core samples at different bedding angles. The bedding angle is the angle between the axial direction of the shale core sample and the bedding plane; 0° is taken for the direction parallel to the bedding plane, and 90° is taken for the direction perpendicular to the bedding plane. When performing velocity testing in the direction parallel to the bedding plane, the core sample preparation and velocity measurement can be performed as follows: Figure 3 As shown, sampling and testing are carried out along the bedding direction of the shale reservoir core to obtain P-wave velocity and S-wave velocity test data parallel to the bedding direction.
[0079] Step 202: Establish the scale relationship between core test data and well logging interpretation curves.
[0080] In this embodiment, step 202 can be the data quality control stage. In this step, the core test data can be projected onto the corresponding depth of the well logging interpretation curve to perform depth repositioning of the core test data, analyze the correlation between the core test data and the well logging interpretation curve, and check the consistency between the core test data and the well logging interpretation curve to ensure good consistency between the core test data and the well logging data, so as to establish the scale relationship between the core test data and the well logging interpretation curve.
[0081] In order to control the quality of core test data, such as Figure 4 As shown, core test data is projected onto the corresponding depth of the well logging interpretation curve to ensure good consistency between the core test data and the well logging interpretation curve. In this embodiment, the core test data is projected onto the corresponding depth position on the well logging interpretation curve. When the core test data and the well logging interpretation curve basically overlap, it indicates that the difference between the two is small, and the core test data is considered to have good consistency with the well logging interpretation curve. Therefore, the core test data can be used as a training dataset. Figure 4 As shown, mathematical statistics methods can be used to establish the scale relationship between core test data and well logging interpretation curves based on the depth of the shale rock experimental sample.
[0082] There are two methods for extracting data from well logging interpretation curves. Method 1: Manually read the data directly from the well logging interpretation curves using graph paper. Method 2: Use computer technology to digitize the graphs corresponding to the well logging interpretation curves, complete the data conversion, and then extract the information. After extracting the data from the well logging interpretation curves, the data can be cleaned to remove outliers.
[0083] Step 203: Divide the core test data into a training set and a test set according to a preset ratio.
[0084] In this step, the training set is used for model training. It contains data samples used to adjust the model's internal parameters (such as neuron weights and biases). During training, the model continuously adjusts these parameters to minimize training error, thereby improving its fitting ability. The test set is used to evaluate the final generalization ability of the trained model. It is only used after model training and hyperparameter tuning are complete, and is used to test the model's performance on unseen data. The test set data does not participate in the model training and tuning process, thus providing an objective evaluation standard. Core test data can be divided into training and test sets in an 8:2 sample size ratio. This division ensures that the model has sufficient data for learning during training, as well as sufficient data for validation and testing. In other embodiments, the ratio of training to test sets can be set to 7:3. The division ratio can be adjusted according to the actual situation, and there is no specific limitation.
[0085] Step 204: Use the mineral content test data and TOC content test data in the training set as input signals for the neural network algorithm, and the velocity test data in the parallel bedding direction as output signals for the neural network algorithm to train the model and obtain the training model.
[0086] Artificial neural networks possess characteristics such as massively parallel processing, distributed information storage, and excellent self-organizing and self-learning capabilities. They are function models based on the structure and function of the human brain, such as... Figure 5 and Figure 6 As shown, this example illustrates the basic structure and algorithm of an artificial neural network. It includes an input layer, hidden layers, and an output layer. The hidden layer can be one or more layers. Neurons between layers are unidirectionally connected, and neurons within a layer are independent of each other. The output layer can take any value. The backpropagation error is calculated based on the network's expected output, and the weights are adjusted accordingly. The network's learning and training process is essentially a weight adjustment process, which ends when the output error is less than a given value. Artificial neural network models can approximate any nonlinear mapping with any precision, possess a certain degree of fault tolerance, and are suitable for handling complex problems.
[0087] like Figure 6 As shown, this example illustrates the computational principle of a neural network model containing one hidden layer, assuming the number of samples is n, x i (1≤i≤n) is the input value, w i (1≤i≤n) represents the input weights, y i (1≤i≤n) represents the output value. Each input neuron in the neural network has a corresponding weight, thus obtaining the input signal u. i =∑ n w i x i -θ i , where θ iLet f(t) represent the threshold for the i-th sample. Assume the system transfer function is f(t) = 1 / (1+e^(-t / t)). t Then the system output signal
[0088] In this embodiment, for example, the mineral content test data and TOC content test data in the training set are used as the input signals of the neural network algorithm, and the velocity test data in the parallel bedding direction is used as the output signal of the neural network algorithm for training. For example, the initial weight is given as 1, that is, the weight of each sample is equal, and the number of hidden neuron layers is 10, thus obtaining the training model.
[0089] Step 205: Apply the trained model to the well logging interpretation curve to obtain the velocity prediction curve of the well logging interpretation curve for the parallel bedding direction.
[0090] In this embodiment, step 205 is the model application stage. Since the well logging interpretation curve does not include the velocity curve in the parallel bedding direction, the trained model is used to predict the obtained velocity curve in the parallel bedding direction.
[0091] In one example, the trained model is applied to well logging interpretation curves to obtain velocity prediction curves for the parallel bedding direction, including:
[0092] Using the mineral content curve and TOC content curve in the well logging interpretation curve as the input signal for training the model, the predicted P-wave velocity and / or S-wave velocity curves of the well logging interpretation curve for the parallel bedding direction are obtained.
[0093] like Figure 8 As shown, the clay content curve, silica content curve, calcium content curve, and TOC content curve in the well logging interpretation curve are used as input signals for training the model. This yields the predicted P-wave velocity and S-wave velocity curves for the parallel bedding direction. Furthermore, the accuracy of the obtained predicted P-wave velocity and S-wave velocity curves for the parallel bedding direction can be determined by comparing the velocity test data at different depths of the core test data with the predicted velocity curves, thus ensuring prediction accuracy.
[0094] The method for predicting velocities in the parallel bedding direction of shale reservoirs provided in this application can establish a training model for predicting velocities in the parallel bedding direction based solely on the lithological information of the target shale reservoir, such as mineral content and TOC content, without requiring data on the microstructure and pore structure of the target shale reservoir. This overcomes the problems of numerous input parameters and complex calculation processes in traditional rock physics models, thereby saving labor costs and calculation time while ensuring prediction accuracy.
[0095] Example 3:
[0096] The method for predicting shale reservoir velocities along the bedding direction in this embodiment can be applied to electronic devices with communication, computing, and data storage capabilities. The specific process can be as follows: Figure 10 As shown, it includes:
[0097] Step 301: Collect core test data and well logging interpretation curves of the target shale reservoir; the core test data shall include at least: mineral content test data, TOC content test data, and velocity test data in the direction of parallel bedding; correspondingly, the well logging interpretation curves shall include at least: mineral content curves and TOC content curves.
[0098] Specifically, step 301 is the data preparation stage, during which core test data and well logging interpretation curves of the target shale reservoir are collected. The core test data of the target shale reservoir can be obtained from core analysis experiments, and includes mineral content test data and TOC content test data from sampling points at different depths. Sampling points are the locations where core analysis experiments are conducted on the cored sections. Core test data is generally a set of discrete data with unequal intervals. Well logging interpretation curves are an important curve in petroleum geological exploration; they reflect not only the mineral composition of the target shale reservoir but also its structure and porosity. Mineral content test data can be obtained by testing the core samples collected from the well. Minerals include clay, silica, and calcareous materials. Total organic carbon (TOC) is the most commonly used indicator of organic matter content.
[0099] In one embodiment, the mineral content test data includes at least one of the following: clay content test data, silica content test data, and calcium content test data; correspondingly, the mineral content curve includes at least one of the following: clay content curve, silica content curve, and calcium content curve.
[0100] Velocity test data parallel to the bedding plane includes one of the following: longitudinal wave velocity test data parallel to the bedding plane and transverse wave velocity test data parallel to the bedding plane. Specifically, sonic logging instruments can be used to obtain velocity test data parallel to the bedding plane. Sonic logging instruments perform sonic measurements parallel to the wellbore. When the wellbore trajectory is perpendicular to the formation bedding plane, the wave velocity obtained in sonic logging is perpendicular to the formation; when the wellbore trajectory is parallel to the formation bedding plane, the wave velocity obtained in sonic logging is parallel to the formation. For example, refer to... Figures 2-3 The diagram illustrates, for example, the preparation and velocity measurement of shale core samples. Velocity measurements can be performed by drilling shale core samples at different bedding angles. The bedding angle is the angle between the axial direction of the shale core sample and the bedding plane; 0° is taken for the direction parallel to the bedding plane, and 90° is taken for the direction perpendicular to the bedding plane. When performing velocity testing in the direction parallel to the bedding plane, the core sample preparation and velocity measurement can be performed as follows: Figure 2 As shown, sampling and testing are carried out along the bedding direction of the shale reservoir core to obtain P-wave velocity and S-wave velocity test data parallel to the bedding direction.
[0101] Step 302: Establish the scale relationship between core test data and well logging interpretation curves.
[0102] In this embodiment, step 302 can be the data quality control stage. In this step, the core test data can be projected onto the corresponding depth of the well logging interpretation curve to perform depth repositioning of the core test data, analyze the correlation between the core test data and the well logging interpretation curve, and check the consistency between the core test data and the well logging interpretation curve to ensure good consistency between the core test data and the well logging data, so as to establish the scale relationship between the core test data and the well logging interpretation curve.
[0103] In order to control the quality of core test data, such as Figure 4 As shown, core test data is projected onto the corresponding depth of the well logging interpretation curve to ensure good consistency between the core test data and the well logging interpretation curve. In this embodiment, the core test data is projected onto the corresponding depth position on the well logging interpretation curve. When the core test data and the well logging interpretation curve basically overlap, it indicates that the difference between the two is small, and the consistency between the core test data and the well logging interpretation curve is considered good. Therefore, the core test data can be used as a training dataset. Figure 4 As shown, mathematical statistics methods can be used to establish the scale relationship between core test data and well logging interpretation curves based on the depth of the shale rock experimental sample.
[0104] There are two methods for extracting data from well logging interpretation curves. Method 1: Manually read the data directly from the well logging interpretation curves using graph paper. Method 2: Use computer technology to digitize the graphs corresponding to the well logging interpretation curves, complete the data conversion, and then extract the information. After extracting the data from the well logging interpretation curves, the data can be cleaned to remove outliers.
[0105] Step 303: Divide the core test data into a training set and a test set according to a preset ratio.
[0106] In this step, the training set is used for model training. It contains data samples used to adjust the model's internal parameters (such as neuron weights and biases). During training, the model continuously adjusts these parameters to minimize training error, thereby improving its fitting ability. The test set is used to evaluate the final generalization ability of the trained model. It is only used after model training and hyperparameter tuning are complete, and is used to test the model's performance on unseen data. The test set data does not participate in the model training and tuning process, thus providing an objective evaluation standard. Core test data can be divided into training and test sets in an 8:2 sample size ratio. This division ensures that the model has sufficient data for learning during training, as well as sufficient data for validation and testing. In other embodiments, the ratio of training to test sets can be set to 7:3. The division ratio can be adjusted according to the actual situation, and there is no specific limitation.
[0107] Step 304: Use the mineral content test data and TOC content test data in the training set as input signals for the neural network algorithm, and the velocity test data in the parallel bedding direction as output signals for the neural network algorithm to train the model and obtain the training model.
[0108] Artificial neural networks possess characteristics such as massively parallel processing, distributed information storage, and excellent self-organizing and self-learning capabilities. They are function models based on the structure and function of the human brain, such as... Figure 5 and Figure 6 As shown, this example illustrates the basic structure and algorithm of an artificial neural network. It includes an input layer, hidden layers, and an output layer. The hidden layer can be one or more layers. Neurons between layers are unidirectionally connected, and neurons within a layer are independent of each other. The output layer can take any value. The backpropagation error is calculated based on the network's expected output, and the weights are adjusted accordingly. The network's learning and training process is essentially a weight adjustment process, which ends when the output error is less than a given value. Artificial neural network models can approximate any nonlinear mapping with any precision, possess a certain degree of fault tolerance, and are suitable for handling complex problems.
[0109] like Figure 6 As shown, this example illustrates the computational principle of a neural network model containing one hidden layer, assuming the number of samples is n, x i (1≤i≤n) is the input value, w i (1≤i≤n) represents the input weights, y i (1≤i≤n) represents the output value. Each input neuron in the neural network has a corresponding weight, thus obtaining the input signal u. i =∑ n w i x i -θ i , where θ iLet f(t) represent the threshold for the i-th sample. Assume the system transfer function is f(t) = 1 / (1+e^(-t / t)). t Then the system output signal
[0110] In this embodiment, for example, the mineral content test data and TOC content test data in the training set are used as the input signals of the neural network algorithm, and the velocity test data in the parallel bedding direction is used as the output signal of the neural network algorithm for training. The initial weight is given as 1, that is, the weight of each sample is equal, and the number of hidden neuron layers is 10, thus obtaining the training model.
[0111] Step 305: Use a test set to test whether the trained model meets the preset conditions.
[0112] If the training model meets the preset conditions, proceed to step 306; if the training model does not meet the preset conditions, return to step 305.
[0113] Among them, using a test set to test whether the trained model meets the preset conditions includes:
[0114] The performance of the trained model is evaluated using the test set to obtain performance evaluation metrics.
[0115] When the performance evaluation metrics meet the preset conditions, the trained model is obtained; the performance evaluation metrics include the training set correlation coefficient, the training set validation correlation coefficient, and the test set correlation coefficient.
[0116] In one embodiment, the preset conditions include: the correlation coefficient of the training set, the correlation coefficient of the training set, and the correlation coefficient of the test set are all greater than a preset value, which is greater than or equal to 0.9.
[0117] In this embodiment, the prediction results of the training model obtained in step 304 are calculated using the test set. If the preset conditions are met, the process proceeds to step 306; otherwise, it returns to step 304 until the preset conditions are met.
[0118] In this embodiment, the preset conditions can be set as follows: the correlation coefficient R of the training set is greater than 0.9, the correlation coefficient R of the training set validation set is greater than 0.9, and the correlation coefficient R of the test set is greater than 0.9. For example... Figure 7 As shown, an exemplary diagram illustrates the predicted shear wave velocity and longitudinal wave velocity obtained using the initial parameters in step 304. If the preset conditions are met, the process can proceed to step 306.
[0119] Step 306: Apply the trained model to the well logging interpretation curve to obtain the velocity prediction curve of the well logging interpretation curve for the parallel bedding direction.
[0120] In this embodiment, step 306 is the model application stage. Since the well logging interpretation curve does not include the velocity curve in the parallel bedding direction, the trained model is used to predict the obtained velocity curve in the parallel bedding direction.
[0121] In one example, the trained model is applied to well logging interpretation curves to obtain velocity prediction curves for the parallel bedding direction, including:
[0122] Using the mineral content curve and TOC content curve in the well logging interpretation curve as the input signal for training the model, the predicted P-wave velocity and / or S-wave velocity curves of the well logging interpretation curve for the parallel bedding direction are obtained.
[0123] like Figure 8 As shown, the clay content curve, silica content curve, calcium content curve, and TOC content curve in the well logging interpretation curve are used as input signals for training the model. This yields the predicted P-wave velocity and S-wave velocity curves for the parallel bedding direction. Furthermore, the accuracy of the obtained predicted P-wave velocity and S-wave velocity curves for the parallel bedding direction can be determined by comparing the velocity test data at different depths of the core test data with the predicted velocity curves, thus ensuring prediction accuracy.
[0124] The method for predicting shale reservoir velocities along the bedding plane provided in this application uses an artificial neural network to predict the P-wave and S-wave velocities along the bedding plane of the target shale reservoir. This method reduces the number of input parameters while maintaining prediction accuracy. The method is simple and easy to implement, effectively saving time and labor costs.
[0125] Example 4:
[0126] Another embodiment of this application relates to a device for predicting the velocity in the parallel bedding direction of shale reservoirs. The implementation details of this device are described below. The following details are provided for ease of understanding and are not essential for implementing this solution. A schematic diagram of the device for predicting the velocity in the parallel bedding direction of shale reservoirs in this embodiment can be seen as follows: Figure 11 As shown, it includes a collection module 901, a setup module 902, a training module 903, and an application module 904.
[0127] The collection module 901 is used to collect core test data and well logging interpretation curves of the target shale reservoir; the core test data includes at least: mineral content test data, TOC content test data, and velocity test data parallel to the bedding direction; correspondingly, the well logging interpretation curves include at least: mineral content curves and TOC content curves;
[0128] Module 902 is established to establish the calibration relationship between core test data and well logging interpretation curves;
[0129] Training module 903 is used to train a training model based on core test data using a neural network algorithm. In the training model, mineral content test data and TOC content test data are used as the input layer dataset, and velocity test data in the parallel bedding direction is used as the output layer dataset.
[0130] Application module 904 is used to apply the trained model to the well logging interpretation curve to obtain the velocity prediction curve of the well logging interpretation curve for the parallel bedding direction.
[0131] In some embodiments, the mineral content test data includes at least one of the following: clay content test data, silica content test data, and calcium content test data;
[0132] Accordingly, the mineral content curves include at least one of the following: clay content curve, silica content curve, and calcium content curve.
[0133] In some embodiments, the velocity test data in the parallel layering direction includes at least one of the following: longitudinal wave velocity test data in the parallel layering direction and transverse wave velocity test data in the parallel layering direction.
[0134] In some embodiments, the application module 904 is further configured to use the mineral content curve and TOC content curve in the well logging interpretation curve as input signals for training the model, to obtain the predicted P-wave velocity curve and / or S-wave velocity curve of the well logging interpretation curve for the parallel bedding direction.
[0135] In some embodiments, the training module 903 is further configured to divide the core test data into a training set and a test set according to a preset ratio; use the mineral content test data and TOC content test data in the training set as input signals for the neural network algorithm, and use the velocity test data in the parallel bedding direction as output signals for the neural network algorithm to train the model and obtain a training model.
[0136] In some embodiments, the training module 903 is further configured to use a test set to test whether the training model meets preset conditions, and when the training model meets the preset conditions, the training model is obtained.
[0137] In some embodiments, the training module 903 is further configured to use a test set to evaluate the performance of the trained model and obtain performance evaluation metrics.
[0138] When the performance evaluation metrics meet the preset conditions, the trained model is obtained; the performance evaluation metrics include the training set correlation coefficient, the training set validation correlation coefficient, and the test set correlation coefficient.
[0139] In some embodiments, the preset conditions include: the correlation coefficient of the training set, the correlation coefficient of the training set, and the correlation coefficient of the test set are all greater than a preset value, and the preset value is a value greater than or equal to 0.9.
[0140] In some embodiments, the training module 903 is further configured to divide the core test data into training set and test set samples in a ratio of 8:2.
[0141] It is worth mentioning that all modules involved in this embodiment are logical modules. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. Furthermore, to highlight the innovative aspects of this application, this embodiment does not introduce units that are not closely related to solving the technical problems proposed in this application; however, this does not mean that other units are absent in this embodiment.
[0142] Example 5:
[0143] Another embodiment of this application relates to an electronic device, such as... Figure 12 As shown, it includes: at least one processor 100; and a memory 200 communicatively connected to at least one processor 100; wherein the memory 200 stores instructions executable by at least one processor 100, the instructions being executed by at least one processor 100 to enable at least one processor 100 to execute the shale reservoir parallel bedding direction velocity prediction method in the above embodiments.
[0144] The memory and processor are connected via a bus, which can include any number of interconnecting buses and bridges, connecting various circuits of one or more processors and memories. The bus can also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and will not be described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over the wireless medium via an antenna, which further receives data and transmits it to the processor.
[0145] The processor manages the bus and general processing, and also provides various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory is used to store data used by the processor during operation.
[0146] Example 6:
[0147] Another embodiment of this application relates to a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the method embodiments described above.
[0148] That is, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.
[0149] Those skilled in the art will understand that the above embodiments are specific embodiments for implementing this application, and in practical applications, various changes can be made to them in form and detail without departing from the spirit and scope of this application.
Claims
1. A method for predicting velocities in shale reservoirs parallel to bedding direction, characterized in that, include: Collect core test data and well logging interpretation curves from the target shale reservoir; The core test data includes at least: mineral content test data, TOC content test data, and velocity test data parallel to the bedding direction; correspondingly, the well logging interpretation curves include at least: mineral content curves and TOC content curves. Establish the calibration relationship between the core test data and the well logging interpretation curve; Based on the core test data, a training model is obtained by training a neural network algorithm. In the training model, the mineral content test data and the TOC content test data are used as the input layer dataset, and the velocity test data in the parallel bedding direction is used as the output layer dataset. The training model is applied to the well logging interpretation curve to obtain the velocity prediction curve of the well logging interpretation curve for the parallel bedding direction.
2. The method for predicting velocities in the parallel bedding direction of shale reservoirs according to claim 1, characterized in that, The mineral content test data includes at least one of the following: clay content test data, silica content test data, and calcium content test data; Accordingly, the mineral content curve includes at least one of the following: clay content curve, silica content curve, and calcium content curve.
3. The method for predicting velocities in the parallel bedding direction of shale reservoirs according to claim 2, characterized in that, The velocity test data in the parallel layering direction includes at least one of the following: longitudinal wave velocity test data in the parallel layering direction and transverse wave velocity test data in the parallel layering direction.
4. The method for predicting shale reservoir velocity in the direction parallel to bedding as described in claim 3, characterized in that, Applying the trained model to the well logging interpretation curve to obtain the velocity prediction curve of the well logging interpretation curve for the parallel bedding direction includes: Using the mineral content curve and the TOC content curve in the well logging interpretation curve as input signals to the training model, the predicted P-wave velocity and / or S-wave velocity of the well logging interpretation curve for the parallel bedding direction are obtained.
5. The method for predicting shale reservoir velocities parallel to bedding direction according to any one of claims 1 to 4, characterized in that, Based on the core test data, a training model was obtained by training using a neural network algorithm, including: The core test data is divided into a training set and a test set according to a preset ratio; The mineral content test data and TOC content test data in the training set are used as input signals to the neural network algorithm, and the velocity test data in the parallel bedding direction is used as the output signal to train the neural network algorithm to obtain the training model.
6. The method for predicting velocities in the parallel bedding direction of shale reservoirs according to claim 5, characterized in that, The training model obtained by training based on the core test data using a neural network algorithm also includes: The training model is tested using the test set to determine whether it meets preset conditions. When the training model meets the preset conditions, the training model is obtained.
7. The method for predicting velocities in the parallel bedding direction of shale reservoirs according to claim 6, characterized in that, Testing whether the trained model meets the preset conditions using the test set includes: The trained model is evaluated using the test set to obtain performance evaluation metrics. The training model is obtained when the performance evaluation index meets the preset conditions; the performance evaluation index includes the training set correlation coefficient, the training set validation correlation coefficient, and the test set correlation coefficient.
8. The method for predicting velocities in the parallel bedding direction of shale reservoirs according to claim 7, characterized in that, The preset conditions include: the correlation coefficient of the training set, the correlation coefficient of the training set, and the correlation coefficient of the test set are all greater than a preset value, and the preset value is greater than or equal to 0.
9.
9. The method for predicting velocities in the parallel bedding direction of shale reservoirs according to claim 5, characterized in that, The core test data is divided into a training set and a test set according to a preset ratio, including: The core test data are divided into sets with a ratio of 8:2 between the number of samples in the training set and the number of samples in the test set.
10. A device for predicting velocities in shale reservoirs parallel to bedding direction, characterized in that, include: The collection module is used to collect core test data and well logging interpretation curves of the target shale reservoir; The core test data includes at least: mineral content test data, TOC content test data, and velocity test data parallel to the bedding direction; correspondingly, the well logging interpretation curves include at least: mineral content curves and TOC content curves. A module is established to establish the scale relationship between the core test data and the well logging interpretation curve; The training module is used to train a training model based on the core test data using a neural network algorithm. The training model uses the mineral content test data and the TOC content test data as the input layer dataset and the velocity test data in the parallel bedding direction as the output layer dataset. An application module is used to apply the training model to the well logging interpretation curve to obtain the velocity prediction curve of the well logging interpretation curve for the parallel bedding direction.
11. An electronic device, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method for predicting the velocity in the parallel bedding direction of shale reservoirs as described in any one of claims 1 to 9.
12. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for predicting the velocity in the parallel bedding direction of shale reservoirs as described in any one of claims 1 to 9.