Method and device for determining shale gas sorption ratio and electronic equipment

Abstract

This invention provides a method, apparatus, and electronic device for determining the shale gas free-float / absorber ratio, belonging to the field of shale gas exploration and development technology. The method includes: acquiring geological data of gas-bearing shale layers in the area to be tested; inputting the geological data into a shale gas free-float / absorber ratio prediction model to obtain the predicted shale gas free-float / absorber ratio output by the model; wherein, the shale gas free-float / absorber ratio prediction model is trained based on multiple sample geological data and the corresponding shale gas free-float / absorber ratio label for each sample geological data; the sample geological data are obtained by screening the basic geological data of gas-bearing shale layers in the target area based on factors influencing the free-float / absorber ratio. This invention implements a method for determining the shale gas free-float / absorber ratio based on an artificial intelligence model, achieving both low cost and high accuracy, and the determination of the shale gas free-float / absorber ratio provides important technical support for the intelligent decision-making process of oil and gas fields.

Description

Technical Field

[0001] This invention relates to the field of shale gas exploration and development technology, specifically to a method for determining the shale gas free-float / free-float ratio, an apparatus for determining the shale gas free-float / free-float ratio, an electronic device, a machine-readable storage medium, and a computer program product. Background Technology

[0002] Shale gas exists primarily in two states: adsorbed gas and free gas in the shale reservoir. The shale gas free-to-adsorbed ratio (the ratio of free gas to adsorbed gas) has a significant impact on shale gas resource assessment and development methods, and is an important consideration in evaluating the "sweet spot" of shale gas.

[0003] In existing technologies, experimental analysis and conventional logging methods are commonly used to evaluate shale gas quantity and the shale gas-to-gas ratio. The former is costly, while the latter is significantly affected by sample heterogeneity, and its accuracy needs improvement, which is detrimental to determining the shale gas-to-gas ratio and consequently affects the prediction and evaluation of shale gas "sweet spots." In other words, existing technologies cannot achieve a balance between low cost and high accuracy in determining the shale gas-to-gas ratio. Summary of the Invention

[0004] The purpose of this invention is to provide a method, apparatus, and electronic device for determining the shale gas free-float / absorber ratio, in order to overcome the shortcomings of existing technologies in determining the shale gas free-float / absorber ratio, which cannot achieve both low cost and high accuracy.

[0005] To achieve the above objectives, embodiments of the present invention provide a method for determining the shale gas free-floating-absorbing ratio, comprising:

[0006] Obtain the geological data of the gas-bearing shale layer in the area to be tested;

[0007] The geological data to be measured is input into the shale gas uptake ratio prediction model to obtain the shale gas uptake ratio prediction result output by the shale gas uptake ratio prediction model;

[0008] The shale gas resupply-to-supply ratio prediction model is trained based on multiple sample geological data and the shale gas resupply-to-supply ratio label corresponding to each sample geological data; the sample geological data are the basic geological data of gas-bearing shale layers in the target area, which are selected based on the factors affecting the resupply-to-supply ratio.

[0009] Optionally, the shale gas free-travel ratio prediction model is trained through the following steps:

[0010] Acquire basic geological data of gas-bearing shale formations in the target area; the basic geological data includes various types of geological data.

[0011] A multiple regression model is constructed based on the basic geological data as independent variables and the shale gas respiration ratio corresponding to the basic geological data as dependent variables.

[0012] Based on the regression coefficients of the aforementioned multiple regression model, factors influencing the reabsorption-suction ratio are screened from the basic geological data to obtain sample geological data;

[0013] The shale gas flow-to-absorption ratio prediction model is trained using the multiple sample geological data and the shale gas flow-to-absorption ratio label corresponding to each sample geological data.

[0014] Optionally, the regression coefficients based on the multiple regression model are used to screen out factors influencing the inflow-extraction ratio from the basic geological data to obtain sample geological data, including:

[0015] Based on the significance test of the regression coefficients of the multivariate regression model, and / or based on the relationship between the absolute value of the regression coefficients of the multivariate regression model and a set threshold, the influencing factors of the swimming-absorbing ratio are screened from the basic geological data to obtain sample geological data.

[0016] Optionally, the shale gas refining-to-suction ratio prediction model is constructed using an ensemble learning model; the step of training the shale gas refining-to-suction ratio prediction model using the multiple sample geological data and the shale gas refining-to-suction ratio label corresponding to each sample geological data includes:

[0017] Repeat the following steps until the set iteration stop condition is met:

[0018] The sample geological data is input into the current decision tree model to obtain the model output value of the current decision tree model;

[0019] The residual of the current decision tree model is calculated based on the model output value and the shale gas respiration ratio label of the sample geological data;

[0020] Use the residuals as the new target variable to train a new decision tree model;

[0021] The new decision tree model is weighted and fused with the current decision tree model to obtain an iterative decision tree model.

[0022] Optionally, the shale gas refining-suction ratio prediction model is constructed using a neural network model; the step of training the shale gas refining-suction ratio prediction model using the multiple sample geological data and the shale gas refining-suction ratio label corresponding to each sample geological data includes:

[0023] Repeat the following steps until the set iteration stop condition is met:

[0024] The sample geological data is input into the shale gas upwelling-to-downwell ratio prediction model to obtain the model output value of the shale gas upwelling-to-downwell ratio prediction model;

[0025] The loss function is calculated based on the model output value and the shale gas adsorption / resorption ratio label of the sample geological data;

[0026] The model parameters of the shale gas free-travel ratio prediction model are adjusted based on the loss function.

[0027] Optionally, the basic geological data includes at least two of the following: formation temperature, formation pressure, organic carbon content, porosity, mineral composition content, water saturation, and free-to-absorbance ratio.

[0028] On the other hand, embodiments of the present invention also provide a device for determining the shale gas free-float / absorption ratio, comprising:

[0029] The acquisition module is used to acquire the geological data of the gas-bearing shale layer in the area to be tested;

[0030] The prediction module is used to input the geological data to be measured into the shale gas upwelling-to-downwell ratio prediction model to obtain the shale gas upwelling-to-downwell ratio prediction result output by the shale gas upwelling-to-downwell ratio prediction model;

[0031] The shale gas resupply-to-supply ratio prediction model is trained based on multiple sample geological data and the shale gas resupply-to-supply ratio label corresponding to each sample geological data; the sample geological data are the basic geological data of gas-bearing shale layers in the target area, which are selected based on the factors affecting the resupply-to-supply ratio.

[0032] On the other hand, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-mentioned method for determining the shale gas uptake ratio.

[0033] On the other hand, the present invention also provides a machine-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for determining the shale gas uptake ratio.

[0034] On the other hand, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-mentioned method for determining the shale gas uptake ratio.

[0035] Through the above technical solution, this embodiment of the invention trains a model using sample geological data obtained from screening factors influencing the shale gas respiration ratio based on basic geological data, thus obtaining a shale gas respiration ratio prediction model. Therefore, this embodiment of the invention implements a method for determining the shale gas respiration ratio based on an artificial intelligence model, achieving a balance between low cost and high accuracy. Furthermore, the determination of the shale gas respiration ratio provides important technical support for the intelligent decision-making process of oil and gas fields.

[0036] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0037] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings:

[0038] Figure 1 This is one of the flowcharts illustrating the method for determining the shale gas free-travel ratio provided by the present invention;

[0039] Figure 2 This is the second flowchart illustrating the method for determining the shale gas free-travel ratio provided by the present invention;

[0040] Figure 3 This is a schematic diagram of the vertical distribution characteristics of the adsorption-to-float ratio of the black shale in the Longmaxi Formation of Well W204, obtained based on an artificial intelligence model provided by the present invention.

[0041] Figure 4 This is a schematic diagram of the planar distribution characteristics of the aquatic-to-hydrodynamic ratio of the black shale in the Longmaxi Formation in southern Sichuan, obtained based on a gradient boosting decision tree, as provided by this invention.

[0042] Figure 5 This is a schematic diagram of the planar distribution characteristics of the aquatic-to-hydrodynamic ratio of the black shale in the Longmaxi Formation in southern Sichuan, obtained based on a BP neural network, provided by the present invention.

[0043] Figure 6 This is one of the structural schematic diagrams of the device for determining the shale gas adsorption-resorption ratio provided by the present invention;

[0044] Figure 7 This is the second schematic diagram of the shale gas adsorption-resorption ratio determination device provided by the present invention;

[0045] Figure 8 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0046] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0047] Method Implementation Examples

[0048] Please refer to Figure 1 This invention provides a method for determining the shale gas upwelling-to-downwell ratio, comprising:

[0049] Step 100: Obtain the geological data of the gas-bearing shale layer in the area to be tested.

[0050] Electronic equipment acquires geological data of the gas-bearing shale layer in the area to be tested. This geological data may include at least one of the following: formation temperature, formation pressure, organic carbon content, porosity, mineral composition content, and water saturation of the gas-bearing shale layer in the area to be tested.

[0051] Step 200: Input the geological data to be tested into the shale gas free-float-absorption ratio prediction model to obtain the shale gas free-float-absorption ratio prediction result output by the shale gas free-float-absorption ratio prediction model.

[0052] The electronic device inputs the geological data to be measured into the shale gas uptake ratio prediction model, and obtains the shale gas uptake ratio prediction result output by the shale gas uptake ratio prediction model, thereby clarifying the spatial distribution characteristics of the uptake ratio of gas-bearing shale strata in the area to be measured.

[0053] The shale gas swim-up / swim-down ratio prediction model can be built based on various artificial intelligence models. For example, it can be based on ensemble learning models and neural network models. Ensemble learning models can include random forest models, gradient boosting decision trees, etc. Neural network models can include convolutional neural networks, backpropagation neural networks, etc.

[0054] The shale gas free-float / absorbance ratio prediction model is trained based on multiple sample geological data and the corresponding shale gas free-float / absorbance ratio label for each sample geological data. The sample geological data are obtained by screening the basic geological data of gas-bearing shale formations in the target area based on factors influencing the free-float / absorbance ratio. The basic geological data includes at least two of the following: formation temperature, formation pressure, organic carbon content, porosity, mineral composition content, water saturation, and free-float / absorbance ratio. The basic geological data can be historical well logging data of gas-bearing shale formations in the target area. This embodiment of the invention obtains sample geological data from the basic geological data based on factors influencing the free-float / absorbance ratio, thereby identifying the main controlling factors of the free-float / absorbance ratio from the basic geological data. This facilitates the construction of a dataset of shale gas free-float / absorbance ratio and main controlling factors, which is used to train the shale gas free-float / absorbance ratio prediction model, improving the prediction accuracy of the shale gas free-float / absorbance ratio prediction model.

[0055] In one embodiment, correlation analysis can be performed between basic geological data and adsorbed gas and free gas respectively, thereby obtaining sample geological data based on the influencing factors of the adsorbed-free gas-to-free gas ratio. In another embodiment, this invention uses a multivariate regression analysis method to determine the main controlling factors of the adsorbed-free gas-to-free gas ratio from basic geological data to obtain sample geological data.

[0056] For details, please refer to Figure 2 The shale gas free-to-absorption ratio prediction model is obtained through the following steps:

[0057] Step 10: Obtain basic geological data of gas-bearing shale layers in the target area.

[0058] Electronic equipment acquires basic geological data of gas-bearing shale layers in the target area. The gas-bearing shale layers in the target area can be the area to be measured or gas-bearing shale layers in areas adjacent to the area to be measured. The basic geological data includes various types of historical geological data. The basic geological data includes at least two of the following: formation temperature, formation pressure, organic carbon content, porosity, mineral composition content, water saturation, and free-floating / absorbing ratio.

[0059] Step 20: Construct a multiple regression model based on the basic geological data as the independent variable and the shale gas recirculation ratio corresponding to the basic geological data as the dependent variable.

[0060] The electronic device constructs a multiple regression model based on basic geological data as the independent variable and the shale gas respiration ratio corresponding to the basic geological data as the dependent variable. In one embodiment, the multiple regression model is expressed by the following formula:

[0061] Y = β0 + β1X1 + β2X2 + ... + β p X p +∈;

[0062] Where: Y is the dependent variable (i.e., the shale gas upwelling-downwell ratio). X1, X2, ..., X p These are the independent variables (basic geological data, i.e., formation temperature, formation pressure, organic carbon content, porosity, mineral composition, water saturation, etc.). β0 is the intercept term, representing the expected value of the dependent variable when all independent variables are 0. β1, β2, ..., β p These are regression coefficients, representing the degree of influence of each independent variable on the dependent variable. ∈ represents the error term, indicating the portion of variation in the model that is not explained; it is usually assumed to follow a normal distribution with a mean of 0. Therefore, in this embodiment of the invention, multiple basic geological data and the corresponding shale gas refining-absorption ratios are input into the above formula, and then the least squares method is used to solve for the regression coefficients of the multiple regression model.

[0063] Step 30: Based on the regression coefficients of the multivariate regression model, filter out the influencing factors of the swim-up-absorption ratio from the basic geological data to obtain sample geological data.

[0064] This invention provides an embodiment that can screen out factors influencing the swim-up-absorption ratio from basic geological data based on the regression coefficients of a multiple regression model, thereby obtaining sample geological data. Specifically, screening out these factors based on the regression coefficients of the multiple regression model includes: performing a significance test on the regression coefficients of the multiple regression model, and / or screening out these factors based on the relationship between the absolute value of the regression coefficients and a set threshold, thereby obtaining sample geological data. That is, this invention can screen out factors influencing the swim-up-absorption ratio from basic geological data based on the regression coefficients of the multiple regression model; or it can screen out these factors based on the relationship between the absolute value of the regression coefficients and a set threshold, thereby obtaining sample geological data; or it can combine the methods of performing a significance test on the regression coefficients of the multiple regression model and the relationship between the absolute value of the regression coefficients and a set threshold to screen out these factors and obtain sample geological data.

[0065] Taking basic geological data, including formation temperature, formation pressure, organic carbon content, and water saturation, as an example, let's assume the calculated multiple regression model is as follows:

[0066] Y = 4.985 + 0.212 * Formation temperature + 0.351 * Formation pressure + 0.237 * Organic carbon content + 0.051 * Water saturation; where the p-values ​​corresponding to regression coefficients 0.212, 0.351, and 0.237 are all less than 0.05; the p-value corresponding to regression coefficient 0.051 is greater than 0.05.

[0067] Taking the comparison of the absolute values ​​of regression coefficients to select sample geological data as an example, by comparing the absolute values ​​of all regression coefficients, it can be seen that the regression coefficients for formation temperature, formation pressure, and organic carbon content are all higher than the set threshold of 0.1. The regression coefficient for water saturation is lower than the set threshold of 0.1. Since a larger absolute value of the regression coefficient indicates that the independent variable has a greater impact on the dependent variable, formation temperature, formation pressure, and organic carbon content can be considered as independent variables affecting the shale gas uptake ratio. Therefore, formation temperature, formation pressure, and organic carbon content were selected as sample geological data.

[0068] Taking the significance test of regression coefficients to select sample geological data as an example, a significance level is set at p-value equal to 0.05. The p-values ​​corresponding to regression coefficients of 0.212, 0.351, and 0.237 are all less than 0.05. The p-value corresponding to regression coefficient 0.051 is greater than 0.05. Since the p-value is less than the significance level, it indicates that the independent variable has a significant effect on the dependent variable. Formation temperature, formation pressure, and organic carbon content can be considered as independent variables affecting the shale gas uptake ratio. Therefore, formation temperature, formation pressure, and organic carbon content are selected as sample geological data.

[0069] Taking the comparison of the absolute values ​​of the comprehensive regression coefficients and the significance test of the regression coefficients as an example to select sample geological data, the regression coefficients of formation temperature, formation pressure, and organic carbon content are all higher than the set threshold of 0.1, and the p-values ​​corresponding to regression coefficients of 0.212, 0.351, and 0.237 are all less than 0.05. Therefore, formation temperature, formation pressure, and organic carbon content can be considered as independent variables affecting the shale gas uptake ratio. Thus, formation temperature, formation pressure, and organic carbon content were selected as sample geological data.

[0070] Through the above method, this embodiment of the invention can obtain sample geological data. For example, the sample geological data can be geological data such as formation temperature, formation pressure, organic carbon content, porosity, and quartz content. Therefore, this embodiment of the invention can construct sample labels and sample geological datasets. Specifically, the sample geological dataset includes a training dataset and a test dataset. Each dataset is composed of sample geological data with sample features, including but not limited to formation temperature, formation pressure, organic carbon content, porosity, mineral composition content, and water saturation. For example, the sample features of each dataset are formation temperature, formation pressure, organic carbon content, porosity, and quartz content. The sample label is the shale gas adsorption-venting ratio.

[0071] Step 40: Train the shale gas flow-to-absorption ratio prediction model using the multiple sample geological data and the shale gas flow-to-absorption ratio label corresponding to each sample geological data.

[0072] Based on the training and test datasets, a training model based on artificial intelligence algorithms is constructed using the training dataset. The shale gas refining-absorption ratio prediction model is obtained through training, and the reliability of the shale gas refining-absorption ratio prediction model is verified using the test dataset.

[0073] In one embodiment, the shale gas swim-up / swim-down ratio prediction model is constructed using an ensemble learning model. For example, the shale gas swim-up / swim-down ratio prediction model is constructed using a gradient boosting decision tree. That is, based on the training dataset, a training model based on a gradient boosting decision tree is constructed, and the shale gas swim-up / swim-down ratio prediction model is obtained through training.

[0074] When constructing the shale gas gas migration ratio prediction model using a gradient boosting decision tree, step 40, training the shale gas migration ratio prediction model using the multiple sample geological data and the shale gas migration ratio label corresponding to each sample geological data, includes:

[0075] Repeat the following steps until the set iteration stop condition is met:

[0076] Step 41: Input the sample geological data into the current decision tree model to obtain the model output value of the current decision tree model.

[0077] Step 43: Calculate the residual of the current decision tree model based on the model output value and the shale gas respiration ratio label of the sample geological data.

[0078] Step 45: Use the residual as the new target variable to train a new decision tree model.

[0079] Step 47: Perform a weighted fusion of the new decision tree model and the current decision tree model to obtain an iterative decision tree model.

[0080] Specifically, training dataset Initialize the first decision tree model f0(x):

[0081]

[0082] Among them, L(y) i c) is the loss function, which calculates the residual between the predicted value and the target value, that is, the residual between the model output value and the shale gas respiration ratio label. In this embodiment of the invention, the mean squared error loss function is used; N is the number of samples; c is a constant that minimizes the loss function

[240] .

[0083] Training begins for decision trees t = 1, 2, ..., T. For each decision tree, the negative gradient r is calculated for the training samples i = 1, 2, ..., N. ti Using (x) i r ti The t-th decision tree is obtained by fitting the data, and the number of leaf nodes in this decision tree is J, with a corresponding leaf node region of R. tj Let j = 1, 2, ..., J. Then, calculate R. tj The output value c that minimizes the loss function. tj .

[0084]

[0085] Based on the fitting results of each iteration, the decision tree model is updated, and finally, the T decision tree models are combined to obtain the final shale gas uptake-to-downtake ratio prediction model f(x).

[0086]

[0087] Among them, f t-1 (x) represents the (t-1)th decision tree model; f t (x) represents the t-th decision tree model; k represents the learning efficiency.

[0088] This invention employs a grid search method to optimize key parameters (number of decision trees, maximum decision tree depth, minimum number of samples required for leaf nodes, and minimum number of samples required for further partitioning within leaf nodes) of the shale gas swim-to-absorption ratio prediction model. The reliability of the shale gas swim-to-absorption ratio prediction model is then verified using a test dataset. Therefore, this invention constructs a shale gas swim-to-absorption ratio prediction model based on a gradient boosting decision tree.

[0089] In another embodiment, the shale gas swim-up / swallowing ratio prediction model is constructed using a neural network model. For example, the shale gas swim-up / swallowing ratio prediction model is constructed using a backpropagation (BP) neural network. That is, a training model based on a BP neural network is constructed based on the training dataset, and the shale gas swim-up / swallowing ratio prediction model is obtained through training.

[0090] When constructing the shale gas gas migration ratio prediction model using a BP neural network, step 40, training the shale gas migration ratio prediction model using the multiple sample geological data and the shale gas migration ratio label corresponding to each sample geological data, includes:

[0091] Repeat the following steps until the set iteration stop condition is met:

[0092] Step 42: Input the sample geological data into the shale gas upwelling-to-downwell ratio prediction model to obtain the model output value of the shale gas upwelling-to-downwell ratio prediction model;

[0093] Step 44: Calculate the loss function based on the model output value and the shale gas respiration ratio label of the sample geological data;

[0094] Step 46: Adjust the model parameters of the shale gas upwelling-to-downwell ratio prediction model based on the loss function.

[0095] A backpropagation (BP) neural network consists of three layers: an input layer, hidden layers, and an output layer. The hidden layers can be single or multiple. During the training of the neural network, the training dataset... After being fed into the network for learning from the input layer, the activation values ​​contained in the neurons are passed from the hidden layers to the output layer. Then, based on the error direction between the model output value and the shale gas uptake / downtake ratio label, the error is propagated back from the output layer through the intermediate layers to the input layer via the backpropagation algorithm. The weights and biases of each connection are adjusted sequentially to improve the network's prediction accuracy. Specifically:

[0096] Taking a single hidden layer as an example, the weights and biases in the BP neural network are first randomly initialized, where the weights and biases from the input layer to the hidden layer are w and b1, and the weights and biases from the hidden layer to the output layer are v and b2; w T Let v be the weight matrix. T This is the bias vector.

[0097] Activate the forward propagation, setting net1 as the node input, with activation function g1 from the input layer to the hidden layer, and output h. Set net2 as the node output, with activation function g2 from the hidden layer to the output layer, and output h. N is the number of samples; the expected value of the loss function is E(θ):

[0098] net1 = w T x i +b1;

[0099] h = g1(net1);

[0100] net2 = v T h + b2 = v T g1(net1)+b2;

[0101]

[0102] Backpropagation involves calculating the error term of the output unit based on the loss function, which is equivalent to calculating the gradient of the loss function with respect to the output unit.

[0103]

[0104] Backpropagation involves calculating the error term of the hidden unit based on the loss function, which is equivalent to calculating the gradient of the loss function with respect to the hidden unit.

[0105]

[0106] Update the weights and biases of the output units in the network:

[0107]

[0108] Update the weights and biases of the hidden units in the network:

[0109]

[0110] Where η is the learning rate and k represents the number of iterations.

[0111] The forward and backward propagation processes are repeated until the loss function reaches its minimum or the number of iterations is completed. The output parameters at this point are the optimal parameters of the BP neural network. If there are multiple hidden layers, the weights and biases of the hidden layers are calculated step by step, using the same formula as above. Finally, the reliability of the shale gas flow-to-absorption ratio prediction model is verified using a test dataset.

[0112] This invention, in its embodiments, acquires basic geological data of gas-bearing shale formations in a target area; determines the main controlling factors of the shale gas resupply-supply ratio using a multivariate regression analysis method, and constructs a dataset of shale gas resupply-supply ratio and main controlling factors; based on the dataset, constructs a training model based on an artificial intelligence algorithm; based on the training model, obtains a shale gas resupply-supply ratio prediction model; and based on the shale gas resupply-supply ratio prediction model, obtains the distribution characteristics of the resupply-supply ratio of gas-bearing shale formations in the target area. In other words, this invention, in its embodiments, trains a model using sample geological data obtained from basic geological data based on factors influencing the resupply-supply ratio, and obtains a shale gas resupply-supply ratio prediction model. Therefore, this invention, in its embodiments, implements a method for determining the shale gas resupply-supply ratio based on an artificial intelligence model, achieving both low cost and high accuracy, and providing important technical support for the intelligent decision-making process of oil and gas fields.

[0113] The following section will use the Longmaxi Formation shale system in southern Sichuan as a complete example, combined with specific implementation methods, to introduce the above-mentioned method for determining the shale gas free-travel ratio based on gradient boosting decision trees, as follows:

[0114] Step 1: Obtain basic geological data of the Longmaxi Formation shale system in the Sichuan Basin, including at least two of the following: formation temperature, formation pressure, organic carbon content, porosity, mineral composition content, water saturation, adsorbed gas volume, and free gas volume.

[0115] Step 2: Determine the main controlling factors of the shale gas flow-to-absorption ratio using the multivariate regression analysis method, and construct a dataset of the shale gas flow-to-absorption ratio and its main controlling factors. Specifically, the dataset includes a training dataset and a test dataset. Each dataset consists of two parts: sample features and sample labels. The sample features are the main controlling factors of the shale gas flow-to-absorption ratio, which include at least one of the following: formation temperature, formation pressure, organic carbon content, porosity, mineral composition content, and water saturation. The sample labels are the shale gas flow-to-absorption ratio.

[0116] Based on the steps of this embodiment of the invention, the main controlling factors of shale gas flow-to-absorption ratio are: formation temperature, formation pressure, organic carbon content, porosity and quartz content. A dataset of shale gas flow-to-absorption ratio and main controlling factors is then constructed, consisting of 80 sample data, of which the first 70 samples constitute the training dataset and the last 10 samples constitute the test dataset, as shown in Table 1.

[0117] Table 1

[0118]

[0119]

[0120]

[0121]

[0122]

[0123] Step 3: Based on the training dataset, construct a training model based on gradient boosting decision trees, and obtain a shale gas refining-absorption ratio prediction model through training. The process is as follows:

[0124] Initialize the first decision tree model f0(x):

[0125]

[0126] Among them, L(y) i c) is the loss function, which calculates the residual between the predicted value and the target value. In this embodiment of the invention, the mean squared error loss function is used. N is the number of samples, which is 70 in this embodiment of the invention. c is a constant that minimizes the loss function

[240] .

[0127] Training begins for decision trees t = 1, 2, ..., T. For each decision tree, the negative gradient r is calculated for training samples i = 1, 2, ..., 70. ti Using (x) i r ti The t-th decision tree is obtained by fitting the data, and the number of leaf nodes in this decision tree is J, with a corresponding leaf node region of R. tj Let j = 1, 2, ..., J. Then, calculate R. tj The output value c that minimizes the loss function. tj .

[0128]

[0129] Based on the fitting results of each iteration, the decision tree model is updated, and finally, the T decision tree models are combined to obtain the final gradient boosting decision tree model f(x).

[0130]

[0131]

[0132] Among them, f t-1 (x) represents the (t-1)th decision tree model; f t (x) represents the t-th decision tree model; k represents the learning efficiency.

[0133] In this embodiment of the invention, a grid search method is used to optimize the key parameters of the prediction model, determining the number of decision trees, the maximum depth of the decision trees, the minimum number of samples required for leaf nodes, and the minimum number of samples required for further partitioning within leaf nodes to be 19, 4, 1, and 12, respectively. Finally, the prediction model is validated using a test dataset, and the coefficient of determination (R²) of the prediction model is obtained. 2 The mean square error (MSE) and root mean square error (RMSE) are 0.99 and 0.011, respectively, proving that the prediction model is reliable.

[0134] Step 4: Input the feature dataset of the Longmaxi Formation shale in well W204 into the shale gas-to-gas ratio prediction model to obtain the vertical distribution characteristics of the single-well gas-to-gas ratio, such as... Figure 3 As shown. Figure 3 The midstream gas uptake ratio of 1 was calculated based on a gradient boosting decision tree. Further, the characteristic datasets of the Longmaxi Formation shale at each well point in the plane were input into the shale gas uptake ratio prediction model to obtain the planar distribution characteristics of the uptake ratio, such as... Figure 4 As shown.

[0135] The following section uses the Longmaxi Formation shale system in southern Sichuan as a complete example, combined with specific implementation methods, to introduce the above-mentioned method for determining the shale gas uptake ratio based on BP neural network, as follows:

[0136] Step 1: Obtain basic geological data of the Longmaxi Formation shale system in the Sichuan Basin, including at least two of the following: formation temperature, formation pressure, TOC content, porosity, mineral composition content, water saturation, adsorbed gas volume, and free gas volume.

[0137] Step 2: Determine the main controlling factors of the shale gas flow-to-absorption ratio using the multivariate regression analysis method, and construct a dataset of the shale gas flow-to-absorption ratio and its main controlling factors. Specifically, the dataset includes a training dataset and a test dataset. Each dataset consists of two parts: sample features and sample labels. The sample features are the main controlling factors of the shale gas flow-to-absorption ratio, which include at least one of the following: formation temperature, formation pressure, organic carbon content, porosity, mineral composition content, water saturation, etc. The sample labels are the shale gas flow-to-absorption ratio.

[0138] Based on the steps of this embodiment of the invention, the main controlling factors of shale gas flow-to-absorption ratio are: formation temperature, formation pressure, organic carbon content, porosity and quartz content. A dataset of shale gas flow-to-absorption ratio and main controlling factors is then constructed, consisting of 80 sample data, of which the first 56 samples constitute the training dataset and the 24 samples constitute the test dataset, as shown in Table 1.

[0139] Step 3: Based on the training dataset, construct a training model based on a BP neural network, and obtain a shale gas upwelling-to-downwell ratio prediction model through training. The process is as follows:

[0140] A backpropagation (BP) neural network consists of three layers: an input layer, hidden layers, and an output layer. The hidden layers can be single or multiple. During the training of the neural network, the training dataset... After being fed into the network for learning from the input layer, the activation values ​​contained in the neurons are passed from the hidden layers to the output layer. Then, based on the error direction between the target output sample and the actual output, the error is propagated back from the output layer through the intermediate layers to the input layer. The weights and biases of each connection are adjusted sequentially to improve the prediction accuracy of the network. Specifically:

[0141] In this embodiment of the invention, when there are two hidden layers, the first layer has 200 neurons and the second layer has 100 neurons. First, the weights and biases in the network are randomly initialized. The weights and biases from the input layer to the first hidden layer are w1 and b1, from the first hidden layer to the second hidden layer are w2 and b2, and from the second hidden layer to the output layer are v and b3; w1 T w2 T Let v be the weight matrix. T This is the bias vector.

[0142] Activate the forward propagation, setting net1 as the node input, with activation function g1 from the input layer to the first hidden layer, and output h1. Net2 is the input from the first hidden layer to the second hidden layer, with activation function g2 and output h2. Net3 is the node output, with activation function g3 from the hidden layer to the output layer, and output h1. The expected value of the loss function, E(θ):

[0143] net1 = w1 T x i +b1;

[0144] h1 = g1(net1);

[0145] net2 = w2 T h1+b2=w2 T g1(net1)+b2;

[0146] h2 = g2(net2);

[0147] net3 = v T h² + b³ = v T g2(net2)+b3;

[0148]

[0149] Backpropagation involves calculating the error term of the output unit based on the loss function, which is equivalent to calculating the gradient of the loss function with respect to the output unit.

[0150]

[0151] Backpropagation involves calculating the error term of the hidden unit based on the loss function, which is equivalent to calculating the gradient of the loss function with respect to the hidden unit.

[0152]

[0153] Update the weights and biases of the output units in the network:

[0154]

[0155] Update the weights and biases of the hidden units in the network:

[0156]

[0157]

[0158] In the formula, η is the learning rate, and k represents the number of iterations;

[0159] The forward and backward propagation processes are repeated until the loss function reaches its minimum or the iteration count is completed. The parameters at this point are the optimal parameters for the neural network. In this embodiment, the learning rate is 0.0052 and the number of iterations is 3000. Finally, the shale gas flow-to-absorption ratio prediction model is validated using a test dataset, and the coefficient of determination (R²) of the shale gas flow-to-absorption ratio prediction model is obtained. 2 The mean absolute error (MAE) and the mean absolute error (MAE) are 0.808 and 0.338, respectively, proving that the shale gas flow-to-absorption ratio prediction model is reliable.

[0160] Step 4: Input the feature dataset of the Longmaxi Formation shale in well W204 into the shale gas-to-gas ratio prediction model to obtain the vertical distribution characteristics of the single-well gas-to-gas ratio, such as... Figure 3 As shown. Figure 3 The midstream gas uptake ratio 2 was calculated based on a BP neural network. Further, the characteristic datasets of the Longmaxi Formation shale from each well point in the plane were input into the shale gas uptake ratio prediction model to obtain the planar distribution characteristics of the uptake ratio, such as... Figure 5 As shown.

[0161] This invention acquires basic geological data of gas-bearing shale formations in a target area; determines the main controlling factors of the shale gas resupply-supply ratio using multivariate regression analysis, and constructs a dataset of shale gas resupply-supply ratio and main controlling factors; based on the dataset, constructs a training model based on artificial intelligence models such as gradient boosting decision trees or BP neural networks; based on the training model, obtains a shale gas resupply-supply ratio prediction model; and based on the shale gas resupply-supply ratio prediction model, obtains the distribution characteristics of the resupply-supply ratio of gas-bearing shale formations in the target area. This invention can accurately calculate the resupply-supply ratio of gas-bearing shale formations in a target area and clarify the spatial distribution characteristics of the resupply-supply ratio. The results of this invention are important parameters for predicting and evaluating shale gas "sweet spots," providing important technical support for the intelligent decision-making process of oil and gas fields.

[0162] Device Examples

[0163] Please refer to Figure 6 On the other hand, embodiments of the present invention also provide a device for determining the shale gas free-float / absorption ratio, comprising:

[0164] The acquisition module 601 is used to acquire the geological data of the gas-bearing shale layer in the area to be tested;

[0165] Prediction module 602 is used to input the geological data to be measured into the shale gas uptake ratio prediction model and obtain the shale gas uptake ratio prediction result output by the shale gas uptake ratio prediction model;

[0166] The shale gas resupply-to-supply ratio prediction model is trained based on multiple sample geological data and the shale gas resupply-to-supply ratio label corresponding to each sample geological data; the sample geological data are the basic geological data of gas-bearing shale layers in the target area, which are selected based on the factors affecting the resupply-to-supply ratio.

[0167] Optional, please refer to Figure 7 The shale gas free-to-absorption ratio prediction model is trained using the following modules:

[0168] The first acquisition module 701 is used to acquire basic geological data of gas-bearing shale layers in the target area; the basic geological data includes various types of geological data.

[0169] Module 702 is used to construct a multiple regression model based on the basic geological data as independent variables and the shale gas respiration ratio corresponding to the basic geological data as dependent variables.

[0170] The screening module 703 is used to screen out the factors affecting the swim-up ratio from the basic geological data based on the regression coefficients of the multiple regression model, and obtain sample geological data.

[0171] Training module 704 is used to train the shale gas flow-to-absorption ratio prediction model using the multiple sample geological data and the shale gas flow-to-absorption ratio label corresponding to each sample geological data.

[0172] Optionally, the regression coefficients based on the multiple regression model are used to screen out factors influencing the inflow-extraction ratio from the basic geological data to obtain sample geological data, including:

[0173] Based on the significance test of the regression coefficients of the multivariate regression model, and / or based on the relationship between the absolute value of the regression coefficients of the multivariate regression model and a set threshold, the influencing factors of the swimming-absorbing ratio are screened from the basic geological data to obtain sample geological data.

[0174] Optionally, the shale gas refining-to-suction ratio prediction model is constructed using an ensemble learning model; the step of training the shale gas refining-to-suction ratio prediction model using the multiple sample geological data and the shale gas refining-to-suction ratio label corresponding to each sample geological data includes:

[0175] Repeat the following steps until the set iteration stop condition is met:

[0176] The sample geological data is input into the current decision tree model to obtain the model output value of the current decision tree model;

[0177] The residual of the current decision tree model is calculated based on the model output value and the shale gas respiration ratio label of the sample geological data;

[0178] Use the residuals as the new target variable to train a new decision tree model;

[0179] The new decision tree model is weighted and fused with the current decision tree model to obtain an iterative decision tree model.

[0180] Optionally, the shale gas refining-suction ratio prediction model is constructed using a neural network model; the step of training the shale gas refining-suction ratio prediction model using the multiple sample geological data and the shale gas refining-suction ratio label corresponding to each sample geological data includes:

[0181] Repeat the following steps until the set iteration stop condition is met:

[0182] The sample geological data is input into the shale gas upwelling-to-downwell ratio prediction model to obtain the model output value of the shale gas upwelling-to-downwell ratio prediction model;

[0183] The loss function is calculated based on the model output value and the shale gas adsorption / resorption ratio label of the sample geological data;

[0184] The model parameters of the shale gas free-travel ratio prediction model are adjusted based on the loss function.

[0185] Optionally, the basic geological data includes at least two of the following: formation temperature, formation pressure, organic carbon content, porosity, mineral composition content, water saturation, and free-to-absorbance ratio.

[0186] The device for determining the shale gas uptake ratio includes a processor and a memory. The aforementioned acquisition module 601, prediction module 602, first acquisition module 701, construction module 702, screening module 703, and training module 704 are all stored in the memory as program units. The processor executes the aforementioned program units stored in the memory to achieve the corresponding functions.

[0187] A processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured.

[0188] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0189] Figure 8 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 8 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communication interface 820, and the memory 830 communicate with each other through the communication bus 840. The processor 810 can call logical instructions in the memory 830 to execute a method for determining the shale gas resupply-to-gas ratio, including: acquiring the geological data to be measured of the gas-bearing shale layer in the area to be measured; inputting the geological data to be measured into a shale gas resupply-to-gas ratio prediction model to obtain the shale gas resupply-to-gas ratio prediction result output by the shale gas resupply-to-gas ratio prediction model; wherein the shale gas resupply-to-gas ratio prediction model is trained based on multiple sample geological data and the shale gas resupply-to-gas ratio label corresponding to each sample geological data; the sample geological data are the basic geological data of the gas-bearing shale layer in the target area, which are selected based on the factors affecting the resupply-to-gas ratio.

[0190] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0191] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a machine-readable storage medium. When the computer program is executed by a processor, the computer can perform a method for determining the shale gas resupply-to-supply ratio, including: acquiring geological data of the gas-bearing shale layer in the area to be tested; inputting the geological data to be tested into a shale gas resupply-to-supply ratio prediction model to obtain a shale gas resupply-to-supply ratio prediction result output by the shale gas resupply-to-supply ratio prediction model; wherein, the shale gas resupply-to-supply ratio prediction model is trained based on multiple sample geological data and the shale gas resupply-to-supply ratio label corresponding to each sample geological data; the sample geological data are the basic geological data of the gas-bearing shale layer in the target area, which are selected based on factors affecting the resupply-to-supply ratio.

[0192] In another aspect, the present invention also provides a machine-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a method for determining the shale gas resupply-to-supply ratio, comprising: acquiring geological data of gas-bearing shale layers in a test area; inputting the geological data to a shale gas resupply-to-supply ratio prediction model to obtain a shale gas resupply-to-supply ratio prediction result output by the shale gas resupply-to-supply ratio prediction model; wherein the shale gas resupply-to-supply ratio prediction model is trained based on multiple sample geological data and shale gas resupply-to-supply ratio labels corresponding to each sample geological data; the sample geological data are basic geological data of gas-bearing shale layers in the target area, selected based on factors influencing the resupply-to-supply ratio.

[0193] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0194] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments. Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for determining the shale gas uptake ratio, characterized in that, include: Obtain the geological data of the gas-bearing shale layer in the area to be tested; The geological data to be measured is input into the shale gas uptake ratio prediction model to obtain the shale gas uptake ratio prediction result output by the shale gas uptake ratio prediction model; The shale gas resupply-to-supply ratio prediction model is trained based on multiple sample geological data and the shale gas resupply-to-supply ratio label corresponding to each sample geological data; the sample geological data are the basic geological data of gas-bearing shale layers in the target area, which are selected based on the factors affecting the resupply-to-supply ratio.

2. The method for determining the shale gas uptake ratio according to claim 1, characterized in that, The shale gas free-to-absorption ratio prediction model is obtained through the following steps: Acquire basic geological data of gas-bearing shale formations in the target area; the basic geological data includes various types of geological data. A multiple regression model is constructed based on the basic geological data as independent variables and the shale gas respiration ratio corresponding to the basic geological data as dependent variables. Based on the regression coefficients of the aforementioned multiple regression model, factors influencing the reabsorption-suction ratio are screened from the basic geological data to obtain sample geological data; The shale gas flow-to-absorption ratio prediction model is trained using the multiple sample geological data and the shale gas flow-to-absorption ratio label corresponding to each sample geological data.

3. The method for determining the shale gas uptake ratio according to claim 2, characterized in that, The regression coefficients based on the multiple regression model are used to screen out the influencing factors of the inflow-extraction ratio from the basic geological data to obtain sample geological data, including: The significance of the regression coefficients of the multivariate regression model is tested, and / or the relationship between the absolute value of the regression coefficients of the multivariate regression model and a set threshold is used to screen out the influencing factors of the swimming-absorbing ratio from the basic geological data to obtain sample geological data.

4. The method for determining the shale gas uptake ratio according to claim 2, characterized in that, The shale gas gas migration ratio prediction model is constructed using an ensemble learning model; the training of the shale gas migration ratio prediction model using the multiple sample geological data and the shale gas migration ratio label corresponding to each sample geological data includes: Repeat the following steps until the set iteration stop condition is met: The sample geological data is input into the current decision tree model to obtain the model output value of the current decision tree model; The residual of the current decision tree model is calculated based on the model output value and the shale gas respiration ratio label of the sample geological data; Use the residuals as the new target variable to train a new decision tree model; The new decision tree model is weighted and fused with the current decision tree model to obtain an iterative decision tree model.

5. The method for determining the shale gas uptake ratio according to claim 2, characterized in that, The shale gas respiration ratio prediction model is constructed using a neural network model; the training of the shale gas respiration ratio prediction model using the multiple sample geological data and the shale gas respiration ratio label corresponding to each sample geological data includes: Repeat the following steps until the set iteration stop condition is met: The sample geological data is input into the shale gas upwelling-to-downwell ratio prediction model to obtain the model output value of the shale gas upwelling-to-downwell ratio prediction model; The loss function is calculated based on the model output value and the shale gas adsorption / resorption ratio label of the sample geological data; The model parameters of the shale gas free-travel ratio prediction model are adjusted based on the loss function.

6. The method for determining the shale gas free-floating-absorption ratio according to any one of claims 1 to 5, characterized in that, The basic geological data includes at least two of the following: formation temperature, formation pressure, organic carbon content, porosity, mineral composition content, water saturation, and free-flow ratio.

7. A device for determining the shale gas uptake ratio, characterized in that, include: The acquisition module is used to acquire the geological data of the gas-bearing shale layer in the area to be tested; The prediction module is used to input the geological data to be measured into the shale gas upwelling-to-downwell ratio prediction model to obtain the shale gas upwelling-to-downwell ratio prediction result output by the shale gas upwelling-to-downwell ratio prediction model; The shale gas resupply-to-supply ratio prediction model is trained based on multiple sample geological data and the shale gas resupply-to-supply ratio label corresponding to each sample geological data; the sample geological data are the basic geological data of gas-bearing shale layers in the target area, which are selected based on the factors affecting the resupply-to-supply ratio.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method for determining the shale gas free-travel ratio as described in any one of claims 1 to 6.

9. A machine-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for determining the shale gas free-travel ratio as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for determining the shale gas free-travel ratio as described in any one of claims 1 to 6.

Images