Total organic carbon content prediction method, electronic equipment, storage medium and device

Abstract

The invention discloses a total organic carbon content prediction method, electronic equipment, a storage medium and a device. The method comprises the following steps: acquiring original logging data of a training well; calculating a correlation coefficient between the logging parameters at different depths in the original logging data and the total organic carbon content corresponding to the depths; selecting a plurality of related logging parameters meeting a set condition from the logging parameters based on the correlation coefficient; constructing training samples based on the related logging parameters of the same depth, taking the corresponding total organic carbon content as a label of the training samples, and constructing a training database and a test database based on the multiple training samples and the corresponding labels; training a total organic carbon content prediction model based on the training database; testing the trained total organic carbon content prediction model based on the test database; and predicting the total organic carbon content based on the tested total organic carbon content prediction model. According to the invention, the total organic carbon content can be predicted continuously, quickly and precisely through the total organic carbon content prediction model.

Description

Technical Field

[0001] The present invention belongs to the field of geophysical technologies, and more specifically, relates to a total organic carbon content prediction method, an electronic device, a storage medium, and a device. Background Art

[0002] Shale gas belongs to unconventional natural gas, which is often adsorbed on organic shale and in tiny pores on the surface of clay particles, stored in natural fractures and pores in a free form, or stored in kerogen in a dissolved form (Shi Wenrui et al., 2014; Pu Boling et al., 2008; Pan Renfang et al., 2009; Zhang Jinyan et al., 2012). The total organic carbon content (TOC) in shale reservoirs is an important parameter for evaluating shale gas (Meng Zhaoping et al., 2015); as one of the most important parameters in the shale "geological sweet spot", the TOC content can not only reflect the hydrocarbon generation capacity of shale reservoirs, but also indirectly determine the free gas content by affecting the effective pores, thus determining the gas content of shale (Xu Jie et al., 2019). When the total organic carbon content (TOC) in the shale layer is greater than 2%, shale gas has commercial exploitation value (Jarvie, 2007; Ross, 2008; Zhang Jinchuan, 2004). In conventional core data analysis, organic geochemical evaluation can only obtain limited and discrete TOC values, which cannot fully reflect the organic matter abundance of shale gas reservoirs. In addition, there is a problem of high analysis and testing costs (Qu Yansheng, 2011). Experts and scholars at home and abroad have conducted a series of studies to more accurately predict the TOC value and its variation law of shale reservoirs.

[0003] Due to the characteristics of good continuity and high vertical resolution of logging data, scholars at home and abroad have explored the relationship between geochemical parameters and logging information (Xiong Lei, 2014): Fertle (1988) proposed a linear regression method for natural gamma ray spectrometry logging and core data; Passey (1990) et al. proposed the ΔLogR method based on resistivity and porosity logging and the improved ΔLogR method proposed by Zhu Guangyou (2003) et al.; Schmoler (1983) et al. proposed a linear regression method based on density logging and core analysis data; Witkowsky et al. proposed a linear regression of pyrite content and core analysis data and a neural network to predict the TOC content (Huang, 1996; Guo Long, 2009). However, using a single method and parameter brings many difficulties to the logging evaluation method of shale reservoirs under the background of different organic matter content differences, organic matter abundance, and thermal evolution degree of shale;

[0004] With the rapid development of artificial intelligence, some scholars have cited artificial intelligence algorithms to predict the TOC content: Xiong Lei (2014), Meng Zhaoping (2015), and Lu Pengyu (2021) predicted the organic carbon content based on the BP neural network and well logging data; Liu Cheng (2021) predicted the organic carbon content of the Qiongzhusi Formation shale reservoir in the Weiyuan area based on the RBF neural network. The BP neural network has a strong non-linear approximation ability and can describe the complex non-linear relationship between input parameters and output parameters. However, this method has the problem of easily leading to the formation of local minima optimization and not obtaining the global optimal solution (Zhou Zheng, 2008).

[0005] It can be seen that there are many methods for predicting TOC in well logging, but the parameters considered in well logging methods are relatively single. When the organic matter content, organic matter abundance, and thermal evolution degree of shale are different, it will bring many difficulties to the well logging evaluation method of shale reservoirs; the TOC obtained by laboratory core sample measurement and analysis is the most accurate, but this method can only obtain the TOC of the core sampling section.

[0006] The information disclosed in the background art part of the present invention is only intended to deepen the understanding of the general background art of the present invention, and should not be regarded as an admission or any form of implication that this information constitutes the prior art known to those skilled in the art. Summary of the Invention

[0007] The object of the present invention is to propose a total organic carbon content prediction method, electronic device, storage medium, and device, which can continuously, quickly, and accurately predict the total organic carbon content of shale gas reservoirs.

[0008] To achieve the above object, the present invention proposes a total organic carbon content prediction method, electronic device, storage medium, and device.

[0009] According to the first aspect of the present invention, a total organic carbon content prediction method is proposed, including:

[0010] Obtain the original well logging data of the training well;

[0011] Calculate the correlation coefficient between the well logging parameters at different depths in the original well logging data and the total organic carbon content corresponding to that depth;

[0012] Based on the correlation coefficient and the well logging parameters used to calculate the total organic carbon content, select multiple relevant well logging parameters that meet the set conditions from the well logging parameters;

[0013] Construct training samples based on the relevant well logging parameters at the same depth, use the total organic carbon content corresponding to the relevant well logging parameters as the labels of the training samples, and construct a training database and a test database based on multiple training samples and the corresponding labels;

[0014] Train a total organic carbon content prediction model based on the training database;

[0015] Test the trained total organic carbon content prediction model based on the test database;

[0016] Predict the total organic carbon content based on the total organic carbon content prediction model that has passed the test.

[0017] Optionally, the expression for calculating the correlation coefficient is:

[0018]

[0019] where inx q is the value of the input logging parameter; y is the total organic carbon content; inx q,i is the value of the logging parameter at the i-th depth of the input; is the average value of the input logging parameter; y i is the value of the total organic carbon content at the i-th depth; is the average value of the total organic carbon content.

[0020] Optionally, the set conditions include:

[0021] The absolute value of the correlation coefficient is greater than 0.4.

[0022] Optionally, the constructing of the training database and the test database based on the multiple training samples and the corresponding labels includes:

[0023] Form a first matrix with all the training samples, and form the row vectors of the first matrix with the relevant logging parameters of each training sample by depth;

[0024] Form a second matrix with all the labels, and form the column vectors of the second matrix with the total organic carbon content of each label by the depth;

[0025] Perform normalization processing on the first matrix to obtain a normalized first matrix;

[0026] Construct the training database based on part of the data of the normalized first matrix and the corresponding part of the data of the second matrix;

[0027] Construct the test database based on another part of the data of the normalized first matrix and the corresponding another part of the data of the second matrix.

[0028] Optionally, the testing of the trained total organic carbon content prediction model based on the test database includes:

[0029] Input the data of the normalized first matrix in the test database into the total organic carbon content prediction model to obtain the predicted value of the total organic carbon content;

[0030] Calculate the mean absolute error and mean absolute percentage error between the predicted value of the total organic carbon content and the data of the second matrix in the corresponding test database;

[0031] If the mean absolute error is less than 0.5 and the mean absolute percentage error is less than 20%, the total organic carbon content prediction model passes the test.

[0032] Optionally, the total organic carbon content prediction model includes:

[0033] A BP neural network model based on the cuckoo algorithm.

[0034] Optionally, the expression of the normalization process is:

[0035]

[0036] where x represents the element in the column vector, x' represents the element after normalization processing, x max and x min represent the maximum and minimum values in the column vector respectively.

[0037] According to the second aspect of the present invention, a total organic carbon content prediction device is proposed, including:

[0038] An acquisition module for acquiring the original logging data of the training well;

[0039] A calculation module for calculating the correlation coefficient between the logging parameters at different depths in the original logging data and the total organic carbon content corresponding to that depth;

[0040] A selection module for selecting multiple relevant logging parameters that meet the set conditions from the logging parameters based on the correlation coefficient and the logging parameters used to calculate the total organic carbon content;

[0041] A construction module for constructing training samples based on the relevant logging parameters at the same depth, using the total organic carbon content corresponding to the relevant logging parameters as the labels of the training samples, and constructing a training database and a test database with multiple training samples and corresponding labels;

[0042] A training module for training the total organic carbon content prediction model based on the training database;

[0043] A test module for testing the trained total organic carbon content prediction model based on the test database;

[0044] A prediction module for predicting the total organic carbon content based on the total organic carbon content prediction model that has passed the test.

[0045] According to a third aspect of the present invention, an electronic device is provided, which includes:

[0046] At least one processor; and,

[0047] A memory communicatively connected to the at least one processor; wherein,

[0048] The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute any of the total organic carbon content prediction methods described in the first aspect.

[0049] According to a fourth aspect of the present invention, a non-transitory computer-readable storage medium is provided, which stores computer instructions for causing a computer to execute any of the total organic carbon content prediction methods described in the first aspect.

[0050] The beneficial effects of the present invention are as follows: The present invention obtains logging parameters with a high correlation with the total organic carbon content through the correlation coefficient, constructs a training database and a test database based on the logging parameters and the corresponding total organic carbon content, trains the total organic carbon content prediction model through the training database, tests the trained total organic carbon content prediction model through the test database, ensures the accuracy of the prediction, improves the precision of the prediction, and the total organic carbon content prediction model that passes the test can continuously, quickly and accurately predict the total organic carbon content.

[0051] The system of the present invention has other characteristics and advantages, which will be obvious from the accompanying drawings incorporated herein and the subsequent specific embodiments, or will be described in detail in the accompanying drawings incorporated herein and the subsequent specific embodiments, and these accompanying drawings and specific embodiments are jointly used to explain the specific principles of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] By describing the exemplary embodiments of the present invention in more detail in conjunction with the accompanying drawings, the above and other objects, features and advantages of the present invention will become more obvious. In the exemplary embodiments of the present invention, the same reference numerals generally represent the same components.

[0053] Figure 1 A flowchart showing the steps of a method for predicting the total organic carbon content according to the present invention.

[0054] Figure 2 A distribution diagram showing the original total organic carbon content TOC value of the training well with respect to depth according to Embodiment 2 of the present invention.

[0055] Figure 3a 、 3b Figures 3c, 3d, 3e and 3f respectively show cross - plot diagrams of acoustic travel - time AC vs. total organic carbon content TOC, neutron porosity CNL vs. total organic carbon content TOC, compensated density DEN vs. total organic carbon content TOC, natural gamma ray GR vs. total organic carbon content TOC, potassium K vs. total organic carbon content TOC, and uranium U vs. total organic carbon content TOC according to Embodiment 2 of the present invention.

[0056] Figure 4 Figure showing the comparison between the predicted value and the true value of the total organic carbon content TOC according to Embodiment 2 of the present invention.

[0057] Figure 5 Figure showing the schematic diagram of the comparison between the predicted value and the true value of the total organic carbon content TOC of Well Y1 in Work Area according to Embodiment 2 of the present invention.

[0058] Figure 6 Figure showing the schematic diagram of the comparison between the predicted value and the true value of the total organic carbon content TOC of Well Y2 in Work Area according to Embodiment 2 of the present invention.

[0059] Figure 7 Figure showing the schematic diagram of a device for predicting total organic carbon content according to Embodiment 3 of the present invention. Detailed implementation manners

[0060] The present invention will be described in more detail below with reference to the accompanying drawings. Although the preferred embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to fully convey the scope of the present invention to those skilled in the art.

[0061] A method for predicting total organic carbon content according to the present invention includes:

[0062] Obtaining the original logging data of the training well;

[0063] Calculating the correlation coefficient between the logging parameters at different depths in the original logging data and the total organic carbon content corresponding to that depth;

[0064] Based on the correlation coefficient and the logging parameters used to calculate the total organic carbon content, selecting a plurality of relevant logging parameters that meet the set conditions from the logging parameters;

[0065] Constructing training samples based on the relevant logging parameters at the same depth, using the total organic carbon content corresponding to the relevant logging parameters as the labels of the training samples, and constructing a training database and a test database based on the plurality of training samples and the corresponding labels;

[0066] Train a total organic carbon content prediction model based on a training database;

[0067] Test the trained total organic carbon content prediction model based on a test database;

[0068] Predict the total organic carbon content based on the total organic carbon content prediction model that has passed the test.

[0069] Specifically, the present invention obtains the original logging data of the training wells. The method for selecting the training wells is as follows: compare the data volume of the total organic carbon content TOC of each well, and select the well with a large data volume of the total organic carbon content TOC as the training well; the original logging data includes the logging parameters and the total organic carbon content TOC at m different depths in the well; the logging parameters include acoustic travel time AC, neutron porosity CNL, compensated density DEN, natural gamma GR, potassium K, uranium U, and resistivity RT; perform a correlation analysis on each logging parameter in the training well with the total organic carbon content TOC respectively, and sort the absolute values of the correlations in descending order, and then combine the logging parameters used by predecessors to calculate the total organic carbon content TOC, and select multiple logging parameters that meet the set conditions as the relevant logging parameters. For example, select the logging parameters with an absolute value of the correlation coefficient greater than 0.4 among the acoustic travel time, neutron porosity, compensated density, natural gamma, potassium, uranium, and resistivity as the relevant logging parameters; form a training sample with all the relevant logging parameters at the same depth, and use the total organic carbon content corresponding to this depth as the label of this training sample. A part of the training samples and the corresponding labels form a training database for training the total organic carbon content prediction model, and another part of the training samples and the corresponding labels form a test database to test the trained total organic carbon content prediction model, and judge whether the predicted value output by the total organic carbon content prediction model meets the requirements to ensure the accuracy of the prediction of the total organic carbon content prediction model, improve the precision of the prediction of the total organic carbon content prediction model, and then perform a prediction based on the total organic carbon content prediction model that has passed the test to achieve continuous, rapid, and high-precision prediction of the total organic carbon content.

[0070] In one example, the expression for calculating the correlation coefficient is:

[0071]

[0072] where inx q is the value of the input logging parameter; y is the total organic carbon content; inx q,i is the value of the logging parameter at the i-th depth of the input; is the average value of the input logging parameter; y i is the value of the total organic carbon content at the i-th depth; is the average value of the total organic carbon content.

[0073] In one example, the set conditions include:

[0074] The absolute value of the correlation coefficient is greater than 0.4.

[0075] Specifically, the logging parameters with the absolute value of the correlation coefficient greater than 0.4 are used as relevant logging parameters to improve the prediction accuracy.

[0076] In one example, constructing a training database and a test database based on multiple training samples and corresponding labels includes:

[0077] All training samples are formed into a first matrix, and the relevant logging parameters of each training sample are formed into the row vectors of the first matrix according to depth;

[0078] All labels are formed into a second matrix, and the total organic carbon content of each label is formed into the column vectors of the second matrix according to depth;

[0079] The first matrix is normalized to obtain a normalized first matrix;

[0080] A training database is constructed based on part of the data of the normalized first matrix and the corresponding part of the data of the second matrix;

[0081] A test database is constructed based on another part of the data of the normalized first matrix and the corresponding another part of the data of the second matrix.

[0082] Specifically, all training samples are formed into a first matrix, and the relevant logging parameters of each training sample are formed into the row vectors of the first matrix in the order of depth. For example, if the relevant logging parameters are acoustic travel time AC, neutron porosity CNL, and compensated density DEN, then the first row of this matrix is the acoustic travel time AC, neutron porosity CNL, and compensated density DEN of the training sample at the shallowest depth, the second row is the acoustic travel time AC, neutron porosity CNL, and compensated density DEN of the training sample at the second shallowest depth, and so on; all labels are formed into a second matrix, and the total organic carbon content of each label is also formed into the column vectors of the second matrix in the order of the depth of the first matrix, so that each label is in the same row as the corresponding training sample; the first matrix is normalized to obtain a normalized first matrix, and part of the data of the normalized first matrix and the corresponding part of the data of the second matrix are used as training data to construct a training database, and another part of the data of the normalized first matrix and the corresponding another part of the data of the second matrix are used as test data to construct a test database.

[0083] In one example, testing the trained total organic carbon content prediction model based on the test database includes:

[0084] The data of the normalized first matrix in the test database is input into the total organic carbon content prediction model to obtain the predicted total organic carbon content;

[0085] Calculate the mean absolute error and mean absolute percentage error between the predicted value of the total organic carbon content and the data in the second matrix in the corresponding test database;

[0086] If the mean absolute error is less than 0.5 and the mean absolute percentage error is less than 20%, the test of the total organic carbon content prediction model is qualified.

[0087] Specifically, input the data of the normalized first matrix in the test database into the total organic carbon content prediction model to obtain the predicted value of the total organic carbon content. Calculate the mean absolute error MAE and mean absolute percentage error MAPE between the predicted value and the data in the second matrix (actual total organic carbon content) in the test database at the corresponding depth. When the calculation results show that MAE is less than 0.5 and MAPE is less than 20%, it meets the high-precision prediction effect, and the test of the total organic carbon content prediction model is qualified.

[0088] In one example, the total organic carbon content prediction model includes:

[0089] A BP neural network model based on the cuckoo algorithm.

[0090] In one example, the expression for normalization processing is:

[0091]

[0092] where x represents the elements in the column vector, x' represents the elements after normalization processing, x max and x min respectively represent the maximum and minimum values in the column vector.

[0093] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but it is not intended to limit the present invention. It should be noted that, without conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0094] Embodiment 1

[0095] This embodiment provides a method for predicting the total organic carbon content, including:

[0096] (1) The original logging data of several known wells are provided. The original logging data include logging parameters and total organic carbon content TOC at n different depths in the well. The logging parameters include acoustic travel time AC, neutron porosity CNL, compensated density DEN, natural gamma GR, potassium K, uranium U, and resistivity RT. Select one well as the training well and the rest as the test wells;

[0097] (2) Perform a correlation analysis on each logging parameter in the training wells with the total organic carbon content (TOC) respectively, sort the absolute values of the correlations in descending order, then combine the logging parameters used by predecessors to calculate the total organic carbon content (TOC), and finally select 6 logging parameters with an absolute value of the correlation coefficient greater than 0.4 as the input parameters of the BP neural network improved by the cuckoo search (CS) algorithm, which are respectively labeled as independent variable A, independent variable B, independent variable C, independent variable D, independent variable E, and independent variable F, and the total organic carbon content (TOC) is used as the output parameter;

[0098] (3) Construct training samples and a training database;

[0099] Construct a training sample x from the 6 independent variables at the same depth in the training well i = {x Ai , x Bi , x Ci , x Di , x Ei , x Fi}, where i represents the i-th depth, i = 1 to n, and x Ai , x Bi , x Ci , x Di , x Ei , x Fi represent the values of independent variable A, independent variable B, independent variable C, independent variable D, independent variable E, and independent variable F at the i-th depth respectively, and the total organic carbon content (TOC) corresponding to this depth is y i , and use y i as the label of this training sample. All training samples form a training database;

[0100] (4) Use the cuckoo search (CS) and the BP neural network training database to construct a CSBP total organic carbon content prediction model; including steps (41)-(46);

[0101] (41) Form all training samples into a matrix X = {X A , X B , X C , X D , X E , X F}, where X A , X B , X C , X D , X E , X F are all x Ai , x Bi , x Ci , x Di , x Ei , x FiThe column vectors formed, and all total organic carbon contents TOC are formed into a column vector Y by depth;

[0102] (42) Divide 80% of the data in the column vector Y into the training data of the CSBP total organic carbon content prediction model to establish the CSBP total organic carbon content prediction model; divide 20% of the data in the column vector Y into the test data for self-training prediction.

[0103] (43) Normalize the elements in the matrix X' respectively to obtain the normalized matrix X″;

[0104] (44) Calculate the predicted values of the matrix Y, including (a1)-(a2);

[0105] (a1) Input the data of the matrix X″ corresponding to 80% of the data in the column vector Y into the cuckoo (CS) improved BP neural network algorithm to train this algorithm, thereby establishing the CSBP total organic carbon content prediction model;

[0106] (a2) Use the established CSBP total organic carbon content prediction model to predict the remaining 20% of the data in the column vector Y (denoted as the true value y zi , and the column vector formed by it is Y z ), and denote the column vector formed by the obtained predicted value y ti as Y t ;

[0107] (45) Calculate the mean absolute error MAE and the mean absolute percentage error MAPE between the predicted value Y t and the true value Y z , and require that MAE is less than 0.5 and MAPE is less than 20% to meet the high-precision prediction effect;

[0108] (46) If the established CSBP total organic carbon content prediction model meets (45), then the established prediction model meets the requirements for predicting the total organic carbon content TOC;

[0109] (5) Select a well to be measured, obtain the logging data corresponding to the matrix X in its original logging data, input it into the CSBP total organic carbon content prediction model, and output its predicted value.

[0110] In step (1), the method for selecting the training well is: compare the total organic carbon content TOC data volume of each well, and select the well with a large total organic carbon content TOC data volume as the training well.

[0111] The calculation formula of the correlation coefficient is:

[0112]

[0113] Where: inx qis the value of the input logging parameter; y is the total organic carbon content TOC; inx q,i is the value of the logging parameter at the i-th depth of the input; is the average value of the input logging parameter; y i is the value of the total organic carbon content at the i-th depth; is the average value of the total organic carbon content.

[0114] In the step (43), the normalization is performed by the following formula:

[0115]

[0116] x represents an element in the vector, x' represents the element after normalization processing, x max and x min respectively represent the maximum and minimum values in the vector.

[0117] In step (45), the predicted value Y t and the true value Yz are calculated for the mean absolute error MAE and the mean absolute percentage error MAPE, specifically calculated using the following formulas:

[0118]

[0119]

[0120] In formulas (1) and (2), y zi is the total organic carbon content TOC value at the i-th depth in step (a2) for Y z , y ti is the predicted value Y t of the i-th element, N is the amount of predicted data 。

[0121] Compared with the prior art, the advantages of the present invention are as follows:

[0122] For the first time, the cuckoo search (CS) improved BP neural network algorithm is adopted, with multiple logging parameters as inputs, training the neural network to establish a prediction model for the total organic carbon content TOC;

[0123] By further dividing the training database into 80% of the data volume as the training data for the cuckoo search (CS) improved BP neural network algorithm and 20% of the data volume as the test data for the training effect, self-prediction is carried out;

[0124] Finally, the predicted value Y t and the true value Y zThe mean absolute error MAE and the mean absolute percentage error MAPE, with MAE less than 0.5 and MAPE less than 20%, that is, a continuous, fast, and high-precision CSBP total organic carbon content prediction model can be obtained.

[0125] Example 2

[0126] This example provides a method for predicting total organic carbon content, including:

[0127] The original logging data of several known wells, where the original logging data includes logging parameters and total organic carbon content TOC at n different depths in the well; the logging parameters include acoustic travel time AC, neutron porosity CNL, compensated density DEN, natural gamma ray GR, potassium K, uranium U, and resistivity RT. Select one well as the training well and the rest as test wells; regarding the selection of the training well, as Figure 2 shown, Figure 2 where the abscissa is the total organic carbon content TOC and the ordinate is the depth. As can be seen from Figure 2 , this well has 39 total organic carbon content TOC data volumes, and the data is suitable and can be used as the training data for the prediction model. The training well and the well to be tested are located in the same work area. By performing a correlation analysis on each logging parameter in the training well with the total organic carbon content TOC respectively and combining the logging parameters used by predecessors to calculate the total organic carbon content TOC, we found through analysis that: among the collected logging parameters (acoustic travel time AC, neutron porosity CNL, compensated density DEN, natural gamma ray GR, potassium K, uranium U, and resistivity RT), resistivity RT can be excluded first because resistivity logging mainly reflects the formation resistivity and has little correlation with the organic carbon content. Through correlation analysis, it is found that the absolute values of the correlation coefficients between uranium U, potassium K, compensated neutron CNL, acoustic travel time AC, natural gamma ray GR, and compensated density DEN and the total organic carbon content TOC are 0.7744, 0.7639, 0.7160, 0.6869, 0.5927, and 0.4673 in sequence, meeting the requirements for selecting input parameters. Therefore, these 6 logging parameters are selected as the independent variables of the prediction model. The cross plots of these 6 logging parameters and the total organic carbon content TOC are as Figure 3a - Figure 3fAs shown in the 6 figures, the vertical coordinate is the total organic carbon content TOC, and the horizontal coordinates are the acoustic time difference AC, neutron porosity CNL, compensated density DEN, natural gamma ray GR, potassium K, and uranium U in sequence. It can be seen that there is a negative correlation between the acoustic time difference AC, neutron porosity CNL, compensated density DEN, and potassium K and the total organic carbon content TOC; there is a positive correlation between the natural gamma ray GR and uranium U and the total organic carbon content TOC. First, 20% (8 groups) of the data volume in the training wells is used as test data; 80% (31 groups) of the data volume is used as training data. Then, a total organic carbon content prediction model, a BP neural network total organic carbon content prediction model, and a multiple linear regression prediction model are established using the cuckoo (CS) improved BP neural network respectively. Finally, the predicted values of the test data are compared with the true values, and the comparison graph of the predicted values and true values of the total organic carbon content TOC is as Figure 4 shown. The formula of the established multiple linear regression prediction model is as follows:

[0128] TOC-DY = -8.8651 + 0.0780×AC - 0.0011×CNL + 2.4734×DEN - 0.0169×GR - 0.6578×K + 0.3821×U;

[0129] Figure 4 where the horizontal coordinate in the figure is the test sample data number, and the vertical coordinate is the total organic carbon content TOC. From Figure 4 it can be seen that the predicted values of the CSBP total organic carbon content prediction model are basically consistent with the true values, and the change trends are the same; the predicted values of the BP total organic carbon content prediction model deviate greatly from the true values; the predicted values of the multiple linear regression prediction model are basically consistent with the true values. In addition, the MAE and MAPE of the prediction results of the CSBP total organic carbon content prediction model are 0.45817 and 15.7672% respectively, and the prediction error is small; the MAE and MAPE of the prediction results of the BP total organic carbon content prediction model are 1.6787 and 36.9792% respectively; the MAE and MAPE of the prediction results of the multiple linear regression prediction model are 0.9338 and 21.48% respectively. In view of the above results, it can be known that only the prediction results of the CSBP total organic carbon content prediction model satisfy MAE < 0.5 and MAPE < 20%; to verify the effect of this embodiment, we select two wells to be measured, obtain the well logging parameters in their original well logging data that are the same as those in the training wells, and input these well logging parameters into the CSBP total organic carbon content prediction model of this embodiment. The distribution of the predicted total organic carbon content values with depth is as Figure 5 and Figure 6 shown. In the two figures, the horizontal coordinate is the total organic carbon content TOC, and the vertical coordinate is the depth. From Figure 5 it can be seen that the prediction results are basically consistent with the true values, and the average value of the total organic carbon content TOC in the predicted value depth segment is 1.68%. From Figure 6It can be seen that the predicted values are roughly consistent with the true values, and the average total organic carbon content TOC in the predicted value depth section is 2.14%. Through the prediction of the total organic carbon content TOC of other wells in the work area, the good performance of the established CSBP total organic carbon content prediction model is confirmed.

[0130] Example 3

[0131] As Figure 7 shown, this embodiment provides a total organic carbon content prediction device, including:

[0132] An acquisition module, configured to acquire the original logging data of the training well;

[0133] A calculation module, configured to calculate the correlation coefficient between the logging parameters at different depths in the original logging data and the total organic carbon content corresponding to that depth;

[0134] A selection module, configured to select multiple relevant logging parameters that meet the set conditions from the logging parameters based on the correlation coefficient and the logging parameters used to calculate the total organic carbon content;

[0135] A construction module, configured to construct training samples based on the relevant logging parameters at the same depth, use the total organic carbon content corresponding to the relevant logging parameters as the labels of the training samples, and construct a training database and a test database with multiple training samples and corresponding labels;

[0136] A training module, configured to train the total organic carbon content prediction model based on the training database;

[0137] A testing module, configured to test the trained total organic carbon content prediction model based on the test database;

[0138] A prediction module, configured to predict the total organic carbon content based on the total organic carbon content prediction model that has passed the test.

[0139] Example 4

[0140] This embodiment provides an electronic device, which includes:

[0141] At least one processor; and,

[0142] A memory communicatively connected to the at least one processor; wherein,

[0143] The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the total organic carbon content prediction method in Example 1.

[0144] An electronic device according to an embodiment of the present disclosure includes a memory and a processor. The memory is used to store non-transitory computer-readable instructions. Specifically, the memory may include one or more computer program products, and the computer program products may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory, etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0145] The processor may be a central processing unit (CPU) or other forms of processing units with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions. In an embodiment of the present disclosure, the processor is used to run the computer-readable instructions stored in the memory.

[0146] Those skilled in the art should understand that, in order to solve the technical problem of how to obtain good user experience effects, well-known structures such as communication buses and interfaces may also be included in this embodiment, and these well-known structures should also be included in the protection scope of the present disclosure.

[0147] For a detailed description of this embodiment, reference may be made to the corresponding descriptions in the foregoing embodiments, and details will not be repeated here.

[0148] Embodiment 5

[0149] This embodiment provides a non-transitory computer-readable storage medium that stores computer instructions for causing a computer to execute the total organic carbon content prediction method in Embodiment 1.

[0150] A computer-readable storage medium according to an embodiment of the present disclosure stores non-transitory computer-readable instructions thereon. When the non-transitory computer-readable instructions are run by a processor, all or part of the steps of the methods of the foregoing embodiments of the present disclosure are executed.

[0151] The above-mentioned computer-readable storage media include, but are not limited to: optical storage media (such as CD-ROM and DVD), magneto-optical storage media (such as MO), magnetic storage media (such as magnetic tape or removable hard disk), media with built-in rewritable non-volatile memory (such as memory card), and media with built-in ROM (such as ROM cartridge).

[0152] The embodiments of the present invention have been described above. The above description is exemplary and not exhaustive, and is also not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments.

[0153] The embodiments of the present invention have been described above. The above description is exemplary and not exhaustive, and is also not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments.

Claims

1. A total organic carbon content prediction method, characterized in that, it includes: Obtain the original logging data of the training wells; Calculate the correlation coefficient between the logging parameters at different depths in the original logging data and the total organic carbon content corresponding to that depth; Based on the correlation coefficient and the logging parameters used to calculate the total organic carbon content, select multiple relevant logging parameters that meet the set conditions from the logging parameters; Construct training samples based on the relevant logging parameters at the same depth, use the total organic carbon content corresponding to the relevant logging parameters as the labels of the training samples, and construct a training database and a test database based on multiple training samples and the corresponding labels; Train a total organic carbon content prediction model based on the training database; Test the trained total organic carbon content prediction model based on the test database; Predict the total organic carbon content based on the total organic carbon content prediction model that has passed the test.

2. The total organic carbon content prediction method according to claim 1, characterized in that , the expression for calculating the correlation coefficient is: where inx q is the value of the input logging parameter; y is the total organic carbon content corresponding to the input logging parameter; inx q,i is the value of the logging parameter at the i-th depth of the input; is the average value of the input logging parameter; y i is the value of the total organic carbon content at the i-th depth; is the average value of the total organic carbon content.

3. The total organic carbon content prediction method according to claim 1, characterized in that, the set conditions include: The absolute value of the correlation coefficient is greater than 0.

4.

4. The total organic carbon content prediction method according to claim 1, characterized in that, the constructing a training database and a test database based on multiple training samples and the corresponding labels includes: Form all the training samples into a first matrix, and form the relevant logging parameters of each training sample by depth into the row vectors of the first matrix; Form all the labels into a second matrix, and form the total organic carbon content of each label by depth into the column vectors of the second matrix; Perform normalization processing on the first matrix to obtain a normalized first matrix; Construct the training database based on part of the data of the normalized first matrix and the corresponding part of the data of the second matrix; Construct the test database based on another part of the data of the normalized first matrix and the corresponding another part of the data of the second matrix.

5. The total organic carbon content prediction method according to claim 4, characterized in that, the testing the trained total organic carbon content prediction model based on the test database includes: Input the data of the normalized first matrix in the test database into the total organic carbon content prediction model to obtain a total organic carbon content prediction value; Calculate the mean absolute error and mean absolute percentage error between the total organic carbon content prediction value and the data of the second matrix in the corresponding test database; If the mean absolute error is less than 0.5 and the mean absolute percentage error is less than 20%, the total organic carbon content prediction model passes the test.

6. The total organic carbon content prediction method according to claim 1, characterized in that, the total organic carbon content prediction model includes: A BP neural network model based on the cuckoo algorithm.

7. The total organic carbon content prediction method according to claim 4, characterized in that, the expression for the normalization processing is: where x represents an element in the column vector, x' represents the element after normalization, x max and x min represent the maximum value and the minimum value in the column vector, respectively.

8. An electronic device, characterized in that, The electronic device includes: 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, and the instructions are executed by the at least one processor to enable the at least one processor to execute the mature shale anisotropy analysis method according to any one of claims 1-7.

9. A non-transitory computer-readable storage medium, characterized in that, the non-transitory computer-readable storage medium stores computer instructions for causing a computer to execute the mature shale anisotropy analysis method according to any one of claims 1-7.

10. A total organic carbon content prediction device, characterized in that, it includes: an acquisition module for acquiring original logging data of a training well; a calculation module for calculating the correlation coefficient between the logging parameters at different depths in the original logging data and the total organic carbon content corresponding to that depth; a selection module for selecting a plurality of relevant logging parameters that meet the set conditions from the logging parameters based on the correlation coefficient and the logging parameters used for calculating the total organic carbon content; a construction module for constructing training samples based on the relevant logging parameters at the same depth, using the total organic carbon content corresponding to the relevant logging parameters as the labels of the training samples, and constructing a training database and a test database with the plurality of training samples and the corresponding labels; a training module for training the total organic carbon content prediction model based on the training database; a test module for testing the trained total organic carbon content prediction model based on the test database; a prediction module for predicting the total organic carbon content based on the total organic carbon content prediction model that has passed the test.

Images