Method for predicting solvability of coal rock peak intensity change under action of CO2

Abstract

The invention provides a solvability prediction method for coal rock peak intensity change under the action of CO2, and relates to the field of geological exploration and development, and the method comprises the steps: collecting experimental data and coal rock industrial component data, and constructing a data set; preprocessing the data set; establishing a random forest prediction model, and training the random forest prediction model through the preprocessed data set to obtain a prediction value; explaining the trained random forest prediction model to obtain the importance ranking of each input variable to the output variable; and according to the importance ranking and the predicted value, drawing a cross plot of the predicted value and a real value in the data set, a schematic diagram of influence of each input variable on an output variable and an importance ranking diagram of the input variables, and completing accurate prediction and explanation of the peak intensity change of the coal rock under the action of CO2. According to the technical scheme, the flexibility and applicability of the model are greatly improved, the calculation time is shortened, the calculation cost is reduced, and the method has important guiding significance for long-term safety and stability of carbon geological sequestration.

Description

Technical Field

[0001] This application relates to the field of geological exploration and development, in particular to a CO 2 A solvable prediction method for the change of peak strength of coal and rock under the action of Background Art

[0002] Peak strength (S) is an important parameter of coal rock mechanical properties, which can accurately predict and evaluate CO 2 The decrease in the peak strength of coal rock under the action of gravity (ΔS) has important guiding significance for the long-term safety and stability of carbon geological storage.

[0003] Accurately obtain CO 2 The ΔS under the action of CO often needs to be analyzed through sampling and experimental analysis methods, which are not only time-consuming and labor-intensive but also costly. In addition, the Langmuir formula can accurately characterize the relationship between pressure and gas adsorption. Previous studies have found that the variation law of coal rock peak intensity and gas adsorption is similar. Some scholars have used the Langmuir formula to analyze CO 2 However, empirical formulas such as Langmuir also have many limitations, such as the inability to predict the changes in the coal matrix under high pressure conditions.

[0004] With the development of technology, machine learning algorithms such as BP neural network and support vector machine algorithm can realize CO 2 The rapid prediction of ΔS under the action of CO can improve accuracy and efficiency compared with traditional methods. However, the above-mentioned technologies still have shortcomings and it is difficult to explain how the input variables affect the output variables, which limits the operability of the model and makes it difficult to adapt to the changing geological environment. In the field of geology, accurate prediction and verification of the model should not be the ultimate goal of the research process. For scholars, it is particularly important to provide explanations for the model to promote its promotion and use. Therefore, there is an urgent need for a method to explain and analyze the prediction model to clarify how the input variables affect the CO 2 The output of the ΔS prediction model under the action of Summary of the invention

[0005] The purpose of the present invention is to solve the existing CO 2 The rapid prediction model of ΔS under the action of CO 2 A solvable prediction method for the change of peak strength of coal and rock under the action of

[0006] The above-mentioned purpose of the present application is achieved through the following technical solutions: S1: Collect experimental data and coal rock industry component data and construct data sets; S2: preprocess the data set; S3: Establish a random forest prediction model, train the random forest prediction model with the preprocessed data set, and obtain the prediction value; S4: Explain the trained random forest prediction model and get the importance ranking of each input variable to the output variable; The input variable is CO 2 Injection pressure, CO 2 Saturation temperature, CO 2 saturation time, moisture, ash, volatile matter, and fixed carbon; The output variable is CO 2 The decrease in the peak strength of coal and rock under the action of S5: Based on the importance ranking and predicted values, draw the intersection diagram of the predicted values ​​and the true values ​​in the data set, the schematic diagram of the impact of each input variable on the output variable, and the input variable importance ranking diagram to complete CO 2 Accurate prediction and explanation of the changes in peak strength of coal and rock under load.

[0007] Optionally, the experimental data include: CO 2 The decrease in peak strength of coal and rock under the action of CO 2 Injection pressure, CO 2 Saturation temperature and CO 2 Saturation time; Coal petrographic industrial component data include: moisture, ash, volatile matter and fixed carbon.

[0008] Optionally, step S2 includes: The preprocessing includes: outlier removal and data correction.

[0009] Optionally, step S3 includes: S31: Establish a random forest prediction model using Python software; S32: Import the data set into the random forest prediction model for training; The CO in the dataset 2 Injection pressure, CO 2 Saturation temperature, CO 2 Saturation time, moisture, ash, volatile matter, and fixed carbon were set as independent variables; CO 2 The decrease in the peak strength of coal rock under the action of the load is set as the dependent variable.

[0010] Optionally, step S3 further includes: Using R 2 The error index evaluates the performance of the random forest prediction model, and the formula is as follows:

[0011] in is the correlation coefficient of determination, is the number of samples in the dataset, is the actual value, is the predicted value, is the average of the actual values.

[0012] Optionally, step S4 includes: The SHAP tool is used to interpret the trained random forest prediction model to obtain the importance ranking of the input variables to the output variables. The SHAP value calculation formula is as follows:

[0013] Where: For the random forest prediction model; , when the input feature is observed hour, =1, otherwise 0; is the number of features, Features The contribution value of is expressed as follows:

[0014] Where: For input features SHAP value of is the set of sample features that need to be explained, To include The set of non-zero indices in , is the total number of sample features, For input features Contribution to output variables; The SHAP value calculation formula is used to calculate the SHAP value of each input variable, and the average value of the SHAP value is used as the influence of the input variable on the prediction result. The input features in the input variable are explained and ranked in importance to obtain the importance ranking of the input variable to the output variable. The output variable is the prediction result of the random forest prediction model.

[0015] An electronic device comprises a processor, a memory, a user interface and a network interface, wherein the memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes a CO 2 A solvable prediction method for the change of peak strength of coal and rock under the action of

[0016] A computer-readable storage medium stores instructions, which, when executed, execute a CO 2A solvable prediction method for the change of peak strength of coal and rock under the action of

[0017] The beneficial effects of the technical solution provided by this application are: Based on experimental data under different experimental conditions and coal rock industry component data, the random forest algorithm was used to predict CO 2 The ΔS of coal rock under the action of CO2 can be accurately predicted. Compared with traditional algorithms, machine learning prediction has the advantages of convenience, efficiency and accuracy. In addition, the SHAP tool is used to explain the random forest prediction model, clarify the impact of each input variable on the prediction result, and derive the importance ranking of the input variables. 2 The ΔS of coal rock under the action of CO2 can be accurately predicted and interpreted, which has the advantages of effectively reducing costs, and the model has strong explanatory power and applicability. The technical solution of this application shortens the time for calculating data compared with conventional methods, has substantial characteristics and progress, and the invention is easy to operate, has high accuracy, reduces human errors, and produces positive effects. It can accurately interpret CO2 2 The prediction model of the peak strength change of coal and rock under the action of water clarifies how the input variables affect the output variables, is more flexible and applicable, and greatly reduces the research cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The present application will be further described below with reference to the accompanying drawings and embodiments, in which: Figure 1 is a step diagram in an embodiment of the present application; Figure 2 It is a prediction result diagram in the embodiment of the present application; Figure 3 It is a scatter diagram in the embodiment of the present application; Figure 4 It is a ranking diagram of the importance of each variable in the embodiment of the present application; Figure 5 It is a schematic diagram of the structure of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0019] In order to have a clearer understanding of the technical features, purposes and effects of the present application, the specific implementation methods of the present application are now described in detail with reference to the accompanying drawings.

[0020] The embodiment of the present application provides a CO 2 A solvable prediction method for the change of peak strength of coal and rock under the action of

[0021] Please refer to Figure 1 , Figure 1 It is a CO in the embodiment of this application. 2 The step diagram of the solvability prediction method of the peak strength change of coal and rock under action includes: S1: Collect experimental data and coal rock industry component data and construct data sets; S2: preprocess the data set; S3: Establish a random forest prediction model, train the random forest prediction model with the preprocessed data set, and obtain the prediction value; S4: Explain the trained random forest prediction model and get the importance ranking of each input variable to the output variable; The input variable is CO 2 Injection pressure, CO 2 Saturation temperature, CO 2 saturation time, moisture, ash, volatile matter, and fixed carbon; The output variable is CO 2 The decrease in the peak strength of coal and rock under the action of S5: Based on the importance ranking and predicted values, draw the intersection diagram of the predicted values ​​and the true values ​​in the data set, the schematic diagram of the impact of each input variable on the output variable, and the input variable importance ranking diagram to complete CO 2 Accurate prediction and explanation of the changes in peak strength of coal and rock under load.

[0022] As an example, by constructing CO based on different experimental conditions and coal rock industry components 2 Machine learning prediction model of ΔS under the action of CO 2 Injection pressure, CO 2 Saturation temperature and CO 2 Saturation time, industrial components include moisture, ash, volatile matter and fixed carbon. Finally, the SHAP tool is used to interpret the prediction model, analyze how the input variables affect the output results of the prediction model, and clarify the importance ranking of the input variables to the output variables in the process, such as Figure 4 shown.

[0023] As an example, Figure 2 As shown, ΔS(CO 2 The decrease in the peak strength of coal rock under the action of 2 =0.97. 2 Injection pressure, CO 2 Saturation time, CO 2 Saturation temperature and fixed carbon have a positive impact on the prediction results, among which CO 2 The injection pressure has the greatest impact. Volatile matter, ash and moisture have a negative impact on the prediction results, among which volatile matter has the greatest impact. The fitting accuracy of the measured ΔS value and the predicted value is high, and R 2 = 0.97, the model achieves CO 2 Accurate prediction of coal rock ΔS under the action of

[0024] As an example, Figure 3 As shown in the figure, red represents high eigenvalues ​​and blue represents low eigenvalues. If the eigenvalue is larger, the corresponding SHAP value is larger, which means that the feature has a positive impact on the output variable, otherwise it has a negative impact. Therefore, it can be concluded that CO 2 Injection pressure, CO 2 Saturation time, CO 2 Saturation temperature and fixed carbon have a positive impact on the prediction results, among which CO 2 The injection pressure has the greatest impact. Volatile matter, ash and moisture have a negative impact on the prediction results, among which volatile matter has the greatest impact. This result is consistent with the existing research rules, indicating that the model can be accurately interpreted by the SHAP tool. In addition, Figure 3 By sorting the input features by the absolute value of the SHAP average, it can be found that among all the input features, CO 2 Effect of injection pressure on CO 2 The prediction of coal rock ΔS under the action of ΔS has the greatest impact.

[0025] The experimental data include: CO 2 The decrease in peak strength of coal and rock under the action of CO 2 Injection pressure, CO 2 Saturation temperature and CO 2 Saturation time; Coal petrographic industrial component data include: moisture, ash, volatile matter and fixed carbon.

[0026] As an example, CO 2 The decrease in peak strength of coal rock under the action of CO (ΔS,%), 2 Injection pressure (P ci , MPa), CO 2 Saturation temperature (T cs ,℃)、CO 2 Saturation time (t cs , h), moisture (M ad ,%), ash content (A d ,%), volatile matter (V daf ,%) and fixed carbon (FC ad , %).

[0027] Step S2 includes: The preprocessing includes: outlier removal and data correction.

[0028] As an embodiment, the data set is preprocessed to ensure the continuity of the data.

[0029] Step S3 includes: S31: Establish a random forest prediction model using Python software; S32: Import the data set into the random forest prediction model for training; The CO in the dataset 2 Injection pressure, CO 2 Saturation temperature, CO 2 Saturation time, moisture, ash, volatile matter, and fixed carbon were set as independent variables; CO 2 The decrease in the peak strength of coal rock under the action of the load is set as the dependent variable.

[0030] As an embodiment, the specific operation steps of the random forest prediction model are as follows: first, clear the environmental variables and import the data. Divide the independent variables (experimental condition data and industrial component data) and the dependent variable (ΔS) data into a training set and a test set: randomly shuffle the data, take the first 80% of the samples as the training set, and the remaining samples as the test set; and normalize the input and output data of the training set and the test set so that their values ​​are between 0 and 1. Establish a random forest model: adjust the number of decision trees, the minimum number of samples for each leaf node, and enable OOB (Out-Of-Bag Error) error prediction to calculate the accuracy of the random forest model. Random forest model training: open the random forest training window of Python, use the training set to train the random forest model, and then use the trained random forest to predict the test set.

[0031] Step S3 also includes: Using R 2 The error index evaluates the performance of the random forest prediction model, and the formula is as follows:

[0032] in is the correlation coefficient of determination, is the number of samples in the dataset, is the actual value, is the predicted value, is the average of the actual values.

[0033] Step S4 includes: The SHAP tool is used to interpret the trained random forest prediction model to obtain the importance ranking of the input variables to the output variables. The SHAP value calculation formula is as follows:

[0034] Where: For the random forest prediction model; , when the input feature is observed hour, =1, otherwise 0; is the number of features, Features The contribution value of is expressed as follows:

[0035] Where: For input features SHAP value of is the set of sample features that need to be explained, To include The set of non-zero indices in , is the total number of sample features, For input features Contribution to output variables; The SHAP value calculation formula is used to calculate the SHAP value of each input variable, and the average value of the SHAP value is used as the influence of the input variable on the prediction result. The input features in the input variable are explained and ranked in importance to obtain the importance ranking of the input variable to the output variable. The output variable is the prediction result of the random forest prediction model.

[0036] The present application also discloses an electronic device. Figure 5 , Figure 5 The electronic device 500 may include: at least one processor 501 , at least one network interface 504 , a user interface 503 , a memory 505 , and at least one communication bus 502 .

[0037] The communication bus 502 is used to realize the connection and communication between these components.

[0038] The user interface 503 may include a display screen, and the optional user interface 503 may also include a standard wired interface or a wireless interface.

[0039] The network interface 504 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0040] The present application also discloses a computer-readable storage medium, wherein the computer-readable storage medium stores a plurality of instructions, and the instructions are suitable for a processor to load to execute the above-mentioned CO 2 A solvable prediction method for the change of peak strength of coal and rock under the action of

[0041] The above are only exemplary embodiments of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure.

[0042] This application is intended to cover any variation, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the art not described in the present disclosure. The description and examples are to be regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A method for predicting the solvability of coal rock peak strength changes, characterized in that: The method comprises the following steps: S1: Collect experimental data and coal rock industry component data and construct data sets; S2: preprocess the data set; S3: Establish a random forest prediction model, train the random forest prediction model with the preprocessed data set, and obtain the prediction value; S4: Explain the trained random forest prediction model and get the importance ranking of each input variable to the output variable; The input variables are CO2 injection pressure, CO2 saturation temperature, CO2 saturation time, moisture, ash, volatile matter, and fixed carbon; The output variable is the decrease in the peak strength of coal rock under the action of CO2; S5: Based on the importance ranking and predicted values, draw the intersection diagram of the predicted value and the true value in the data set, the schematic diagram of the impact of each input variable on the output variable, and the input variable importance ranking diagram to complete the accurate prediction and explanation of the change in the peak strength of coal rock under the action of CO2.

2. The method for predicting the solvability of the peak strength change of coal rock under the action of CO2 as claimed in claim 1 is characterized in that: The experimental data include: the decrease in peak strength of coal rock under the action of CO2, CO2 injection pressure, CO2 saturation temperature and CO2 saturation time; Coal petrographic industrial component data include: moisture, ash, volatile matter and fixed carbon.

3. The method for predicting the solvability of the peak strength change of coal rock under the action of CO2 as claimed in claim 1 is characterized in that: Step S2 includes: The preprocessing includes: outlier removal and data correction.

4. The method for predicting the solvability of the peak strength change of coal rock under the action of CO2 as claimed in claim 1 is characterized in that: Step S3 includes: S31: Establish a random forest prediction model using Python software; S32: Import the data set into the random forest prediction model for training; The CO2 injection pressure, CO2 saturation temperature, CO2 saturation time, moisture, ash, volatile matter and fixed carbon in the data set are set as independent variables; the decrease in the peak strength of coal rock under the action of CO2 is set as the dependent variable.

5. The method for predicting the solvability of the peak strength change of coal rock under the action of CO2 as claimed in claim 1, characterized in that: Step S3 also includes: Using R 2 The error index evaluates the performance of the random forest prediction model, and the formula is as follows: in is the correlation coefficient of determination, is the number of samples in the dataset, is the actual value, is the predicted value, is the average of the actual values.

6. The method for predicting the solvability of the peak strength change of coal rock under the action of CO2 as claimed in claim 1, characterized in that: Step S4 includes: The SHAP tool is used to interpret the trained random forest prediction model to obtain the importance ranking of the input variables to the output variables. The SHAP value calculation formula is as follows: Where: For the random forest prediction model; , when the input feature is observed hour, =1, otherwise 0; is the number of features, Features The contribution value of is expressed as follows: Where: For input features SHAP value of is the set of sample features that need to be explained, To include The set of non-zero indices in , is the total number of sample features, For input features Contribution to output variables; The SHAP value calculation formula is used to calculate the SHAP value of each input variable, and the average value of the SHAP value is used as the influence of the input variable on the prediction result. The input features in the input variable are explained and ranked in importance to obtain the importance ranking of the input variable to the output variable. The output variable is the prediction result of the random forest prediction model.

7. An electronic device, characterized in that: The electronic device comprises a processor (501), a memory (505), a user interface (503) and a network interface (504), wherein the memory (505) is used to store instructions, the user interface (503) and the network interface (504) are used to communicate with other devices, and the processor (501) is used to execute the instructions stored in the memory (505) so that the electronic device executes the method according to any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed by a computer, the method according to any one of claims 1 to 6 is executed.

Images