Method for constructing an analysis model to predict rock properties by simulating a rock drilling process

Abstract

This invention provides a method for constructing an analytical model to predict rock properties by simulating the rock drilling process. It relates to the fields of geotechnical and tunnel engineering. The method uses similarity criteria to determine the similarity parameters between drilling parameters and drilling mode parameters and the actual drilling process. A rock drilling process simulation platform is used to simulate the actual engineering rock drilling process to obtain measured drilling parameter data. The drilling parameter data is segmented according to a time series, and features are extracted. Using different combinations of feature vector matrices as input and rock properties as output, the model is trained. Sensitive data on rock properties are obtained using feature importance analysis or ablation experiments. Based on this sensitive data, an analytical model is constructed, and the model is used to predict rock properties. This method reduces the workload of data analysis, improves analytical efficiency and accuracy, and solves the problem of low accuracy in existing rock property prediction methods. This invention is applicable to rock property analysis during the drilling process.

Description

Technical Field

[0001] This invention relates to the field of geotechnical and tunnel engineering technology, and in particular to a method for constructing an analytical model to predict rock properties by simulating the rock drilling process. Background Technology

[0002] In geotechnical engineering, tunnel engineering, and scientific research, rock drilling parameters, such as temperature, vibration, drilling pressure, rotational speed, torque, and drilling depth, contain rich information about rock properties. Accurately predicting rock properties is crucial for reducing construction risks and improving drilling efficiency. However, effectively obtaining rock drilling parameters during actual engineering construction remains very difficult. Sensors need to be deployed on-site in advance, which is both challenging and costly. Furthermore, site conditions and equipment limitations can interfere with sensor data acquisition, affecting the accuracy of data and analysis results. Therefore, conducting more efficient, economical, and accurate simulations of the rock drilling process based on actual engineering contexts is an effective way to overcome the difficulties of on-site measurement. On the other hand, predicting rock properties based on drilling parameters requires extensive sensor data analysis. Conventional predictive analysis models typically require a large amount of sample data for training. This sample data often reflects the influence of the original engineering and experimental conditions. Under different conditions, the sensitivity of rock properties to different data varies. Ignoring the differences in sample data and the sensitivity of rock properties to different data will lead to low accuracy of the trained analysis model, and may even result in unreliable analytical conclusions, making it difficult to effectively guide actual rock drilling operations. Summary of the Invention

[0003] The technical problem solved by this invention: This invention provides a method for constructing an analytical model to predict rock properties by simulating the rock drilling process, thereby solving the problem of low accuracy in existing rock property prediction methods.

[0004] The technical solution adopted by this invention to solve the above-mentioned technical problems is a method for constructing an analytical model to predict rock properties by simulating the rock drilling process, comprising the following steps: S01. Determine the similarity parameters between drilling parameters and drilling mode parameters and the actual engineering rock drilling process by using similarity criteria; S02. Set the drilling mode parameters, simulate the drilling process through the rock drilling process simulation platform, and obtain the measured drilling parameter data. S03. Perform data cleaning on the measured drilling parameter data to obtain the cleaned drilling parameter data. S04. After cleaning, the drilling parameter data is segmented according to the time series, and the time domain, frequency domain and time-frequency domain features of each segment of drilling parameter data are extracted to form the feature vector matrix corresponding to each segment of data. S05. Using different combinations of feature vector matrices as input and rock properties as output, perform model training, and use feature importance analysis or ablation experiment to obtain sensitive data of rock properties. S06. Based on sensitive data of rock properties, construct an analysis model and use the analysis model to predict rock properties.

[0005] Furthermore, the rock drilling process simulation platform includes a rock sample clamp, a drilling motor module, and a data acquisition module; the rock sample clamp is used to fix the rock sample, the drilling motor module is used to control the drill bit to drill the rock sample, and the data acquisition module is used to collect drilling parameter data during the drilling process.

[0006] Furthermore, the data acquisition module includes a torque sensor, a speed sensor, a temperature sensor, an acceleration sensor, and a displacement sensor. The torque sensor is used to collect torque data during the drilling process, the speed sensor is used to collect drill bit speed data during the drilling process, the temperature sensor is used to collect temperature data during the drilling process, the acceleration sensor is used to collect vibration data during the drilling process, and the displacement sensor is used to collect drill bit displacement data during the drilling process, thereby obtaining the drill bit advance.

[0007] Furthermore, the drilling mode parameters include drill bit diameter and strength; the drilling parameters include temperature, vibration, drilling pressure, rotational speed, torque, and depth.

[0008] Furthermore, the data cleaning includes outlier handling, duplicate value handling, and normalization.

[0009] Furthermore, the rock properties include one of rock type, rock grade, and elastic modulus.

[0010] Furthermore, sensitive data on rock properties are obtained using eigenvalue importance analysis or ablation experiments. This includes using eigenvalue importance analysis or ablation experiments to obtain the importance scores of data on rock properties for different drilling parameters at different time periods, and selecting sensitive data on rock properties based on the importance scores.

[0011] The beneficial effects of this invention are as follows: This invention provides a method for constructing an analytical model to predict rock properties by simulating the rock drilling process. It obtains similarity parameters between drilling parameters and drilling mode parameters and the actual drilling process through similarity criteria. The drilling process is simulated using a rock drilling process simulation platform to obtain measured drilling parameter data. This measured data reflects the actual drilling process and can be used as the basis for analysis. The drilling parameter data is segmented according to time series, and features are extracted. Using different combinations of feature vector matrices as input and rock properties as output, the model is trained. Sensitive data on rock properties are obtained using feature importance analysis or ablation experiments. Based on this sensitive data, an analytical model is constructed, and the model is used to predict rock properties. This reduces the workload of data analysis, improves analytical efficiency and accuracy, and solves the problem of low accuracy in existing rock property prediction methods. Attached Figure Description

[0012] Figure 1 This is a flowchart illustrating a method for constructing an analytical model to predict rock properties by simulating the rock drilling process, as provided by the present invention. Detailed Implementation

[0013] This invention addresses the problem of low accuracy in existing rock property prediction methods by providing a method for constructing an analytical model based on simulating the rock drilling process to predict rock properties. Figure 1 As shown, it includes the following steps: S01. Obtain similarity parameters between drilling parameters and drilling mode parameters and the actual drilling process through similarity criteria.

[0014] Specifically, the similarity criterion is an implicit equation, which can be obtained using dimensional analysis or equation analysis based on similarity theory. The essence of the similarity criterion equation is the reflection of the similarity between the simulated rock drilling process and the actual rock drilling process. The similarity criterion equation is affected by the dimensionless combination of measured key parameters, as well as by the specific drill bit specifications and strength of the test platform.

[0015] The drilling mode parameters include drill bit diameter and strength; the drilling parameters include temperature, vibration, drilling pressure, rotational speed, torque, and depth.

[0016] S02. Set the drilling mode parameters, simulate the drilling process through the rock drilling process simulation platform, and obtain the measured drilling parameter data.

[0017] Specifically, drilling mode parameters are set to determine the drill bit diameter and strength during the simulation process. The measured drilling parameter data from the simulated drilling process can reflect the actual drilling process, thus serving as the basis for analysis.

[0018] The rock drilling process simulation platform includes a rock sample clamp, a drilling motor module, and a data acquisition module. The rock sample clamp is used to fix the rock sample, the drilling motor module is used to control the drill bit to drill into the rock sample, and the data acquisition module is used to collect drilling parameter data during the drilling process. The data acquisition module includes a torque sensor, a speed sensor, a temperature sensor, an acceleration sensor, and a displacement sensor. The torque sensor is used to collect torque data during the drilling process, the speed sensor is used to collect drill bit speed data during the drilling process, the temperature sensor is used to collect temperature data during the drilling process, the acceleration sensor is used to collect vibration data during the drilling process, and the displacement sensor is used to collect drill bit displacement data during the drilling process, thereby obtaining the drill bit footage.

[0019] S03. Perform data cleaning on the measured drilling parameter data to obtain the cleaned drilling parameter data.

[0020] Specifically, the data cleaning includes outlier handling, duplicate value handling, and normalization. Outlier handling can employ noise filtering or noise smoothing methods; duplicate value handling mainly uses statistical methods to detect duplicates based on the time series of the measured dataset; normalization mainly normalizes the data based on the value range of the measured dataset.

[0021] S04. After cleaning, the drilling parameter data is segmented according to the time series, and the time domain, frequency domain, and time-frequency domain features of each segment of drilling parameter data are extracted to form the feature vector matrix corresponding to each segment of data.

[0022] Specifically, since different drilling parameters have varying degrees of importance to rock characteristics at different times, the drilling parameter data is segmented according to the time series to refine the data and lay the foundation for subsequent searches for sensitive data on rock characteristics.

[0023] S05. Using different combinations of feature vector matrices as input and rock properties as output, train the model and obtain sensitive data of rock properties using feature importance analysis or ablation experiments.

[0024] Specifically, the rock properties include one of rock type, rock grade, and elastic modulus. Using different combinations of feature vector matrices—that is, data from different drilling parameters at different time periods—a suitable model is selected for training. Feature importance analysis or ablation testing methods are used to obtain the importance scores of different drilling parameters to rock properties at different time periods. Sensitive data for rock properties are then selected based on these importance scores.

[0025] S06. Based on sensitive data of rock properties, construct an analysis model and use the analysis model to predict rock properties.

[0026] Specifically, the analysis model built using sensitive data on rock properties can quickly and accurately analyze rock properties. Compared to data from all drilling parameters across all time periods, this reduces the workload of data analysis and improves the efficiency and accuracy of the analysis model.

Claims

1. A method for constructing an analytical model to predict rock properties by simulating the rock drilling process, characterized in that, Includes the following steps: S01. Determine the similarity parameters between drilling parameters and drilling mode parameters and the actual engineering rock drilling process by using similarity criteria; S02. Set the drilling mode parameters, simulate the drilling process through the rock drilling process simulation platform, and obtain the measured drilling parameter data. S03. Perform data cleaning on the measured drilling parameter data to obtain the cleaned drilling parameter data. S04. After cleaning, the drilling parameter data is segmented according to the time series, and the time domain, frequency domain and time-frequency domain features of each segment of drilling parameter data are extracted to form the feature vector matrix corresponding to each segment of data. S05. Using different combinations of feature vector matrices as input and rock properties as output, perform model training, and use feature importance analysis or ablation experiment to obtain sensitive data of rock properties. S06. Based on sensitive data of rock properties, construct an analysis model and use the analysis model to predict rock properties.

2. The method for constructing an analytical model to predict rock properties by simulating the rock drilling process according to claim 1, characterized in that, The rock drilling process simulation platform includes a rock sample clamp, a drilling motor module, and a data acquisition module; the rock sample clamp is used to fix the rock sample, the drilling motor module is used to control the drill bit to drill the rock sample, and the data acquisition module is used to collect drilling parameter data during the drilling process.

3. The method for constructing an analytical model to predict rock properties by simulating the rock drilling process according to claim 2, characterized in that, The data acquisition module includes a torque sensor, a speed sensor, a temperature sensor, an acceleration sensor, and a displacement sensor. The torque sensor is used to collect torque data during the drilling process, the speed sensor is used to collect drill bit speed data during the drilling process, the temperature sensor is used to collect temperature data during the drilling process, the acceleration sensor is used to collect vibration data during the drilling process, and the displacement sensor is used to collect drill bit displacement data during the drilling process, thereby obtaining the drill bit advance.

4. The method for constructing an analytical model to predict rock properties by simulating the rock drilling process according to claim 1, characterized in that, The drilling mode parameters include drill bit diameter and strength; the drilling parameters include temperature, vibration, drilling pressure, rotational speed, torque, and depth.

5. The method for constructing an analytical model to predict rock properties by simulating the rock drilling process according to claim 1, characterized in that, The data cleaning includes outlier handling, duplicate value handling, and normalization.

6. The method for constructing an analytical model to predict rock properties by simulating the rock drilling process according to claim 1, characterized in that, The rock properties include one of the following: rock type, rock grade, and elastic modulus.

7. The method for constructing an analytical model to predict rock properties by simulating the rock drilling process according to claim 1, characterized in that, Sensitive data on rock properties are obtained using eigenvalue importance analysis or ablation experiments. This includes using eigenvalue importance analysis or ablation experiments to obtain the importance scores of data on rock properties for different drilling parameters at different time periods. Sensitive data on rock properties are then selected based on these importance scores.

Images