Surrounding rock engineering characteristic prediction method, device and equipment for digital drilling and medium

Abstract

The invention provides a surrounding rock engineering characteristic prediction method and device for digital drilling, equipment and a medium, and the method comprises the steps: obtaining a drilling parameter of a target drilling section, and determining an auxiliary parameter based on the drilling parameter; the drilling parameters and the auxiliary parameters are input into a surrounding rock engineering characteristic prediction model, and a surrounding rock engineering characteristic result output by the surrounding rock engineering characteristic prediction model is obtained; the surrounding rock engineering prediction model comprises a plurality of prediction sub-models, and the prediction sub-models are obtained through machine learning training by taking respective parameter samples as inputs; the plurality of parameter samples includes an initial parameter sample and an enhanced parameter sample constructed based on the initial parameter sample. The method can better adapt to complex and changeable actual working conditions, so that the surrounding rock engineering characteristics are sensed more accurately and reliably, and timely decision support is provided for tunnel engineering construction.

Description

Technical Field

[0001] The present invention relates to the field of geological engineering technology, and in particular to a method, device, equipment and medium for predicting surrounding rock engineering characteristics through digital drilling. Background Art

[0002] Technologies related to the simultaneous sensing of tunnel surrounding rock engineering characteristics are used in fields such as geological engineering and geotechnical mechanics. Surrounding rock engineering characteristics refer to the various properties and states of surrounding rock during engineering activities. These characteristics are crucial for engineering design, construction, and safety assessment. For example, tunnel surrounding rock engineering characteristics include surrounding rock strength, integrity coefficient, and abrasiveness index.

[0003] Currently, the most commonly used evaluation method for surrounding rock engineering properties (surrounding rock strength, integrity coefficient, and abrasibility index) is traditional indoor rock mechanics testing, such as uniaxial compression testing and abrasibility testing. These tests involve collecting rock core samples from the field and subjecting them to loading tests in a laboratory environment to determine parameters such as rock strength and abrasibility. However, these tests cannot accurately simulate the stress state and environmental conditions of the in-situ rock mass, require significant material and financial resources, and lack real-time performance. Summary of the Invention

[0004] In view of the problems existing in the prior art, the present invention provides a method, device, equipment and medium for predicting surrounding rock engineering characteristics in digital drilling.

[0005] The present invention provides a method for predicting surrounding rock engineering characteristics by digital drilling, comprising: Acquiring drilling parameters of a target drilling section, and determining auxiliary parameters based on the drilling parameters; Inputting the drilling parameters and the auxiliary parameters into a surrounding rock engineering characteristic prediction model to obtain a surrounding rock engineering characteristic result output by the surrounding rock engineering characteristic prediction model; Among them, the surrounding rock engineering prediction model includes multiple prediction sub-models, and the prediction sub-models are all obtained through machine learning training based on their respective parameter samples as input; the multiple parameter samples include initial parameter samples and enhanced parameter samples constructed based on the initial parameter samples.

[0006] According to a method for predicting surrounding rock engineering characteristics of digital drilling provided by the present invention, the method includes inputting the drilling parameters and the auxiliary parameters into a surrounding rock engineering characteristics prediction model to obtain surrounding rock engineering characteristics results output by the surrounding rock engineering characteristics prediction model, including: The drilling parameters and the auxiliary parameters are processed by respective prediction sub-models to obtain a plurality of surrounding rock engineering characteristic results; Based on the weight information corresponding to each prediction sub-model and a plurality of surrounding rock engineering characteristic results, the surrounding rock engineering characteristic results corresponding to the drilling parameters are determined.

[0007] According to a method for predicting surrounding rock engineering characteristics of digital drilling provided by the present invention, the method further includes a step of acquiring a surrounding rock engineering prediction model, including: Acquire an initial parameter sample, wherein the initial parameter sample includes initial drilling parameters, initial auxiliary parameters, and initial surrounding rock engineering characteristic data; the initial auxiliary parameters are calculated based on the initial drilling parameters; Based on the initial parameter sample and a plurality of preset enhancement strategies, generating a plurality of enhancement parameter samples; Performing model training based on the initial parameter samples and the enhanced parameter samples to obtain multiple prediction sub-models; The surrounding rock engineering prediction model is obtained by fusing multiple prediction sub-models.

[0008] According to a method for predicting surrounding rock engineering characteristics of digital drilling provided by the present invention, the enhancement strategy is data translation. Accordingly, based on the initial parameter sample and multiple preset enhancement strategies, multiple enhanced parameter samples are generated, including: Determining the change amount corresponding to each data type in the initial drilling parameters and initial surrounding rock engineering characteristic data; generating corresponding enhanced drilling parameters and enhanced surrounding rock engineering characteristic data according to the initial drilling parameters, the initial surrounding rock engineering characteristic data and the variation; Calculating enhanced auxiliary parameters based on the enhanced drilling parameters; The enhanced drilling parameters, enhanced auxiliary parameters and enhanced surrounding rock engineering characteristic data are taken as enhanced parameter samples.

[0009] According to a method for predicting surrounding rock engineering characteristics of digital drilling provided by the present invention, the enhancement strategy is logarithmic exchange. Accordingly, based on the initial parameter sample and multiple preset enhancement strategies, multiple enhanced parameter samples are generated, including: Determine a logarithmic transformation formula corresponding to each data type in the initial drilling parameters and initial surrounding rock engineering characteristic data; generating corresponding enhanced drilling parameters and enhanced surrounding rock engineering characteristic data according to the initial drilling parameters and initial surrounding rock engineering characteristic data and the logarithmic transformation formula; Calculating enhanced auxiliary parameters based on the enhanced drilling parameters; The enhanced drilling parameters, enhanced auxiliary parameters and enhanced surrounding rock engineering characteristic data are taken as enhanced parameter samples.

[0010] According to a method for predicting surrounding rock engineering characteristics of digital drilling provided by the present invention, the enhancement strategy is noise enhancement. Accordingly, based on the initial parameter sample and multiple preset enhancement strategies, multiple enhancement parameter samples are generated, including: performing noise enhancement processing on the initial drilling parameters to obtain enhanced drilling parameters; Enhanced auxiliary parameters are calculated based on the enhanced drilling parameters; The enhanced drilling parameters, enhanced auxiliary parameters and initial surrounding rock engineering property data are taken as enhanced parameter samples.

[0011] According to a method for predicting surrounding rock engineering characteristics of digital drilling provided by the present invention, the auxiliary parameters are determined based on the drilling parameters, including: Statistical indicators of preset types are calculated according to various types of data values ​​in the drilling parameters of the target drilling section, and the statistical indicators of each preset type are used as auxiliary parameters.

[0012] The present invention also provides a device for predicting surrounding rock engineering characteristics for digital drilling, comprising: an acquisition module, configured to acquire drilling parameters of a target drilling section and determine auxiliary parameters based on the drilling parameters; a prediction module, configured to input the drilling parameters and the auxiliary parameters into a surrounding rock engineering characteristic prediction model to obtain a surrounding rock engineering characteristic result output by the surrounding rock engineering characteristic prediction model; Among them, the surrounding rock engineering prediction model includes multiple prediction sub-models, and the prediction sub-models are all obtained through machine learning training based on their respective parameter samples as input; the multiple parameter samples include initial parameter samples and enhanced parameter samples constructed based on the initial parameter samples.

[0013] The present invention also provides an electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, any of the above-mentioned methods for predicting surrounding rock engineering characteristics by digital drilling is implemented.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for predicting surrounding rock engineering characteristics of any of the above-mentioned digital drilling methods is implemented.

[0015] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned methods for predicting surrounding rock engineering characteristics by digital drilling.

[0016] The present invention provides a method, device, equipment and medium for predicting surrounding rock engineering characteristics for digital drilling. The method analyzes real-time drilling parameters and auxiliary parameters through a surrounding rock engineering characteristic model obtained by fusing models trained with multiple parameter samples, and predicts the surrounding rock engineering characteristic results under the current working conditions. The method can better adapt to complex and changeable actual working conditions, thereby more accurately and reliably perceiving surrounding rock engineering characteristics and providing timely decision support for tunnel engineering construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 It is a flow chart of the method for predicting surrounding rock engineering characteristics by digital drilling provided by the present invention.

[0019] Figure 2 It is a flow chart of drilling parameter preprocessing and storage management provided by the present invention.

[0020] Figure 3 It is a structural schematic diagram of the device for predicting surrounding rock engineering characteristics for digital drilling provided by the present invention.

[0021] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0022] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0023] The following combination Figure 1-Figure 4 The present invention describes the method, device, equipment and medium for predicting surrounding rock engineering characteristics by digital drilling.

[0024] Figure 1 A flow chart showing a method for predicting surrounding rock engineering characteristics by digital drilling provided by the present invention is shown in FIG. Figure 1 , the method comprises the following steps: Step 11: Obtain drilling parameters of the target drilling section, and determine auxiliary parameters based on the drilling parameters.

[0025] Step 12: Input the drilling parameters and auxiliary parameters into the surrounding rock engineering characteristics prediction model to obtain the surrounding rock engineering characteristics results output by the surrounding rock engineering characteristics prediction model, wherein the surrounding rock engineering characteristics prediction model includes multiple prediction sub-models, and the prediction sub-models are all obtained through machine learning training based on their respective parameter samples as input; the multiple parameter samples include initial parameter samples and enhanced parameter samples constructed based on the initial parameter samples.

[0026] Regarding steps 11 and 12, it should be noted that the technical field to which this invention belongs relates to technologies related to the simultaneous sensing of tunnel surrounding rock engineering characteristics in fields such as geological engineering and geotechnical mechanics. Surrounding rock engineering characteristics refer to the various properties and states of surrounding rock during engineering activities. These characteristics are of great significance to engineering design, construction, and safety assessment. For example, tunnel surrounding rock engineering characteristics include surrounding rock strength, integrity coefficient, and abrasiveness index.

[0027] In the present invention, drilling parameters during the drilling process can reflect the engineering properties of the surrounding rock. For example, the strength of the surrounding rock can be reflected in parameters such as the drilling speed and pressure of the drill pipe. To this end, the drilling parameters of the present invention include data such as drilling speed, torque, and pressure.

[0028] In the present invention, high-precision sensors, including speed sensors, torque sensors, and pressure sensors, are installed on the drilling equipment to collect various parameters during the drilling process in real time. The sensor data is transmitted via a high-speed data transmission line to a data processing terminal. The data processing terminal utilizes an advanced microprocessor and data processing software to process, analyze, and store the collected data in real time. To ensure high data quality, it should be further explained that preprocessing of the collected drilling parameters is required. This preprocessing can include data format unification, outlier processing, missing value processing, stable drilling process data extraction, and full drilling process data splicing.

[0029] A. Unified data format: Establish a standard format for data collection, including data units, sampling frequency, and data recording methods. For example, torque data for all drilling rigs is standardized in Newton-meters (N-m), with a sampling frequency set at 10 times per second. Data is recorded with a timestamp and corresponding measurement value to ensure consistent formatting across different drilling rigs.

[0030] B. Outlier processing: The collected data may contain outliers due to sensor failure, interference, etc. A reasonable threshold range can be set to identify and eliminate outliers. For example, when the torque data exceeds the normal operating range by ±30%, it is determined to be an outlier.

[0031] C. Missing value processing: For missing values ​​in the data, choose an appropriate handling method based on the data characteristics. If the number of missing values ​​is small, fill them with adjacent values. If the number of missing values ​​is large (e.g., more than 50% of the data collected from a single drill rod is missing), discard the drill rod data and mark the corresponding depth of the drill rod to facilitate subsequent data splicing during stable drilling.

[0032] D. Data extraction during stable drilling: The drilling process is divided into three phases: preparation, stable drilling, and lifting. Since the preparation and lifting phases do not involve rock breaking, state transition logic is used to automatically determine the stable drilling process. Specifically, the state change is obtained from the drilling rig's operating status. Based on this state change and the state transition logic, the drilling rig's phase transition information is determined. This determines the stable drilling phase, facilitating the collection of drilling parameters. For example, a geological drill rig's operating state undergoes significant changes between the start and end of drilling, whereas during normal drilling, the state remains relatively stable. By monitoring the drill rig's start and stop signals and other relevant state parameters, combined with changes in parameters such as current and speed during the drilling process, the stable drilling phase can be determined. For example, if a geological drill rig enters a stable speed and torque state after a period of acceleration after starting and maintaining this state for a period of time, it can be considered a stable drilling process. Based on this, the relevant state transition logic function is set to extract data related to the stable drilling process.

[0033] (1) (2) (3) in, P Drilling pressure is the axial force applied to the drill bit by the drill rig during drilling, which directly affects the drilling efficiency and rock crushing effect; I Drilling current is the current consumed by the electric drill when it is working. It reflects the load of the drill. This item is ignored when the hydraulic drive drill is used. T Drilling torque refers to the moment that makes the drill bit rotate and break the rock. It is one of the important indicators to measure the working capacity of the drilling rig. Different types of drilling rigs have different ways of expressing torque. N Drilling speed refers to the number of revolutions per minute of the drill bit. It works together with drilling pressure and torque to affect drilling efficiency and rock crushing effect. Get the values ​​of drilling pressure, drilling current, drilling torque and drilling speed.

[0034] At the same time, in order to avoid the problem of abnormal fluctuations in the parameter product results during the start-up and stop process of the drilling rig and when encountering sudden rock changes, making it difficult to accurately judge the stable drilling process. For each parameter, a sliding window technique is used to calculate the parameter mean over a period of time (such as 2 seconds, 5 seconds, 10 seconds, and 20 seconds) and input it into formulas (1), (2), and (3). A comprehensive judgment is made based on the parameter mean results of different windows instead of using the parameter value at a single moment. This can smooth the instantaneous fluctuations of the drilling parameters and reduce the impact of abnormal scenarios on the parameter product results.

[0035] F. Data splicing of the entire drilling process: First, ensure that each segment of stable drilling phase data, based on drill rods, has a unique and unambiguous identifier. This identifier includes information such as the drill rod number (e.g., drill rod 1, drill rod 2, etc.), the time range of data collection (start and end time), and the drilling rig operation project to which it belongs. Next, verify whether any drilling parameters within each segment contain missing or outliers. For problematic data segments, repair and address them using steps A through D. For discarded drill rod data with excessively high missingness rates, fill in their positions during subsequent splicing. Finally, chronologically sort all segments of stable drilling phase data to ensure the data is arranged in the order of the actual drilling process. For drilling displacement, set the initial displacement of the second drill rod to the same as the final displacement of the first drill rod, and continue this process to complete the splicing of drilling displacements. For the remaining drilling parameters, simply connect the beginning of the subsequent segment to the end of the previous segment. Finally, the continuity and consistency of the complete drilling process data obtained by splicing are checked to ensure that the changes in parameter values ​​at the splicing points of adjacent data segments are reasonable and there are no sudden changes.

[0036] After the above operations, the drilling displacement in the spliced ​​and verified complete drilling process data is combined with the corresponding time information in the data to calculate the displacement change rate in different time periods, thereby obtaining the drilling speed V (mm / s). All the acquired drilling data is then stored in a database or file system. During storage, the necessary metadata is added to the data, including the data acquisition time range, drilling rig type, operation location, and processing performed during the splicing process, to facilitate subsequent data query, analysis, and management.

[0037] See also Figure 2 The above-mentioned pre-processing steps and data storage and management are clearly presented in the form of a flow chart.

[0038] In this invention, since drilling parameters are process parameters of the drilling process, auxiliary parameters are determined from the drilling parameters of the drilling section, taking into account both point-in-time data collection and time-period data collection. These auxiliary parameters fully consider the interrelationships and synergy between various parameters, and can more comprehensively and accurately reflect the characteristics of the surrounding rock engineering. For example, auxiliary parameters can be derived parameters (such as mean, standard deviation, and peak value) of multiple parameters such as drilling speed, torque, and pressure.

[0039] In this invention, to improve the efficiency of prediction results, parameters such as drilling speed, torque, and pressure are collected in real time. By establishing a surrounding rock engineering property prediction model based on these parameters, and using advanced algorithms (such as neural network algorithms and multivariate regression analysis algorithms), the collected parameters are correlated with surrounding rock engineering properties (such as surrounding rock strength, integrity coefficient, and abrasiveness index). A mapping relationship between drilling parameters and surrounding rock engineering properties is established, and a surrounding rock engineering property prediction model is constructed. The model is then used to analyze the real-time collected drilling data and infer values ​​such as surrounding rock strength, integrity coefficient, and abrasiveness index.

[0040] In the present invention, in order to improve the accuracy of the prediction results, it is necessary to improve the performance and generalization ability of the surrounding rock engineering characteristics prediction model. To this end, it is necessary to obtain enhanced parameter samples based on the initial parameter samples of the training model, and use various forms of parameter samples to train different forms of models. This can not only significantly increase the amount of sample data, but also enable the model to learn the complex relationship between different parameter value combinations and surrounding rock engineering characteristics, thereby improving the model's adaptability and anti-interference ability to parameter changes. In other words, different parameter samples can adapt to different working conditions. For this reason, the trained surrounding rock engineering prediction model is formed by integrating multiple prediction sub-models, and each prediction sub-model is obtained through machine learning training based on its own parameter sample as input.

[0041] In the present invention, drilling parameters and auxiliary parameters are input into a surrounding rock engineering characteristic prediction model to obtain surrounding rock engineering characteristic results output by the surrounding rock engineering characteristic prediction model.

[0042] The method for predicting surrounding rock engineering characteristics of digital drilling provided by the present invention analyzes real-time drilling parameters and auxiliary parameters through the surrounding rock engineering characteristic model obtained by fusing models trained with multiple parameter samples, and predicts the surrounding rock engineering characteristic results under the current working conditions. It can better adapt to complex and changeable actual working conditions, thereby more accurately and reliably perceiving the surrounding rock engineering characteristics and providing timely decision support for tunnel engineering construction.

[0043] In a further method of the above method, the processing of inputting drilling parameters and auxiliary parameters into the surrounding rock engineering property prediction model to obtain the surrounding rock engineering property results output by the surrounding rock engineering property prediction model is mainly explained as follows: Each prediction sub-model processes the drilling parameters and auxiliary parameters respectively to obtain multiple surrounding rock engineering characteristic results.

[0044] Based on the weight information corresponding to each prediction sub-model and multiple surrounding rock engineering characteristic results, the surrounding rock engineering characteristic results corresponding to the drilling parameters are determined.

[0045] It's important to note that different parameter samples are suitable for different operating environments. Therefore, the various prediction sub-models have different priorities for prediction results in different operating environments. Therefore, an adaptive weighted averaging method is used to fuse multiple prediction sub-models. To fully leverage the strengths of each model, weights are assigned based on their performance on the test set during training. Models with superior performance are given higher weights, while models with relatively weaker performance are given lower weights.

[0046] The present invention enables the adaptive weighted average model to better adapt to complex and changeable actual working conditions, thereby perceiving surrounding rock engineering characteristics more accurately and reliably.

[0047] In the further method of the above method, the process of obtaining the surrounding rock engineering prediction model is mainly explained as follows: An initial parameter sample is obtained, which includes initial drilling parameters, initial auxiliary parameters, and initial surrounding rock engineering characteristic data; the initial auxiliary parameters are calculated based on the initial drilling parameters.

[0048] Based on the initial parameter sample and multiple preset enhancement strategies, multiple enhancement parameter samples are generated.

[0049] Model training is performed based on the initial parameter samples and the enhanced parameter samples to obtain multiple prediction sub-models.

[0050] The surrounding rock engineering prediction model is obtained by fusing multiple prediction sub-models.

[0051] To improve the performance and generalization of the rock mass engineering property prediction model, this paper employs a series of data augmentation strategies, including data translation, logarithmic transformation, and noise addition. Alternatively, data rotation and scaling can be used instead of translation, linear transformation can be used instead of logarithmic transformation, and impulse noise can be used instead of Gaussian noise. These data augmentation strategies significantly increase the amount of parameter sample data, allowing the model to learn the complex relationships between different combinations of parameter values ​​and rock mass engineering properties, thereby improving the model's adaptability to parameter changes and its ability to resist interference.

[0052] For further explanation, if the enhancement strategy is data shift, then based on the initial parameter sample and the preset multiple enhancement strategies, multiple enhancement parameter samples are generated, including: Determine the change amount corresponding to each data type in the initial drilling parameters and initial surrounding rock engineering characteristic data; Generate corresponding enhanced drilling parameters and enhanced surrounding rock engineering characteristic data according to initial drilling parameters and initial surrounding rock engineering characteristic data and changes; Enhanced auxiliary parameters are calculated based on enhanced drilling parameters; The enhanced drilling parameters, enhanced auxiliary parameters and enhanced surrounding rock engineering characteristic data are taken as enhanced parameter samples.

[0053] It's important to note that the initial drilling parameters and initial surrounding rock engineering property data are previously collected sample data. These data can be obtained through drilling tests. Generally speaking, for this engineering field, the collected data tends to be small sample data from a model training perspective. Therefore, data augmentation is required to enhance the data's adaptability to various working conditions and increase the sample data volume.

[0054] In actual drilling operations, drilling data is often collected based on time. However, surrounding rock engineering parameters, such as strength, integrity coefficient, and abrasiveness index, are measured based on spatial distance. This difference in data collection methods creates a mismatch between the temporal and spatial dimensions. To eliminate this mismatch and better meet actual engineering needs, drilling data must be converted into a format that is compatible with space. Especially when surrounding rock characteristics change frequently or when the project requires high precision, drilling data must not only be formatted correctly but also reflect the dynamic changes in surrounding rock properties in greater detail, providing a more accurate and reliable basis for subsequent engineering design and construction.

[0055] Based on this, for the drilling data obtained at a certain spatial distance, it is necessary to separately perform statistics on each drilling parameter. Specifically, according to the various types of data values ​​in the drilling parameters of the target drilling section, a preset type of statistical index is calculated, and each preset type of statistical index is used as an auxiliary parameter. For example, for each 1 meter (or 2 meters, 5 meters, 8 meters) of drilling data segment, the mean, standard deviation and other statistical indicators of each drilling parameter are calculated as auxiliary parameters. The auxiliary parameters can fully consider the mutual relationship and collaborative operation between the various parameters. P For example, suppose that within this 1-meter data segment there is n pressure measurements, in order P 1, P 2,…, P n , then the mean and standard deviation of the drilling pressure in this section are: Other statistical indicators, such as the median and coefficient of variation, can also be calculated as needed. The median reflects the middle level of the data and is unaffected by extreme values. The coefficient of variation, the ratio of the standard deviation to the mean, is used to compare the degree of dispersion between different parameters. Thus, the drilling data arranged in time has been converted into drilling data within a 1-meter interval.

[0056] For data translation, drilling parameters and surrounding rock engineering characteristics are integrated into vector or matrix form. For example, multiple drilling parameters are arranged into vectors in a specific order. Based on this, these vectors or matrices can be translated.

[0057] Taking the translation operation as an example, the translation of the drilling parameter vector is achieved by adding a fixed value (i.e., the variation) to each dimension of the vector. When there is a vector consisting of three parameters: drilling pressure, torque, and speed [ V , N , P ], and after performing translation operation, we get [ V+ △ V , N+ △ N , P+ △ P ], where △ V ,△ N ,△ P To ensure the rationality of the translation, the translation is selected as 0.5 times the standard deviation of each drilling parameter. At the same time, to avoid generating data that exceeds practical significance or is unreasonable, the translation is limited to 5% of the theoretical maximum and minimum boundaries of each parameter.

[0058] Then, enhanced auxiliary parameters are calculated based on the enhanced drilling parameters.

[0059] Finally, the enhanced drilling parameters and auxiliary parameters after translation processing are combined with the enhanced surrounding rock engineering characteristic data to form a new training sample. In this way, the corresponding surrounding rock engineering characteristic parameter prediction model is obtained with the new training sample.

[0060] If the enhancement strategy is logarithmic exchange, accordingly, based on the initial parameter sample and the preset multiple enhancement strategies, multiple enhancement parameter samples are generated, including: Determine the logarithmic transformation formula corresponding to each data type in the initial drilling parameters and initial surrounding rock engineering characteristic data; Generate corresponding enhanced drilling parameters and enhanced surrounding rock engineering characteristic data according to the initial drilling parameters, initial surrounding rock engineering characteristic data and logarithmic transformation formula; Enhanced auxiliary parameters are calculated based on enhanced drilling parameters; The enhanced drilling parameters, enhanced auxiliary parameters and enhanced surrounding rock engineering characteristic data are taken as enhanced parameter samples.

[0061] Logarithmic transformation is an effective method for processing drilling parameters or surrounding rock engineering characteristic data when there are large differences in numerical ranges or when the distribution is skewed. In practical applications, depending on the specific conditions of the data and research needs, logarithmic transformation can be performed only on drilling parameters, only on surrounding rock engineering characteristic parameters, or both.

[0062] The drilling parameters and surrounding rock engineering characteristic indicators are logarithmically transformed, specifically through the formula y = log( x ) implementation, where x represents the original data, y The transformed data is shown in Figure 2. Taking drilling torque data as an example, during the actual drilling process, torque values ​​can fluctuate widely, with some values ​​being extremely large. Directly using this raw data for analysis can mask the influence of other parameters, resulting in poor model training results. By performing a logarithmic transformation on the drilling torque data, large values ​​can be compressed, allowing for better integration with other parameters for analysis.

[0063] Then, enhanced auxiliary parameters are calculated based on the enhanced drilling parameters.

[0064] Finally, the enhanced drilling parameters and auxiliary parameters after translation processing are combined with the enhanced surrounding rock engineering characteristic data to form a new training sample. In this way, the corresponding surrounding rock engineering characteristic parameter prediction model is obtained with the new training sample.

[0065] If the enhancement strategy is noise enhancement, accordingly, based on the initial parameter sample and the preset multiple enhancement strategies, multiple enhancement parameter samples are generated, including: Performing noise enhancement processing on the initial drilling parameters to obtain enhanced drilling parameters; Enhanced auxiliary parameters are calculated based on the enhanced drilling parameters; The enhanced drilling parameters, enhanced auxiliary parameters and initial surrounding rock engineering property data are taken as enhanced parameter samples.

[0066] Regarding noise enhancement, in actual drilling operations, drilling parameters can vary due to a variety of factors, such as fluctuations in equipment performance and operator habits. These fluctuations often exhibit a degree of randomness, and in most cases, their probability distribution approximates a normal (Gaussian) distribution.

[0067] Therefore, we add Gaussian noise to the drilling parameter data for data enhancement. When adding noise, we set the appropriate mean and variance according to the characteristics of the data and the expected level of noise. For example, for the drilling pressure data, assuming its original value is P. You can add a mean of 0 and a variance of σ 2 Gaussian noise N (0, σ 2 ). Get new data, P ′= P + N (0, σ 2 ), by adjusting the variance σ 2 The size of can control the intensity of the noise. σ 2 Solve the problem by using the standard deviation of historical drilling parameter data, such as P The standard deviation of is 0.3, then the variance σ 2 =0.09.

[0068] Then, enhanced auxiliary parameters are calculated based on the enhanced drilling parameters.

[0069] Finally, the processed enhanced drilling parameters and enhanced auxiliary parameters are combined with the initial surrounding rock engineering characteristic data to form a new training sample. In this way, the corresponding surrounding rock engineering characteristic parameter prediction model is obtained with the new training sample.

[0070] In the present invention, the model training includes the following steps: 1) Determine model input and output: Calculated auxiliary parameters of drilling parameters (mean, standard deviation, peak value, etc.) serve as model input variables. Output variables are set to the acquired surrounding rock engineering property data: surrounding rock strength: Uniaxial compressive strength (UCS) corresponding to the number of meters drilled is determined on-site using a point load test; abrasiveness index (CAI) is calculated using an abrasiveness test device, obtaining the mean value of steel needle wear measurements; and rock integrity factor (RQD) is assessed by calculating the core recovery rate during the on-site drilling process.

[0071] 2) Dataset Partitioning: Divide the existing samples into a training set and a test set. The training set is used to train the model, while the test set is used to adjust model parameters and select the optimal model. The ratio of the training set to the test set is 9:1.

[0072] 3) Model training: The engineering characteristic parameters of surrounding rock (UCS, CAI, and RQD) vary significantly in both data dimensions and mechanical meaning. Therefore, a multi-parameter fusion analysis model for UCS, CAI, and RQD was independently established.

[0073] For small data volumes, complex machine learning models are prone to overfitting, so relatively simple models such as linear regression, decision trees, integrated models (such as random forests, AdaBoost, etc.) and other machine learning algorithms are selected to train the data set.

[0074] The structural parameters of the machine learning algorithm have a decisive influence on the prediction results. Therefore, Bayesian optimization is used to find the optimal structural parameters of the machine learning model for surrounding rock engineering characteristics. The training goal is to accurately predict the surrounding rock strength, integrity coefficient, and abrasive index. Therefore, the mean absolute error (MAE) of the prediction model on the test set is used. MAE ) as the objective function. Then, the model structural parameters to be optimized and their value ranges are determined to form a parameter space. Bayesian optimization then uses a Gaussian process as a proxy model. The Gaussian process constructs the probability distribution of the objective function based on existing sample points (parameter combinations and their corresponding objective function values), thereby understanding the possible value range and uncertainty of the objective function at different parameter points. Then, a suitable acquisition function is selected to guide the selection of parameter points. Then, iterative optimization is performed. When the stopping condition is met, the parameter combination that optimizes the objective function value is selected from all sampled parameter points as the final result. This parameter combination is the optimal structural parameters of the machine learning model of surrounding rock engineering characteristics found by Bayesian optimization. The optimal structural parameters of these machine learning models are then used for model training.

[0075] 4) Model evaluation: After the machine learning model training is completed, the test set data is input into the trained model to obtain the model's prediction results. Since the prediction effect of a single point is more important for the surrounding rock engineering characteristics, the mean absolute error ( MAE ) Evaluate the prediction performance of different machine learning algorithms.

[0076] in, n is the sample size, y i is the measured value, is the predicted value. MAE Indicators are used to screen out the machine learning algorithm with the best prediction effect and obtain the prediction model of surrounding rock engineering characteristic parameters.

[0077] After a series of steps, multiple models related to surrounding rock engineering characteristics were successfully constructed. For example, four models were obtained, namely surrounding rock engineering characteristics prediction model 1, model 2, model 3 and model 4. Given that different models have different prediction performances when facing diverse geological conditions, differences in drilling equipment performance and changes in operating parameters, in order to achieve more accurate predictions of surrounding rock engineering characteristics, it is necessary to assign corresponding weights to each model based on its actual performance on the test set. Models with superior performance will be assigned higher weights, while models with relatively poor performance will be assigned lower weights. The determination of weights can be based on model evaluation indicators, such as mean square error, accuracy, etc. Through this strategy of dynamically adjusting weights, it is possible to effectively deal with the inconsistency of model prediction results under different working conditions, thereby significantly improving the accuracy and reliability of the prediction.

[0078] Abrasiveness index of surrounding rock CAI Taking the prediction of as an example, the adaptive weighted average calculation method is adopted, and the calculation formula is as follows: Where, It is the confidence function of the prediction model of surrounding rock engineering characteristic parameters.

[0079] The confidence function assigns a reasonable weight to each model based on multiple factors such as the reliability, accuracy and uncertainty of the surrounding rock engineering characteristic parameter prediction model, so as to accurately reflect the importance of each model in the prediction process. The higher the confidence of the model, the higher the credibility of its prediction results, and the greater the contribution of the surrounding rock engineering characteristic parameter prediction model to the final prediction results; conversely, the lower the confidence of the model, the smaller its contribution to the prediction results. Mainly based on MAE The inverse of determines the confidence level of each rock mass engineering characteristic parameter prediction model, which serves as an important basis for weight allocation. confidence (Model-1)=2.07; Model 2 confidence (Model-2)=4.02; Model 3 confidence (Model-3)=7.27; Model 4 confidence (Model-4)=4.52. Finally, CAI The calculation formula of adaptive weighted average is: Where, CAI Model-1 、 CAI Model-2 、 CAI Model-3 、 CAI Model-4 Represents the prediction model of surrounding rock engineering characteristics Model -1. Model -2. Model -3. Model -4 confirmed dolomites CAI value.

[0080] In order to improve the performance and generalization ability of the rock mass engineering property prediction model, a series of data enhancement training methods are used: data translation, logarithmic transformation, and noise addition. Through these data enhancement training methods, the amount of small sample data is significantly increased, allowing the model to learn the complex relationship between different parameter value combinations and rock mass engineering properties, thereby improving the model's adaptability to parameter changes and anti-interference ability. The method for predicting surrounding rock engineering characteristics of digital drilling provided by the present invention analyzes real-time drilling parameters and auxiliary parameters through the surrounding rock engineering characteristic model obtained by fusing models trained with multiple parameter samples, and predicts the surrounding rock engineering characteristic results under the current working conditions. It can better adapt to complex and changeable actual working conditions, thereby more accurately and reliably perceiving the surrounding rock engineering characteristics and providing timely decision support for tunnel engineering construction.

[0081] The digital drilling surrounding rock engineering characteristic prediction device provided by the present invention is described below. The digital drilling surrounding rock engineering characteristic prediction device described below and the digital drilling surrounding rock engineering characteristic prediction method described above can be referenced to each other.

[0082] Figure 3 The present invention provides a digital drilling surrounding rock engineering characteristics prediction device structure diagram, see Figure 3 , the device includes an acquisition module 31 and a prediction module 32, wherein: An acquisition module, configured to acquire drilling parameters of a target drilling section and determine auxiliary parameters based on the drilling parameters; A prediction module is used to input drilling parameters and auxiliary parameters into the surrounding rock engineering characteristics prediction model to obtain the surrounding rock engineering characteristics results output by the surrounding rock engineering characteristics prediction model; Among them, the surrounding rock engineering prediction model includes multiple prediction sub-models, and the prediction sub-models are all obtained through machine learning training based on their respective parameter samples as input; the multiple parameter samples include initial parameter samples and enhanced parameter samples constructed based on the initial parameter samples.

[0083] Since the principles of the apparatus of the embodiment of the present invention are the same as those of the method of the above embodiment, more detailed explanations are omitted here.

[0084] It should be noted that, in the embodiment of the present invention, relevant functional modules may be implemented by a hardware processor.

[0085] The digital drilling surrounding rock engineering characteristics prediction device provided by the present invention analyzes real-time drilling parameters and auxiliary parameters through the surrounding rock engineering characteristics model obtained by fusing models trained with multiple parameter samples, and predicts the surrounding rock engineering characteristics results under the current working conditions. It can better adapt to complex and changeable actual working conditions, thereby more accurately and reliably perceiving the surrounding rock engineering characteristics and providing timely decision support for tunnel engineering construction.

[0086] Figure 4 An example of a physical structure diagram of an electronic device is shown below. Figure 4 As shown, the electronic device may include: a processor 41 (processor), a communication interface 42 (Communications Interface), a memory 43 (memory), and a communication bus 44, wherein the processor 41, the communication interface 42, and the memory 43 communicate with each other via the communication bus 44. The processor 41 may call the logic instructions in the memory 43 to execute a method for predicting surrounding rock engineering characteristics for digital drilling, which includes: obtaining drilling parameters of a target drilling section, determining auxiliary parameters based on the drilling parameters; inputting the drilling parameters and the auxiliary parameters into a surrounding rock engineering characteristics prediction model, and obtaining surrounding rock engineering characteristics results output by the surrounding rock engineering characteristics prediction model, wherein the surrounding rock engineering prediction model includes multiple prediction sub-models, each of which is obtained through machine learning training using its own parameter samples as input; the multiple parameter samples include initial parameter samples and enhanced parameter samples constructed based on the initial parameter samples.

[0087] Furthermore, the logic instructions in the aforementioned memory 43 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, or the portion that contributes to the prior art, or a portion 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 for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0088] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the digital drilling surrounding rock engineering characteristics prediction method provided by the above-mentioned methods, the method including: obtaining the drilling parameters of the target drilling section, and determining auxiliary parameters based on the drilling parameters; inputting the drilling parameters and auxiliary parameters into the surrounding rock engineering characteristics prediction model to obtain the surrounding rock engineering characteristics results output by the surrounding rock engineering characteristics prediction model, wherein the surrounding rock engineering prediction model includes multiple prediction sub-models, and the prediction sub-models are all obtained through machine learning training based on their respective parameter samples as input; the multiple parameter samples include initial parameter samples and enhanced parameter samples constructed based on the initial parameter samples.

[0089] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the digital drilling surrounding rock engineering characteristics prediction method provided by the above-mentioned methods, the method comprising: obtaining drilling parameters of the target drilling section, and determining auxiliary parameters based on the drilling parameters; inputting the drilling parameters and the auxiliary parameters into a surrounding rock engineering characteristics prediction model, and obtaining surrounding rock engineering characteristics results output by the surrounding rock engineering characteristics prediction model, wherein the surrounding rock engineering prediction model includes multiple prediction sub-models, and the prediction sub-models are all obtained through machine learning training based on their respective parameter samples as input; the multiple parameter samples include initial parameter samples and enhanced parameter samples constructed based on the initial parameter samples.

[0090] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0091] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion 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, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for predicting surrounding rock engineering characteristics by digital drilling, characterized in that: include: Acquiring drilling parameters of a target drilling section, and determining auxiliary parameters based on the drilling parameters; Inputting the drilling parameters and the auxiliary parameters into a surrounding rock engineering characteristic prediction model to obtain a surrounding rock engineering characteristic result output by the surrounding rock engineering characteristic prediction model; The surrounding rock engineering prediction model includes multiple prediction sub-models, each of which is obtained through machine learning training based on its own parameter samples as input; The multiple parameter samples include initial parameter samples and enhanced parameter samples constructed based on the initial parameter samples.

2. The method for predicting surrounding rock engineering characteristics by digital drilling according to claim 1, characterized in that: Inputting the drilling parameters and the auxiliary parameters into a surrounding rock engineering characteristic prediction model to obtain surrounding rock engineering characteristic results output by the surrounding rock engineering characteristic prediction model includes: The drilling parameters and the auxiliary parameters are processed by respective prediction sub-models to obtain a plurality of surrounding rock engineering characteristic results; Based on the weight information corresponding to each prediction sub-model and a plurality of surrounding rock engineering characteristic results, the surrounding rock engineering characteristic results corresponding to the drilling parameters are determined.

3. The method for predicting surrounding rock engineering characteristics by digital drilling according to claim 1, characterized in that: The method further includes a step of obtaining a surrounding rock engineering prediction model, including: Acquire an initial parameter sample, wherein the initial parameter sample includes initial drilling parameters, initial auxiliary parameters, and initial surrounding rock engineering characteristic data; the initial auxiliary parameters are calculated based on the initial drilling parameters; Based on the initial parameter sample and a plurality of preset enhancement strategies, generating a plurality of enhancement parameter samples; Performing model training based on the initial parameter samples and the enhanced parameter samples to obtain multiple prediction sub-models; The surrounding rock engineering prediction model is obtained by fusing multiple prediction sub-models.

4. The method for predicting surrounding rock engineering characteristics by digital drilling according to claim 3, characterized in that: The enhancement strategy is data translation. Accordingly, based on the initial parameter sample and multiple preset enhancement strategies, multiple enhancement parameter samples are generated, including: Determining the change amount corresponding to each data type in the initial drilling parameters and initial surrounding rock engineering characteristic data; generating corresponding enhanced drilling parameters and enhanced surrounding rock engineering characteristic data according to the initial drilling parameters, the initial surrounding rock engineering characteristic data and the variation; Calculating enhanced auxiliary parameters based on the enhanced drilling parameters; The enhanced drilling parameters, enhanced auxiliary parameters and enhanced surrounding rock engineering characteristic data are taken as enhanced parameter samples.

5. The method for predicting surrounding rock engineering characteristics by digital drilling according to claim 3, characterized in that: The enhancement strategy is logarithmic exchange. Accordingly, based on the initial parameter sample and multiple preset enhancement strategies, multiple enhancement parameter samples are generated, including: Determine a logarithmic transformation formula corresponding to each data type in the initial drilling parameters and initial surrounding rock engineering characteristic data; generating corresponding enhanced drilling parameters and enhanced surrounding rock engineering characteristic data according to the initial drilling parameters and initial surrounding rock engineering characteristic data and the logarithmic transformation formula; Calculating enhanced auxiliary parameters based on the enhanced drilling parameters; The enhanced drilling parameters, enhanced auxiliary parameters and enhanced surrounding rock engineering characteristic data are taken as enhanced parameter samples.

6. The method for predicting surrounding rock engineering characteristics by digital drilling according to claim 3, characterized in that: The enhancement strategy is noise enhancement. Accordingly, based on the initial parameter sample and multiple preset enhancement strategies, multiple enhancement parameter samples are generated, including: performing noise enhancement processing on the initial drilling parameters to obtain enhanced drilling parameters; Enhanced auxiliary parameters are calculated based on the enhanced drilling parameters; The enhanced drilling parameters, enhanced auxiliary parameters and initial surrounding rock engineering property data are taken as enhanced parameter samples.

7. The method for predicting surrounding rock engineering characteristics by digital drilling according to claim 1, characterized in that: The determining of auxiliary parameters based on the drilling parameters includes: Statistical indicators of preset types are calculated according to various types of data values ​​in the drilling parameters of the target drilling section, and the statistical indicators of each preset type are used as auxiliary parameters.

8. A device for predicting surrounding rock engineering characteristics for digital drilling, characterized in that: include: an acquisition module, configured to acquire drilling parameters of a target drilling section and determine auxiliary parameters based on the drilling parameters; a prediction module, configured to input the drilling parameters and the auxiliary parameters into a surrounding rock engineering characteristic prediction model to obtain a surrounding rock engineering characteristic result output by the surrounding rock engineering characteristic prediction model; The surrounding rock engineering prediction model includes multiple prediction sub-models, each of which is obtained through machine learning training based on its own parameter samples as input; The multiple parameter samples include initial parameter samples and enhanced parameter samples constructed based on the initial parameter samples.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method for predicting surrounding rock engineering characteristics by digital drilling as described in any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for predicting surrounding rock engineering characteristics by digital drilling as claimed in any one of claims 1 to 7 is implemented.

Images