Roadway surrounding rock cuttable intelligent evaluation method and device

Abstract

The invention relates to the technical field of rock tunneling, and discloses a method and a device for intelligently evaluating the cuttable property of roadway surrounding rock. The method comprises the following steps: acquiring initial drilling main control parameter data of roadway surrounding rock, preprocessing the initial drilling main control parameter data, and dividing the preprocessed drilling main control parameter data into a training data set and a test data set; constructing a multi-source information fusion prediction model according to the training data set, and verifying the theoretical solution of the rock mass strength to obtain a predicted solution of the rock mass strength; constructing a rock mass parameter model according to the predicted solution of the rock mass strength, and predicting the roadway surrounding rock to obtain predicted values of rock mass parameters; and constructing a rock mass cuttability prediction model according to the predicted values of the rock mass parameters, and predicting the test data set to obtain a rock mass cuttability grade value prediction result of the roadway surrounding rock. According to the method, rapid real-time grading of the cuttability of the rock mass in the drilling test area can be achieved, engineering construction is guided, the construction efficiency is improved, and the construction cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of rock mass tunneling, and particularly to an intelligent evaluation method and device for the cuttability of roadway surrounding rock. Background Art

[0002] At present, the technical process of measuring the cuttability of rock is to first conduct laboratory standard tests and in-situ tests on rock strength, and then judge the excavability of rock under mechanical mining conditions according to rock strength and the principle of cuttability. However, laboratory standard tests need to be completed by professional technicians in a professional laboratory. These tests not only have high requirements for rock samples, but also have a long time, high cost, and high labor intensity from the complete processes of sampling, transportation, rock sample preparation, and testing, and are destructive tests. Therefore, it is not practical for work scenarios that require timely knowledge of the rock strength in the construction area and rapid assessment and decision-making.

[0003] At present, there are problems in the excavation of mine rock tunnels, such as low efficiency, unreasonable cuttability evaluation methods, high manual labor intensity, and low equipment intelligence level. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to overcome the deficiencies in the prior art and provide an intelligent evaluation method and device for the cuttability of roadway surrounding rock.

[0005] The present invention provides the following technical solutions:

[0006] In a first aspect, the present invention provides an intelligent evaluation method for the cuttability of roadway surrounding rock, and the method includes:

[0007] Obtain the initial drilling main control parameter data of the roadway surrounding rock, preprocess the initial drilling main control parameter data to obtain the preprocessed drilling main control parameter data, and divide the preprocessed drilling main control parameter data into a training data set and a test data set;

[0008] Construct a multi-source information fusion prediction model according to the training data set, and use the multi-source information fusion prediction model to verify the theoretical solution of rock mass strength to obtain the predicted solution of rock mass strength;

[0009] Construct a rock mass parameter model according to the predicted solution of rock mass strength, and use the rock mass parameter model to predict the roadway surrounding rock to obtain the predicted value of rock mass parameters;

[0010] Construct a rock mass cuttability prediction model according to the predicted value of rock mass parameters, and use the rock mass cuttability prediction model to predict the test data set to obtain the predicted result of the rock mass cuttability level value of the roadway surrounding rock.

[0011] In an alternative embodiment, the initial drilling main control parameter data includes at least one of rotational speed, torque, thrust, drilling speed, noise, and vibration. The rock mass strength includes rock mass cutting strength and rock mass intrusion strength. The rock mass parameter model includes a rock mass mechanical parameter model and a rock mass structure parameter model. The rock mass parameters include rock mass mechanical parameters and rock mass structure parameters. The rock mass mechanical parameters include at least one of strength, elastic modulus, cohesion, and internal friction angle. The rock mass structure parameters include at least one of bedding, interface, angle, and thickness.

[0012] In an alternative embodiment, the preprocessing of the initial drilling main control parameter data to obtain the preprocessed drilling main control parameter data includes:

[0013] Performing wavelet denoising on the initial drilling main control parameter data to obtain the denoised drilling main control parameter data;

[0014] Performing outlier removal processing on the denoised drilling main control parameter data to obtain the drilling main control parameter data after removal;

[0015] Performing normalization processing on the drilling main control parameter data after removal to obtain the preprocessed drilling main control parameter data.

[0016] In an alternative embodiment, the performing wavelet denoising on the initial drilling main control parameter data to obtain the denoised drilling main control parameter data includes:

[0017] Performing wavelet decomposition on the initial drilling main control parameter data to obtain high-frequency components and low-frequency components. Removing the noise in the high-frequency components according to a preset noise threshold, and fusing the removed high-frequency components and the low-frequency components to obtain a denoised signal;

[0018] Performing local statistics on the denoised signal using a sliding window, calculating the noise power spectral density, dynamically adjusting the parameters of the filter according to the noise power spectral density, and using the adjusted filter to smooth the low-frequency components in the denoised signal. Performing spectral analysis on the smoothed denoised signal, identifying the interference frequency signal components, and using a band-stop filter to remove the interference frequency signal components to obtain a filtered signal;

[0019] Fusing the denoised signal and the filtered signal to obtain the denoised drilling main control parameter data.

[0020] In an alternative embodiment, the performing outlier removal processing on the denoised drilling main control parameter data to obtain the drilling main control parameter data after removal includes:

[0021] Calculate the mean, standard deviation, and quartiles of each parameter data in the noise-reduced drilling main control parameter data, set an initial threshold range according to the mean, the standard deviation, and the quartiles, and use the sliding window technique to adjust the initial threshold range in real time to obtain a single-parameter dynamic threshold range;

[0022] Obtain historical drilling main control parameter data, construct a linear regression model and a non-linear model according to the historical drilling main control parameter data, and use the linear regression model and the non-linear model to predict each parameter data in the noise-reduced drilling main control parameter data to obtain predicted values, obtain the observed values of each parameter data in the noise-reduced drilling main control parameter data, calculate the residual values of each parameter data in the noise-reduced drilling main control parameter data according to the observed values and the predicted values, and set a multi-parameter residual range;

[0023] If any parameter data in the noise-reduced drilling main control parameter data is outside the single-parameter dynamic threshold range, and the residual value of any parameter data in the noise-reduced drilling main control parameter data is outside the multi-parameter residual range, then determine the corresponding parameter data as abnormal data;

[0024] Remove the abnormal data, and use the linear regression model and the non-linear model to predict reasonable values of the abnormal data for substitution to obtain the drilling main control parameter data after removal.

[0025] In an alternative embodiment, the normalizing the drilling main control parameter data after removal to obtain the preprocessed drilling main control parameter data includes:

[0026] Set a dynamic range for the mean, standard deviation, and quartiles of each parameter data in the noise-reduced drilling main control parameter data;

[0027] Normalize each parameter data in the drilling main control parameter data after removal according to the dynamic range to obtain the normalized drilling main control parameter data, obtain the gradient difference of the contribution of each parameter data in the normalized drilling main control parameter data to a preset target value, and calculate the gradient balance coefficient of each parameter data in the normalized drilling main control parameter data according to the gradient difference, and multiply each parameter data in the normalized drilling main control parameter data by its corresponding gradient balance coefficient to obtain the preprocessed drilling main control parameter data.

[0028] In an alternative embodiment, the constructing a multi-source information fusion prediction model according to the training data set and using the multi-source information fusion prediction model to verify the theoretical solution of the rock mass strength to obtain the predicted solution of the rock mass strength includes:

[0029] Input the training data set into the computer algorithm library, perform multi-source information cross-training using a preset multi-source information cross-training algorithm to obtain a training result, and construct the multi-source information fusion prediction model according to the training result, where the multi-source information cross-training algorithm includes BP network, SVM, BP-NN, K-NN, C-NN, and Catboost;

[0030] Use the multi-source information fusion prediction model to verify the theoretical solutions of the rock cutting strength and the rock intrusion strength of the rock mass, and obtain the predicted solutions of the rock cutting strength and the rock intrusion strength of the rock mass.

[0031] In an alternative embodiment, constructing a rock mass cuttability prediction model according to the predicted values of the rock mass parameters, and using the rock mass cuttability prediction model to predict the test data set to obtain the predicted result of the rock mass cuttability grade value of the roadway surrounding rock, including:

[0032] Define a loss function according to the theoretical solution and the predicted solution of the rock mass strength, and optimize the predicted values of the rock mass parameters and their contribution weights according to the loss function to obtain the optimized predicted values of the rock mass parameters and their optimized weight coefficients;

[0033] Calculate the comprehensive cuttability index of the roadway surrounding rock according to the optimized predicted values of the rock mass parameters and their optimized weight coefficients;

[0034] Use a preset grade division rule and according to the value range of the comprehensive cuttability index of the roadway surrounding rock, determine the predicted result of the rock mass cuttability grade value of the roadway surrounding rock.

[0035] In an alternative embodiment, the method further includes:

[0036] Evaluate the predicted result of the rock mass cuttability grade value of the roadway surrounding rock using a preset evaluation index, where the preset evaluation index includes at least one of mean absolute percentage error, balanced accuracy rate, and F1-Score;

[0037] Judge whether the predicted result of the rock mass cuttability grade value meets the preset conditions. If so, directly output the predicted result of the rock mass cuttability grade value of the roadway surrounding rock. If not, retrain and evaluate each model until the predicted result of the rock mass cuttability grade value of the rock mass cuttability prediction model on the test data set meets the preset conditions.

[0038] In a second aspect, the present invention provides an intelligent evaluation device for the cuttability of a roadway surrounding rock, and the device includes:

[0039] A processing module, configured to obtain initial drilling main control parameter data of roadway surrounding rock, preprocess the initial drilling main control parameter data to obtain preprocessed drilling main control parameter data, and divide the preprocessed drilling main control parameter data into a training data set and a test data set;

[0040] A first construction module, configured to construct a multi-source information fusion prediction model according to the training data set, verify the theoretical solution of rock mass strength by using the multi-source information fusion prediction model, and obtain a predicted solution of rock mass strength;

[0041] A second construction module, configured to construct a rock mass parameter model according to the predicted solution of rock mass strength, predict the roadway surrounding rock by using the rock mass parameter model, and obtain a predicted value of rock mass parameters;

[0042] A prediction module, configured to construct a rock mass cuttability prediction model according to the predicted value of rock mass parameters, predict the test data set by using the rock mass cuttability prediction model, and obtain a predicted result of rock mass cuttability of the roadway surrounding rock.

[0043] Advantages of the present application:

[0044] The intelligent evaluation method for the cuttability of roadway surrounding rock provided by the embodiment of the present application obtains the initial drilling main control parameter data of the roadway surrounding rock, preprocesses the initial drilling main control parameter data to obtain preprocessed drilling main control parameter data, and divides the preprocessed drilling main control parameter data into a training data set and a test data set; constructs a multi-source information fusion prediction model according to the training data set, verifies the theoretical solution of rock mass strength by using the multi-source information fusion prediction model, and obtains a predicted solution of rock mass strength; constructs a rock mass parameter model according to the predicted solution of rock mass strength, predicts the roadway surrounding rock by using the rock mass parameter model, and obtains a predicted value of rock mass parameters; constructs a rock mass cuttability prediction model according to the predicted value of rock mass parameters, predicts the test data set by using the rock mass cuttability prediction model, and obtains a predicted result of the rock mass cuttability grade of the roadway surrounding rock. The present application can realize rapid real-time grading of the cuttability of rock mass in the drilling test area, guide engineering construction, improve construction efficiency, and reduce construction costs.

[0045] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following specifically lists preferred embodiments and, in conjunction with the accompanying drawings, makes the following detailed description. Description of the Drawings

[0046] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the embodiments. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings. In each drawing, similar components are numbered similarly.

[0047] Figure 1 The flowchart of an intelligent evaluation method for the cuttability of roadway surrounding rock provided by an embodiment of the present application is shown;

[0048] Figure 2 The schematic diagram of an intelligent evaluation method for the cuttability of roadway surrounding rock provided by an embodiment of the present application is shown;

[0049] Figure 3 The flowchart of a data preprocessing method provided by an embodiment of the present application is shown;

[0050] Figure 4 The schematic diagram of another intelligent evaluation method for the cuttability of roadway surrounding rock provided by an embodiment of the present application is shown;

[0051] Figure 5 The structural schematic diagram of an intelligent evaluation device for the cuttability of roadway surrounding rock provided by an embodiment of the present application is shown. Detailed implementation manners

[0052] The following will describe in detail the embodiments of the present invention. The examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention and should not be construed as limiting the present invention.

[0053] It should be noted that the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, "a plurality" means two or more, unless otherwise specifically defined.

[0054] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs. The terms used in the description of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0055] Example 1

[0056] As Figure 1 shown, it is a flowchart of an intelligent evaluation method for the cuttability of roadway surrounding rock in the embodiments of the present application. As Figure 2 shown, it is a schematic diagram of an intelligent evaluation method for the cuttability of roadway surrounding rock in the embodiments of the present application. The intelligent evaluation method for the cuttability of roadway surrounding rock provided by the embodiments of the present application includes the following steps:

[0057] Step S110: Obtain the initial drilling main control parameter data of the roadway surrounding rock, preprocess the initial drilling main control parameter data to obtain the preprocessed drilling main control parameter data, and divide the preprocessed drilling main control parameter data into a training data set and a test data set.

[0058] In this embodiment, a portable test instrument can be used to carry out in-situ drilling tests on the selected area of the roadway surrounding rock to obtain the initial drilling main control parameter data of the roadway surrounding rock, including but not limited to rotational speed, torque, thrust, drilling speed, noise, vibration, etc. The obtained initial drilling main control parameter data is used to construct a data set, denoted as κ(x,y,z)={(x1,y1,z1),(x1,y2,z2)…(x n ,y n ,z n )}, where x, y, and z respectively represent different initial drilling main control parameter data, and n is the number of data points.

[0059] Optionally, the portable test instrument in this embodiment can be a small portable hand-held drill equipped with an intelligent drilling test system. The total machine mass does not exceed 10 kg, the drill rotational speed is 0 - 600 rpm, the peak torque is 5 N·m, and the maximum thrust is 1 kN. The specific settings can be determined according to the actual situation, and the embodiments of the present application do not limit this.

[0060] Since there may be some problems such as noise, outliers, and inconsistent data dimensions in the obtained initial drilling main control parameter data, it is necessary to preprocess the initial drilling main control parameter data. The specific steps are as Figure 3 shown, including steps S111 - S113:

[0061] Step S111: Perform wavelet denoising on the initial drilling main control parameter data to obtain the denoised drilling main control parameter data.

[0062] It can be understood that first, wavelet decomposition is performed on the initial drilling main control parameter data. Wavelet decomposition can decompose the signal into multiple frequency band components to obtain high-frequency components and low-frequency components. The decomposition result can be expressed as:

[0063]

[0064] where a j is the low-frequency component of the j-th layer, d j is the high-frequency component of the j-th layer, and N is the number of decomposition layers.

[0065] For each high-frequency component d j , a preset noise threshold needs to be set to screen the noise:

[0066]

[0067] where T j is the preset noise threshold of the high-frequency component of the j-th layer, σ j is the standard deviation of noise estimation, and n is the length of the high-frequency component.

[0068] Soft threshold processing is performed on the high-frequency component according to the preset noise threshold to remove the noise. The soft threshold processing formula is:

[0069]

[0070] where d j ′ is the high-frequency component after removal.

[0071] The high-frequency component d j ′ after removal and the low-frequency component a j are recombined to obtain the noise-reduced signal κ′(t).

[0072] Then, a sliding window is used to perform local statistics on the above noise-reduced signal κ′(t) to calculate the Power Spectral Density (PSD). The size of the sliding window can be adjusted according to the characteristics of the signal and the nature of the noise. The calculation of PSD helps to understand the intensity change of the noise at different time periods and provides a basis for dynamically adjusting the filtering parameters in the follow-up.

[0073] According to the calculated noise power spectral density, the parameters of the filter are dynamically adjusted so that the filter can adapt to the change of the noise intensity, thereby improving the filtering effect. These parameters may include the type, order, cut-off frequency, etc. of the filter.

[0074] The low-frequency part of the signal is smoothed using the adjusted filter. The smoothing process can reduce the unnecessary fluctuations and noise in the signal, making the signal smoother and more stable. And spectral analysis is performed on the smoothed noise-reduced signal to identify the interference frequency signal components. Spectral analysis can reveal the intensity distribution of different frequency components in the signal. By identifying the interference frequency signal components, it can be determined which frequency components are the interference signals that need to be filtered out. A band-stop filter is used to remove the identified interference frequency signal components. The band-stop filter can suppress the transmission of the signal within the specified frequency range, thereby removing the interference and obtaining the filtered signal.

[0075] Finally, the denoised signal processed by wavelet and the filtered signal processed by dynamic filtering are fused to generate the main control parameter data of drilling after denoising, ensuring that as much useful information as possible is retained in the signal during the processing, while removing as much noise as possible.

[0076] Optionally, the denoising effect is evaluated by indicators such as the signal-to-noise ratio (SNR) and the mean squared error (MSE). The signal-to-noise ratio reflects the relative intensity between the signal and the noise, while the mean squared error measures the difference between the denoised signal and the original signal. These indicators help to quantify the effectiveness of the denoising method and provide a reliable basis for subsequent signal processing and analysis.

[0077] Step S112: Perform outlier rejection processing on the main control parameter data of drilling after denoising to obtain the main control parameter data of drilling after rejection.

[0078] Understandably, first calculate the mean μ, standard deviation σ, and quartiles Q1 and Q3 of each parameter data in the main control parameter data of drilling after denoising, and set the initial threshold range as:

[0079]

[0080] In the formula, α and β are dynamic adjustment coefficients used to adjust the threshold range according to the data characteristics.

[0081] During the data acquisition process, the sliding window technique is used to update the data in real time, and the mean data within the threshold range is readjusted according to each batch of new data to obtain the single-parameter dynamic threshold range:

[0082] μ t = λ·μ t-1 +(1 - λ)·μ new

[0083] In the formula, λ is the time decay factor used to balance the influence of historical data and new data. μ t is the mean at time t, μ t-1 is the mean at time t - 1, and μ new represents the mean of the new data.

[0084] Then, obtain the historical main control parameters data of drilling, and construct a linear regression model and non-linear models (such as neural networks, support vector machines, etc.) based on the historical main control parameters data of drilling to capture the correlation relationships between the parameters. Use the linear regression model and non-linear models to predict each parameter data in the noise-reduced main control parameters data of drilling, obtain the predicted values, and calculate the residual values of each parameter data in the noise-reduced main control parameters data of drilling according to the obtained observed values and predicted values:

[0085] ε = y observed - y predicted

[0086] In the formula, ε is the residual value, y observed is the observed value, and y predicted is the predicted value.

[0087] Set a multi-parameter residual range [-ε threshold , +ε threshold . If each parameter data in the noise-reduced main control parameters data of drilling is within the single-parameter dynamic threshold range, and the residual value of each parameter data in the noise-reduced main control parameters data of drilling is within the multi-parameter residual range (ε ∈ [-ε threshold , +ε threshold ), then determine the corresponding parameter data as normal data; if any parameter data in the noise-reduced main control parameters data of drilling is outside the single-parameter dynamic threshold range, and the residual value of any parameter data in the noise-reduced main control parameters data of drilling is outside the multi-parameter residual range, then determine the corresponding parameter data as abnormal data; if any parameter data in the noise-reduced main control parameters data of drilling is outside the single-parameter dynamic threshold range, but the residual value of each parameter data in the noise-reduced main control parameters data of drilling is within the multi-parameter residual range, then determine whether it is a temporary fluctuation according to the time trend (such as checking whether the data points before and after this data point also show an abnormal trend). If it is a temporary fluctuation, then temporarily retain the parameter data and mark it as an abnormal point for subsequent analysis; if it is not a temporary fluctuation, then determine the parameter data as abnormal data.

[0088] Finally, for the parameter data determined to be abnormal data, use the constructed linear regression model and non-linear models, or perform linear or polynomial interpolation using the normal data points before and after, predict the reasonable value of the abnormal data, and perform substitution to obtain the main control parameters data of drilling after elimination. It can effectively avoid the influence of abnormal data on subsequent model training and improve the stability and accuracy of the model.

[0089] Step S113, perform normalization processing on the main control parameters data of drilling after elimination to obtain the preprocessed main control parameters data of drilling.

[0090] Understandably, first, the mean μ, standard deviation σ, and quartiles Q1 and Q3 of each parameter data in the noise-reduced main drilling control parameter data are calculated. Based on the interquartile range (IQR = Q3 - Q1) and a dynamically adjusted range expansion factor δ, a dynamic range is set:

[0091] [L, U] = [Q1 - δ·IQR, Q3 + δ·IQR]

[0092] This range is used for subsequent data normalization, where L = Q1 - δ·IQR, U = Q3 + δ·IQR, and the value of δ is dynamically adjusted according to the specific application scenario to adapt to different distribution characteristics of the data.

[0093] Considering that there may be significant gradient differences between different main drilling control parameter data, each parameter data τ in the filtered main drilling control parameter data needs to be normalized into a dimensionless value τ′, ensuring that the range of the normalized main drilling control parameter data is [0, 1]. The normalization formula is:

[0094]

[0095] Then, the gradient difference of each parameter data's contribution to the preset target value in the normalized main drilling control parameter data is obtained, and the gradient balance coefficient g of each parameter data is calculated according to the gradient difference. The normalized main drilling control parameter data is further adjusted using the gradient balance coefficient:

[0096] τ″ = τ′·g

[0097] Finally, the preprocessed main drilling control parameter data τ″ is obtained. Converting the main drilling control parameter data with different dimensions into dimensionless values facilitates subsequent model processing and ensures that the data can be compared and analyzed on a unified scale.

[0098] Preferably, in this embodiment, a global optimization function can be introduced to test the preprocessed main drilling control parameter data to ensure the balance of normalization:

[0099]

[0100] In the formula, is the gradient (i.e., the influence degree) of the preprocessed main drilling control parameter data on each parameter τ i , is the mean of all gradient balance coefficients. This testing process aims to ensure that the normalized data maintains a reasonable gradient distribution as a whole and does not introduce imbalance due to the normalization process.

[0101] Further, the preprocessed main drilling parameter data is divided into a training data set and a test data set according to a preset ratio (e.g., 8:2), providing a data basis for subsequent model training and detection.

[0102] The above step S110 effectively ensures the accuracy and consistency of the input data, providing a reliable basis for subsequent model training.

[0103] Step S120: Construct a multi-source information fusion prediction model based on the training data set, and use the multi-source information fusion prediction model to verify the theoretical solution of the rock mass strength to obtain the predicted solution of the rock mass strength.

[0104] Understandably, as Figure 2 shown, the training data set is input into the computer algorithm library, and multi-source information cross-training is performed using a preset multi-source information cross-training algorithm. The multi-source information cross-training algorithm includes six types: BP network, SVM, BP-NN, K-NN, C-NN, and Catboost. The training data set is randomly divided into six subsets with equal data volumes, and 6-fold cross-validation is performed using different meta-feature functions respectively to obtain the training results. According to the training results, a multi-source information fusion prediction model is constructed.

[0105] Use the multi-source information fusion prediction model to verify the theoretical solution of the rock mass cutting strength and the theoretical solution of the rock mass intrusion strength to obtain the predicted solution of the rock mass cutting strength and the predicted solution of the rock mass intrusion strength. Compare the results of the multi-source information fusion prediction model with the theoretical solutions of the rock mass cutting strength and the intrusion strength to verify the prediction accuracy of the model, ensuring that the comparison accuracy between the prediction model results and the theoretical solutions is not less than 80%.

[0106] It should be noted that the multi-source information cross-training algorithm adopted in this embodiment can be determined according to the actual situation, and this embodiment does not limit it. Other types of multi-source information cross-training algorithms are also included in the protection scope of this application.

[0107] The above step S120 uses a multi-source information cross-training algorithm to fuse the information of multiple data sources, improving the accuracy, robustness, and generalization ability of rock mass strength prediction.

[0108] Step S130: Construct a rock mass parameter model based on the predicted solution of the rock mass strength, and use the rock mass parameter model to predict the surrounding rock of the roadway to obtain the predicted value of the rock mass parameters.

[0109] Further, according to the predicted solution of the rock mass strength predicted in step S120, construct a rock mass parameter model, including a rock mass mechanical parameter model and a rock mass structure parameter model. Use the rock mass mechanical parameter model and the rock mass structure parameter model to predict the surrounding rock of the roadway to obtain the rock mass mechanical parameters and the rock mass structure parameters.

[0110] Preferably, the rock mass mechanical parameters include but are not limited to strength, elastic modulus, cohesion, and internal friction angle, and the rock mass structure parameters include but are not limited to bedding, interface, angle, and thickness.

[0111] Based on the prediction solution of the rock mass strength, the above step S130 further constructs a rock mass mechanical parameter model and a rock mass structure parameter model, providing detailed rock mass mechanical parameters and structure parameters for the prediction of the roadway surrounding rock, and laying a foundation for the subsequent prediction of the rock mass cuttability.

[0112] Step S140: Construct a rock mass cuttability prediction model according to the predicted values of the rock mass parameters, and use the rock mass cuttability prediction model to predict the test data set to obtain the predicted result of the rock mass cuttability level value of the roadway surrounding rock.

[0113] It can be understood that through the multi-task learning framework (Multi-Task Learning, MTL), the loss function is defined according to the prediction errors (the differences between the theoretical solutions and the prediction solutions) of the rock mass cutting strength, the rock mass intrusion strength, and the rock mass integrity index:

[0114]

[0115] In the formula, is the loss function, is the prediction error of the rock mass cutting strength, is the prediction error of the rock mass intrusion strength, is the prediction error of the rock mass integrity index, and λ1, λ2, and λ3 are the corresponding weight coefficients, with the initial values set according to expert experience or pre-experiments and dynamically adjusted through backpropagation during the training process.

[0116] Optimize the predicted values of the rock mass parameters and their contribution weights according to the above loss function, output the optimized predicted values of the rock mass parameters and their optimized weight coefficients, and calculate the comprehensive cuttability index according to the above output:

[0117] K c = ω1·C s + ω2·I s - ω3·R i

[0118] In the formula, K c is the comprehensive cuttability index, C s is the optimized rock mass cutting strength, I s is the optimized rock mass intrusion strength, R i is the optimized rock mass integrity index, where C s and I s belong to the rock mass mechanical parameters and contribute positively to the cutting difficulty, and Ri It belongs to the rock mass structure parameters and contributes inversely to the cutting difficulty (the more complete the structure, the more difficult it is to cut). ω1, ω2, and ω3 are the optimized weight coefficients, which are determined by experimental calibration (such as orthogonal experiments or regression analysis).

[0119] Finally, using the preset level division rules and based on the value range of the comprehensive index of the cutability of the roadway surrounding rock, the predicted result of the rock mass cutability level of the roadway surrounding rock is determined. The preset level division rules are as follows:

[0120]

[0121] In the formula, G is the predicted result of the rock mass cutability level, K threshold1 , K threshold2 , …, K threshold6 are empirical thresholds, which can be determined by historical data calibration and usually correspond to the working ability boundaries of different cutting equipment.

[0122] In step S140 above, using the multi-task learning framework, comprehensively considering the rock mass cutting strength, rock mass intrusion strength, and rock mass integrity index, a rock mass cutability prediction model is constructed to achieve rapid real-time grading of the cutability of the roadway surrounding rock.

[0123] In an optional implementation manner, as Figure 4 shown, the intelligent evaluation method for the cutability of the roadway surrounding rock in this embodiment further includes:

[0124] Evaluating the predicted result of the rock mass cutability level of the roadway surrounding rock using preset evaluation indicators, where the preset evaluation indicators include but are not limited to Mean Absolute Percentage Error (MAPE), balanced accuracy rate, and F1-Score (F1-Score).

[0125] Judge whether the predicted result of the rock mass cutability level meets the preset conditions. If it meets, directly output the predicted result of the rock mass cutability level of the roadway surrounding rock; if it does not meet, for example, if the MAPE value of the predicted result is high and the balanced accuracy rate and F1-Score are low, it indicates that there are deficiencies in the current model when predicting the rock mass cutability level and does not meet the preset conditions. At this time, it is necessary to adjust the multi-source information cross-training steps (such as adjusting the algorithm combination, optimizing the algorithm parameters, increasing the training data, etc.). After adjusting the multi-source information cross-training steps, re-train and evaluate the model. This process needs to be carried out in a loop until the predicted result of the rock mass cutability level of the rock mass cutability prediction model on the test data set meets the preset conditions. During the loop process, the model can be gradually optimized according to the results of each evaluation until satisfactory prediction performance is obtained.

[0126] The above method evaluates the prediction results through preset evaluation indicators, discovers problems existing in the model in a timely manner and optimizes them, ensuring the accuracy and reliability of the prediction results. The model training and evaluation are carried out cyclically to gradually optimize the model performance until the preset conditions are met, improving the overall efficiency and accuracy of the evaluation method.

[0127] The intelligent evaluation method for the cuttability of roadway surrounding rock provided by the embodiment of the present application obtains the initial drilling main control parameter data of the roadway surrounding rock, preprocesses the initial drilling main control parameter data to obtain the preprocessed drilling main control parameter data, and divides the preprocessed drilling main control parameter data into a training data set and a test data set; constructs a multi-source information fusion prediction model according to the training data set, uses the multi-source information fusion prediction model to verify the theoretical solution of rock mass strength, and obtains the predicted solution of rock mass strength; constructs a rock mass parameter model according to the predicted solution of rock mass strength, uses the rock mass parameter model to predict the roadway surrounding rock, and obtains the predicted value of rock mass parameters; constructs a rock mass cuttability prediction model according to the predicted value of rock mass parameters, and uses the rock mass cuttability prediction model to predict the test data set to obtain the predicted result of the rock mass cuttability level value of the roadway surrounding rock. The present application can realize the rapid and real-time grading of the cuttability of the rock mass in the drilling test area, guide the engineering construction, improve the construction efficiency, and reduce the construction cost.

[0128] Embodiment 2

[0129] As Figure 5 shown, it is a schematic structural diagram of an intelligent evaluation device 500 for the cuttability of roadway surrounding rock in an embodiment of the present application. The device includes:

[0130] A processing module 510, configured to obtain the initial drilling main control parameter data of the roadway surrounding rock, preprocess the initial drilling main control parameter data to obtain the preprocessed drilling main control parameter data, and divide the preprocessed drilling main control parameter data into a training data set and a test data set;

[0131] A first construction module 520, configured to construct a multi-source information fusion prediction model according to the training data set, and use the multi-source information fusion prediction model to verify the theoretical solution of rock mass strength to obtain the predicted solution of rock mass strength;

[0132] A second construction module 530, configured to construct a rock mass parameter model according to the predicted solution of rock mass strength, and use the rock mass parameter model to predict the roadway surrounding rock to obtain the predicted value of rock mass parameters;

[0133] A prediction module 540 is configured to construct a rock mass cuttability prediction model based on the predicted values of the rock mass parameters, and use the rock mass cuttability prediction model to predict the test data set, so as to obtain the rock mass cuttability prediction result of the surrounding rock of the roadway.

[0134] The intelligent evaluation device for the cuttability of the surrounding rock of the roadway provided by the embodiments of the present application can implement each process of the intelligent evaluation method for the cuttability of the surrounding rock of the roadway corresponding to Embodiment 1, and can achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0135] The intelligent evaluation device for the cuttability of the surrounding rock of the roadway provided by the embodiments of the present application can realize the rapid real-time grading of the cuttability of the rock mass in the drilling test area, guide the engineering construction, improve the construction efficiency and reduce the construction cost.

[0136] In several embodiments provided by the present application, it should be understood that the disclosed device and method can also be implemented in other ways. The device embodiments described above are only illustrative. For example, the flowcharts and structure diagrams in the drawings show the possible architectures, functions and operations of the device, method and computer program product according to multiple embodiments of the present invention. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code, and the module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in the alternative implementation, the functions marked in the blocks may occur in a different order from that marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the structure diagram and / or flowchart, as well as the combination of blocks in the structure diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0137] In addition, each functional module or unit in various embodiments of the present invention may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0138] When the above-mentioned functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a smart phone, a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium can be a non-volatile storage medium or a volatile storage medium. For example, the storage medium can be: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc, etc., which can store program codes.

[0139] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention.

Claims

1. An intelligent evaluation method for the cutability of roadway surrounding rock, characterized in that, The method includes: Obtaining the initial drilling main control parameter data of the roadway surrounding rock, preprocessing the initial drilling main control parameter data to obtain the preprocessed drilling main control parameter data, and dividing the preprocessed drilling main control parameter data into a training data set and a test data set; Constructing a multi-source information fusion prediction model according to the training data set, and using the multi-source information fusion prediction model to verify the theoretical solution of the rock mass strength to obtain the predicted solution of the rock mass strength; Constructing a rock mass parameter model according to the predicted solution of the rock mass strength, and using the rock mass parameter model to predict the roadway surrounding rock to obtain the predicted value of the rock mass parameters; Constructing a rock mass cuttability prediction model according to the predicted value of the rock mass parameters, and using the rock mass cuttability prediction model to predict the test data set to obtain the predicted result of the rock mass cuttability level value of the roadway surrounding rock.

2. The intelligent evaluation method for the cutability of roadway surrounding rock according to claim 1, wherein The initial drilling main control parameter data includes at least one of rotational speed, torque, thrust, drilling speed, noise, and vibration. The rock mass strength includes rock mass cutting strength and rock mass intrusion strength. The rock mass parameter model includes a rock mass mechanical parameter model and a rock mass structural parameter model. The rock mass parameters include rock mass mechanical parameters and rock mass structural parameters. The rock mass mechanical parameters include at least one of strength, elastic modulus, cohesion, and internal friction angle. The rock mass structural parameters include at least one of bedding, interface, angle, and thickness.

3. The intelligent evaluation method for the cutability of roadway surrounding rock according to claim 1, characterized in that The preprocessing of the initial drilling main control parameter data to obtain the preprocessed drilling main control parameter data includes: Performing wavelet denoising on the initial drilling main control parameter data to obtain the denoised drilling main control parameter data; Performing outlier removal processing on the denoised drilling main control parameter data to obtain the removed drilling main control parameter data; Performing normalization processing on the removed drilling main control parameter data to obtain the preprocessed drilling main control parameter data.

4. The intelligent evaluation method for the cutability of roadway surrounding rock according to claim 3, characterized in that, The performing wavelet denoising on the initial drilling main control parameter data to obtain the denoised drilling main control parameter data includes: Performing wavelet decomposition on the initial drilling main control parameter data to obtain high-frequency components and low-frequency components, removing the noise in the high-frequency components according to a preset noise threshold, and fusing the removed high-frequency components and the low-frequency components to obtain a denoised signal; Performing local statistics on the denoised signal using a sliding window, calculating the noise power spectral density, dynamically adjusting the parameters of the filter according to the noise power spectral density, smoothing the low-frequency components in the denoised signal using the adjusted filter, performing spectral analysis on the smoothed denoised signal, identifying the interference frequency signal components, and removing the interference frequency signal components using a band-stop filter to obtain a filtered signal; Fusing the denoised signal and the filtered signal to obtain the denoised drilling main control parameter data.

5. The intelligent evaluation method for the cutability of roadway surrounding rock according to claim 3, characterized in that The performing outlier removal processing on the denoised drilling main control parameter data to obtain the removed drilling main control parameter data includes: Calculate the mean, standard deviation, and quartiles of each parameter data in the noise-reduced main drilling control parameter data. Set an initial threshold range based on the mean, the standard deviation, and the quartiles, and use the sliding window technique to adjust the initial threshold range in real time to obtain a single-parameter dynamic threshold range; Obtain historical main drilling control parameter data, construct a linear regression model and a non-linear model based on the historical main drilling control parameter data, and use the linear regression model and the non-linear model to predict each parameter data in the noise-reduced main drilling control parameter data to obtain predicted values. Obtain the observed values of each parameter data in the noise-reduced main drilling control parameter data, calculate the residual values of each parameter data in the noise-reduced main drilling control parameter data based on the observed values and the predicted values, and set a multi-parameter residual range; If any parameter data in the noise-reduced main drilling control parameter data is outside the single-parameter dynamic threshold range, and the residual value of any parameter data in the noise-reduced main drilling control parameter data is outside the multi-parameter residual range, then determine the corresponding parameter data as abnormal data; Remove the abnormal data, and use the linear regression model and the non-linear model to predict reasonable values of the abnormal data for substitution to obtain the main drilling control parameter data after removal.

6. The intelligent evaluation method for the cutability of roadway surrounding rock according to claim 5, characterized in that, The normalization process for the main drilling control parameter data after removal to obtain the preprocessed main drilling control parameter data includes: Set a dynamic range for the mean, standard deviation, and quartiles of each parameter data in the noise-reduced main drilling control parameter data; Normalize each parameter data in the main drilling control parameter data after removal according to the dynamic range to obtain the normalized main drilling control parameter data. Obtain the gradient difference of the contribution of each parameter data in the normalized main drilling control parameter data to a preset target value, and calculate the gradient balance coefficient of each parameter data in the normalized main drilling control parameter data according to the gradient difference. Multiply each parameter data in the normalized main drilling control parameter data by its corresponding gradient balance coefficient to obtain the preprocessed main drilling control parameter data.

7. The intelligent evaluation method for the cutability of roadway surrounding rock according to claim 1, characterized in that, Construct a multi-source information fusion prediction model based on the training data set, and use the multi-source information fusion prediction model to verify the theoretical solution of rock mass strength to obtain the predicted solution of rock mass strength, including: Input the training data set into a computer algorithm library, perform multi-source information cross-training using a preset multi-source information cross-training algorithm to obtain a training result, and construct the multi-source information fusion prediction model according to the training result, where the multi-source information cross-training algorithm includes BP network, SVM, BP-NN, K-NN, C-NN, and Catboost; Use the multi-source information fusion prediction model to verify the theoretical solution of the rock cutting strength and the theoretical solution of the rock intrusion strength of the rock mass to obtain the predicted solution of the rock cutting strength and the predicted solution of the rock intrusion strength of the rock mass.

8. The intelligent evaluation method for the cutability of roadway surrounding rock according to claim 1, characterized in that Constructing a rock mass cutability prediction model based on the predicted values of the rock mass parameters, and using the rock mass cutability prediction model to predict the test data set to obtain the prediction result of the rock mass cutability level value of the roadway surrounding rock, including: Defining a loss function according to the theoretical solution and the predicted solution of the rock mass strength, and optimizing the predicted values of the rock mass parameters and their contribution weights according to the loss function to obtain the optimized predicted values of the rock mass parameters and their optimized weight coefficients; Calculating the comprehensive cutability index of the roadway surrounding rock according to the optimized predicted values of the rock mass parameters and their optimized weight coefficients; Determining the prediction result of the rock mass cutability level value of the roadway surrounding rock by using a preset level division rule and according to the value range of the comprehensive cutability index of the roadway surrounding rock.

9. The intelligent evaluation method for the cuttability of roadway surrounding rock according to claim 8, characterized in that The method further includes: Evaluating the prediction result of the rock mass cutability level value of the roadway surrounding rock by using a preset evaluation index, where the preset evaluation index includes at least one of mean absolute percentage error, balanced accuracy rate, and F1-Score; Judging whether the prediction result of the rock mass cutability level value meets the preset conditions. If so, directly outputting the prediction result of the rock mass cutability level value of the roadway surrounding rock. If not, retraining and evaluating each model until the prediction result of the rock mass cutability level value of the rock mass cutability prediction model on the test data set meets the preset conditions.

10. An intelligent evaluation device for the cutability of roadway surrounding rock, characterized in that, The device includes: A processing module, configured to obtain the initial drilling main control parameter data of the roadway surrounding rock, preprocess the initial drilling main control parameter data to obtain the preprocessed drilling main control parameter data, and divide the preprocessed drilling main control parameter data into a training data set and a test data set; A first construction module, configured to construct a multi-source information fusion prediction model according to the training data set, and use the multi-source information fusion prediction model to verify the theoretical solution of the rock mass strength to obtain the predicted solution of the rock mass strength; A second construction module, configured to construct a rock mass parameter model according to the predicted solution of the rock mass strength, and use the rock mass parameter model to predict the roadway surrounding rock to obtain the predicted values of the rock mass parameters; A prediction module, configured to construct a rock mass cutability prediction model according to the predicted values of the rock mass parameters, and use the rock mass cutability prediction model to predict the test data set to obtain the prediction result of the rock mass cutability of the roadway surrounding rock.

Images