Construction decision-making method and device based on rock joint roughness coefficient

Abstract

This invention discloses a construction decision-making method and apparatus based on the rock joint roughness coefficient (JRC), relating to the field of geological engineering technology. The method includes: acquiring a first dataset; substituting the statistical parameters of each rock joint surface into various empirical regression formulas to obtain multiple JRC estimates; fusing the multiple JRC estimates for each rock joint surface to obtain JRC labels; training a model using the statistical parameters and JRC labels as training data to obtain a JRC prediction model; acquiring the statistical parameters of the target rock joint surface during construction, inputting them into the JRC prediction model, and outputting JRC prediction values; generating a new construction decision when the absolute deviation between the JRC prediction value and historical JRC prediction values ​​exceeds a preset threshold; otherwise, maintaining the current construction decision. This invention integrates complementary information from multiple empirical regression formulas, improving the accuracy of JRC prediction, realizing dynamic adaptive adjustment of construction decisions, and enhancing construction efficiency and safety.

Description

Technical Field

[0001] This invention relates to the field of geological engineering technology, specifically to a construction decision-making method and apparatus based on the rock joint roughness coefficient. Background Technology

[0002] The Joint Roughness Coefficient (JRC) is an important indicator characterizing the roughness of rock joint surfaces, directly affecting the rock-breaking efficiency, cutter wear, and tunneling energy consumption of tunnel boring machines (TBMs). Specifically, a higher JRC value indicates a rougher joint surface, resulting in greater impact loads on the cutter during rock breaking, faster cutter wear, increased tunneling specific energy, and decreased tunneling speed. Conversely, a lower JRC value indicates a smoother joint surface and relatively higher rock-breaking efficiency. Therefore, obtaining the JRC value in real-time and accurately is fundamental for optimizing tunneling parameters, predicting cutter life, and developing support schemes, directly impacting construction efficiency, cost, and safety.

[0003] Currently, there are two main methods for obtaining JRC values:

[0004] The first method is a prediction method based on empirical regression formulas. By establishing empirical regression formulas between various statistical parameters (such as root mean square of the profile, profile structure function, etc.) and the JRC, JRC estimates can be quickly obtained by inputting the statistical parameters. However, these empirical formulas only use a single parameter for regression analysis and do not consider information from other statistical parameters, leading to inconsistent prediction results from different empirical regression formulas.

[0005] The second approach is based on neural network prediction methods. By establishing a nonlinear prediction model between statistical parameters and JRC, multi-parameter information can be fused, resulting in higher prediction accuracy than a single empirical formula. However, this type of method requires calibrating a large number of JRC labels through shearing experiments, which are costly, time-consuming, and difficult to obtain large-scale training data, thus limiting the model's generalization ability and prediction accuracy.

[0006] Furthermore, existing technologies lack an intelligent mechanism for updating construction decisions after obtaining JRC prediction values. Typically, a decision is made only once before construction begins, failing to dynamically adjust construction parameters based on real-time changes in JRC. This results in difficulties in timely responses to significant changes in joint surface roughness, impacting tunneling efficiency and construction safety.

[0007] Therefore, how to efficiently and accurately obtain JRC values ​​and dynamically adjust construction decisions based on JRC values ​​in real time is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0008] The technical problem to be solved by the present invention is that the existing technology cannot efficiently and accurately obtain JRC values ​​and dynamically adjust construction decisions in real time based on JRC. The purpose is to provide a construction decision-making method and device based on rock joint roughness coefficient, which solves the above-mentioned problem.

[0009] This invention is achieved through the following technical solution:

[0010] In a first aspect, the present invention provides a construction decision-making method based on the rock joint roughness coefficient, comprising:

[0011] A first dataset is obtained, and the statistical parameters of each rock joint surface in the first dataset are substituted into various types of empirical regression formulas to obtain multiple joint roughness coefficients (JRC) estimates for each rock joint surface; wherein, the first dataset includes statistical parameters of multiple rock joint surfaces; the statistical parameters are used to quantitatively describe the surface morphology characteristics of the rock joint surfaces;

[0012] Multiple JRC estimates for each rock joint surface in the first dataset are fused to obtain the JRC label for each rock joint surface;

[0013] The model is trained using the statistical parameters of each rock joint surface and the corresponding JRC labels in the first dataset as training data to obtain the JRC prediction model.

[0014] During construction, statistical parameters of the target rock joint surface are obtained and input into the JRC prediction model, and the JRC prediction value of the target rock joint surface is output.

[0015] When the absolute deviation between the JRC prediction value of the target rock joint surface and the historical JRC prediction value is greater than a preset threshold, a new construction decision is generated based on the JRC prediction value of the target rock joint surface; otherwise, the current construction decision is maintained.

[0016] Optionally, the step of fusing multiple JRC estimates for each rock joint surface in the first dataset to obtain the JRC label for each rock joint surface includes:

[0017] Obtain the second dataset; the second dataset includes statistical parameters of multiple rock joint surfaces and corresponding JRC measured values;

[0018] Based on the prediction error of each empirical regression formula on the second dataset, assign weight coefficients to each empirical regression formula.

[0019] Based on the weighting coefficients, a weighted average is performed on multiple JRC estimates of each rock joint surface in the first dataset to obtain the JRC label of each rock joint surface in the first dataset.

[0020] Optionally, assigning weight coefficients to each empirical regression formula based on its prediction error on the second dataset includes:

[0021] For each type of empirical regression formula, the statistical parameters of each rock joint surface in the second dataset are substituted into the empirical regression formula to obtain the JRC estimate of each rock joint surface, as well as the statistical error between the JRC estimate of each rock joint surface and the corresponding measured JRC value, and the prediction error of the empirical regression formula is calculated.

[0022] Based on the prediction error of each empirical regression formula, a weight coefficient is assigned to each empirical regression formula; among them, the empirical regression formula with the smaller prediction error has a larger weight coefficient.

[0023] Optionally, assigning weight coefficients to each empirical regression formula based on its prediction error includes:

[0024] The reciprocal of the prediction error of each empirical regression formula is normalized and used as the weight coefficient of each empirical regression formula.

[0025] Optionally, generating a new construction decision based on the JRC prediction value of the target rock joint surface includes:

[0026] The objective function is constructed with the optimization objectives of maximizing tunneling speed, minimizing tool wear rate, and minimizing tunneling specific energy.

[0027] Based on the JRC prediction value of the target rock joint surface and other rock mass parameters, an optimization algorithm is used to search within the feasible region of the tunneling parameters to find the optimal combination of tunneling parameter values ​​that makes the objective function optimal; the tunneling parameters include thrust, torque, penetration depth and cutterhead rotation speed;

[0028] The optimal combination of tunneling parameters is used as the construction decision for the target rock joint surface.

[0029] Optionally, the objective function is as follows:

[0030] ;

[0031] Where F is the comprehensive evaluation value of various construction effects; v is the tunneling speed. This refers to the maximum propulsion speed of the equipment. For tool wear rate, SE represents the maximum allowable wear rate of the cutting tool; SE is the tunneling specific energy. This is the preset maximum tunneling specific energy reference value; , and These are the weighting coefficients for each construction effect.

[0032] Optionally, the JRC prediction value based on the target rock joint surface and other rock mass parameters is used to search within the feasible region of the tunneling parameters using an optimization algorithm to find the optimal combination of tunneling parameter values ​​that makes the objective function optimal, including:

[0033] Initialize the population by randomly generating N combinations of tunneling parameter values ​​as individuals; where N is the preset population size; each individual includes values ​​for thrust, torque, penetration depth, and cutterhead rotation speed;

[0034] Based on the JRC prediction value of the target rock joint surface and other rock mass parameters, the fitness value of each individual is calculated; wherein, the fitness value is the value of the objective function;

[0035] Select superior individuals based on their fitness values, perform crossover and mutation operations, and generate a new generation of population.

[0036] Repeat the iteration until the preset maximum number of iterations or the fitness value converges. Select the individual with the highest fitness from the final population as the optimal combination of tunneling parameters.

[0037] Optionally, the statistical parameters include: average relative height, height standard deviation, average tilt angle, maximum relative height, tilt angle standard deviation, roughness profile index, profile structure function, and profile root mean square.

[0038] Optionally, each type of empirical regression formula is used to reflect the mapping relationship between one of the statistical parameters and JRC; before substituting the statistical parameters of each rock joint surface in the first dataset into the various types of empirical regression formulas, the method further includes:

[0039] When any type of empirical regression formula includes multiple candidate formulas, the formula with the highest fitting accuracy is selected as the empirical regression formula of that type; the fitting accuracy is the goodness of fit index between the JRC estimate calculated based on the validation data of each candidate formula and the JRC measured value.

[0040] Secondly, the present invention provides a construction decision-making device based on the rock joint roughness coefficient, comprising:

[0041] The JRC tagging module is used to acquire a first dataset and substitute the statistical parameters of each rock joint surface in the first dataset into various types of empirical regression formulas to obtain multiple joint roughness coefficient JRC estimates for each rock joint surface; the multiple JRC estimates for each rock joint surface in the first dataset are fused to obtain the JRC label for each rock joint surface; wherein, the first dataset includes statistical parameters of multiple rock joint surfaces; the statistical parameters are used to quantitatively describe the surface morphology characteristics of the rock joint surfaces;

[0042] The model training module is used to train the model using the statistical parameters of each rock joint surface in the first dataset and the corresponding JRC labels as training data to obtain the JRC prediction model.

[0043] The JRC prediction module is used to obtain statistical parameters of the target rock joint surface during construction, input them into the JRC prediction model, and output the JRC prediction value of the target rock joint surface.

[0044] The construction decision module is used to generate a new construction decision based on the JRC prediction value of the target rock joint surface when the absolute deviation between the JRC prediction value of the target rock joint surface and the historical JRC prediction value is greater than a preset threshold; otherwise, the current construction decision is maintained.

[0045] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0046] This application provides a construction decision-making method based on the rock joint roughness coefficient. By fusing the prediction results of multiple empirical regression formulas, a JRC label is generated, effectively integrating complementary information from different statistical parameters and eliminating the systematic bias of a single formula across different lithologies and roughness ranges. This results in a JRC label with higher accuracy than any single empirical regression formula. The fused JRC label is used as the training target to train a JRC prediction model. During construction, only the statistical parameters of the target rock joint surface need to be obtained to quickly output high-precision JRC prediction values ​​without destructive sampling. Finally, by comparing the absolute deviation between the current JRC prediction value and historical JRC prediction values, it is determined whether the joint surface roughness has changed significantly. When the deviation exceeds a preset threshold, a new construction decision is automatically generated; otherwise, the current decision is maintained. This avoids frequent and invalid decision updates, ensuring construction stability and enabling timely response to geological changes. The high-precision JRC prediction value provides a reliable basis for construction decisions, thereby improving construction efficiency and safety. Attached Figure Description

[0047] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0048] Figure 1 A flowchart illustrating a construction decision-making method based on rock joint roughness coefficient, provided for an embodiment of this application;

[0049] Figure 2This is a schematic diagram of the structure of the JRC prediction model provided in the embodiments of this application;

[0050] Figure 3 This is a schematic diagram of a construction decision-making device based on the rock joint roughness coefficient, provided as an embodiment of this application. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0052] Please refer to Figure 1 This is a flowchart illustrating a construction decision-making method based on rock joint roughness coefficient, provided in an embodiment of this application. The following is a further explanation. Figure 1 The construction decision-making method based on the rock joint roughness coefficient is introduced.

[0053] S1. Obtain the first dataset, and substitute the statistical parameters of each rock joint surface in the first dataset into various types of empirical regression formulas to obtain the estimated values ​​of multiple joint roughness coefficients (JRC) for each rock joint surface.

[0054] In the specific implementation process, the first dataset includes a large number of statistical parameters of rock joint surfaces. The statistical parameters of each rock joint surface can be obtained in the following ways:

[0055] The first method, on-site data acquisition and calculation: Three-dimensional laser scanning or photogrammetry technology is used to obtain profile data of rock joint surfaces, and statistical parameters are calculated based on the profile data.

[0056] The second method, using existing data: directly adopt statistical parameters of rock joint surfaces from publicly available literature, databases, or historical engineering projects.

[0057] This application does not limit the source of the statistical parameters, as long as the statistical parameters of each rock joint surface can be obtained. The two methods can be used individually or in combination.

[0058] Statistical parameters are mathematical indicators used to quantitatively describe the surface morphological characteristics of rock joint surfaces. They are obtained through mathematical calculations on the cross-sectional data of rock joint surfaces. There can be multiple statistical parameters, which are described below.

[0059] (1) Average relative height :

[0060] The average relative height is equal to the ratio of the average height of the rock joint surface to the length of the projected profile, representing the average amplitude characteristic along the discontinuity profile. The specific calculation formula is as follows:

[0061]

[0062] in, The projected length of the discontinuous section along the x-axis; The average fluctuation height of the discontinuous profile is calculated using the following formula:

[0063]

[0064] Where N is the number of sampling points on each discontinuous profile; For the first discontinuous section The y-coordinates of the points.

[0065] (2) Standard deviation of height :

[0066] The standard deviation of height represents the height distribution characteristics of the roughness elevation angle of a discontinuous surface profile. The specific calculation formula is as follows:

[0067]

[0068] (3) Mean angle of inclination :

[0069] The mean dip angle represents the characteristic of the mean dip angle of the profile. The specific calculation formula is as follows:

[0070]

[0071] in, For the first discontinuous section The y-coordinates of the points; arctan() represents the arctangent function; For the first discontinuous section points coordinate; For the first discontinuous section points coordinate.

[0072] (4) Maximum relative height :

[0073] The maximum relative height represents the ratio of the maximum profile height to the projected profile length, indicating the maximum amplitude characteristic of the profile. The specific calculation formula is as follows:

[0074]

[0075] in, The maximum value of the y-coordinate; It is the minimum value of the y-coordinate.

[0076] (5) Standard deviation of tilt angle :

[0077] The standard deviation of the dip angle represents the distribution characteristics of the roughness dip angle of a discontinuous surface profile. The specific calculation formula is as follows:

[0078]

[0079] (6) Roughness profile index :

[0080] The surface roughness index is equal to the ratio of the actual length of the discontinuous surface trajectory to its projected length. The specific calculation formula is as follows:

[0081]

[0082] in, For the first discontinuous section points coordinate; For the first discontinuous section points coordinate;

[0083] (7) Profile structure function :

[0084] The profile structure function represents the variation characteristics of the texture of discontinuous profiles. The specific calculation formula is as follows:

[0085]

[0086] (8) Root mean square of cross-section :

[0087] The root mean square (RMS) of the profile represents the small-angle characteristic of a discontinuous section. The specific calculation formula is as follows:

[0088]

[0089] In the embodiments of this application, the above eight statistical parameters comprehensively characterize the surface morphology features of rock joint surfaces from different dimensions. By combining these eight parameters, the roughness information of joint surfaces can be comprehensively and systematically characterized, providing rich and complete input features for the subsequent training of JRC prediction models, thereby improving the accuracy and robustness of JRC prediction.

[0090] After obtaining the first dataset, the statistical parameters of each rock joint surface in the dataset were substituted into various types of empirical regression formulas to obtain multiple JRC estimates for each rock joint surface. The empirical regression formulas are mathematical expressions obtained by fitting a large amount of rock joint surface sample data, reflecting the mapping relationship between statistical parameters and JRC. Previous studies, based on a large amount of measured data, established these formulas through regression analysis, enabling the rapid acquisition of JRC estimates simply by inputting statistical parameters.

[0091] In one possible embodiment, each type of empirical regression formula is used to reflect the mapping relationship between one of the statistical parameters and JRC; before performing S1, the method further includes: when any type of empirical regression formula includes multiple candidate formulas, the formula with the highest fitting accuracy is selected as the empirical regression formula of that type; the fitting accuracy is the goodness of fit index between the JRC estimate calculated based on its validation data when each candidate formula is proposed and the JRC measured value.

[0092] In the specific implementation process, the coefficient of determination (COP) can be used. The accuracy of the fit is measured by mean squared error (MSE), mean absolute error (MAE), or mean absolute percentage error (MAPE). The closer the value is to 1, the higher the fitting accuracy of the formula; the closer MSE, MAE, and MAPE are to 0, the higher the fitting accuracy of the formula.

[0093] Because different researchers may propose different empirical regression formulas, even those based on the same statistical parameter (such as...). Formulas from different sources also differ in form, coefficients, and fitting accuracy. A comparison of the fitting accuracy of some empirical regression formulas proposed by previous researchers is shown in Table 1.

[0094] Table 1

[0095]

[0096] It should be noted that the empirical regression formulas in Table 1 are all existing formulas obtained from published literature, and their coefficients of determination (COPs) are... The mean squared error (MSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) are calculated based on their validation data. The embodiments of this application can directly use these existing formulas without recalibration or validation.

[0097] When there are multiple candidate formulas for a certain type, such as those in Table 1... There are three formulas available; you can choose one. This serves as an empirical regression formula for this type of problem.

[0098] In this embodiment, there is no need for trial and error or manual judgment. The empirical regression formula with the highest fitting accuracy, which has been verified in published literature, is directly adopted. Recalibration or verification is unnecessary, simplifying the implementation process, reducing the difficulty of implementation, and fully utilizing existing research results, thus improving the feasibility and reliability of the technical solution. Furthermore, if new formulas with even higher fitting accuracy emerge in the future, this embodiment can also adopt these new formulas, demonstrating good scalability and foresight.

[0099] S2. The multiple JRC estimates of each rock joint surface in the first dataset are fused to obtain the JRC labels of each rock joint surface.

[0100] In one possible embodiment, the median or average of multiple JRC estimates for each rock joint surface in the first dataset can be selected as the JRC label for each rock joint surface.

[0101] In another possible embodiment, considering the differences in fitting accuracy among different empirical regression formulas, a weighted average method is used for fusion processing. The specific steps are as follows:

[0102] Obtain the second dataset; assign weight coefficients to each empirical regression formula based on the prediction error of each empirical regression formula on the second dataset; based on the weight coefficients, perform a weighted average of multiple JRC estimates for each rock joint surface in the first dataset to obtain the JRC label for each rock joint surface in the first dataset.

[0103] The second dataset includes statistical parameters of multiple rock joint surfaces and corresponding measured JRC values. The statistical parameters are calculated by acquiring profile data through three-dimensional laser scanning, while the measured JRC values ​​are obtained directly through indoor shear experiments.

[0104] A single empirical regression formula uses only a single statistical parameter, which introduces systematic bias. In this embodiment, the prediction results of multiple formulas are fused by weighted averaging, which effectively offsets the biases of each formula, making the accuracy of the fused JRC label better than any single formula. Furthermore, the weights are dynamically assigned based on the actual prediction errors of each formula on the second dataset, achieving data-driven adaptive fusion and avoiding biases caused by subjective manual assignment.

[0105] In one possible embodiment, the step of assigning weight coefficients to each empirical regression formula based on the prediction error of each empirical regression formula on the second dataset includes:

[0106] For each type of empirical regression formula, the statistical parameters of each rock joint surface in the second dataset are substituted into the empirical regression formula to obtain the JRC estimate of each rock joint surface, as well as the statistical error between the JRC estimate of each rock joint surface and the corresponding measured JRC value. The prediction error of the empirical regression formula is then calculated. Based on the prediction error of each empirical regression formula, a weight coefficient is assigned to each empirical regression formula. The empirical regression formula with the smaller prediction error has a larger weight coefficient.

[0107] In the specific implementation process, for the m-th type of empirical regression formula, each rock joint surface sample in the second dataset is iterated through, and the following processing is performed: First, the statistical parameters of the k-th rock joint surface are substituted into the empirical regression formula to calculate the JRC estimate. Then, the statistical error between the estimated value of the k-th rock joint surface and the measured JRC value is calculated as the prediction error of the m-th type of empirical regression formula. The statistical error can be, for example, mean absolute error, root mean square error, or sum of squared errors. Next, the reciprocal of the prediction error of each empirical regression formula is normalized and used as the weight coefficient of each empirical regression formula. The normalization formula is as follows:

[0108]

[0109] in, For the first The prediction error of various types of empirical regression formulas; Let be the prediction error of the empirical regression formula for the m-th type; This represents the total number of empirical regression formulas; denoted as the weight coefficient of the empirical regression formula for the m-th type.

[0110] Finally, for each rock joint surface in the first dataset, the JRC estimates obtained through each empirical regression formula are weighted and averaged according to the above weight coefficients to obtain the JRC label of the rock joint surface.

[0111] In this embodiment, the formula with smaller prediction error has a larger reciprocal and a larger weight coefficient after normalization, ensuring that the formula with high fitting accuracy dominates the weighted average. Conversely, the formula with larger prediction error has a smaller reciprocal and a smaller weight coefficient, automatically reducing the impact of unreliable formulas on the fusion result. Through this adaptive weight allocation strategy, the prediction results of multiple formulas can be objectively and accurately fused without manual intervention, thereby ensuring the accuracy of the JRC label.

[0112] It should be noted that although the shearing experiment is costly and time-consuming, this embodiment only requires a small number (e.g., 20-50) of shearing experiment samples to construct the second dataset for determining the weight coefficients of each empirical regression formula. The JRC labels for the remaining large number of training samples (e.g., hundreds of samples in the first dataset) can be generated through weighted averaging, eliminating the need for shearing experiments on each sample individually. Compared to existing technologies that require shearing experiments for every training sample, this embodiment significantly reduces data acquisition costs and time, while ensuring the accuracy of the JRC labels by fusing the prediction results of multiple empirical regression formulas through weighted averaging.

[0113] S3. Using the statistical parameters of each rock joint surface and the corresponding JRC labels in the first dataset as training data, the model is trained to obtain the JRC prediction model.

[0114] In the specific implementation process, the statistical parameters of each rock joint surface in the first dataset are used as input features, and their corresponding JRC labels are used as training targets to train the deep learning model, thereby obtaining the JRC prediction model.

[0115] In one possible embodiment, please refer to Figure 2 This is a schematic diagram of the JRC prediction model provided in an embodiment of this application. The model includes an input layer, a squeezing and excitation network module, a convolutional layer, a fully connected layer, and an output layer connected in sequence.

[0116] The input layer is used to receive various statistical parameters;

[0117] The squeeze and excitation network module is used to capture the intrinsic correlation between various statistical parameters and assign importance weights to obtain weighted statistical parameters;

[0118] Convolutional layers are used to extract deep features from weighted statistical parameters to obtain feature maps;

[0119] Fully connected layers are used to perform non-linear combinations and dimensional transformations on feature maps to obtain JRC prediction values;

[0120] The output layer is used to output JRC prediction values.

[0121] The squeezing and incentive network module includes a squeezing layer, an incentive layer, and a scaling layer. The squeezing layer is used to perform average pooling on each statistical parameter and convert it into a scalar value that represents the importance of each statistical parameter. The incentive layer is used to adaptively assign weights to each statistical parameter through a gating mechanism. The scaling layer is used to multiply the weights with the corresponding statistical parameters and output the weighted statistical parameters.

[0122] S4. During construction, obtain the statistical parameters of the target rock joint surface, input them into the JRC prediction model, and output the JRC prediction value of the target rock joint surface.

[0123] During tunnel boring machine construction, three-dimensional laser scanning or photogrammetry technology can be used to obtain cross-sectional data of the target rock joint surface. Based on the cross-sectional data, eight statistical parameters of the target rock joint surface can be calculated. , , , , , , Then, these eight statistical parameters are input into the trained JRC prediction model. After forward propagation, the model outputs the JRC prediction value of the target rock joint surface.

[0124] S5. When the absolute deviation between the JRC prediction value of the target rock joint surface and the historical JRC prediction value is greater than the preset threshold, a new construction decision is generated based on the JRC prediction value of the target rock joint surface; otherwise, the current construction decision is maintained.

[0125] In the actual implementation process, after each new JRC prediction value is obtained, the following judgment process is executed:

[0126] If this is the first JRC prediction, the construction decision will be generated directly based on the JRC prediction value, and the JRC prediction value will be recorded as a historical JRC prediction value.

[0127] If this is not the first JRC forecast, then calculate the absolute deviation between the current JRC forecast and the historical JRC forecast. ,Compare The relationship between the magnitude of the value and the preset threshold T. If... This indicates that the joint surface roughness does not change significantly, and the current construction decision remains applicable; therefore, the current construction decision should be maintained. This indicates a significant change in the roughness of the joint surface, necessitating an update to the construction decision. A new construction decision should be generated based on the current JRC prediction value, and the current JRC prediction value should be updated to the historical JRC prediction value for the next comparison.

[0128] Construction decisions specifically involve the combination of tunnel boring machine (TBM) parameters, including thrust, torque, penetration depth, and cutterhead speed. New or current construction decisions are sent to the TBM's control system, which either executes them automatically or prompts the operator for manual adjustments.

[0129] Furthermore, there are several ways to generate construction decisions based on the JRC prediction values ​​of the target rock joint surfaces, which will be introduced below.

[0130] The first method is the empirical formula method: Substitute the JRC prediction value into the known empirical formula in the field to calculate the combination of tunneling parameters, and use the combination of tunneling parameters as the new construction decision.

[0131] The second method is the mapping table method: based on the numerical range of the JRC predicted value, the preset decision mapping table is queried to obtain the combination of tunneling parameter values, and the combination of tunneling parameter values ​​is used as the new construction decision.

[0132] The decision mapping table is established based on engineering experience, historical construction data or expert knowledge, and divides the JRC value into several continuous numerical intervals. Each JRC numerical interval corresponds to a set of tunneling parameter value intervals, including the value intervals of thrust, torque, penetration, and cutterhead speed.

[0133] After obtaining the JRC prediction value of the target rock joint surface, follow these steps to query the mapping table: first determine which range the JRC prediction value falls into, then read the range of tunneling parameters corresponding to that range from the mapping table, and finally, within the range of tunneling parameters, choose a conservative strategy (lower limit), an aggressive strategy (upper limit), or an intermediate strategy (middle value) according to the project requirements.

[0134] In the embodiments of this application, the empirical formula method and the mapping table method have low computational load, can quickly obtain recommended values ​​of tunneling parameters, and complete decisions in milliseconds, meeting the real-time decision-making needs of the construction site.

[0135] The third method is the multi-objective optimization method: the objective function is constructed with the optimization objectives of maximizing tunneling speed, minimizing tool wear rate, and minimizing tunneling specific energy; based on the JRC prediction value of the target rock joint surface and other rock mass parameters, the optimization algorithm is used to search within the feasible domain of the tunneling parameters to find the optimal combination of tunneling parameter values ​​that makes the objective function optimal; the optimal combination of tunneling parameter values ​​is used as the new construction decision.

[0136] In the specific implementation process, the feasible range of tunneling parameters is limited by the equipment performance, including: the thrust does not exceed the maximum thrust of the equipment, the torque does not exceed the maximum torque of the equipment, the penetration depth is within the preset range, and the cutterhead speed is within the preset range.

[0137] The objective function is as follows:

[0138]

[0139] Where F is the comprehensive evaluation value of various construction effects; v is the tunneling speed. This refers to the maximum propulsion speed of the equipment. For tool wear rate, SE represents the maximum allowable wear rate of the cutting tool; SE is the tunneling specific energy. This is the preset maximum tunneling specific energy reference value; , and These are the weighting coefficients for each construction effect.

[0140] In this embodiment, the collaborative optimization of three conflicting objectives—tunneling speed, tool wear, and tunneling energy—is achieved. High-precision JRC predictions and other rock mass parameters are fully utilized. Global search using optimization algorithms (genetic algorithms, ant colony algorithms, etc.) avoids the limitations of human experience. The output results can be directly executed, realizing an automated closed loop from prediction to decision-making, and improving the scientificity and accuracy of construction decisions.

[0141] In one possible embodiment, the optimization algorithm is a genetic algorithm, and the optimization process is as follows:

[0142] Initialize the population by randomly generating N combinations of tunneling parameter values ​​as individuals; where N is the preset population size; each individual includes values ​​for thrust, torque, penetration depth, and cutterhead rotation speed; calculate the fitness value of each individual based on the JRC prediction value of the target rock joint surface and other rock mass parameters; where the fitness value is the value of the objective function; select excellent individuals based on their fitness values, perform crossover and mutation operations to generate a new generation of population; repeat the iteration until the preset maximum number of iterations is reached or the fitness value converges, and select the individual with the highest fitness from the final population as the optimal combination of tunneling parameter values.

[0143] In this embodiment, a genetic algorithm is used to seek the optimal solution. Through operations such as population initialization, selection, crossover, and mutation, a global search can be performed within the feasible region of the tunneling parameters, avoiding getting trapped in local optima and obtaining a better combination of tunneling parameters. Simultaneously, by setting parameters such as population size, crossover probability, and mutation probability, a balance can be achieved between search efficiency and solution quality, adapting to the real-time decision-making needs of the construction site.

[0144] Based on the same inventive concept, please refer to Figure 3 This application also provides a construction decision-making device based on the rock joint roughness coefficient, the device comprising:

[0145] The JRC tagging module is used to acquire the first dataset and substitute the statistical parameters of each rock joint surface in the first dataset into various types of empirical regression formulas to obtain multiple joint roughness coefficient JRC estimates for each rock joint surface. The statistical parameters are used to quantitatively describe the surface morphology characteristics of the rock joint surface. The multiple JRC estimates of each rock joint surface in the first dataset are fused to obtain the JRC labels for each rock joint surface.

[0146] The model training module is used to train the model using the statistical parameters of each rock joint surface and the corresponding JRC labels in the first dataset as training data to obtain the JRC prediction model.

[0147] The JRC prediction module is used to obtain statistical parameters of the target rock joint surface during construction, input them into the JRC prediction model, and output the JRC prediction value of the target rock joint surface.

[0148] The construction decision module is used to generate a new construction decision based on the JRC prediction value of the target rock joint surface when the absolute deviation between the JRC prediction value of the target rock joint surface and the historical JRC prediction value is greater than a preset threshold; otherwise, the current construction decision is maintained.

[0149] It should be noted that each module in the construction decision-making device based on the rock joint roughness coefficient in this embodiment corresponds one-to-one with each step in the construction decision-making method based on the rock joint roughness coefficient in the aforementioned embodiment. Therefore, the specific implementation of this embodiment can refer to the implementation of the aforementioned construction decision-making method based on the rock joint roughness coefficient, and will not be repeated here.

[0150] Based on the same inventive concept, this application also provides a computer device, which includes a processor, a memory, and a computer program stored in the memory. The computer program is executed by the processor to implement the aforementioned construction decision method based on the rock joint roughness coefficient.

[0151] Based on the same inventive concept, this application also provides a computer storage medium storing a computer program, which is executed by a processor to implement the aforementioned construction decision-making method based on rock joint roughness coefficient.

[0152] In some embodiments, the computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; or it may be a device including one or any combination of the above-mentioned memories. The computer may be a variety of computing devices, including smart terminals and servers.

[0153] In some embodiments, executable instructions may take the form of a program, software, software module, script, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a standalone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0154] As an example, executable instructions may, but do not necessarily, correspond to files in the file system. They may be stored as part of a file that holds other programs or data, for example, in one or more scripts in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple collaborative files (e.g., a file that stores one or more modules, subroutines, or code sections).

[0155] As an example, executable instructions can be deployed to execute on a single computing device, or on multiple computing devices located in one location, or on multiple computing devices distributed across multiple locations and interconnected via a communication network.

[0156] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0157] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0158] The above specific embodiments further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A construction decision-making method based on rock joint roughness coefficient, characterized in that, include: A first dataset is obtained, and the statistical parameters of each rock joint surface in the first dataset are substituted into various types of empirical regression formulas to obtain multiple joint roughness coefficient JRC estimates for each rock joint surface; wherein, the statistical parameters are used to quantitatively describe the surface morphology characteristics of the rock joint surface. Multiple JRC estimates for each rock joint surface in the first dataset are fused to obtain the JRC label for each rock joint surface; The model is trained using the statistical parameters of each rock joint surface and the corresponding JRC labels in the first dataset as training data to obtain the JRC prediction model. During construction, statistical parameters of the target rock joint surface are obtained and input into the JRC prediction model, and the JRC prediction value of the target rock joint surface is output. When the absolute deviation between the JRC prediction value of the target rock joint surface and the historical JRC prediction value is greater than a preset threshold, a new construction decision is generated based on the JRC prediction value of the target rock joint surface; otherwise, the current construction decision is maintained. The process of generating new construction decisions based on the JRC prediction values ​​of the target rock joint surfaces includes: An objective function is constructed with the optimization objectives of maximizing tunneling speed, minimizing cutter wear rate, and minimizing tunneling specific energy. Based on the JRC prediction value of the target rock joint surface and other rock mass parameters, an optimization algorithm is used to search within the feasible region of the tunneling parameters to find the optimal combination of tunneling parameter values ​​that makes the objective function optimal. The tunneling parameters include thrust, torque, penetration depth, and cutterhead rotation speed. The optimal combination of tunneling parameter values ​​is used as the construction decision for the target rock joint surface.

2. The construction decision-making method based on rock joint roughness coefficient according to claim 1, characterized in that, The process of fusing multiple JRC estimates for each rock joint surface in the first dataset to obtain the JRC label for each rock joint surface includes: Obtain the second dataset; the second dataset includes statistical parameters of multiple rock joint surfaces and corresponding JRC measured values; Based on the prediction error of each empirical regression formula on the second dataset, assign weight coefficients to each empirical regression formula. Based on the weighting coefficients, a weighted average is performed on multiple JRC estimates of each rock joint surface in the first dataset to obtain the JRC label of each rock joint surface in the first dataset.

3. The construction decision-making method based on rock joint roughness coefficient according to claim 2, characterized in that, The step of assigning weight coefficients to each empirical regression formula based on its prediction error on the second dataset includes: For each type of empirical regression formula, the statistical parameters of each rock joint surface in the second dataset are substituted into the empirical regression formula to obtain the JRC estimate of each rock joint surface, as well as the statistical error between the JRC estimate of each rock joint surface and the corresponding measured JRC value, and the prediction error of the empirical regression formula is calculated. Based on the prediction error of each empirical regression formula, a weight coefficient is assigned to each empirical regression formula; among them, the empirical regression formula with the smaller prediction error has a larger weight coefficient.

4. The construction decision-making method based on rock joint roughness coefficient according to claim 3, characterized in that, The process of assigning weight coefficients to each empirical regression formula based on its prediction error includes: The reciprocal of the prediction error of each empirical regression formula is normalized and used as the weight coefficient of each empirical regression formula.

5. The construction decision-making method based on rock joint roughness coefficient according to claim 1, characterized in that, The JRC prediction value based on the target rock joint surface and other rock mass parameters is used to search within the feasible region of the tunneling parameters using an optimization algorithm to find the optimal combination of tunneling parameter values ​​that makes the objective function optimal, including: Initialize the population by randomly generating N combinations of tunneling parameter values ​​as individuals; where N is the preset population size; each individual includes values ​​for thrust, torque, penetration depth, and cutterhead rotation speed; Based on the JRC prediction value of the target rock joint surface and other rock mass parameters, the fitness value of each individual is calculated; wherein, the fitness value is the value of the objective function; Select superior individuals based on their fitness values, perform crossover and mutation operations, and generate a new generation of population. Repeat the iteration until the preset maximum number of iterations or the fitness value converges. Select the individual with the highest fitness from the final population as the optimal combination of tunneling parameters.

6. The construction decision-making method based on rock joint roughness coefficient according to claim 1, characterized in that, The objective function is as follows: ； Where F is the comprehensive evaluation value of various construction effects; v is the tunneling speed. WR represents the maximum feed rate of the equipment; WR represents the tool wear rate. SE represents the maximum allowable wear rate of the cutting tool; SE is the tunneling specific energy. This is the preset maximum tunneling specific energy reference value; , and These are the weighting coefficients for each construction effect.

7. The construction decision-making method based on rock joint roughness coefficient according to claim 1, characterized in that, The statistical parameters include: average relative height, height standard deviation, average tilt angle, maximum relative height, tilt angle standard deviation, roughness profile index, profile structure function, and profile root mean square.

8. The construction decision-making method based on rock joint roughness coefficient according to claim 7, characterized in that, Each type of empirical regression formula is used to reflect the mapping relationship between one of the statistical parameters and the JRC; Before substituting the statistical parameters of each rock joint surface in the first dataset into various types of empirical regression formulas, the method further includes: When any type of empirical regression formula includes multiple candidate formulas, the formula with the highest fitting accuracy is selected as the empirical regression formula of that type; wherein, the fitting accuracy is the goodness of fit index between the JRC estimate calculated based on its validation data and the JRC measured value when each candidate formula is proposed.

9. A construction decision-making device based on the rock joint roughness coefficient, characterized in that, include: The JRC tagging module is used to acquire a first dataset and substitute the statistical parameters of each rock joint surface in the first dataset into various types of empirical regression formulas to obtain multiple joint roughness coefficient JRC estimates for each rock joint surface; the multiple JRC estimates for each rock joint surface in the first dataset are fused to obtain the JRC label for each rock joint surface; wherein, the statistical parameters are used to quantitatively describe the surface morphology characteristics of the rock joint surface. The model training module is used to train the model using the statistical parameters of each rock joint surface in the first dataset and the corresponding JRC labels as training data to obtain the JRC prediction model. The JRC prediction module is used to obtain statistical parameters of the target rock joint surface during construction, input them into the JRC prediction model, and output the JRC prediction value of the target rock joint surface. The construction decision module is used to generate a new construction decision based on the JRC prediction value of the target rock joint surface when the absolute deviation between the JRC prediction value of the target rock joint surface and the historical JRC prediction value is greater than a preset threshold; otherwise, the current construction decision is maintained. The process of generating new construction decisions based on the JRC prediction values ​​of the target rock joint surfaces includes: An objective function is constructed with the optimization objectives of maximizing tunneling speed, minimizing cutter wear rate, and minimizing tunneling specific energy. Based on the JRC prediction value of the target rock joint surface and other rock mass parameters, an optimization algorithm is used to search within the feasible region of the tunneling parameters to find the optimal combination of tunneling parameter values ​​that makes the objective function optimal. The tunneling parameters include thrust, torque, penetration depth, and cutterhead rotation speed. The optimal combination of tunneling parameter values ​​is used as the construction decision for the target rock joint surface.

Images