Geological risk-torque prediction driving stuck machine risk evaluation method and system

Abstract

The invention belongs to the technical field related to tunnels, and provides a geological risk-torque prediction driven jamming risk evaluation method and system in order to solve the problems that risk prediction is inaccurate and early warning lags and the like due to the fact that an existing monitoring method is difficult to achieve collaborative perception of geological inducements and equipment responses in the tunneling process, and the geological risk-torque prediction driven jamming risk evaluation method and system are provided. The method is realized by constructing a standard cloud model, and a dynamic adjustment mechanism based on real-time data fluctuation and risk sensitivity is introduced, so that the cloud model not only can more accurately respond to complex and dynamically changing geological conditions, but also can keep efficient and accurate risk assessment capability during geological sudden change; by adopting a geological risk prediction and torque prediction dual-drive prediction method, collaborative perception of geological inducements and equipment response is realized, the cause and evolution process of the jamming risk are comprehensively and dynamically mastered, and the prediction accuracy is ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field related to tunnel excavation, and in particular relates to a geological risk-torque prediction-driven card machine risk assessment method and system. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Tunnel boring machines (TBMs), widely used in the construction of long and large tunnels for highways, railways, and water conservancy projects, offer advantages such as high efficiency, safety, and environmental friendliness. However, with the expansion of construction scale and increasing geological complexity, TBMs often face the risk of machine jamming during excavation, especially in complex strata and fault zones.

[0004] Existing technologies for assessing the risk of tunnel boring machine jams generally suffer from single models and delayed responses. On the one hand, risk assessment models based on geological parameters typically rely on static geological exploration data, making it difficult to promptly reflect dynamic risk factors such as sudden changes in strata or the evolution of weak surrounding rock during tunneling. Furthermore, they lack quantitative modeling of the uncertainty and volatility within geological data, resulting in a crude classification of risk levels and insufficient reliability of prediction results. On the other hand, while monitoring methods based on key control parameters such as cutterhead torque can achieve real-time perception of equipment status, they are still essentially "results-driven" and typically only issue warnings after abnormal fluctuations in equipment parameters. This makes it difficult to proactively identify potential jam risks caused by geological factors, resulting in significant warning lags. Furthermore, current prevention and control strategies rely heavily on the experience of on-site personnel for parameter adjustment and emergency response, lacking a systematic, standardized intelligent decision-making support platform. This leads to delays in risk identification and response measures, which can easily lead to serious equipment damage and project delays.

[0005] In summary, existing geological parameter risk assessments usually rely on static geological exploration data and empirical models, which have problems such as insufficient prediction accuracy, poor real-time performance, and ineffective response to emergencies. During the excavation process, existing monitoring methods make it difficult to achieve coordinated perception of geological inducements and equipment responses, and are unable to comprehensively and dynamically grasp the causes and evolution of machine jam risks, making it difficult to identify machine jam risks caused by latent geological inducements in advance, resulting in delayed early warning. Summary of the Invention

[0006] In order to overcome the shortcomings of the above-mentioned existing technologies, the present invention provides a geological risk-torque prediction-driven card machine risk assessment method and system, which adopts a prediction method driven by geological risk prediction and torque prediction to achieve coordinated perception of geological inducements and equipment responses, comprehensively and dynamically grasp the causes and evolution process of card machine risks, and ensure prediction accuracy.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a geological risk-torque prediction-driven card machine risk assessment method, comprising: Obtain geological parameters, geophysical parameters and main control parameters of the tunneling section; Based on adverse geology and its corresponding geological parameters and geophysical parameters, initial weights are assigned to risk indicators of the geological parameters and geophysical parameters, and the geology is divided into multiple risk levels to construct a standard cloud model. Based on the standard cloud model, the weights of the geological parameters and / or geophysical parameters are dynamically adjusted according to the fluctuations and risk sensitivity of the geological parameters and / or geophysical parameters during the tunneling process, and the predicted results of all risk indicators are weighted and integrated to obtain the geological risk level of the tunneling section. Based on the geological parameters, the geophysical parameters and the main control parameters of the tunnel boring machine, using the trained torque prediction model, obtaining the tunnel boring machine operation risk level; Based on the machine jam risk prediction model constructed by the gradient boosting tree, the geological risk level and the tunnel boring machine operation risk level are integrated to obtain the final machine jam risk evaluation result.

[0008] In a second aspect, the present invention provides a roadheader stuck risk assessment system, comprising: The data collection module is configured to: obtain geological parameters, geophysical parameters and main control parameters of the tunneling section; The geological risk assessment module is configured to: assign initial weights to risk indicators of the geological parameters and geophysical parameters based on adverse geology and their corresponding geological parameters and geophysical parameters, divide the geology into multiple risk levels, and construct a standard cloud model; based on the standard cloud model, dynamically adjust the weights of the geological parameters and / or geophysical parameters according to the fluctuations and risk sensitivity of the geological parameters and / or geophysical parameters during the tunneling process, and weightedly integrate the predicted results of all risk indicators to obtain the geological risk level of the tunneling section; The torque prediction module is configured to: obtain an operation risk level of the tunnel boring machine using a trained torque prediction model based on the geological parameters, the geophysical parameters and the tunnel boring machine main control parameters; The dual-drive evaluation module is configured to: fuse the geological risk level and the tunnel boring machine operation risk level based on the machine jam risk prediction model constructed by the gradient boosting tree to obtain a machine jam risk evaluation result.

[0009] One or more of the above technical solutions have the following beneficial effects: In the present invention, geological risk assessment is achieved by constructing a comprehensive cloud model. The cloud model has strong advantages in describing the uncertainty, fuzziness and evolution trend among multi-source data, and has the ability to integrate qualitative cognition with quantitative analysis. It can accurately depict the randomness and fuzzy boundaries of risk distribution in complex geological environments with strong parameter volatility and insufficient information integrity, and improve the stability and adaptability of risk level judgment. A prediction method driven by both geological risk prediction and torque prediction is adopted, and a fusion analysis is performed using the gradient boosting tree algorithm to obtain the final machine jam risk assessment result. The gradient boosting tree algorithm is used to automatically learn the nonlinear relationship between the two and the actual risk level, realize the coordinated perception of geological inducements and equipment responses, and comprehensively and dynamically grasp the causes and evolution processes of machine jam risks, while ensuring the accuracy of prediction and achieving the quantification of model results.

[0010] The present invention introduces a dynamic adjustment mechanism based on real-time data fluctuations and risk sensitivity, which enables the comprehensive cloud model to not only respond more accurately to complex and dynamically changing geological conditions, but also maintain efficient and accurate risk assessment capabilities in the event of geological mutations.

[0011] Advantages of additional aspects of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0013] Figure 1 This is a flow chart of the geological risk assessment process from indicator selection, weight adjustment to cloud droplet generation and comprehensive cloud model construction in Example 1 of the present invention; Figure 2 Schematic diagram of a deep learning network structure based on LSTM and attention mechanism and its specific implementation process in torque dynamic prediction in Example 1 of the present invention; Figure 3 Schematic diagram of the process of fusion analysis of geological risk assessment and torque prediction results through gradient boosting tree and the output process of comprehensive risk level in Example 1 of the present invention; Figure 4 This is a flow chart of generating excavation parameter adjustment plans or escape measures at different risk levels in the first embodiment of the present invention; Figure 5 This is a schematic diagram of a geological risk-torque prediction-driven card machine risk assessment system in Example 2 of the present invention; Figure 6 This is a schematic diagram of the risk classification of a card machine in accordance with an embodiment of the present invention. DETAILED DESCRIPTION

[0014] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0015] It should be noted that the terms used herein are for describing particular embodiments only and are not intended to limit the exemplary embodiments according to the present invention.

[0016] In the absence of conflict, the embodiments of the present invention and the features thereof may be combined with each other.

[0017] Example 1 This embodiment discloses a geological risk-torque prediction-driven card machine risk assessment method, including: Obtain geological parameters of the tunneling section and main control parameters of the tunnel boring machine; Based on adverse geology during the excavation process and the corresponding geological parameters and geophysical parameters, initial weights are assigned to the geological parameters and geophysical parameters, and the geology is divided into multiple risk levels to construct a comprehensive cloud model; based on the fluctuations and risk sensitivity of the geological parameters and / or geophysical parameters during the excavation process, the weights of the geological parameters and / or geophysical parameters are dynamically adjusted, and the geological risk level of the excavation section is obtained based on the updated comprehensive cloud model; Based on the geological parameters, the geophysical parameters and the main control parameters of the tunnel boring machine, using the trained torque prediction model, obtaining the tunnel boring machine operation risk level; Based on the machine jam risk prediction model constructed by gradient boosting tree, the geological risk level and the tunnel boring machine operation risk level are integrated to obtain the final machine jam risk assessment result.

[0018] This embodiment is achieved through the construction of a comprehensive cloud model. This model offers significant advantages in describing the uncertainty, ambiguity, and evolving trends across multi-source data. It possesses the ability to integrate qualitative cognition with quantitative analysis. This model can accurately depict the randomness and fuzzy boundaries of risk distribution in complex geological environments characterized by high parameter volatility and insufficient information integrity, thereby improving the stability and adaptability of risk level assessment. Furthermore, the present invention incorporates a dynamic adjustment mechanism based on real-time data fluctuations and risk sensitivity, enabling the comprehensive cloud model to not only more accurately respond to complex and dynamically changing geological conditions but also maintain efficient and accurate risk assessment capabilities during sudden geological changes. In this embodiment, a gradient boosting tree algorithm is used to perform a fusion analysis based on the predicted geological risk level and the roadheader operational risk level to obtain the final machine jam risk assessment result. The gradient boosting tree algorithm automatically learns the nonlinear relationship between the two and the actual risk level, ensuring both prediction accuracy and quantifiable and interpretable model results.

[0019] The following combination Figures 1-4 The geological risk-torque prediction-driven card machine risk assessment method proposed in this embodiment is described in detail: Step 1: Obtain the geological parameters, geophysical parameters and main control parameters of the tunneling section.

[0020] Specifically, geological parameters, geophysical parameters, and excavator control parameters related to the excavation process are acquired. Among these, three types of adverse geological conditions are primarily classified based on typical geological unit types: fracture zones, karst, and alteration zones. For fracture zones, the collected geological parameters include fault width, rock mass integrity coefficient (RQD), joint spacing, water seepage, and rock compressive strength. For karst geology, the collected geological parameters include the spatial distribution and volume of caves, fissure water pressure, permeability, distribution ratio of carbonate rocks, and water inflow. For alteration zone geology, the collected geological parameters cover key indicators such as changes in rock chemical composition, rock mechanical properties, expansibility index, degree of fracture development, and water seepage and inflow.

[0021] Geophysical parameters include resistivity, relaxation time, longitudinal wave velocity, polarizability, etc. Excavator main control parameters include key indicators such as TBM cutterhead diameter, cutterhead thickness, and cutterhead opening ratio.

[0022] Step 2: Based on the adverse geological types and their corresponding geological parameters and geophysical parameters during the excavation process, initial weights are assigned to the geological parameter and geophysical parameter risk indicators, and the geological types are divided into multiple risk levels to construct a standard cloud model. Based on the fluctuations and risk sensitivity of the geological parameters and / or geophysical parameters during the excavation process, the weights of the geological parameters and / or geophysical parameters are dynamically adjusted, and the real-time values ​​are substituted into the cloud generator to generate the cloud to be identified. The membership degree between the cloud to be identified and the standard cloud is calculated to obtain the geological risk level of the excavation section.

[0023] In this embodiment, a comprehensive evaluation of the collected geological parameters and geophysical parameters is performed, using a cloud model as the core evaluation tool. The cloud model has strong advantages in describing the uncertainty, fuzziness, and evolutionary trends among multi-source data. Compared with traditional weighted scoring methods or hierarchical analysis methods, the cloud model has the ability to integrate qualitative cognition with quantitative analysis. It can accurately depict the randomness and fuzzy boundaries of risk distribution in complex geological environments with strong parameter volatility and insufficient information integrity, thereby improving the stability and adaptability of risk level judgment. To further enhance the sensitivity of the cloud model, this embodiment introduces a dynamic weight adjustment mechanism based on the cloud model to achieve feedback correction of real-time monitoring data, thereby enhancing the response capability to geological evolution.

[0024] In practical applications, we first construct a geological risk assessment index system based on the different risk characteristics of typical adverse geological types such as fracture zones, karst, and alteration zones. The geological risk assessment index system includes several key geological parameter indicators, such as rock integrity index , groundwater permeability index , the degree of development of weak interlayer etc., unified as ,in, Indicates the total number of geological parameters and geophysical parameters. In order to reasonably express the degree of influence of each parameter index on the risk assessment results, it is necessary to initially assign an initial weight to each parameter index based on expert experience or historical case data. , and ensure that the sum of weights satisfies:

[0025] In order to characterize the geological risk level, this embodiment constructs a standard cloud model based on cloud model theory. Each parameter indicator is divided into multiple risk level intervals based on actual engineering experience, generally divided into six levels: "stable", "weak risk", "medium risk", "higher risk", "high risk" and "extremely high risk". For each risk level's numerical range, its cloud model triplet parameter is calculated. The triplet parameter includes the expected value Ex, entropy En and excess entropy He. Among them, the expected value Ex represents the typical characteristic value of the risk level; the entropy En reflects the uncertainty within the risk level; and the excess entropy He further characterizes the fluctuation of the entropy itself, reflecting the random perturbation characteristics of the data.

[0026] The calculation formulas for the three are as follows:

[0027] in, and are the maximum and minimum values ​​corresponding to the risk level respectively; It is an empirical constant, and its value generally ranges from 0.1 to 0.5.

[0028] Then, the above parameters are used to construct a standard cloud model and a large number of cloud droplets are generated through the forward cloud generator. The specific process includes: For expectations, Is the variance, generate random entropy value , and then For expectations, Generate random variables for variance , and finally calculate the membership of the cloud droplet , used to describe the degree of match between the data point and the risk level.

[0029] Through large-scale cloud droplet sampling, a cloud map of standard grade can be formed, which can be used for subsequent cloud similarity judgment. The calculation formula is as follows:

[0030]

[0031]

[0032] During the actual excavation process, the monitoring equipment continuously collects the real-time values ​​of various parameters , Indicates the Parameters at time To enable the cloud model to have dynamic response capabilities, this embodiment introduces the "fluctuation analysis" and "sensitivity analysis" mechanisms in the risk identification stage to form a real-time adaptive adjustment system for weights. Fluctuation analysis is used to capture the magnitude of changes in indicators over time series and is defined as:

[0033] If the fluctuation range Exceeding the fluctuation value threshold When , it is considered that the parameter has a sudden change, which is a significant disturbance and its weight should be increased.

[0034] At the same time, sensitivity analysis is used to further evaluate the impact of each parameter index on the overall risk result. The risk assessment function is defined as In this embodiment, the risk assessment function It can be constructed in the form of linear or nonlinear combinations, such as linear weighted models, fuzzy inference functions or neural network outputs, and the specific form can be flexibly set according to the actual engineering scenario.

[0035] Rule No. The sensitivity of an indicator is defined as:

[0036] when This parameter is considered to be a highly sensitive indicator and its impact should be enhanced; is the sensitivity threshold.

[0037] Based on the above analysis, dynamic weight It can be updated according to the following formula:

[0038] in, Indicates time No. The weight of each indicator; Represents the weight adjustment factor, reflecting the impact of volatility and sensitivity; the denominator Used for normalization to ensure that the sum of all weights still equals 1.

[0039] Finally, after completing the identification cloud construction for each parameter indicator, a comprehensive identification cloud map is formed by weighted fusion of the cloud droplets of all parameter indicators. The comprehensive identification cloud map is compared with the cloud maps of each standard risk level, and the principle of maximum membership is used to determine the closest level.

[0040] The specific judgment criteria are as follows:

[0041]

[0042] in, Recognition Cloud and The degree of matching of the levels; Indicates the The cloud droplets are The degree of membership in the risk level standard cloud; Indicates the total number of cloud droplets; The risk level of the current excavation section is determined by the maximum membership principle of the final output.

[0043] Step 3: Based on the geophysical parameters, formation parameters and TBM main control parameters, the trained torque prediction model is used to obtain the TBM operation risk level.

[0044] In this embodiment, a torque prediction model is built based on a deep learning framework, incorporating a long short-term memory (LSTM) network, an attention mechanism, and a geological disturbance-aware gating mechanism. This model aims to accurately model and dynamically predict the changing trends of cutterhead torque during tunneling, enhancing the ability to proactively identify operational risks such as machine jams. This torque prediction model offers significant advantages in handling multi-source TBM input parameters, complex geological coupling effects, and sudden disturbance responses. It is widely applicable to intelligent decision-making support systems for shield tunneling in areas with complex geological conditions.

[0045] The torque prediction model input is multi-dimensional data collected in real time, including the equipment operating status and the surrounding geological environment. It covers main control parameters such as cutterhead speed, propulsion thrust, penetration, shield friction, hydraulic system pressure, etc., and integrates typical geological indicators such as surrounding rock strength (RQD, UCS), cave characteristics, fracture width, and combines geophysical parameters such as resistivity, relaxation time, polarizability, and longitudinal wave velocity.

[0046] After the data are collected, they are preprocessed by the data normalization method. The normalization calculation formula is:

[0047] in, Indicates the first feature vector in the original The value of the sample, for example The original values ​​of propulsion speed, current, etc. at each time point; and They represent the minimum and maximum values ​​of the feature in all samples respectively. Represents the normalized value, that is, Linear mapping to interval The value after .

[0048] All input samples are organized into multi-dimensional tensors in the form of sliding time windows, and the tensor dimensions are ,in, A tensor representing the input data, which is a three-dimensional array; Indicates that each element in the tensor is a real number; Indicates the batch size, represents the time step, Indicates the number of feature dimensions. The preprocessed data is input into the LSTM network to extract the memory and dependency structure of the input sequence in the time dimension.

[0049] The internal structure of an LSTM network consists of an input gate, a forget gate, an output gate, and a state update mechanism. The input gate updates the cell state and controls the input vectors entering the memory cell. The forget gate determines which information should be discarded or retained; its opening and closing determine whether the information can be passed to the next moment. The output gate determines the value of the next hidden state, which contains the previously input information. The update formula is as follows: Input Gate :

[0050] Forget Gate :

[0051] Output Gate :

[0052] Candidate status :

[0053] Cell status update :

[0054] Hide status updates :

[0055] in, Indicates the Input features of time steps; Indicates the The same applies to the hidden state corresponding to the time step is the hidden state of the previous time step; For the The memory unit of the time step is the cell state, is the memory unit of the previous step; are the weight matrices connecting the input and the current state, acting on ; are the weight matrices of input and past hidden states, acting on ; is the bias term; Represents the Sigmoid function; Represents the hyperbolic tangent function. The output of the LSTM network is the hidden state sequence , which is used to describe the impact of each time step feature on the target variable.

[0056] In order to further enhance the model's ability to identify and express key time segments, the attention mechanism is introduced to perform weighted aggregation on the hidden state sequence. The calculation of attention weights first constructs a scoring function to evaluate each hidden state. With global state The correlation between the global state Can be set to Or a trainable vector, the scoring function takes the following form:

[0057] Among them, the calculated results Indicates the The “importance” of each time step to the target; A weight matrix that maps hidden states to the attention space; A weight matrix that maps the global state to the attention space; represents the hyperbolic tangent function; The trainable weight vector is used to compress the nonlinear result into a scalar score. The superscript T indicates transpose.

[0058] Attention weight Through Softmax normalization, we get:

[0059] in, is the total number of time steps.

[0060] Then we get the weighted context vector :

[0061] Based on the traditional LSTM and attention structure, this embodiment further introduces a geological disturbance perception gating mechanism to achieve dynamic response adjustment to sudden geological anomalies. Based on the lithology change rate, fault density, seepage rate slope and other indicators, the disturbance event label is generated when the coefficient of variation of these disturbance indicators in the sample section exceeds the threshold value such as 3σ or the seepage rate change slope is greater than the critical value. With the above vector After splicing, the perturbation regulation factor is generated through the gating function , the expression is as follows:

[0062] in, is the regulation coefficient calculated by the perturbation gating mechanism, ranging from , reflecting whether the geology is stable. Represents vector concatenation operation, and are the gating layer weights and biases respectively. The final prediction output after perturbation adjustment is It is a weighted fusion of the current forecast value and the historical trend value:

[0063] in, Based on the context vector The predicted torque value, that is, the torque value predicted by LSTM and attention mechanism; This represents the sliding mean or weighted trend value of historical torque, providing a robust reference when disturbance signals are significant. By introducing a geological disturbance-aware gating mechanism, the torque prediction model can dynamically correct its prediction output in the face of sudden geological changes, effectively improving its sensitivity to abnormal fluctuations and its robust response.

[0064] During the training process of the torque prediction model, the mean square error is used as the loss function, and the optimization goal is:

[0065] in, is the sample size, and The optimizer uses the Adam algorithm, combined with the learning rate adjustment strategy and the EarlyStopping mechanism to ensure the efficiency and generalization ability of the training process. After training, the model is evaluated on the test set. Common indicators include mean square error (MSE), mean absolute error (MAE) and coefficient of determination ( ), which is used to quantify the prediction accuracy and stability.

[0066] Ultimately, the torque prediction model can be deployed in a real-time TBM tunneling control system. It continuously predicts the changing trends of cutterhead torque based on current input parameters and classifies risk levels based on set thresholds. When the predicted value is within a stable range, it is considered low risk. When the predicted value approaches the upper limit of abnormality or is accompanied by a significant increase in disturbance signals, it is considered medium-to-high risk, triggering an early warning mechanism or recommending adjustments to tunneling parameters. The overall model structure exhibits excellent generalization and field adaptability, making it particularly suitable for intelligent TBM tunneling control systems operating under complex geological conditions and multiple disturbances.

[0067] Step 4: Based on the machine jam risk prediction model constructed by the gradient boosting tree, the geological risk level and the tunnel boring machine operation risk level are integrated to obtain the final machine jam risk assessment result.

[0068] In this embodiment, a TBM jam risk assessment model based on dual-drive information fusion is constructed using the XGBoost algorithm. This model aims to couple geological risk with equipment torque response and output a unified TBM jam risk level. As the system's comprehensive discriminant core, the jam risk assessment model receives key feature information from the aforementioned cloud model and torque prediction model. Using machine learning, it automatically learns the nonlinear relationship between these two models and the actual TBM jam risk level. This ensures predictive accuracy while making the model's results quantifiable and interpretable.

[0069] In the process of building the card machine risk assessment model, various input features are first structured preprocessed. The input features mainly include comprehensive descriptive information of geological evaluation results and equipment operating status. Among them, the geological part covers the risk level output by the comprehensive cloud model and its corresponding representative indicators, such as fault width, water seepage, and expansion index; the torque part introduces characteristic parameters such as torque level, mean, fluctuation range, and penetration obtained by the torque prediction model.

[0070] To enhance the discriminative capabilities of the card machine risk assessment model, this example uses one-hot encoding for categorical features and normalizes continuous data, creating a unified data input format. Each set of samples corresponds to an actual construction scenario. Risk level supervision training is performed using historically annotated data, and the output labels are classified into three levels: low risk, medium risk, and high risk.

[0071] In the dual-drive information fusion card machine risk assessment model based on XGBoost constructed in this embodiment, the input features are composed of geological risk features and equipment operation features, which are unified to form a structured risk feature vector, specifically including: The geological feature subset from the cloud model includes geological risk levels and their corresponding key geological parameters, such as fault width, RQD value, water seepage, expansion index, joint spacing, and fissure water pressure; the equipment feature subset includes the equipment operation risk level output by the torque prediction model and its corresponding tunnel boring machine main control parameters, such as cutterhead torque prediction value, fluctuation amplitude, penetration rate, propulsion thrust, and shield friction resistance.

[0072] Among them, geological risk level and equipment operation risk level are used as classification features and represented by one-hot encoding; the remaining continuous variables are normalized and input into the fusion model, and finally a high-dimensional risk feature vector in a unified format is constructed as the input of the XGBoost model.

[0073] As an enhanced gradient decision tree method, the XGBoost algorithm has strong generalization capabilities for high-dimensional, heterogeneous data and is sensitive to changes in feature importance. During XGBoost model training, an iterative structure automatically fits the coupling between geological and torque characteristics, further optimizing the tree structure and node partitioning.

[0074] The dataset is divided into training, validation, and test sets. By adjusting hyperparameters such as learning rate, tree depth, and the number of weak classifiers, the XGBoost model can avoid overfitting while ensuring fitting performance and improve its ability to discriminate unknown samples. Its loss function consists of two parts: model prediction error and structural complexity:

[0075] in, Represents the set of all parameters of the XGBoost model; Indicates the samples ( ), represents the total number of samples; Indicates the The true label of each sample, i.e., the actual risk level; Indicates the The XGBoost model prediction value of samples; is a loss function, such as squared error or log loss, which measures the difference between the predicted value and the true value; Indicates the samples, , Indicates the total number of decision trees constructed; Indicates the A regression tree; Indicates the The complexity penalty of a tree, such as the number of leaf nodes and the square of the weight.

[0076] During the XGBoost model training process, hyperparameters such as learning rate, tree depth, and number of subtrees are adjusted through GridSearch and K-fold cross-validation methods, and the model performance is evaluated based on the training set and validation set.

[0077] To enhance the interpretability of the XGBoost model, this embodiment introduces a feature importance analysis mechanism to quantify the contribution of each input variable. In the card machine risk assessment model based on XGBoost's dual-driven information fusion, the quantification of variable contribution is not only used for posterior interpretation, but also directly participates in the optimization design of the XGBoost model structure and input space.

[0078] The feature importance analysis mechanism introduces three types of feature importance indicators as follows: The number of times the feature is used as a partition node in all trees; Indicates the decrease in the loss function caused by using this feature for each split; Indicates the weight and coverage of samples by this feature in the split node.

[0079] In view of the differences in physical properties between geological and equipment characteristics, a differentiated indicator priority strategy is adopted. For geological characteristics, Gain is preferentially used to measure their fine division capability; for torque characteristics, Cover is preferentially used to measure their global response capability; for characteristics with frequent structural involvement but large fluctuations, Weight and Gain are used together for comprehensive evaluation.

[0080] The important coefficient of variation (CV) is introduced to monitor feature stability:

[0081] in, Indicates the The set of Gain values ​​obtained by a feature in multiple rounds of training; Indicates the The standard deviation of the characteristic Gain value, used to measure fluctuation; No. The average of the Gain values ​​of the features is used to measure the contribution level. If a feature fluctuates too much in multiple consecutive rounds, it is judged as a "weak stability factor" and is eliminated from the current training round. If a feature has a high and stable contribution in multiple rounds, it will be given priority in node division.

[0082] To capture the cross-coupling effects between geological variables and torque variables, a joint contribution evaluation mechanism is introduced. If a combination of features significantly improves the objective function, it is automatically added as a cross term to participate in the tree construction. The joint contribution evaluation mechanism defines the joint importance function of the interactive feature combination:

[0083] in, Representation characteristics and The joint contribution of Indicates the geological features, Indicates the Torque characteristics. If the value exceeds the set threshold, a cross-feature item is automatically generated , and incorporate it into the candidate split feature set of the current round tree structure to participate in tree construction. This mechanism can effectively capture the nonlinear interaction effects between cross-domain variables and improve the model's ability to identify complex induced mechanisms.

[0084] After training is complete, XGBoost will output the final prediction results through weighted voting. For each set of input features, the XGBoost model will calculate the probability of each category and select the category corresponding to the maximum probability as the final risk level. This process is based on the Softmax function, which is used to convert the prediction value output by XGBoost into a probability value:

[0085] in, Indicates that the XGBoost model Prediction score of class risk level; Represents all category indices, used for normalization; Indicates that the prediction is a category By selecting the class with the highest probability, XGBoost gives the risk level corresponding to the current construction status.

[0086] To enhance the interpretability of the XGBoost model output, the SHAP (Shapley Additive Explanations) mechanism is further introduced. Its calculation formula is as follows:

[0087] Among them, each SHAP value Representation characteristics i The marginal contribution to the final prediction of the XGBoost model; F Represents the set of input features in all XGboost models, including variables from geological prediction and torque prediction, such as water seepage, propulsion speed, torque fluctuation, etc. Represents no features i All feature subsets of .

[0088] This embodiment uses SHAP analysis results to visualize the contribution ranking of variables such as "water seepage," "propulsion force fluctuation," and "fault width," thereby clarifying the weight of each feature's impact on high-risk predictions and assisting construction personnel in tracing risk sources and making targeted adjustments to control strategies.

[0089] The machine jam risk assessment model constructed in this embodiment not only overcomes the bottleneck of traditional empirical rules in characterizing the interactions of complex working conditions, but also establishes a mapping relationship between risk levels and input parameters through machine learning technology, achieving a logical closed loop between risk identification and proactive prevention and control strategies. The model can be deployed locally in the field monitoring system and can also continuously absorb new construction data through a cloud-based training platform, enabling self-learning and continuous optimization of the model, thereby meeting the requirements of dynamic adaptability under complex geological conditions and versatility in large-scale construction environments.

[0090] Step 5: Based on the card machine risk assessment results, realize the dynamic output of risk response strategy.

[0091] In this embodiment, based on the aforementioned comprehensive risk level judgment results, the excavation parameters are dynamically adjusted, escape suggestions are generated, and automatic control and early warning responses are implemented under necessary conditions to achieve closed-loop management of machine jam risks and real-time protection of construction safety.

[0092] Integrating risk response identification, parameter optimization decision-making, emergency response execution, and multi-level information support, its core operating mechanism revolves around "risk-driven parameter behavior adjustment." When the risk level is determined to be low, existing tunneling parameters remain unchanged, while a continuous monitoring mechanism for equipment status is activated. When the risk level is determined to be medium, a preset optimization model is invoked to intelligently generate parameter adjustment suggestions based on torque trends and geological environment evolution. These suggestions may include reducing thrust, reducing cutterhead speed, or optimizing penetration depth to mitigate equipment load and slow risk evolution. Parameter adjustment instructions are then transmitted to the equipment control system in real time via a control link, enabling instant adjustment and feedback of tunneling conditions.

[0093] When a high-risk state is identified, the system enters emergency linkage mode. At this point, based on the current risk type (such as machine jam, water inrush, surrounding rock instability, etc.) and the combination of induced characteristics, a predefined response strategy library is called to automatically generate targeted escape plans. For example, in a typical machine jam risk scenario, the torque fluctuation rate, penetration rate reduction rate, and geological grade information can be combined to determine that it is a "cutterhead jam." This then recommends switching the excavation mode, such as switching from a double shield to a single shield, and simultaneously activating the front face grouting device to reduce surrounding rock resistance. Step-by-step relief is achieved through shield tail thrust feedback. In water inrush risk scenarios, the deceleration mechanism is linked to the face grouting module, and the shield drainage system is used to reduce the face seepage pressure, achieving rapid drainage and risk transfer. All plans can be executed by the control system via a programmable logic controller (PLC), ensuring the system has automated processing capabilities in high-risk scenarios.

[0094] To enhance operational transparency and support decision-making capabilities, this embodiment simultaneously outputs a graphical interface and textual suggestions, including the current risk assessment basis, recommended adjustment strategies, and expected effect evaluation. Through this auxiliary interface, operators can grasp the equipment operating status, risk sources, and control response paths in real time, effectively reducing human decision-making delays and improving response efficiency.

[0095] This embodiment utilizes a complete closed-loop mechanism for parameter adjustment and strategy execution during operation. Upon receiving a parameter adjustment command, the device automatically verifies the execution effect through feedback signals, reassesses the risk status based on real-time monitoring data, and updates the control strategy if necessary. This forms a continuous cycle of "assessment-response-feedback-reassessment," ensuring stable and controlled tunneling conditions.

[0096] In complex geological and high-risk scenarios, the active prevention and control of this embodiment can achieve data-driven risk perception, intelligent algorithm-based parameter optimization, control system-based strategy execution, and continuous adjustment based on feedback mechanisms, forming a full-process dynamic closed-loop control system from "identification" to "response". It significantly improves the adaptability and safe operation level of the tunnel boring machine under complex geological conditions, and provides key control support for the intelligentization of TBM construction.

[0097] For example, let's take the excavation construction in the fracture zone as an example: First, a fracture zone risk database was established. This database contains multi-dimensional information related to fracture zones, such as fault width, rock mass integrity factor (RQD), joint spacing, water seepage, surrounding rock strength (UCS), historical construction records, and corresponding machine jam risk levels. The database also includes classifications of fracture zones, such as high, medium, and low risk, and their corresponding characteristic descriptions, providing data support for subsequent risk assessments.

[0098] Next, data for training was screened and prepared. Geological information and equipment operating data were collected through on-site construction records, geophysical exploration, and drilling. Geological information included parameters such as the fracture width, rock mass integrity factor (RQD), water seepage, and surrounding rock strength (UCS) of the fracture zone. Equipment operating information included cutterhead speed, thrust, penetration, shield friction, and hydraulic system pressure. After cleaning and normalizing the collected data, features with a high correlation with the risk of machine jamming in the fracture zone were selected for use as model input.

[0099] Subsequently, model training is performed. The selected data is input into a dual-driven comprehensive risk analysis model. This model consists of a geological risk assessment model and a torque prediction model. The geological risk assessment model utilizes a cloud model to assess geological information and calculates geological risk levels (such as A, B, and C) through cloud droplet generation and fusion. The torque prediction model uses a deep learning model (LSTM + attention mechanism) to perform time series analysis on equipment operating parameters (such as cutterhead speed and penetration), predict torque values ​​and their fluctuation trends, and assess the equipment's operating risk level. Training is completed by minimizing the loss function. Historical data on geological and torque risks are used to optimize model weights, and a gradient boosting tree (XGBoost) is used to learn the relationship between their contributions to the comprehensive risk.

[0100] Next, a risk assessment is conducted and a risk rating is output. The geological information and equipment operating parameters for the fracture zone area requiring risk assessment are input into the trained dual-driven comprehensive risk analysis model. The geological information is calculated using the cloud model module, outputting a fracture zone geological risk rating (e.g., A, B, or C). The equipment operating parameters are analyzed using the deep learning module, outputting a risk rating (e.g., low, medium, or high). The geological risk and torque risk are fused and analyzed using a gradient boosting tree (XGBoost), outputting a comprehensive risk rating (e.g., low, medium, or high). The comprehensive risk rating is then compared with the results in the fracture zone risk database to verify the reliability of the assessment results.

[0101] Next, proactive control measures are implemented. Based on the comprehensive risk assessment results, proactive control measures are implemented to provide recommendations for adjusting tunneling parameters or solving problems. When the risk is low, current tunneling parameters are maintained and monitored in real time. When the risk is medium, tunneling parameters are adjusted, such as reducing cutterhead speed and thrust, optimizing penetration, etc. When the risk is high, a solution is generated, such as switching tunneling modes (e.g., from double shield to single shield), implementing chemical grouting, or implementing advanced support.

[0102] Finally, the results were checked and optimized. The model output was compared with the actual risk conditions recorded during construction to verify the model's effectiveness in predicting machine jam risks in fractured and broken zones. The fractured and broken zone risk database and model parameters were updated through the continuous accumulation of new construction data, improving the applicability and accuracy of the risk assessment model.

[0103] The proposed method for assessing the risk of a roadheader jam, which is driven by a dual combination of geological risk assessment and torque prediction, builds a dual-drive structure centered on a cloud model and a deep neural network prediction model. This method integrates a gradient boosting tree to achieve comprehensive risk assessment, and links active prevention and control to form a closed-loop control system from risk identification to dynamic response. This method has the following significant benefits: 1. Dynamic Response and Scientific Decision-Making. This embodiment, through linkage with risk assessment, monitors geology and equipment operating conditions in real time. Based on risk levels, it triggers corresponding prevention and control strategies, dynamically adjusting excavation parameters and generating escape plans. This provides data-driven, scientific decision-making support, avoiding the potential for misjudgment caused by relying solely on human experience, and improving the accuracy and reliability of decision-making.

[0104] 2. Active Prevention and Control with Efficient Execution. This embodiment's active prevention and control features optimized tunneling parameters and automated execution. It proactively adjusts tunneling parameters such as thrust and cutterhead speed under medium and high risk levels, or directly activates support equipment such as advanced grouting and small conduit deployment. In high-risk situations, the system triggers an emergency shutdown and alarm, enabling rapid response and fully automated risk prevention and control, significantly reducing errors and delays caused by human intervention.

[0105] 3. Enhanced risk prevention and extrication capabilities. At medium-risk levels, this embodiment provides optimized construction parameter recommendations, proactively reducing equipment operating loads and preventing further escalation of risks. At high-risk levels, targeted extrication plans can be quickly generated and implemented, such as switching excavation modes, chemical grouting support, and drainage hole dredging. Through precise and efficient risk management, the system's extrication capabilities are significantly enhanced, reducing the impact of common construction issues such as machine jams, water inrush, and surrounding rock instability on the construction progress.

[0106] 4. Dynamic Adaptation and Closed-Loop Control. This embodiment utilizes real-time monitoring data to optimize tunneling parameters, forming a closed-loop control mechanism. This allows for dynamic adjustment of control measures based on changing risks, ensuring equipment operation within a safe range. By dynamically linking risk levels with parameter adjustments in real time, the system can adapt to changes in the construction environment, ensuring safety and continuity in complex geological conditions.

[0107] 5. Improved construction safety and efficiency. Through scientific decision-making and proactive prevention and control, this embodiment effectively reduces equipment downtime and the incidence of safety accidents caused by machine jams, water inrush, or surrounding rock instability during construction. It also optimizes tunneling parameters to improve construction efficiency. In complex geological conditions, particularly in high-risk scenarios such as fractured zones, karst, and altered zones, this invention can achieve an optimal balance between construction efficiency and safety.

[0108] 6. Technical Scalability and Applicability. The active prevention and control system of this embodiment has good technical scalability and can adapt to a variety of risk types and construction conditions. For example, it can combine a richer range of equipment operating parameters such as vibration and temperature, and geological monitoring data such as pressure and water seepage to further optimize model accuracy and applicability. In addition, this embodiment can be directly deployed to the on-site control system, supporting dual-end operation on-premises and in the cloud, achieving real-time risk prevention and control and construction optimization in multiple scenarios.

[0109] In summary, this embodiment achieves intelligent monitoring, dynamic adjustment, and efficient control throughout the entire construction process through the active control module. It effectively addresses the safety and efficiency issues of tunnel excavation under complex geological conditions, provides strong technical support for the intelligent and automated development of TBM construction, and has significant engineering application value.

[0110] Example 2 like Figure 5 As shown, the purpose of this embodiment is to provide a roadheader stuck risk assessment system, including: A data acquisition module, which is used to obtain geological parameters, geophysical parameters of the tunneling section and main control parameters of the tunneling machine; A geological risk assessment module is configured to assign initial weights to the geological parameters and geophysical parameters according to adverse geology during tunneling and their corresponding parameters, classify the geology into multiple risk levels, and construct a comprehensive cloud model; dynamically adjust the weights of the geological parameters and / or geophysical parameters according to fluctuations and risk sensitivity during tunneling, and determine the geological risk level of the tunneling section based on the updated comprehensive cloud model; a torque prediction module for obtaining an operation risk level of the tunnel boring machine using a trained torque prediction model based on the geological parameters, the geophysical parameters, and the tunnel boring machine main control parameters; The dual-drive evaluation module is used to integrate the geological risk level and the tunnel boring machine operation risk level based on the machine jam risk prediction model constructed by the gradient boosting tree to obtain the final machine jam risk evaluation result.

[0111] In this embodiment, an active prevention and control module is also included. The active prevention and control module is used to enter the emergency linkage mode when the system identifies a high-risk state. At this time, the active prevention and control module calls the predefined response strategy library based on the current risk type such as machine jamming, water bursting, surrounding rock instability, etc. and the combination of induced characteristics, and automatically generates a targeted escape plan. For example, in a typical machine jamming risk scenario, the system can combine the torque fluctuation rate, penetration drop rate and geological grade information to determine that it is a "cutterhead obstruction type jamming machine", and then recommend switching the excavation mode such as switching from a double shield to a single shield, and simultaneously activating the front face grouting device to reduce the surrounding rock resistance, and achieve step-by-step relief through the thrust feedback of the shield tail. In the water burst risk scenario, the system links the deceleration mechanism with the face grouting module, and reduces the face seepage pressure through the shield drainage system to achieve rapid drainage and risk transfer.

[0112] In further embodiments, there is also provided: An electronic device includes a memory and a processor, and computer instructions stored in the memory and executed by the processor. When the computer instructions are executed by the processor, the method described in Example 1 is performed. For the sake of brevity, no further details are given here.

[0113] It should be understood that in this embodiment, the processor may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), off-the-shelf field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0114] The memory may include a read-only memory and a random access memory, and provides instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0115] A computer-readable storage medium is used to store computer instructions, and when the computer instructions are executed by a processor, the method described in embodiment 1 is performed.

[0116] The method in Example 1 can be directly implemented as being executed by a hardware processor, or by a combination of hardware and software modules within the processor. The software module can be located in a storage medium well-established in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. The storage medium is located in the memory, and the processor reads the information in the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, a detailed description is not given here.

[0117] Those skilled in the art will appreciate that the units and algorithm steps of the various examples described in conjunction with this embodiment can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0118] Although the above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without any creative work are still within the scope of protection of the present invention.

Claims

1. A geological risk-torque prediction-driven card machine risk assessment method, characterized in that: include: Obtain geological parameters, geophysical parameters and main control parameters of the tunneling section; According to the adverse geology and its corresponding geological parameters and geophysical parameters, initial weights are assigned to the risk indicators of the geological parameters and the geophysical parameters, and the geology is divided into multiple risk levels to construct a standard cloud model; Based on the standard cloud model, according to the fluctuation and risk sensitivity of the geological parameters and / or the geophysical parameters during the tunneling process, the weights of the geological parameters and / or the geophysical parameters are dynamically adjusted, and the prediction results of all risk indicators are weighted and integrated to obtain the geological risk level of the tunneling section; Based on the geological parameters, the geophysical parameters and the main control parameters of the tunnel boring machine, using the trained torque prediction model, obtaining the tunnel boring machine operation risk level; Based on the machine jam risk prediction model constructed by the gradient boosting tree, the geological risk level and the tunnel boring machine operation risk level are integrated to obtain a machine jam risk evaluation result.

2. The geological risk-torque prediction-driven card machine risk assessment method according to claim 1, characterized in that: Based on the adverse geology and its corresponding geological parameters and geophysical parameters, initial weights are assigned to the risk indicators of the geological parameters and geophysical parameters, and the geology is divided into multiple risk levels to construct a standard cloud model, specifically: Establish a corresponding risk indicator system based on three types of unfavorable geological conditions: fracture zones, karst zones, and alteration zones; Each risk indicator is divided into multiple risk level intervals, and the triplet parameters of the cloud model are calculated based on the numerical range of each risk level; According to the calculated triplet parameters, cloud droplets are generated using a forward cloud generator to establish a standard cloud model.

3. The geological risk-torque prediction-driven card machine risk assessment method according to claim 1, characterized in that: Based on the standard cloud model, the weights of the geological parameters and / or the geophysical parameters are dynamically adjusted according to the fluctuations and risk sensitivity of the geological parameters and / or the geophysical parameters during the tunneling process, specifically: determining fluctuation values ​​of the geological parameters and the geophysical parameters according to the variation amplitudes of the geological parameters and the geophysical parameters in the time dimension during the tunneling process; determining the sensitivity of the geological parameters and the geophysical parameters according to the correlation between the geological parameters and the geophysical parameters and the risk assessment results; The initial weights of the geological parameters and / or the geophysical parameters are dynamically adjusted based on the comparison results of the fluctuation values ​​and sensitivities of the geological parameters and the geophysical parameters with the corresponding fluctuation value thresholds and sensitivity thresholds.

4. The geological risk-torque prediction-driven card machine risk assessment method according to claim 1 or 3, characterized in that: The prediction results of all risk indicators are weighted and integrated to obtain the geological risk level of the tunneling section, which is as follows: Based on the standard cloud model corresponding to each risk indicator and the geological parameters and geophysical parameters of the tunneling section, cloud droplets for each risk indicator are obtained, and the cloud droplets for all risk indicators are weighted and fused to obtain a comprehensive identification cloud map; The comprehensive identification cloud map is compared with the cloud maps of each standard risk level, and the maximum membership principle is used to determine the risk level that the comprehensive identification cloud map is closest to, so as to determine the geological risk level of the excavation section.

5. The geological risk-torque prediction-driven card machine risk assessment method according to claim 1, characterized in that: The torque prediction model processes the geological parameters, the geophysical parameters, and the tunnel boring machine main control parameters as follows: Use long short-term memory neural networks to extract long-term dependencies of input data of time series; The attention mechanism is used to perform weighted aggregation on the hidden states of the long short-term memory neural network to obtain a weighted context vector; A geological disturbance perception gating mechanism is introduced to perform disturbance adjustment on the weighted context vector through the disturbance feature to obtain the prediction result of the torque prediction model.

6. The geological risk-torque prediction-driven card machine risk assessment method according to claim 5, characterized in that: The geological disturbance perception gating mechanism is introduced to perform disturbance adjustment on the weighted context vector through the disturbance feature to obtain the prediction result of the torque prediction model, specifically: Concatenating the disturbance feature with the weighted context vector to obtain a disturbance adjustment factor; The disturbance adjustment factor is used to perform weighted fusion on the torque values ​​predicted by the trained long short-term memory network based on the attention mechanism, and combined with the historical torque sliding mean or weighted trend value to obtain the prediction result of the torque prediction model.

7. The geological risk-torque prediction-driven card machine risk assessment method according to claim 1, characterized in that: In the process of constructing the card machine risk prediction model, a joint contribution evaluation mechanism is introduced. Based on the gain value of the geological feature and the coverage value of the torque feature, the joint contribution of the geological feature and the torque feature is calculated; it is judged whether the joint contribution of the geological feature and the torque feature exceeds the set threshold. If the joint contribution of the geological feature and the torque feature exceeds the set threshold, the cross features generated by the geological feature and the torque feature are involved in the candidate split feature set construction of the current tree structure.

8. The geological risk-torque prediction-driven card machine risk assessment method according to claim 1, characterized in that: The machine jam risk prediction model is constructed using the geological risk level and the corresponding risk indicators, as well as the tunnel boring machine operation risk level and the corresponding tunnel boring machine main control parameters.

9. The geological risk-torque prediction-driven card machine risk assessment method according to claim 2, characterized in that: The geological parameters of the fracture zone include fault width, rock mass integrity coefficient, joint spacing, water seepage, and rock compressive strength; the geological parameters of the karst include the spatial distribution and volume of caves, fissure water pressure, permeability coefficient, distribution ratio of carbonate rocks, and water inflow; the geological parameters of the alteration zone include changes in rock chemical composition, rock mechanical properties, expansion index, degree of fissure development, and water seepage and inflow; The geophysical parameters include resistivity, relaxation time, compressional wave velocity and polarizability; The main control parameters of the tunnel boring machine include the TBM cutterhead diameter, cutterhead thickness and cutterhead opening ratio.

10. The geological risk-torque prediction-driven card machine risk assessment system is characterized by: include: The data collection module is configured to: obtain geological parameters, geophysical parameters and main control parameters of the tunneling section; The geological risk assessment module is configured to: assign initial weights to risk indicators of the geological parameters and geophysical parameters according to adverse geology and their corresponding geological parameters and geophysical parameters, divide the geology into multiple risk levels, and construct a standard cloud model; Based on the standard cloud model, according to the fluctuation and risk sensitivity of the geological parameters and / or the geophysical parameters during the tunneling process, the weights of the geological parameters and / or the geophysical parameters are dynamically adjusted, and the prediction results of all risk indicators are weighted and integrated to obtain the geological risk level of the tunneling section; The torque prediction module is configured to: obtain an operation risk level of the tunnel boring machine using a trained torque prediction model based on the geological parameters, the geophysical parameters and the tunnel boring machine main control parameters; The dual-drive evaluation module is configured to: fuse the geological risk level and the tunnel boring machine operation risk level based on the machine jam risk prediction model constructed by the gradient boosting tree to obtain a machine jam risk evaluation result.

Images