A method for predicting and actively reinforcing a TBM tunnel soft broken zone stuck risk

Abstract

The present application relates to the technical field of tunnel monitoring, in particular to a TBM tunnel soft and broken zone jam risk prediction and active reinforcement decision method. The present application predicts by using a TBM shield pressure prediction sub-model and multiple TBM characteristic data, introduces a Dropout operation to reduce model overfitting; risk grading is carried out by using a bias degree, the risk grade can be quickly and effectively obtained, different support methods are selected according to different risk grades, and the problem that reinforcement treatment methods cannot be effectively matched with surrounding rock conditions in the case of jam risk is solved; the system structure is simple, the applicability is strong, the suitable active reinforcement decision is made, the resource waste is reduced, the tunnel construction is completed by the staff, the TBM equipment damage is prevented, and the incidence of engineering accidents is reduced.

Description

Technical Field

[0001] This invention relates to the field of tunnel monitoring technology, and in particular to a method for predicting risks and making proactive reinforcement decisions for TBM tunnels with weak and broken structures and jamming. Background Technology

[0002] Tunnel boring machines (TBMs), as high-end intelligent equipment for tunnel and underground engineering construction, are widely used in tunnel construction and underground space development in various fields such as transportation, energy, and water conservancy, demonstrating excellent tunneling performance even in good rock masses. However, in recent years, with the increasing complexity of tunnel construction environments, the geological conditions faced by TBMs during tunneling have also become increasingly complex, with weak and fractured surrounding rock, represented by altered rocks, faults, and densely jointed zones, being commonplace. TBMs are highly susceptible to large deformation and jamming when tunneling in such surrounding rock. Once jamming occurs, it can result in significant time, manpower, and material costs for handling, or even damage to the TBM equipment, causing substantial economic losses and engineering accidents. Secondly, in surrounding rock with a risk of jamming, reinforcement treatment is often necessary to reduce the risk of jamming. However, if the reinforcement method is not suitable for the surrounding rock conditions, it can lead to poor reinforcement or over-reinforcement.

[0003] The prior art discloses a method for assessing and handling the risk of TBM jamming in open-face tunnels (publication number: CN114692457A), which includes the following steps: obtaining the physical and mechanical parameters of the surrounding rock of the tunnel; constructing a stratum-structure model and extracting the settlement deformation value u0 of the surrounding rock at the tunnel arch; if the settlement deformation value u0 of the surrounding rock at the tunnel arch is less than the reserved deformation amount δR, it is determined that there is no risk of jamming in the open-face TBM; otherwise, there may be a risk of jamming; then, the horizontal stress σx and the vertical stress σz are calculated, and the radial stress σr, the surrounding rock pressure P, and the frictional resistance Rp between the TBM shield and the surrounding rock are obtained; the ratio k of the frictional resistance Rp between the TBM shield and the surrounding rock to the TBM tunneling thrust FI is calculated; and the risk of jamming in the open-face TBM is determined based on the P value and the k value. By utilizing the physical and mechanical parameters of the surrounding rock at the site, a correlation between the frictional resistance of the tunnel surrounding rock and the shield and the thrust of the TBM is established. A quantitative TBM method for assessing the risk of tunnel jamming is proposed, which has high accuracy and strong practicality.

[0004] However, existing technologies only assess the risk of machine jamming, and the risk assessment and subsequent handling methods are not consistent enough. Faced with increasingly complex geological conditions during tunneling, large deformation and machine jamming problems are likely to occur. Summary of the Invention

[0005] The purpose of this invention is to overcome the problems of large deformation and machine jamming that existing technologies face with increasingly complex geological conditions, and to provide a method for predicting the risk of machine jamming in weak and fractured TBM tunnels and for proactive reinforcement decision-making.

[0006] In a first aspect, the present invention provides a method for predicting and actively reinforcing risks associated with TBM tunneling in weak and fractured conditions, comprising: S1. Input the characteristic data of the full-face tunnel boring machine (TBM) into the TBM shield pressure prediction sub-model to obtain the shield cylinder pressure value of the TBM in the next process. S2. Input the shield cylinder pressure value of the next process of the full-face tunnel boring machine (TBM) to the jamming processing module to obtain the active reinforcement decision; S3. Based on the aforementioned active reinforcement decision, adjust the operating parameters of the full-face tunnel boring machine (TBM) for the next process. Preferably, the characteristic data of the full-face tunnel boring machine (TBM) specifically include: the advance speed, cutterhead rotation speed, thrust, torque, and penetration depth of the full-face tunnel boring machine (TBM); Before inputting the feature data of the full-face tunnel boring machine (TBM) into the TBM shield pressure prediction sub-model, preprocessing is required, specifically including normalization.

[0007] Preferably, a risk decision model needs to be constructed before step S1. The risk decision model specifically includes: the TBM shield pressure prediction sub-model and the card processing module.

[0008] More preferably, the TBM shield pressure prediction sub-model includes at least two residual modules connected in series; each residual module includes a first convolutional layer and a convolutional sub-module connected in parallel; each convolutional sub-module includes a first convolutional unit and a second convolutional unit connected in series; both the first and second convolutional units include a causal convolutional layer, a weight normalization layer, an activation function layer, and a Dropout block; the activation function layer uses the ReLU function. The card processing module includes: a deviation calculation submodule, a risk classification submodule, and a decision scheme processing submodule connected in series.

[0009] Preferably, S1 specifically includes: S11. The initial causal convolution result is calculated in the first residual module. ; S12, the initial causal convolution result We perform weight normalization to obtain the normalized initial causal convolution result. ; S13. Based on the normalized initial causal convolution result The initial causal convolution result after processing is obtained by using the activation function. ; S14. Process the initial causal convolution result. The input is fed into the Dropout block for random regularization to obtain the initial causal convolution result after regularization; S15. Repeat S11 to S14 to obtain the final causal convolution result; S16. Perform nonlinear mapping on the normalized feature data of the full-face tunnel boring machine (TBM) and concatenate it with the final causal convolution result to obtain the initial training prediction result. S17. Input the initial training prediction result into the next residual module connected to the first residual module, and repeat S11 to S16 to obtain the shield cylinder pressure value of the next process of the full-face tunnel boring machine (TBM).

[0010] Preferably, S2 specifically includes: S21. Obtain the shield cylinder pressure value of the current process; S22. Input the shield cylinder pressure value of the current process and the shield cylinder pressure value of the next process into the deviation calculation submodule for calculation to obtain the deviation. ; S23. Based on the risk assessment criteria, the deviation is assessed using the risk grading submodule. Make a judgment and obtain the risk outcome; S24. Based on the risk results, the decision processing submodule is used to process the data to obtain the proactive reinforcement decision.

[0011] More preferably, the risk assessment criterion in S43 is: If the deviation If so, the risk result is determined to be at the risk-free level; If the deviation If so, the risk result is determined to be of low risk level; If the deviation If so, the risk result is determined to be of a high-risk level; If the deviation If so, the risk result is determined to be the surrounding rock level.

[0012] More preferably, the proactive hardening decision corresponding to the risk outcome includes: If the risk result is the risk-free level, then the corresponding active hardening strategy is to maintain the current process's operating parameters; If the risk result is at the low risk level, then the corresponding active reinforcement strategy is to adjust the tunneling parameters; If the risk result is the high risk level and the surrounding rock level of IV and V, then a three-dimensional tunnel numerical model needs to be constructed to simulate the tunneling process of the full-face tunnel boring machine (TBM) and calculate the difference between the thrust of the TBM and the frictional resistance of the shield. If the difference is greater than 0, the corresponding active reinforcement strategy is to maintain the operating parameters of the current process. If the difference is less than 0, the corresponding active reinforcement strategy is to select the appropriate support form. If the surrounding rock level is not IV or V, the corresponding active reinforcement strategy is to maintain the operating parameters of the current process.

[0013] Preferably, S3 specifically includes: If the active hardening strategy is to maintain the current process's running parameters, then the current running parameters will not be changed; If the active reinforcement strategy involves adjusting tunneling parameters, then reducing the rotational speed is: The thrust is controlled between 4 and 6 MN, and the torque is controlled at 50% of the normal tunneling torque; If the active reinforcement strategy involves selecting an appropriate support form, then a surrounding rock-advanced support model is constructed using FLAC3. Based on this model, the deformation of the full-face tunnel boring machine (TBM) during its excavation process and the settlement value of the surrounding rock inside the shield are calculated. A surrounding rock deformation diagram is obtained based on the deformation and settlement value. The void value between the shield tail and the tunnel roof surrounding rock at the current progress is obtained. The appropriate support form is selected based on the surrounding rock deformation diagram and the void value.

[0014] More preferably, the support method includes: advanced small pipe, advanced medium pipe shed, advanced large pipe shed, and double-layer advanced large pipe shed.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention provides a method for predicting and actively reinforcing the risk of TBM jamming in weak and fractured tunnels. Compared with existing technologies, this invention reduces model overfitting by using characteristic data of full-face TBMs and a TBM shield pressure prediction sub-model. Furthermore, this invention calculates the deviation through a jamming processing module to classify the risk, enabling rapid and effective acquisition of risk levels. Different active reinforcement strategies can be selected for different risk levels, solving the problem of not being able to effectively match reinforcement methods with surrounding rock conditions when jamming risk exists. Attached Figure Description

[0016] Figure 1This is a flowchart of a TBM tunnel weak and broken machine risk prediction and active reinforcement decision-making method in Example 1.

[0017] Figure 2 This is a schematic diagram of the TBM shield pressure prediction sub-model in Example 1.

[0018] Figure 3 The diagram shows the deformation of the surrounding rock acting on the shield under different support conditions in Example 1.

[0019] Figure 4 This is a diagram showing the deformation of the surrounding rock acting on the shield under support condition 1 in Example 1.

[0020] Figure 5 This is a diagram showing the deformation of the surrounding rock acting on the shield under support condition 2 in Example 1.

[0021] Figure 6 The figure shows the results of surrounding rock settlement and void values ​​under different support conditions in Example 2.

[0022] Figure 7 The diagram shows the original and optimized operating parameters of the TBM in Example 2. Detailed Implementation

[0023] The present invention will now be described in further detail with reference to specific embodiments. However, this should not be construed as limiting the scope of the present invention to the following embodiments; all technologies implemented based on the content of the present invention fall within the scope of the present invention.

[0024] Unless otherwise specified, the use of terms such as "upper," "lower," "left," "right," "center," "inner," and "outer" to indicate orientation or positional relationships in the description of specific embodiments of the present invention is based on the orientation or positional relationships shown in the accompanying drawings, or the orientation or positional relationship in which the product / equipment / device is typically placed during use. These terms are merely for the purpose of facilitating the description of the present invention or simplifying the description in specific embodiments, enabling those skilled in the art to quickly understand the solution, and do not indicate or imply that a particular device / component / element must have a specific orientation, or be constructed and operated in a specific positional relationship. Therefore, they should not be construed as limitations on the present invention.

[0025] Furthermore, the use of terms such as "horizontal," "vertical," "suspended," and "parallel" does not imply that the corresponding device / component / element must be absolutely horizontal, vertical, suspended, or parallel, but rather that it can be slightly tilted or have a deviation. For example, "horizontal" merely means that its direction is more horizontal relative to "vertical," not that the structure must be completely horizontal, but that it can be slightly tilted. Alternatively, it can be simplified to mean that the corresponding device / component / element, when set in a "horizontal," "vertical," "suspended," or "parallel" direction, can have an error / deviation of ±10% relative to the corresponding direction, more preferably within ±8%, more preferably within ±6%, more preferably within ±5%, and more preferably within ±4%. As long as the corresponding device / component / element is within the error / deviation range, it can still achieve its function in the present invention.

[0026] Furthermore, the use of terms such as "first," "second," and "third" in terminology is merely for distinguishing descriptions of identical or similar components and should not be interpreted as emphasizing or implying the relative importance of a particular component.

[0027] Furthermore, in the description of the embodiments of the present invention, "several", "more than", and "a number of" represent at least two. The number can be any number, such as 2, 3, 4, 5, 6, 7, 8, or 9, and can even exceed nine.

[0028] Furthermore, in the description of the technical solution of this invention, unless otherwise explicitly specified / limited / restricted, the terms "set up," "install," "connect," "link," "equipped with," "laid out," and "arranged" should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to connection methods commonly used in the field of tunnel monitoring technology, such as welding, riveting, bolting, and threaded connections. Such connections can be mechanical, electrical, or communication connections; they can be direct connections or indirect connections through an intermediate medium; and they can refer to the internal communication between two components.

[0029] Example 1 A flowchart of a risk prediction and proactive reinforcement decision-making method for TBM-driven crushing machines in weak and fractured zones is shown below. Figure 1 As shown, the specific steps include: S1. Obtain the feature data of the full-face tunnel boring machine (TBM) in the current process and perform preprocessing; The feature data in step S1 includes the advance speed, cutterhead rotation speed, thrust, torque, and penetration depth of the full-face tunnel boring machine (TBM). In well-stable surrounding rock, there is a good correlation between the shield pressure and the advance speed, cutterhead rotation speed, thrust, torque, and penetration depth. However, in poorly stable surrounding rock, the shield pressure will change. Therefore, the advance speed, cutterhead rotation speed, thrust, torque, and penetration depth of the full-face tunnel boring machine (TBM) are selected as the inputs to the prediction model.

[0030] The preprocessing in step S1 includes normalization, and the corresponding formula is:

[0031] In the formula: This represents the normalized feature data. Representing feature data, This represents the minimum value of the feature data. This represents the maximum value of the feature data.

[0032] S2. Construct a risk decision-making model; the risk decision-making model includes a TBM shield pressure prediction sub-model and a jamming processing module; The structural diagram of the TBM shield pressure prediction sub-model in step S2 is shown below. Figure 2 As shown, the TBM shield pressure prediction sub-model includes at least two residual modules connected in series. Each residual module consists of a first convolutional layer and a convolutional sub-module connected in parallel. Each convolutional sub-module consists of a first convolutional unit and a second convolutional unit connected in series. Both the first and second convolutional units include a causal convolutional layer, a weight normalization layer, an activation function layer, and a Dropout block connected in series. The activation function layer uses the ReLU function. Causal convolution ensures the causal relationship between data, meaning that the input information at the current moment is only related to information before the current moment and is unrelated to information at future moments, thus avoiding information leakage. The model weights are standardized (weight normalization), and a Dropout block is added to avoid overfitting. During model training, the Dropout block deactivates some neurons with a certain probability, achieving a degree of regularization. When the input and output dimensions of the model are different, a non-linear mapping is performed through the first convolutional layer, adjusting the dimension of the input data to ensure the model has input and output of the same dimension.

[0033] The card processing module in step S2 includes a deviation calculation submodule, a risk classification submodule, and a decision scheme processing submodule connected in series.

[0034] S3. Input the preprocessed feature data into the TBM shield pressure prediction sub-model to obtain the shield cylinder pressure value of the next process of the full-face tunnel boring machine (TBM), i.e., the prediction result. The training process for the TBM shield pressure prediction sub-model is as follows: S31. According to the formula:

[0035] In the formula: Represents the initial causal convolution result; where, This indicates the total number of filters. This represents the summation function. This indicates the number of feature data after normalization. This represents the normalized feature data. These represent the normalized propulsion speed, cutterhead speed, thrust, torque, and penetration depth of a full-face tunnel boring machine (TBM). Indicates the total filter. Each represents a filter. Indicates the first One filter, Indicates the first One preprocessed feature data; S32. The initial causal convolution result... Weight normalization is performed to obtain the normalized initial causal convolution result; S33. According to the formula:

[0036] In the formula: This represents the initial causal convolution result after processing; This represents the activation function. These represent the weight parameters of the TBM shield pressure prediction sub-model. This represents the bias term of the TBM shield pressure prediction sub-model. This represents the initial causal convolution result after normalization; S34. Process the initial causal convolution result. The input is fed into the Dropout block for random regularization to obtain the initial causal convolution result after regularization; S35. Repeat steps S31 to S34 to obtain the final causal convolution result; S36. Perform nonlinear mapping on the normalized feature data and concatenate it with the final causal convolution result to obtain the initial training prediction result. S37. Input the initial training prediction result into the next residual module, and repeat steps S31 to S36 until all residual modules complete data processing and obtain the final training prediction result, which is the shield cylinder pressure value of the next process of the full-face tunnel boring machine (TBM). S38. Adjust the parameters of the TBM shield pressure prediction sub-model based on the final training prediction results to complete the training of the TBM shield pressure prediction sub-model.

[0037] S4. Input the shield cylinder pressure value of the next process into the jamming processing module to obtain the active reinforcement decision; Step S4 includes the following steps: S41. Obtain the shield cylinder pressure value of the current process; S42. Input the shield cylinder pressure value of the current process and the shield cylinder pressure value of the next process into the deviation calculation submodule for calculation to obtain the deviation. The calculation formula for the deviation calculation submodule in step S42 is as follows:

[0038] In the formula: Indicates the degree of deviation. This indicates the shield cylinder pressure value for the next process. This indicates the current shield cylinder pressure value.

[0039] S43. Based on the risk assessment criteria, the deviation is assessed using the risk grading submodule to obtain the risk result; The risk assessment criteria in step S43 are as follows: If the deviation If so, the risk result is determined to be at the risk-free level; If the deviation If so, the risk result is determined to be of low risk level; If the deviation If so, the risk result is determined to be of a high-risk level; If the deviation If so, the risk result is determined to be the surrounding rock level.

[0040] S44. Based on the risk results, the decision-making scheme processing submodule is used to process the data to obtain the corresponding proactive reinforcement decision.

[0041] Step S44 includes the following steps: S44-1. Determine whether the risk result is at the risk-free level; if so, the proactive reinforcement decision is to maintain the current process operating parameters and proceed to step S5; otherwise, proceed to step S44-2. S44-2. Determine if the risk outcome is low-risk; if so, the proactive reinforcement decision is to adjust the tunneling parameters and reduce the rotation speed. If the thrust is controlled between 4 and 6 MN and the torque is controlled at 50% of the normal tunneling torque, proceed to step S5; otherwise, proceed to step S44-3. S44-3. Determine whether the risk result is a surrounding rock level; if yes, proceed to step S44-4; otherwise, proceed to step S44-5. S44-4. Obtain on-site tunnel geological data; based on the on-site tunnel geological data and the shield cylinder pressure value for the next process, determine whether the surrounding rock level is IV or V; if yes, proceed to step S44-5; otherwise, the risk result is no risk level, and the active reinforcement decision is to maintain the current process's operating parameters, and proceed to step S5; On-site tunnel geological data is obtained through methods such as data surveys and on-site testing. On-site tunnel geological data includes, but is not limited to, stratigraphic information, fault information, lithological interface information, and initial stress and strain information. Specific geological data will be determined based on the actual situation of the project.

[0042] S44-5. Based on the field geological data, a three-dimensional tunnel numerical model is constructed using FLAC3. The three-dimensional tunnel numerical model includes a shield and contact surface elements on the outside of the shield. The three-dimensional tunnel numerical model uses solid elements to build the shield and contact surface elements on the outside of the shield. The shield includes a top shield, top-side shield, side shield, bottom shield, and bottom shield. S44-6. The tunneling process of a full-face tunnel boring machine (TBM) is simulated using a three-dimensional tunnel numerical model to obtain the shield friction resistance. The formula for calculating the frictional resistance of a shield is:

[0043] In the formula: This represents the summation function. Indicates the first The pressure on the shield This represents the area of ​​the contact unit. Indicates the first The normal stress exerted on the contact element by the block shield This indicates the pressure experienced by all the shields. , , , , These represent the pressure experienced by each shield. Indicates the coefficient of friction. This indicates the frictional resistance of the shield.

[0044] S44-7. Calculate the difference between the thrust of the full-face tunnel boring machine (TBM) and the frictional resistance of the shield, and determine whether the difference is less than 0. If it is, it is determined that the full-face tunnel boring machine (TBM) will jam, and proceed to step S44-8. Otherwise, the active reinforcement decision is to maintain the current process operating parameters, and proceed to step S5. S44-8. Use flac3 to construct a surrounding rock-advanced support model. In the surrounding rock-advanced support model, consider different support forms such as advanced small pipe, advanced medium pipe roof, advanced large pipe roof, and double-layer advanced large pipe roof. Use the Hooke equation to calculate the deformation of the surrounding rock under different support conditions during the tunneling process and the settlement value of the surrounding rock within the shield range. S44-9. Based on the deformation and the settlement value of the surrounding rock inside the shield, obtain the deformation map of the surrounding rock; Deformation diagrams of surrounding rock acting on the shield under different support conditions are shown below. Figure 3 As shown, a certain shield length is set at the working face, and the settlement value of the surrounding rock inside the shield varies depending on the support conditions.

[0045] S44-10. Obtain the gap value between the shield tail and the surrounding rock at the top of the tunnel in the current process. ; S44-11. Based on the surrounding rock deformation diagrams under different support conditions and the current void value between the shield tail and the tunnel roof surrounding rock, select the appropriate support form, i.e., active reinforcement decision-making. The settlement value of the surrounding rock inside the shield corresponding to the surrounding rock deformation diagrams under different support conditions is... ,choose The corresponding support method.

[0046] The deformation diagram of the surrounding rock acting on the shield under support condition 1 is shown below. Figure 4 As shown, the deformation diagram of the surrounding rock acting on the shield under support condition 2 is as follows. Figure 5 As shown, the settlement value of the surrounding rock within the shield range varies under different conditions, thus requiring the selection of appropriate support methods to ensure that the full-face tunnel boring machine (TBM) for the next stage is not damaged during operation and does not cause tunnel collapse, thereby ensuring the safety of the staff and reducing the occurrence of accidents.

[0047] The support methods in step S44-11 include advanced small pipe, advanced medium pipe shed, advanced large pipe shed, and double-layer advanced large pipe shed.

[0048] S5. Based on proactive reinforcement decisions, adjust the operating parameters of the full-face tunnel boring machine (TBM) for the next process.

[0049] In summary, this invention utilizes a TBM shield pressure prediction sub-model and multiple TBM feature data for prediction, introducing Dropout to reduce model overfitting. Risk classification using deviation allows for rapid and effective acquisition of risk levels, enabling the selection of different support methods for different risk levels. This solves the problem of ineffective matching of reinforcement methods to surrounding rock conditions in situations where there is a risk of TBM jamming. The system has a simple structure, strong applicability, and enables appropriate proactive reinforcement decisions, reducing resource waste, assisting workers in tunnel construction, preventing TBM equipment damage, and reducing the incidence of engineering accidents.

[0050] Example 2 This embodiment takes a mountain tunnel TBM crossing a Class V weak and fractured zone surrounding rock as an example, and selects data such as the violent fluctuation of shield pressure and the abnormal increase of cutterhead torque that occurred when the TBM crossed a large fault area.

[0051] Acquire and preprocess the working data of the TBM in this section. Perform weighted normalization on the initial TBM tunneling data and assign higher weights to key parameters such as thrust and torque (weighting coefficient ω=1.5).

[0052] Three cascaded residual modules are configured, including a causal convolutional layer, a weight normalization layer, an activation function layer, and a dropout block. The causal convolutional layer has a kernel size of 5, a stride of 1, and 64 channels; Group Normalization with 8 groups is used for weight normalization; ReLU + Swish is used to construct the activation function layer to enhance nonlinear fitting capability; and a dropout block with dynamically adjusted inactivation rate is included, with an initial inactivation rate of 20% that linearly increases to 30% with each training epoch.

[0053] The current hydraulic cylinder pressure is 16.8 MPa. The model predicts the next stage's shield hydraulic cylinder pressure to be 19.5 MPa, a deviation of 16.1%, which is considered high-risk. A three-dimensional tunnel numerical model was constructed using FLAC3D to simulate the tunneling process and calculate the difference between thrust and shield friction resistance: thrust F = 7.0 MN, shield friction resistance Rp = 8.2 MN; the difference F - Rp = -1.2 MN (less than 0, indicating a risk of jamming and requiring support measures).

[0054] A surrounding rock-advance support model was constructed using FLAC3D, and the settlement and porosity values ​​of the surrounding rock under different support conditions were calculated as follows: Figure 6 As shown.

[0055] Considering both economic efficiency and safety, the advanced large-diameter pipe roof + high-pressure grouting + steel arch temporary support scheme was selected. This ensures that the surrounding rock deformation meets the standards and provides a 34% safety margin. Based on the proactive reinforcement decision, the operating parameters of the full-face tunnel boring machine (TBM) for the next phase were adjusted. The original and optimized operating parameters of the TBM are as follows: Figure 7 As shown.

[0056] By using the method of this invention for risk prediction and proactive reinforcement decision-making, abnormal working conditions can be responded to in a timely and effective manner, the accident rate can be reduced, and the tunneling speed can be increased, demonstrating that the method of this invention is effective.

Claims

1. A method for risk prediction and proactive reinforcement decision-making in TBM tunnels with weak and fractured structures, characterized in that, include: S1. Construct a risk decision model, which specifically includes: a TBM shield pressure prediction sub-model and a jamming processing module; input the characteristic data of the full-face tunnel boring machine (TBM) into the TBM shield pressure prediction sub-model to obtain the shield cylinder pressure value of the TBM in the next process; S2. Input the shield cylinder pressure value of the next process of the full-face tunnel boring machine (TBM) to the jamming processing module to obtain the active reinforcement decision; S3. Based on the aforementioned active reinforcement decision, adjust the operating parameters of the full-face tunnel boring machine (TBM) for the next process; The TBM shield pressure prediction sub-model includes at least two residual modules connected in series. Each residual module includes a first convolutional layer and a convolutional sub-module connected in parallel. Each convolutional sub-module includes a first convolutional unit and a second convolutional unit connected in series. Both the first and second convolutional units include a causal convolutional layer, a weight normalization layer, an activation function layer, and a Dropout block. The activation function layer uses the ReLU function. The card processing module includes: a deviation calculation submodule, a risk classification submodule, and a decision scheme processing submodule connected in series. S1 specifically includes: S11. The initial causal convolution result is calculated in the first residual module. ; S12, the initial causal convolution result We perform weight normalization to obtain the normalized initial causal convolution result. ; S13. Based on the normalized initial causal convolution result The initial causal convolution result after processing is obtained by using the activation function. ; S14. Process the initial causal convolution result. The input is fed into the Dropout block for random regularization to obtain the initial causal convolution result after regularization; S15. Repeat S11 to S14 to obtain the final causal convolution result; S16. Perform nonlinear mapping on the normalized feature data of the full-face tunnel boring machine (TBM) and concatenate it with the final causal convolution result to obtain the initial training prediction result. S17. Input the initial training prediction result into the next residual module connected to the first residual module, and repeat S11 to S16 to obtain the shield cylinder pressure value of the next process of the full-face tunnel boring machine (TBM). S2 specifically includes: S21. Obtain the shield cylinder pressure value of the current process; S22. Input the shield cylinder pressure value of the current process and the shield cylinder pressure value of the next process into the deviation calculation submodule for calculation to obtain the deviation. ; S23. Based on the risk assessment criteria, the deviation is assessed using the risk grading submodule. Make a judgment and obtain the risk outcome; S24. Based on the risk results, the decision processing submodule is used to process the data to obtain the proactive reinforcement decision.

2. The method for risk prediction and active reinforcement decision-making of TBM tunnel weak and fractured machine with jamming as described in claim 1, characterized in that, The specific characteristic data of the full-face tunnel boring machine (TBM) include: the advance speed, cutterhead rotation speed, thrust, torque, and penetration depth of the full-face tunnel boring machine (TBM). Before inputting the characteristic data of the full-face tunnel boring machine (TBM) into the TBM shield pressure prediction sub-model, preprocessing is performed, specifically including normalization.

3. The method for risk prediction and active reinforcement decision-making of TBM tunnel weak and fractured machine according to claim 1, characterized in that, The risk assessment criteria in S23 are as follows: If the deviation If so, the risk result is determined to be at the risk-free level; If the deviation If so, the risk result is determined to be of low risk level; If the deviation If so, the risk result is determined to be of a high-risk level; If the deviation If the risk result is determined to be the surrounding rock level, then the on-site tunnel geological data will be obtained. The grade of the surrounding rock is determined based on the on-site tunnel geological data and the shield cylinder pressure value in the next process.

4. The method for risk prediction and active reinforcement decision-making of TBM tunnel weak and fractured machine according to claim 3, characterized in that, The proactive hardening decision derived from the risk outcome includes: If the risk result is the risk-free level, then the corresponding proactive hardening decision is to maintain the current process's operating parameters; If the risk result is at the low risk level, the corresponding proactive reinforcement decision is to adjust the tunneling parameters; If the risk result is the high-risk level or the surrounding rock level is IV or V, then a three-dimensional tunnel numerical model needs to be constructed to simulate the tunneling process of the full-face tunnel boring machine (TBM) and calculate the difference between the thrust of the TBM and the frictional resistance of the shield. If the difference is greater than 0, the corresponding active reinforcement decision is to maintain the operating parameters of the current process. If the difference is less than 0, the corresponding active reinforcement decision is to select the appropriate support form. If the surrounding rock grade is not IV or V, the corresponding active reinforcement decision is to maintain the operating parameters of the current process.

5. The method for risk prediction and active reinforcement decision-making of TBM tunnel weak and fractured machine with jamming as described in claim 1, characterized in that, S3 specifically includes: If the proactive hardening decision is to maintain the current process's operating parameters, then the current operating parameters will not be changed; If the active reinforcement decision is to adjust the tunneling parameters, then reducing the rotational speed is: The thrust is controlled between 4 and 6 MN, and the torque is controlled at 50% of the normal tunneling torque; If the active reinforcement decision is to select an appropriate support method, then a surrounding rock-advanced support model is constructed using flac3d. Based on the surrounding rock-advanced support model, the deformation of the full-face tunnel boring machine (TBM) during the tunneling process and the settlement value of the surrounding rock inside the shield are calculated. Based on the deformation and the settlement value of the surrounding rock inside the shield, a surrounding rock deformation diagram is obtained. The void value between the shield tail and the surrounding rock at the tunnel top is obtained at the current progress. Based on the surrounding rock deformation diagram and the void value, an appropriate support method is selected.

6. The method for risk prediction and active reinforcement decision-making of TBM tunnel weak and fractured machine according to claim 5, characterized in that, Support forms include: advanced small pipe, advanced medium pipe shed, advanced large pipe shed, and double-layer advanced large pipe shed.

Images