A method for predicting tunnel deformation level based on active support force

Abstract

This invention discloses a method for predicting tunnel deformation levels based on active support force, relating to the field of tunnel deformation level prediction technology. The method includes the following steps: establishing a tunnel foundation database; obtaining foundation parameters from the tunnel foundation database, including the tunnel's equivalent circle radius, strength-stress ratio, equivalent active support force, and support stiffness, to form a tunnel calculation database; incorporating the reserved deformation amount and equivalent active support force into a deep learning-based deformation level prediction model; training the deep learning-based deformation level prediction model using the tunnel calculation database to obtain the tunnel deformation level prediction model; and predicting the deformation level of the unexcavated tunnel section based on the tunnel deformation level prediction model. This invention effectively obtains accurate tunnel deformation level predictions by establishing a calculation database to perform prediction model calculations, providing a valuable reference for deformation disaster response measures.

Description

Technical Field

[0001] This invention relates to the field of tunnel deformation level prediction technology, specifically a method for predicting tunnel deformation level based on active support force. Background Technology

[0002] The displacement data obtained in tunnel engineering are all generated under (initial) support conditions. Therefore, reasonable deformation prediction based on this displacement data must be based on the accurate identification of the "deformation-support" mechanism. Only in this way can deformation prediction play its due role in preventing deformation disasters, improving engineering quality, and ensuring the service performance of the structure.

[0003] Currently, in tunnel engineering, deformation level prediction methods based on monitored displacement data mainly fall into four categories: empirical methods, numerical simulation calculation methods, mathematical statistics, and artificial intelligence methods. The deformation (level) prediction methods currently in use are primarily based on a purely passive (initial) support system composed of "system anchors, shotcrete, and steel arch frames," thus exhibiting a "deformation-passive support" (action) mechanism.

[0004] In existing technologies, the active-passive combined support system based on the concept of active deformation control and with the prestressed anchoring system as its core has been widely used and achieved great success, highlighting the significant difference in support effect between active and passive support.

[0005] Therefore, comparing the differences between the active-passive combined support system and the purely passive support system, it can be seen that the prestressed anchoring system is the core. The active support force it provides can improve the self-bearing capacity of the surrounding rock to varying degrees, thereby achieving efficient control of tunnel deformation. Therefore, to make tunnel deformation prediction more reasonable and applicable, it is urgent to propose a tunnel deformation level prediction method that considers active support force to solve this problem. Furthermore, in setting the tunnel deformation level, we should return to the essence of tunnel engineering, namely, that the tunnel structure is used to ensure the clearance cross-section. Therefore, the setting of the tunnel deformation level should also be related to the allowable deformation amount during construction. Thus, the predicted deformation level also needs to effectively reflect the allowable deformation amount during construction to enhance the effectiveness of countermeasures. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of existing technologies in predicting tunnel deformation levels, which suffer from low accuracy due to unreasonable parameters and failure to include reserved deformation amounts, thus failing to provide effective reference for deformation disaster response measures. This invention provides a tunnel deformation level prediction method based on active support force, which calculates the tunnel deformation level using a prediction model by establishing a computational database, thereby effectively obtaining accurate tunnel deformation level predictions and providing effective reference for deformation disaster response measures.

[0007] The objective of this invention is mainly achieved through the following technical solutions:

[0008] A method for predicting tunnel deformation levels based on active support force includes the following steps:

[0009] A tunnel foundation database is established based on the excavated data of the target tunnel. This database includes information on tunnel excavation area, tunnel depth, support design parameters, reserved deformation amount, rock mass quality index, rock mass weight and deformation level.

[0010] The equivalent circle radius of the tunnel is calculated based on the tunnel excavation area.

[0011] The strength-stress ratio is calculated based on the tunnel burial depth.

[0012] The equivalent active support force and support stiffness are calculated based on the support design parameters.

[0013] The corresponding tunnel burial depth, reserved deformation amount, rock mass quality index, tunnel equivalent circle radius, strength stress ratio, equivalent active support force, support stiffness and deformation level are formed into sample pairs, and all sample pairs are integrated into a tunnel calculation database.

[0014] Using the tunnel burial depth, reserved deformation amount, rock mass quality index, tunnel equivalent circle radius, strength-stress ratio, equivalent active support force and support stiffness in the tunnel calculation database as input data, and the corresponding deformation level as output data, a neural network model is trained to form a deformation level prediction model.

[0015] When predicting the deformation level of the unexcavated section of the target tunnel, the tunnel burial depth, reserved deformation amount, rock mass quality index, equivalent circle radius of the tunnel, strength-stress ratio, equivalent active support force and support stiffness of the target section are used as input data, and the deformation level of the target section is used as output data.

[0016] Existing technologies for predicting tunnel deformation typically employ direct calculation. However, due to variations in tunnel foundation data and design, the required allowance for deformation and the equivalent active support force to ensure tunnel safety differ. Consequently, existing tunnel deformation level prediction methods are based on purely passive support systems and lack consideration of the crucial factor of active support force. This makes it difficult for existing prediction methods to be applied to deformation prediction of support structures with active support.

[0017] In this invention, the tunnel foundation database is used to incorporate tunnel foundation information into a global consideration. The database is used to obtain basic data affecting tunnel deformation, such as tunnel excavation area, tunnel depth, support design parameters, rock mass quality indicators, and rock mass density. The equivalent circle radius, strength-stress ratio, equivalent active support force, and support stiffness of the tunnel are then calculated using the tunnel foundation data. This allows for the acquisition of parameter data affecting tunnel deformation as completely as possible.

[0018] After obtaining the parameter data affecting tunnel deformation, all parameters are integrated by forming sample pairs to establish a tunnel calculation database. By incorporating the reserved deformation amount and the equivalent active support force into the overall prediction, a new deep learning-based deformation level prediction model is established, which can then use the data from the tunnel calculation database to predict the tunnel deformation level. The deep learning-based deformation level prediction model is based on a neural network with 7 layers: 1 input layer, 3-5 hidden layers, and 1 output layer. Training the deformation level prediction model using the tunnel calculation database involves using seven data points—tunnel depth, reserved deformation amount, rock mass quality indicators, equivalent tunnel circle radius, strength-stress ratio, equivalent active support force, and support stiffness—as input data, and the deformation level of the target section as output data. Three hidden layers should be selected for training to avoid overfitting. During training, the actual deformation of the excavated section can be referenced, effectively improving the accuracy of the output data after training through the neural network model. This invention enables the accurate prediction of the deformation level of unexcavated tunnel sections.

[0019] Furthermore, based on tunnel design, geological survey data, tunnel face sketches, and monitoring and measurement data, the tunnel excavation area, tunnel depth, support design parameters, reserved deformation amount, rock mass quality indicators, and rock mass density are obtained.

[0020] Furthermore, the calculation formula for the rock mass quality index is as follows:

[0021]

[0022] In the formula, Q is the rock mass quality index, RQD is the rock quality index, and J r J is the joint roughness coefficient. w J is the water reduction factor for water treatment. n J is the number of joint groups. a is the joint alteration influence coefficient, and SRF is the stress reduction coefficient.

[0023] Furthermore, the tunnel deformation level is set by the ratio of the deformation amount to the reserved deformation amount, specifically set as follows:

[0024] When u / u0 ≤ 50%, the tunnel deformation level is level 0 deformation, and the code is 0;

[0025] 50%

[0026] 75%

[0027] 90%

[0028] When u / u0>100%, the tunnel deformation level is level 4, coded as 4;

[0029] Where u is the deformation amount and u0 is the reserved deformation amount.

[0030] To ensure the generalization performance of the tunnel deformation level prediction model, the tunnel calculation database should contain different deformation characteristics, with a minimum of 500 sets, specifically at least 100 sets each for level 0, level 1, level 2, level 3, and level 4 deformation. Each set should contain 60-80% samples with active support technology. The samples for each deformation level in the tunnel calculation database should be divided into a training set:prediction set ratio of 7:3 and then used for model training.

[0031] Furthermore, the formula for calculating the equivalent circle radius of the tunnel is as follows:

[0032]

[0033] The formula for calculating the rock mass strength-stress ratio is as follows:

[0034]

[0035] The formula for calculating the equivalent active support force is as follows:

[0036]

[0037] ​​​In the formula, R is the equivalent circle radius of the tunnel, SSR is the rock mass strength-stress ratio, P is the equivalent active support force, F is the prestress of the anchoring system in N, and a and b are the support spacing and row spacing of the anchoring system in m. θ is the correction factor; θ is the circumferential support range of the anchoring system, in degrees.

[0038] Furthermore, the support stiffness is composed of the anchoring system support stiffness, the shotcrete support stiffness, and the steel arch support stiffness, and the calculation formula is as follows:

[0039]

[0040] In the formula, K is the support stiffness, K b For the support stiffness of the anchoring system, K c For the support stiffness of shotcrete, K s This refers to the support stiffness of the steel arch frame.

[0041] Furthermore, when the anchoring system is a prestressed anchoring system, the support stiffness of the anchoring system is calculated as follows:

[0042]

[0043] In the formula, E b The modulus of elasticity of the anchor bolt (cable), in Pa; d b L is the diameter of the anchor bolt (cable), in meters; L' is the length of the free section of the anchoring system, in meters; Q is the stress-strain constant of the anchoring end.

[0044] When the anchoring system is a non-prestressed anchoring system, i.e., a full-length bonded anchoring system, the support stiffness of the anchoring system is calculated as follows:

[0045]

[0046] In the formula, L is the length of the anchoring system, in meters (m).

[0047] Furthermore, the stiffness of the shotcrete support is calculated as follows:

[0048]

[0049] In the formula, E c The elastic modulus of shotcrete is expressed in Pa; ν c t is the Poisson's ratio of shotcrete; c The thickness of the sprayed concrete is in meters (m).

[0050] Furthermore, when the steel arch frame is a shaped steel arch frame, the support stiffness of the steel arch frame is calculated as follows:

[0051]

[0052] In the formula, E s The modulus of elasticity of the steel arch frame is expressed in Pa; S is the longitudinal spacing of the steel arch frame, expressed in meters; h s The cross-sectional height of the steel arch frame is in meters (m); A s The cross-sectional area of ​​the steel arch frame is expressed in m². 2 ;

[0053] When the steel arch frame is a grid arch frame, the support stiffness of the steel arch frame is calculated as follows:

[0054]

[0055] In the formula, E g ν represents the elastic modulus of the grating steel frame, in Pa; S represents the longitudinal spacing of the grating steel frame, in meters; h represents the distance between the centerlines of the upper and lower rows of reinforcing bars in the cross-section of the grating steel frame; g denoted as Poisson's ratio of the lattice steel frame; r is the equivalent radius of the inner boundary of the lattice steel frame, r = Re; and e is the cross-sectional height of the lattice steel frame, in meters.

[0056] Furthermore, the tunnel calculation database involves basic data and input data, wherein,

[0057] The basic data includes sample number and target section mileage;

[0058] The input data includes the tunnel burial depth, reserved deformation amount, rock mass quality index, equivalent circle radius of the tunnel, strength-stress ratio, equivalent active support force and support stiffness of the target section.

[0059] By inputting the tunnel burial depth, reserved deformation amount, rock mass quality index, equivalent circle radius of the tunnel, strength-stress ratio, equivalent active support force and support stiffness of the target section, the deformation level of the target section is output, along with the corresponding sample number and target section mileage.

[0060] In summary, the present invention has the following advantages compared with the prior art:

[0061] This invention incorporates the reserved deformation amount and the equivalent active support force into the model during model building, thereby achieving deformation prediction for support structures with active support. Furthermore, during model training, the actual deformation of the excavated section can be referenced, effectively improving the accuracy of the output data through neural network model training. Thus, this invention enables accurate prediction of the deformation level of unexcavated tunnel sections. Attached Figure Description

[0062] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings:

[0063] Figure 1 This is a schematic diagram of the process of the present invention;

[0064] Figure 2 This is a diagram of the deformation level prediction model based on deep learning in this invention.

[0065] Figure 3 This is a schematic diagram of a tunnel pure passive support system structure according to one specific embodiment of the present invention;

[0066] Figure 4 This is a schematic diagram of a tunnel active-passive combined support system architecture according to one specific embodiment of the present invention. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0068] Example:

[0069] like Figure 1 As shown, one specific implementation method in this embodiment is as follows:

[0070] S1. Establish a tunnel foundation database, which includes tunnel excavation area A, tunnel burial depth H, support design parameters, reserved deformation amount u0, rock mass quality index Q value, rock mass weight γ, and deformation level information.

[0071] In this embodiment, the tunnel excavation area A, tunnel depth H, support design parameters, reserved deformation amount u0, rock mass quality index Q value, rock mass weight γ, and deformation level information can be determined through design documents, geological survey data, tunnel face sketches, and monitoring measurement data. The deformation level is determined based on the ratio of the maximum displacement of the monitoring point to the reserved deformation amount.

[0072] The basic database data types are shown in Table 1:

[0073] Table 1. Basic Database Structure

[0074]

[0075] like Figure 3 and Figure 4 As shown, a tunnel foundation database is established with reference to the support system actually used.

[0076] S2. Based on the tunnel excavation area A, tunnel burial depth H, support design parameters, rock mass quality index Q value, and rock mass weight γ, calculate the tunnel equivalent circle radius R, strength stress ratio SSR, equivalent active support force P, and support stiffness K.

[0077] The required data is then extracted from the tunnel foundation database, and the calculation parameters are obtained through calculation.

[0078] For example, if we adopt the purely passive support system of XK0+500 and XK0+520, and the active-passive combined support system of XK0+740 and XK0+760, we can obtain the following basic database:

[0079] Table 2. Partial Data from the Basic Database

[0080]

[0081] Based on the parameter data in Table 2, the following calculation parameters are obtained:

[0082] Table 3 Calculation parameters

[0083]

[0084] S3. Based on the calculation parameters obtained in S2, establish a tunnel calculation database on the basis of the tunnel basic database.

[0085] The calculation database includes three types of information: basic data, input data, and output data. The basic data includes the sample number and the target section mileage. The input data includes the tunnel burial depth H, the reserved deformation amount u0, the rock mass quality index Q value, the equivalent circle radius R of the tunnel, the strength-stress ratio SSR, the equivalent active support force P, and the support stiffness K of the target section. The output data is the deformation level of the target section.

[0086] The structure of the tunnel calculation database is as follows:

[0087] Table 4. Composition of the tunnel calculation database

[0088]

[0089] The main indicators used in the prediction of surrounding rock deformation include rock mass strength stress ratio (SSR), tunnel depth (H), tunnel width (B), support stiffness (K), and vertical ground stress (σ). vThe model uses various indicators, such as the comprehensive rock quality index Q (or N), and selects conventional input data including tunnel depth H, rock mass quality index Q value, tunnel equivalent circle radius R, strength-stress ratio SSR, and support stiffness K. It also adds two unique input data points: reserved deformation amount u0 and equivalent active support force P, to establish a deep learning-based tunnel deformation level prediction model. To ensure the generalization performance of the intelligent deformation level prediction model, the tunnel calculation database should contain different deformation characteristics, with a minimum of 500 groups, i.e., at least 100 groups each for level 0, level 1, level 2, level 3, and level 4 deformation. Samples containing active support technology should account for 60-80% of each group. The samples for each deformation level in the calculation database are divided into a training set:prediction set ratio of 7:3 and then used for model training.

[0090] S4. Establish a deformation level prediction model based on deep learning, and train the model by substituting the computational database into the established deformation level prediction model. The established deep learning-based deformation level prediction model has 5 to 7 layers, namely 1 input layer, 3 to 5 hidden layers, and 1 output layer.

[0091] The process of training the deformation level prediction model by substituting the computational database into the model refers to using seven data points as input data: tunnel burial depth H, reserved deformation amount u0, rock mass quality index Q value, tunnel equivalent circle radius R, strength-stress ratio SSR, equivalent active support force P, and support stiffness K. The deformation level of the target section is used as the output data. Figure 2 The hidden layers should preferably be selected for training with 3 layers to avoid data overfitting. Taking 500 sets of data in the database as an example, the prediction accuracy of the deformation level of the prediction set is shown in Table 5.

[0092] Table 5. Prediction Set Deformation Level Accuracy

[0093]

[0094] S5. Based on the established deformation level prediction model, predict the deformation level of the unexcavated section of the tunnel. By inputting the database of the section to be predicted (tunnel face), the deformation level can be output using the prediction model. Taking a certain tunnel as an example, using 15 continuous cross-sections (intervals of 10m) as the test set, the accuracy rate is 93%, as shown in Table 6.

[0095] Table 6 Deformation Prediction Accuracy

[0096]

[0097]

[0098] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for predicting tunnel deformation level based on active support force, characterized in that, Includes the following steps: A tunnel foundation database is established based on the excavated data of the target tunnel. This database includes information on tunnel excavation area, tunnel depth, support design parameters, reserved deformation amount, rock mass quality index, rock mass weight and deformation level. The equivalent circle radius of the tunnel is calculated based on the tunnel excavation area. The strength-stress ratio is calculated based on the tunnel burial depth. The equivalent active support force and support stiffness are calculated based on the support design parameters. The corresponding tunnel burial depth, reserved deformation amount, rock mass quality index, tunnel equivalent circle radius, strength stress ratio, equivalent active support force, support stiffness and deformation level are formed into sample pairs, and all sample pairs are integrated into a tunnel calculation database. Using the tunnel burial depth, reserved deformation amount, rock mass quality index, tunnel equivalent circle radius, strength-stress ratio, equivalent active support force and support stiffness in the tunnel calculation database as input data, and the corresponding deformation level as output data, a neural network model is trained to form a deformation level prediction model. When predicting the deformation level of the unexcavated section of the target tunnel, the tunnel burial depth, reserved deformation amount, rock mass quality index, equivalent circle radius of the tunnel, strength-stress ratio, equivalent active support force and support stiffness of the target section are used as input data, and the deformation level of the target section is used as output data.

2. The method for predicting tunnel deformation level based on active support force according to claim 1, characterized in that, The tunnel excavation area, tunnel depth, support design parameters, reserved deformation, rock mass quality index, and rock mass weight are obtained based on tunnel design, geological survey data, tunnel face sketch, and monitoring and measurement data.

3. The method for predicting tunnel deformation level based on active support force according to claim 2, characterized in that, The calculation formula for the rock mass quality index is as follows: In the formula, Q is the rock mass quality index, RQD is the rock quality index, and J r J is the joint roughness coefficient. w J is the water reduction factor for water treatment. n J is the number of joint groups. a is the joint alteration influence coefficient, and SRF is the stress reduction coefficient.

4. The method for predicting tunnel deformation level based on active support force according to claim 1, characterized in that, The tunnel deformation level is set by the ratio of the deformation amount to the reserved deformation amount, specifically set as follows: When u / u0 ≤ 50%, the tunnel deformation level is level 0 deformation, and the code is 0; When 50%, 75%, 90%, and u / u0 > 100%, the tunnel deformation level is level 4, coded as 4; Where u is the deformation amount and u0 is the reserved deformation amount. The formula for calculating the radius of the equivalent circle of the tunnel is as follows: The formula for calculating the rock mass strength-stress ratio is as follows: The formula for calculating the equivalent active support force is as follows:

5. The method for predicting tunnel deformation level based on active support force according to claim 1, characterized in that, In the formula, R is the equivalent circle radius of the tunnel, SSR is the rock mass strength-stress ratio, and P is the equivalent active support force. F represents the prestress of the anchoring system, expressed in N. The support stiffness is composed of the anchoring system support stiffness, the shotcrete support stiffness, and the steel arch support stiffness, and the calculation formula is as follows: When the anchoring system is a prestressed anchoring system, the support stiffness of the anchoring system is calculated as follows: When the anchoring system is a non-prestressed anchoring system, i.e., a full-length bonded anchoring system, the support stiffness of the anchoring system is calculated as follows: a and b represent the support bays and row spacing of the anchoring system, in meters; θ is the correction factor; θ is the circumferential support range of the anchoring system, in degrees.

6. The method for predicting tunnel deformation level based on active support force according to claim 1, characterized in that, In the formula, L is the length of the anchoring system, in meters (m). In the formula, K is the support stiffness, K b For the support stiffness of the anchoring system, K c For the support stiffness of shotcrete, K s This refers to the support stiffness of the steel arch frame.

7. The method for predicting tunnel deformation level based on active support force according to claim 6, characterized in that, The stiffness of the shotcrete support is calculated as follows: In the formula, E b The modulus of elasticity of the anchor bolt (cable), in Pa; d b L is the diameter of the anchor bolt (cable), in meters; L' is the length of the free section of the anchoring system, in meters; Q is the stress-strain constant of the anchoring end. When the steel arch frame is a shaped steel arch frame, the support stiffness of the steel arch frame is calculated as follows: ​ 8. The method for predicting tunnel deformation level based on active support force according to claim 6, characterized in that, ​ In the formula, E c The elastic modulus of shotcrete is expressed in Pa; ν c t is the Poisson's ratio of shotcrete; c The thickness of the sprayed concrete is in meters (m).

9. The method for predicting tunnel deformation level based on active support force according to claim 6, characterized in that, ​ In the formula, E s The modulus of elasticity of the steel arch frame is expressed in Pa; S is the longitudinal spacing of the steel arch frame, expressed in meters; h s The cross-sectional height of the steel arch frame is in meters (m); A s The cross-sectional area of ​​the steel arch frame is expressed in meters (m²). 2 ; When the steel arch frame is a grid arch frame, the support stiffness of the steel arch frame is calculated as follows: In the formula, E g ν represents the elastic modulus of the grating steel frame, in Pa; S represents the longitudinal spacing of the grating steel frame, in meters; h represents the distance between the centerlines of the upper and lower rows of reinforcing bars in the cross-section of the grating steel frame; g denoted as Poisson's ratio of the lattice steel frame; r is the equivalent radius of the inner boundary of the lattice steel frame, r = Re; and e is the cross-sectional height of the lattice steel frame, in meters.

Images