A cold region tunnel deformation-surrounding rock frost heaving dynamic sensing and prediction system

Abstract

A kind of cold region tunnel deformation-surrounding rock frost heaving dynamic perception and prediction system, including deformation sensing fulcrum module, intelligent gateway module, deformation monitoring data management module and surrounding rock frost heaving prediction module;Deformation sensing fulcrum module is used to obtain the deformation value of lining;Intelligent gateway module and deformation sensing fulcrum module form local area network, receive the monitoring data of deformation sensing fulcrum module through local area network and upload to data management module through wireless network, simultaneously, user instruction is sent to intelligent gateway module through wireless network;Data management module is used to download and process monitoring data, and the data after pre-processing is input to frost heaving prediction module, simultaneously, user instruction is sent to intelligent gateway module;Frost heaving prediction module is used to predict surrounding rock frost heaving rate and frozen depth, to assist in judging tunnel frost heaving safety state.The present application uses tunnel real-time deformation monitoring data to carry out dynamic prediction to surrounding rock frost heaving, to more reasonably evaluate cold region tunnel structure safety state.

Description

Technical Field

[0001] This invention relates to the field of deformation monitoring technology for highway tunnels in cold regions, specifically to a dynamic sensing and prediction system for tunnel deformation and surrounding rock frost heave in cold regions. Background Technology

[0002] Permafrost regions are widely distributed in my country, with perennial permafrost accounting for approximately 22% of the total area and seasonal permafrost accounting for approximately 55%. The number of tunnels being built in cold regions is constantly increasing. China defines cold regions as areas where the average temperature of the coldest month is between -10℃ and 0℃, and the number of days with an average daily temperature ≤5℃ is between 90 and 145 days. Their main characteristics are high altitude and cold climate, with the extreme minimum temperature of the coldest month reaching -52.3℃. Heat transfer between the outside environment and the tunnel wall occurs through convection. When the outside temperature drops below 0℃, heat from the surrounding rock gradually dissipates into the atmosphere, forming a freezing zone within the rock. This generates additional frost heave forces on the tunnel lining, altering the stress distribution and causing deformation and cracking, thus affecting the safe operation of the tunnel.

[0003] Currently, most highway tunnels are inspected using traditional manual methods, such as manual inspections with total stations and other conventional surveying tools. However, this method is inefficient, carries high safety risks, and significantly impacts traffic operations, making it difficult to meet the requirements of large-scale, high-speed, efficient, and high-precision tunnel inspection. Furthermore, traditional manual inspection methods are difficult to perform in extreme weather conditions. Frost heave in cold-region tunnels often occurs in winter, when low temperatures and manual operations are challenging. While newer tunnel inspection vehicles can quickly acquire 3D point cloud data and impact data of the entire tunnel cross-section in a single operation, reconstructing a 3D model to detect defects, this method cannot obtain continuous inspection data over time.

[0004] In monitoring frost heave in surrounding rock, a common method involves drilling holes in the rock and inserting thermometers into the holes to determine the freezing depth by measuring locations where the temperature is below 0°C. However, this method requires drilling into the surrounding rock, which can damage the structure of the operating tunnel. Therefore, a non-destructive or minimally damaging monitoring method is needed to predict frost heave. Summary of the Invention

[0005] To address the aforementioned issues, this invention utilizes wireless sensor network technology to establish a dynamic sensing and prediction system for tunnel deformation and surrounding rock frost heave in cold regions. This system can achieve real-time dynamic sensing of tunnel deformation and, considering parameter uncertainties, establishes a Bayesian network prediction model. Based on the prior distribution of frost heave model parameters and tunnel deformation monitoring data, the Markov chain-Monte Carlo (MCMC) algorithm is used to predict the frost heave of the surrounding rock. The prediction results are compared with the design values ​​to assist tunnel operators in determining the safety status of the tunnel.

[0006] The present invention adopts the following technical solution:

[0007] A dynamic sensing and prediction system for tunnel deformation and surrounding rock frost heave in cold regions includes a deformation sensing pivot module, an intelligent gateway module, a data management module, and a surrounding rock frost heave prediction module.

[0008] The deformation sensing pivot module has multiple components, which are installed in the tunnel lining to obtain the deformation value of the lining and act as a relay node in the connection with the smart gateway module.

[0009] The intelligent gateway module is installed in the tunnel lining and is used to search for nearby deformation sensing support point modules. It forms a local area network with the deformation sensing support point modules, receives monitoring data from the deformation sensing support point modules through the local area network, and uploads the monitoring data to the data management module through the wireless network. It also receives user commands through the wireless network and transmits the commands to the deformation sensing support modules through the local area network.

[0010] The data management module is used to download and process monitoring data, preprocess the monitoring data, and input the preprocessed monitoring data into the frost heave prediction module. At the same time, it sends user instructions to the smart gateway module through the wireless network.

[0011] The frost heave prediction module is used to predict the frost heave rate and freezing depth of the surrounding rock. It adopts a Bayesian network prediction model and uses the Markov chain-Monte Carlo (MCMC) algorithm to predict the frost heave rate and freezing depth of the surrounding rock based on the prior distribution of the frost heave model parameters and the preprocessed monitoring data provided by the data management module. The prediction results are compared with the design values ​​to help determine the safety status of tunnel frost heave.

[0012] The beneficial effects of this invention are:

[0013] This invention enables remote, dynamic, real-time, and precise sensing of tunnel deformation, and utilizes real-time tunnel deformation monitoring data to dynamically predict frost heave of surrounding rock, thereby providing a more reasonable assessment of the structural safety status of tunnels in cold regions. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the composition and data transmission of a dynamic sensing and prediction system for tunnel deformation and surrounding rock frost heave in cold regions.

[0015] Figure 2 This is a flowchart illustrating the implementation method of the dynamic sensing and prediction system for tunnel deformation and surrounding rock frost heave in cold regions according to the present invention.

[0016] Figure 3 This is a schematic diagram of the monitoring section setup in the example.

[0017] Figure 4 This is a schematic diagram of the deformation sensing pivot module layout in the example.

[0018] Figure 5 This is a Bayesian network prediction model that includes surrounding rock parameters, frost heave parameters, and deformation monitoring indicators in the example.

[0019] Figure 6 The prior and posterior distributions of frost heave rate, freezing depth, and deformation monitoring indicators in the example are shown. Detailed Implementation

[0020] The technical solutions provided in this application will be further described below with reference to specific embodiments and accompanying drawings. The advantages and features of this application will become clearer from the following description.

[0021] like Figure 1 As shown, a dynamic sensing and prediction system for tunnel deformation and surrounding rock frost heave in cold regions includes a deformation sensing pivot module, an intelligent gateway module, a data management module, and a surrounding rock frost heave prediction module, wherein:

[0022] The deformation sensing pivot module has multiple components, which are installed in the tunnel lining to obtain the deformation value of the lining and act as a relay node in the connection with the smart gateway module.

[0023] The intelligent gateway module is installed in the tunnel lining and is used to search for nearby deformation sensing support point modules. It forms a local area network with the deformation sensing support point modules, receives monitoring data from the deformation sensing support point modules through the local area network, and uploads the monitoring data to the data management module through the wireless network. It also receives user commands through the wireless network and transmits the commands to the deformation sensing support modules through the local area network.

[0024] The data management module is used to download and process deformation data, preprocess the monitoring data, and input the preprocessed monitoring data into the frost heave prediction module. At the same time, it sends user instructions to the smart gateway module through the wireless network.

[0025] The frost heave prediction module is used to predict the frost heave rate and freezing depth of the surrounding rock. It adopts a Bayesian network prediction model and uses the Markov chain-Monte Carlo (MCMC) algorithm to predict the frost heave rate and freezing depth of the surrounding rock based on the prior distribution of the frost heave model parameters and the processed monitoring data provided by the data management module. The prediction results are compared with the design values ​​to help determine the safety status of tunnel frost heave.

[0026] Furthermore, the Bayesian network prediction model includes a first layer of nodes, a second layer of nodes, and a multi-parameter response surface function, wherein: the first layer of nodes are surrounding rock parameters and frost heave parameters, the second layer of nodes are deformation monitoring indicators, and the edges of the Bayesian network prediction model point from the first layer of nodes to the second layer of nodes;

[0027] The multi-parameter response surface function is designed as a multi-parameter response surface function for the frost heave-tunnel deformation monitoring index, and its expression is as follows:

[0028]

[0029] In the formula, g k (X) is the multi-parameter response surface function corresponding to the kth deformation monitoring index;

[0030] X = [x1, x2, ..., x i ,…,x n [ ] represents the parameter vector for the frost heave model, which includes surrounding rock parameters and frost heave parameters;

[0031] α k0 ,α k1i ,α k2i The coefficients are those of the k-th multi-parameter response surface function;

[0032] ε k Let be the error distribution of the k-th multi-parameter response surface function.

[0033] Based on the above system, the prediction of frost heave in surrounding rock can be implemented. For example... Figure 2 As shown, the specific implementation steps are as follows:

[0034] Step (1): Collect relevant structural design data and geological survey data for the tunnel to be studied; thus, by reviewing the geological survey data of the tunnel and investigating relevant literature, the prior distribution of each parameter is obtained;

[0035] Step (2): Based on the tunnel geological conditions, select the optimal monitoring section, analyze the stress and deformation characteristics of the lining using numerical simulation methods, and select the optimal installation location for the deformation sensing support module; Figure 3 , Figure 4 (As shown)

[0036] Step (3): Install the deformation sensing fulcrum module and the intelligent gateway module, establish the deformation monitoring data management module, and form a wireless sensing system for deformation monitoring of tunnels in cold regions; Figure 1 , Figure 4 (As shown)

[0037] Step (4): Based on the sensing system in step (3), taking into account the battery life of the sensing pivot and the monitoring requirements, determine a reasonable deformation monitoring cycle and monitoring data feedback cycle to realize real-time dynamic monitoring and visualization of tunnel monitoring data.

[0038] Step (5): Considering the uncertainties of surrounding rock and tunnel structure materials, a multi-parameter response surface function of frost heave-tunnel deformation monitoring index is established through numerical simulation, and a Bayesian network prediction model of surrounding rock parameters-frost heave parameters-deformation monitoring index is established.

[0039] Step (6): Combining the prior distribution of the frost heave model parameters and the multi-parameter response surface function of the frost heave-tunnel deformation monitoring index in step (5), the Bayesian network model in step (5) is dynamically updated using the tunnel monitoring data in step (4). The predicted distribution of the frost heave rate and freezing depth of the surrounding rock is calculated using the Markov Chain-Monte Carlo (MCMC) algorithm, and then compared with the design values ​​of the frost heave rate and freezing depth of the surrounding rock to determine the safety status of the tunnel. The Markov Chain-Monte Carlo (MCMC) algorithm process is as follows:

[0040] 1) Use the mean of the prior distribution of parameters as the initial parameter vector X0;

[0041] 2) Select a multidimensional normal distribution as the jump distribution, with the mean of the multidimensional normal distribution being the parameter vector X of the current sample. i-1 And perform the following operations on i = 1, 2, ..., n: ① Extract candidate parameter vector X from the jump distribution * ② Calculate the probability density ratio In the formula, K(X|y) is the maximum likelihood function value based on the observed values; ③ If min(r,1)≥Uniform(0,1), then X i =X * Otherwise X i =X i-1 ;

[0042] 3) Repeat process 2) until i = n.

[0043] Application Examples:

[0044] The tunnel in a certain cold region has an extreme low temperature of -34℃ and an annual precipitation of approximately 850mm. The maximum freezing depth of the surrounding rock is about 4.5m. Due to the frost heave of the surrounding rock, the tunnel has experienced multiple defects such as lining cracking and water leakage during its operation. In order to achieve dynamic sensing of the tunnel deformation and prediction of frost heave of the surrounding rock, this invention is applied to the tunnel project.

[0045] First, based on the geological survey report for this tunnel project, monitoring sections were set up in areas with poor geological conditions. In this example, a total of four monitoring sections were set up, as shown in the schematic diagram below. Figure 3 As shown.

[0046] Then, deformation sensing support point modules are deployed on the monitoring sections. In this example, the deformation sensing support point modules use tilt angle sensing support points to monitor the horizontal tilt angle of the lining. On each monitoring section, based on the stress and deformation characteristics of the lining, the tilt angle sensing support points are installed at locations with larger lining corner deformations to capture larger tilt angle changes and reduce the relative error of the sensors. A schematic diagram of the tilt angle sensing support point deployment is shown below. Figure 4 As shown. In Figure 4Taking monitoring section ③ as an example, a total of four tilt angle sensing supports are arranged, namely the first tilt angle sensing support 1, the second tilt angle sensing support 2, the third tilt angle sensing support 3, and the fourth tilt angle sensing support 4. Among them, the first tilt angle sensing support 1 and the second tilt angle sensing support 2 are arranged at the upper arch waist and are symmetrical about the central axis of the monitoring section. The third tilt angle sensing support 3 and the fourth tilt angle sensing support 4 are arranged at the lower arch waist and are symmetrical about the central axis of the monitoring section.

[0047] The intelligent gateway module is positioned between the second and third monitoring sections to ensure that each tilt sensor pivot point can be directly connected to the intelligent gateway module as much as possible. Figure 3 As shown.

[0048] The tilt sensor pivot and smart gateway module are fixed to the tunnel lining with bolts and mounting brackets. The installation steps are as follows:

[0049] (1) Mark the installation location;

[0050] (2) Drill holes at the installation location using an impact drill;

[0051] (3) Use M10 or M12 bolts to fix the tilt sensor support and the smart gateway module to the lining;

[0052] (4) Turn on the tilt sensor pivot and smart gateway module to put them into working state, and observe the networking quality between the tilt sensor pivot and smart gateway module on the data management module (running on the computer). If the data is uploaded normally 5 times in a row, the networking quality is good. Otherwise, the antenna and installation position of the tilt sensor pivot and smart gateway module need to be adjusted.

[0053] (5) Set the data collection cycle and upload cycle through the data management module.

[0054] When the smart gateway module and the tilt sensor pivot are both enabled, the smart gateway module will automatically search for the tilt sensor pivot within its communication range and, through... The Protocol wireless network is used to form a local area network (LAN). The tilt sensing pivot has a built-in MEMS sensor that monitors the tilt angle along the X, Y, and Z axes. Data is collected according to a set acquisition cycle, and the data is transmitted to the intelligent gateway module via the LAN. The intelligent gateway module can temporarily store the data uploaded by the tilt sensing pivot and uploads it to the data management module via a 4G network according to a set upload cycle. The intelligent gateway module can also send user commands to the tilt sensing pivot via the LAN. In this example, the tilt sensing pivot collects tilt information every 1 hour and uploads tilt monitoring data (including the tilt angle value at the pivot's location) every 8 hours. The data management module (running on a computer) is used for downloading, preprocessing, and sending commands to the monitoring data. It also inputs the preprocessed data (including tilt angle changes) into the frost heave prediction module. After receiving the monitoring information (including tilt angle changes), the frost heave prediction module (running on a computer) uses the Markov Chain Monte Carlo (MCMC) algorithm to predict the frost heave rate and freezing depth of the surrounding rock and outputs the predicted distribution. The data transmission diagrams for the deformation sensing fulcrum module, intelligent gateway module, data management module, and frost heave prediction module are shown below. Figure 1 As shown.

[0055] In this example, taking monitoring section ③ as an example, the surrounding rock grade of this monitoring section is IV, with well-developed joints and severe weathering. The parameters of the surrounding rock frost heave model are: elastic modulus E of the surrounding rock. r Poisson's ratio of the surrounding rock v r Density of surrounding rock ρ, frost heave rate of surrounding rock R f The freezing depth of the surrounding rock, D f The tunnel is buried at a depth of H. The elastic modulus of the surrounding rock is E. r Poisson's ratio of the surrounding rock v r The surrounding rock density ρ and tunnel burial depth H are parameters of the surrounding rock, and the frost heave rate R of the surrounding rock is... f , freezing depth of surrounding rock D f The parameters are frost heave parameters. The prior distribution of each parameter was obtained by reviewing the geological survey data of the tunnel and investigating relevant literature, as shown in Table 1.

[0056] Table 1 Parameter Distribution Table

[0057]

[0058] In this example, the deformation monitoring indicators are the change in the inclination angle of the upper arch waist (Δθ1) and the change in the inclination angle of the lower arch waist (Δθ2). For example... Figure 4As shown, in this example, the upper arch waist inclination angle (θ1) is the angle between the lining at the installation position of the first inclination angle sensing fulcrum 1 and the horizontal line (i.e., the angle formed between the tangent and the horizontal line in the figure); the lower arch waist inclination angle (θ2) is the angle between the lining at the installation position of the third inclination angle sensing fulcrum 3 and the horizontal line (i.e., the angle formed between the tangent and the horizontal line in the figure). Symmetrically, the angle between the lining at the installation position of the second inclination angle sensing fulcrum 2 and the horizontal line is also the upper arch waist inclination angle (θ1), and the angle between the lining at the installation position of the fourth inclination angle sensing fulcrum 4 and the horizontal line is also the lower arch waist inclination angle (θ2).

[0059] The initial value was obtained after the tilt sensor fulcrum was installed and stabilized. When the surrounding rock frosts and heaves, the lining deforms under the action of frost heave force. The value collected by the tilt sensor support at this time is recorded as follows. The change in the inclination angle of the upper arch waist Change in the inclination angle of the lower arch waist

[0060] Using a numerical simulation orthogonal experimental method, the multi-parameter response surface functions for the changes in the inclination angle of the upper arch waist (Δθ1) and the lower arch waist (Δθ2) are constructed as follows:

[0061]

[0062]

[0063] In the formula: ε1 is the error between the multi-parameter response surface function g1(X) and the numerical simulation, ε1~N(0,0.0157); ε2 is the error between the multi-parameter response surface function g2(X) and the numerical simulation, ε2~N(0,0.01).

[0064] The Bayesian network prediction model for the frost heave prediction module in this example is as follows: Figure 5 As shown in the diagram, the frost heave prediction module combines the prior distribution of the surrounding rock frost heave model parameters and the multi-parameter response surface function. It dynamically updates the Bayesian network model using on-site deformation monitoring data and solves for the posterior distribution of the frost heave parameters using the Markov chain-Monte Carlo (MCMC) algorithm. This yields the predicted results of the surrounding rock frost heave rate and freezing depth, which are then compared with the design values ​​to assist in determining the tunnel's safety status.

[0065] In this example, the monitored values ​​for the daily changes in the upper arch waist inclination angle (Δθ1) were 0.0406 and 0.0607; the monitored values ​​for the daily changes in the lower arch waist inclination angle (Δθ2) were 0.0375 and 0.0237. The monitoring data were input into the frost heave prediction module, which automatically calculated and predicted the frost heave rate and freezing depth of the surrounding rock. The results are as follows: the mean frost heave rate of the surrounding rock was updated from 0.01 to 0.0044, and the standard deviation decreased from 0.002 to 0.0019; the mean freezing depth of the surrounding rock... The mean value of the change in the upper arch waist angle (Δθ1) was updated from 2m to 0.62m, and the standard deviation decreased from 0.2 to 0.15; the mean value of the change in the lower arch waist angle (Δθ2) was updated from 0.0454 to 0.0506, and the standard deviation decreased from 0.02 to 0.01; the mean value of the change in the lower arch waist angle (Δθ2) was updated from 0.0389 to 0.0306, and the standard deviation decreased from 0.014 to 0.005. The prior and posterior distributions of the frost heave rate, freezing depth, Δθ1, and Δθ2 of the surrounding rock are as follows: Figure 6 As shown above, the predicted mean values ​​of frost heave rate and freezing depth of the surrounding rock are lower than the mean values ​​of the prior distribution of frost heave rate and freezing depth of the surrounding rock, indicating that the tunnel is in a relatively safe state. At the same time, updating the parameters using the MCMC method reduces the uncertainty of the parameters.

[0066] The above description is merely a description of preferred embodiments of this application and is not intended to limit the scope of this application in any way. Any changes or modifications made by those skilled in the art based on the above-disclosed technical content should be considered as equivalent and valid embodiments and fall within the scope of protection of the technical solution of this application.

Claims

1. A dynamic sensing and prediction system for tunnel deformation and surrounding rock frost heave in cold regions, characterized in that, It includes a deformation sensing fulcrum module, an intelligent gateway module, a data management module, and a surrounding rock frost heave prediction module; The deformation sensing pivot module has multiple components, which are installed in the tunnel lining to obtain the deformation value of the lining and act as a relay node in the connection with the smart gateway module. The intelligent gateway module is installed in the tunnel lining and is used to search for nearby deformation sensing support point modules. It forms a local area network with the deformation sensing support point modules, receives monitoring data from the deformation sensing support point modules through the local area network, and uploads the monitoring data to the data management module through the wireless network. It also receives user commands through the wireless network and transmits the commands to the deformation sensing support modules through the local area network. The data management module is used to download and process monitoring data, preprocess the monitoring data, and input the preprocessed monitoring data into the frost heave prediction module. At the same time, it sends user instructions to the smart gateway module through the wireless network. The frost heave prediction module is used to predict the frost heave rate and freezing depth of the surrounding rock. It adopts a Bayesian network prediction model and uses the Markov chain-Monte Carlo (MCMC) algorithm. Based on the prior distribution of the frost heave model parameters and the preprocessed monitoring data provided by the data management module, it predicts the frost heave rate and freezing depth of the surrounding rock. The prediction results are compared with the design values ​​to help determine the safety status of tunnel frost heave. The Bayesian network prediction model includes first-layer nodes, second-layer nodes, and a multi-parameter response surface function, wherein: The first layer of nodes represents the surrounding rock parameters and frost heave parameters, while the second layer of nodes represents the deformation monitoring indicators. The edges of the Bayesian network prediction model point from the first layer of nodes to the second layer of nodes. The multi-parameter response surface function is designed as a multi-parameter response surface function for the frost heave-tunnel deformation monitoring index, and its expression is as follows: In the formula, For the first Multi-parameter response surface functions corresponding to each deformation monitoring index; This is a parameter vector for the frost heave model, where the frost heave model parameters include surrounding rock parameters and frost heave parameters; For the first The coefficients of a multi-parameter response surface function; For the first Error distribution of a multi-parameter response surface function; The deformation sensing pivot module adopts an angle sensing pivot to monitor the horizontal inclination angle of the lining. Correspondingly, the Bayesian network prediction model uses the change in the horizontal inclination angle of the lining as the deformation monitoring index. The frost heave model parameters include the elastic modulus of the surrounding rock, Poisson's ratio of the surrounding rock, density of the surrounding rock, frost heave rate of the surrounding rock, freezing depth of the surrounding rock, and tunnel burial depth. The elastic modulus of the surrounding rock, Poisson's ratio of the surrounding rock, density of the surrounding rock, and tunnel burial depth are the surrounding rock parameters, while the frost heave rate of the surrounding rock and freezing depth of the surrounding rock are the frost heave parameters.

2. The dynamic sensing and prediction system for tunnel deformation and surrounding rock frost heave in cold regions as described in claim 1, characterized in that, Inclination sensing points are installed at the upper and lower arch sections of the tunnel monitoring section. The inclination angle of the upper arch is... The angle between the lining at the installation position of the tilt sensor pivot point on the upper arch waist and the horizontal line, and the tilt angle of the lower arch waist. The angle between the lining and the horizontal line at the installation position of the tilt sensor fulcrum at the lower arch waist; The initial value was obtained after the tilt sensor fulcrum was installed and stabilized. ; When the surrounding rock frosts and heaves, the lining deforms under the frost heave force. The values ​​collected at the inclination angle sensing supports at the upper and lower arch waists are recorded as follows: The change in the inclination angle of the upper arch waist ; Change in the inclination angle of the lower arch ; The above changes in the arch angle and the change in the inclination angle of the lower arch As a deformation monitoring indicator for Bayesian network prediction models; Based on tunnel geological survey data and relevant literature, the prior distribution of the above-mentioned frost heave model parameters was obtained. A numerical simulation orthogonal experimental method was used to construct the variation of the upper arch waist inclination angle. ) and the change in the inclination angle of the lower arch ( The multi-parameter response surface function is as follows: In the formula: The elastic modulus of the surrounding rock. Poisson's ratio of the surrounding rock The density of the surrounding rock, The frost heave rate of the surrounding rock. This represents the freezing depth of the surrounding rock. For tunnel burial depth, For multi-parameter response surface function Error between numerical simulation and ; For multi-parameter response surface function Error between numerical simulation and ; The frost heave prediction module combines the prior distribution of surrounding rock and frost heave parameters with multi-parameter response surface functions. It uses on-site deformation monitoring index data to dynamically update the Bayesian network model and uses the Markov chain-Monte Carlo (MCMC) algorithm to solve the posterior distribution of frost heave parameters, obtaining the predicted results of surrounding rock frost heave rate and surrounding rock freezing depth. These results are compared with the design values ​​to help determine the safety status of the tunnel.

3. The dynamic sensing and prediction system for tunnel deformation and surrounding rock frost heave in cold regions as described in claim 1, characterized in that, The implementation process includes the following steps: Step (1): Collect relevant structural design data and geological survey data for the tunnel to be studied; thus, by reviewing the geological survey data of the tunnel and investigating relevant literature, the prior distribution of each parameter is obtained; Step (2): Select the monitoring section according to the tunnel geological conditions, analyze the stress and deformation characteristics of the lining using numerical simulation methods, and select the installation position of the deformation sensing support module. Step (3): Install the deformation sensing fulcrum module and the intelligent gateway module, establish the deformation monitoring data management module, and form a wireless sensing system for deformation monitoring of tunnels in cold regions; Step (4): Based on the sensing system in step (3), taking into account the battery life of the sensing pivot and the monitoring requirements, determine a reasonable deformation monitoring cycle and monitoring data feedback cycle to realize real-time dynamic monitoring and visualization of tunnel monitoring data. Step (5): Considering the uncertainty of surrounding rock and tunnel structure materials, establish a multi-parameter response surface function of frost heave-tunnel deformation monitoring index through numerical simulation, and establish a Bayesian network prediction model of surrounding rock parameters-frost heave parameters-deformation monitoring index. Step (6): Combining the prior distribution of the frost heave model parameters and the multi-parameter response surface function of the above-mentioned frost heave-tunnel deformation monitoring index, the Bayesian network model in step (5) is dynamically updated using the tunnel deformation monitoring data in step (4). The predicted distribution of the frost heave rate and freezing depth of the surrounding rock is calculated using the Markov chain-Monte Carlo (MCMC) algorithm, and then compared with the design values ​​of the frost heave rate and freezing depth of the surrounding rock to determine the safety status of the tunnel.

Images