Metro contact channel freezing method parameter optimization method and system

Abstract

The invention relates to the technical field of underground engineering construction, in particular to a subway contact channel freezing method parameter optimization method and system. Through multi-source data fusion and real-time perception, based on a heat transfer theory and soil mechanics coupling model, a three-dimensional stratum stable field distribution diagram is constructed through real-time state parameters, and the stability of the frozen wall is visually presented, so that the limitation that a traditional empirical formula cannot reflect complex geological response is solved; defect features such as non-intersection circles and insufficient thickness are extracted through a multi-scale algorithm, automatic identification and positioning of defects are achieved, subjectivity and hysteresis of manual judgment are overcome, process parameters are optimized in combination with finite element simulation, an optimal parameter set is obtained, and therefore the problem that static parameters are difficult to cope with geological mutation is solved. And finally, a closed-loop optimization mode is formed by comparing the quality evaluation value with a preset threshold value, the'one-time optimization 'mode of a traditional method is broken through, it is ensured that parameters are matched with dynamic geology, and the frozen wall quality and the stratum reinforcement effect are guaranteed.

Description

Technical Field

[0001] This invention relates to the field of underground engineering construction technology, and in particular to a method and system for optimizing parameters of the freezing method for subway connecting passages. Background Technology

[0002] Subway tunnels typically consist of two parallel tunnels (one for the northbound line and one for the southbound line). Connecting passageways, serving as transverse corridors linking the two main tunnels, primarily function as emergency evacuation routes, providing space for equipment maintenance, facilitating pipeline installation, and supporting transportation during construction. However, due to complex geological conditions, the susceptibility to collapse during excavation, and the potential failure of traditional support methods (such as grouting) under high-pressure water conditions, connecting passage construction faces multiple high-risk challenges. To address these issues, the freezing method, through "frozen soil reinforcement" technology, can effectively solve these problems. Its core principle is to use artificial refrigeration technology to freeze the moisture in the soil layer into ice, forming a high-strength, impermeable frozen wall. This frozen wall can simultaneously function as a temporary support structure and a natural waterproof barrier, providing reliable safety assurance for the excavation and construction of connecting passageways.

[0003] The existing method of freezing subway connecting passages generally involves first calculating initial construction parameters based on geological reports and industry-standard empirical formulas, and then carrying out construction according to these parameters. However, during construction, due to the real-time interaction between construction disturbances and geological responses, geological abrupt changes can easily occur. Since the initial construction parameters are determined based on static geological data, they cannot dynamically adapt to such abrupt changes. This leads to problems such as uneven freezing of the frozen wall or insufficient thickness, ultimately resulting in poor reinforcement of the temporary strata and difficulty in ensuring construction safety and stratum stability. Summary of the Invention

[0004] The main objective of this invention is to provide a method and system for optimizing parameters of subway connection passage freezing, aiming to solve the technical problems in the prior art.

[0005] This invention proposes a parameter optimization method for subway connecting passage freezing, comprising: Acquire multi-source real-time monitoring data, which includes formation state parameters and process real-time parameters, and obtain initial state parameters and real-time state parameters based on the formation state parameters; The formation stability field distribution map is obtained based on the initial state parameters and the real-time state parameters; Obtain the freezing defect feature set based on the real-time status parameters; Based on the formation stability field distribution map and the freezing defect feature set, the real-time process parameters are optimized to obtain the optimal parameter set; The freezing efficiency evolution data is obtained based on the optimal parameter set and the real-time state parameters, and the freezing efficiency evaluation feature set is obtained based on the freezing efficiency evolution data. The quality assessment value is obtained based on the frozen efficiency assessment feature set, and it is determined whether the quality assessment value is greater than a preset threshold. If the quality assessment value is greater than the preset threshold, then the optimal parameter set is determined to meet the construction requirements; If the quality assessment value is not greater than the preset threshold, then the defect feature update set is obtained based on the frozen efficiency evolution data and the frozen defect feature set, and the process is returned to the step of optimizing the real-time process parameters until the optimal parameter set meets the construction requirements.

[0006] Preferably, the step of obtaining the formation stability field distribution map based on the initial state parameters and the real-time state parameters includes: Based on the real-time state parameters, spatiotemporal temperature data, pore hydrodynamic parameters, and mechanical property parameters of multiple monitoring points are obtained, and the actual compressive strength field is obtained based on the mechanical property parameters. The formation characteristic parameters are obtained based on the initial state parameters and the spatiotemporal temperature data. The thermal stress tensor field and the frost heave pressure tensor field are obtained based on the spatiotemporal temperature data and the formation characteristic parameters, and the total stress tensor field is obtained based on the thermal stress tensor field and the frost heave pressure tensor field. The shear failure field is obtained from the total stress tensor field. The seepage drag force field is obtained by Darcy's law based on the pore hydrodynamic parameters and the formation characteristic parameters, and the combined force field is obtained by the seepage drag force field and the shear failure field. The formation stability field distribution map is obtained based on the combined force field.

[0007] Preferably, the step of obtaining the frozen defect feature set based on the real-time state parameters includes: Based on the phase change latent heat compensation algorithm, the three-dimensional evolution data of the frozen front is obtained according to the real-time state parameters, and the spatial variation coefficients of multiple cross sections are obtained according to the three-dimensional evolution data of the frozen front. Obtain frozen design parameters, and obtain a first threshold and a second threshold based on the frozen design parameters, and determine whether the spatial variation coefficient of each cross section is greater than the first threshold; If the spatial variation coefficient of the cross section is greater than the first threshold, the cross section is determined to be an unconnected defect surface, and the unconnected defect characteristics are obtained based on the unconnected defect surface. The thickness deviation rate of each monitoring point is obtained based on the real-time status parameters, and it is determined whether the thickness deviation rate of each monitoring point is greater than the second threshold. If the thickness deviation rate of the monitoring point is greater than the second threshold, the monitoring point is determined to be a thickness deficiency defect point, and the thickness deficiency defect characteristics are obtained based on the thickness deficiency defect point. Construct a frozen defect feature set based on the insufficient thickness defect feature and the non-intersecting defect feature.

[0008] Preferably, the step of optimizing the real-time process parameters based on the formation stability field distribution map and the freezing defect feature set to obtain the optimal parameter set includes: The severity of defects in each independent defect region is obtained based on the frozen defect feature set, and the geological sensitivity coefficient of each independent defect region is obtained based on the frozen defect feature set and the formation stability field distribution map. The process dynamic control coefficient is obtained based on the real-time status parameters, and the initial adjustment value of the process parameters is obtained based on the process dynamic control coefficient and the real-time process parameters. The process parameter adjustment range is obtained based on the initial adjustment value of the process parameters and the geological sensitivity coefficient, and multiple parameter optimization groups are obtained based on the process parameter adjustment range; Obtain the parameter optimization objective and process optimization constraints, and filter multiple parameter optimization groups according to the parameter optimization objective and process optimization constraints to obtain the optimal parameter group.

[0009] Preferably, the steps of obtaining freezing efficiency evolution data based on the optimal parameter set and the real-time state parameters, and obtaining a freezing efficiency evaluation feature set based on the freezing efficiency evolution data, include: Based on the real-time state parameters and the formation stability field distribution map, basic modeling data is obtained, and a multiphysics coupling body is constructed based on the basic modeling data; The freezing efficiency evolution data is obtained based on the multiphysics coupler and the optimal parameter set, wherein the freezing efficiency evolution data includes temperature field evolution data, formation response data and freezing wall development data; Obtain the freezing design target, and obtain the project schedule compliance based on the freezing design target and the temperature field evolution data; The stability coefficient is obtained based on the freezing design target and the formation response data, and the thickness compliance volume ratio and the cross-ring compliance index are obtained based on the freezing design target and the frozen wall development data. The intersection compliance index, construction period consistency, stability coefficient, thickness compliance volume ratio, and intersection compliance index are used as the feature set for freezing efficiency evaluation.

[0010] Preferably, the step of obtaining the defect feature update set based on the frozen efficiency evolution data and the frozen defect feature set includes: The updated coordinates of the defect center are obtained based on the frozen efficiency evolution data and the real-time status parameters. The update features of unconnected defects are obtained based on the freezing efficiency evolution data and the freezing defect feature set, wherein the update features of unconnected defects include the maximum gap update width and the percentage of unclosed area update; The thickness growth compensation coefficient is obtained based on the freezing efficiency evolution data, and the thickness deficiency defect update feature is obtained based on the thickness growth compensation coefficient and the freezing defect feature set, wherein the thickness deficiency defect update feature includes the minimum thickness update value and the defect spatial distribution update density. A defect feature update set is constructed based on the defect center update coordinates, the non-intersecting defect update features, and the insufficient thickness defect update features.

[0011] This application also provides a parameter optimization system for the subway connection passage freezing method, including: The first acquisition module is used to acquire multi-source real-time monitoring data, which includes formation state parameters and process real-time parameters, and to acquire initial state parameters and real-time state parameters based on the formation state parameters. The second acquisition module is used to acquire a formation stability field distribution map based on the initial state parameters and the real-time state parameters. The feature extraction module is used to obtain a set of freezing defect features based on the real-time status parameters; The parameter optimization module is used to optimize the real-time process parameters based on the formation stability field distribution map and the freezing defect feature set to obtain the optimal parameter set; The efficiency evolution module is used to obtain frozen efficiency evolution data based on the optimal parameter set and the real-time state parameters, and to obtain frozen efficiency evaluation feature set based on the frozen efficiency evolution data. The iterative optimization module is used to obtain a quality assessment value based on the frozen efficiency assessment feature set, and to determine whether the quality assessment value is greater than a preset threshold. If the quality assessment value is greater than the preset threshold, then the optimal parameter set is determined to meet the construction requirements; If the quality assessment value is not greater than the preset threshold, then the defect feature update set is obtained based on the frozen efficiency evolution data and the frozen defect feature set, and the process is returned to the step of optimizing the real-time process parameters until the optimal parameter set meets the construction requirements.

[0012] Preferably, the feature extraction module includes: The acquisition unit is used to acquire three-dimensional evolution data of the frozen front based on the phase change latent heat compensation algorithm according to the real-time state parameters, and to acquire the spatial variation coefficients of multiple cross sections based on the three-dimensional evolution data of the frozen front. The first judgment unit is used to obtain the frozen design parameters, obtain the first threshold and the second threshold according to the frozen design parameters, and determine whether the spatial variation coefficient of each cross section is greater than the first threshold. The first identification unit is used to determine that the cross section is an unconnected defect surface if the spatial variation coefficient of the cross section is greater than the first threshold, and to obtain the unconnected defect features based on the unconnected defect surface. The second judgment unit is used to obtain the thickness deviation rate of each monitoring point according to the real-time status parameters, and to determine whether the thickness deviation rate of each monitoring point is greater than the second threshold. The second identification unit is used to determine that the monitoring point is a thickness deficiency defect point if the thickness deviation rate of the monitoring point is greater than the second threshold, and to obtain the thickness deficiency defect features based on the thickness deficiency defect point. The feature integration unit is used to construct a frozen defect feature set based on the thickness deficiency defect feature and the non-intersecting defect feature.

[0013] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described subway connection channel freezing method parameter optimization method.

[0014] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method for optimizing parameters of the subway connection channel freezing method.

[0015] The beneficial effects of this invention are as follows: By introducing multi-source data fusion and real-time dynamic perception, this invention overcomes the limitations of traditional methods that rely solely on static geological reports, solves the problem of failing to capture dynamic changes in geological conditions during construction, and provides a data foundation for subsequent dynamic optimization. Then, based on a coupled heat transfer and soil mechanics model, combined with initial state parameters, real-time state parameters are mapped to a three-dimensional formation stability field, and a formation stability field distribution map is constructed, thus intuitively presenting the stability state of the frozen wall. This addresses the limitation of traditional empirical formulas in reflecting complex geological responses. Next, through a multi-scale defect feature extraction algorithm, features such as unconnected defect surfaces and insufficient thickness defect points are extracted from the real-time state parameters, and a frozen wall defect feature set is constructed to achieve automatic defect identification and location. To address the shortcomings of traditional methods that rely on human experience to judge defects, which are subjective and lagging, a multi-objective optimization model is constructed. This model, combined with finite element simulation algorithms, optimizes process parameters based on the formation stability field distribution map and defect feature set to obtain the optimal parameter set. This enables adaptive parameter adjustment, solving the problem that traditional static parameters cannot cope with freezing defects caused by geological abrupt changes. Subsequently, a quality assessment value is obtained based on the optimal parameter set and real-time state parameters. This value is compared with a preset threshold; if it does not meet the standard, the freezing defect feature set is updated based on freezing efficiency evolution data, resulting in an updated defect feature set. This update is then returned to the parameter optimization step, forming a closed loop. This breaks through the "one-time optimization, unchanged throughout" model, ensuring that parameters adapt to dynamic geological conditions, guaranteeing the quality of the frozen wall, and achieving the best temporary formation reinforcement effect. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of a method flow according to an embodiment of the present invention.

[0017] Figure 2 This is a schematic diagram of the system structure according to an embodiment of the present invention.

[0018] Figure 3 This is a schematic diagram of the internal structure of a computer device according to an embodiment of this application.

[0019] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0021] like Figure 1 As shown, this application provides a parameter optimization method for the freezing of subway connection channels, including: S1. Obtain multi-source real-time monitoring data, which includes formation state parameters and process real-time parameters, and obtain initial state parameters and real-time state parameters based on the formation state parameters. S2. Obtain the formation stability field distribution map based on the initial state parameters and the real-time state parameters; S3. Obtain the freezing defect feature set based on the real-time status parameters; S4. Optimize the real-time process parameters based on the formation stability field distribution map and the freezing defect feature set to obtain the optimal parameter set; S5. Obtain freezing efficiency evolution data based on the optimal parameter set and the real-time status parameters, and obtain freezing efficiency evaluation feature set based on the freezing efficiency evolution data; S6. Obtain the quality assessment value based on the frozen work efficiency assessment feature set, and determine whether the quality assessment value is greater than a preset threshold. If the quality assessment value is greater than the preset threshold, then the optimal parameter set is determined to meet the construction requirements; If the quality assessment value is not greater than the preset threshold, then the defect feature update set is obtained based on the frozen efficiency evolution data and the frozen defect feature set, and the process is returned to the step of optimizing the real-time process parameters until the optimal parameter set meets the construction requirements.

[0022] As described in steps S1-S6 above, this invention acquires multi-source real-time monitoring data in real time through a distributed sensor network. This multi-source real-time monitoring data refers to a multi-dimensional dynamic data set reflecting the construction environment and technological status, including geological state parameters and real-time technological parameters. Geological state parameters are a set of parameters describing the physical and mechanical properties of the geological soil and rock mass, including initial state parameters and real-time state parameters. Real-time technological parameters refer to the technical parameters executed during the frozen construction process. By introducing multi-source data fusion and real-time dynamic sensing, this invention overcomes the limitations of traditional methods that rely solely on static geological reports, and solves the problem of geological conditions during construction. This addresses the problem of failing to capture dynamic changes in the formation and provides a data foundation for subsequent dynamic optimization. Then, based on the coupled model of heat transfer and soil mechanics, combined with the initial state parameters, the real-time state parameters are mapped into a three-dimensional formation stability field, and a formation stability field distribution map is constructed. The formation stability field distribution map refers to a three-dimensional visualization field map generated by interpolating discrete monitoring data, which reflects the spatial distribution of physical quantities such as formation temperature, stress, and seepage, thus intuitively presenting the stability state of the frozen wall. Furthermore, the visualization of the real-time field distribution provides a quantitative basis for subsequent defect identification, thereby overcoming the limitation that traditional empirical formulas cannot reflect complex geological responses. Next, a multi-scale defect feature extraction algorithm is used to extract features such as unconnected defect surfaces and insufficient thickness defect points from real-time state parameters, and a freezing defect feature set is constructed. The freezing defect feature set refers to the set of feature vectors extracted from real-time data that characterize abnormal freezing effects, thereby transforming the original monitoring data into quantifiable defect indicators and realizing automatic defect identification and location. This solves the problem that traditional methods rely on human experience to judge defects, which is subjective and lagging. This solution uses an intelligent algorithm to capture defect features in real time, providing precise targets for subsequent parameter optimization. Then, a multi-objective optimization model is constructed, and combined with the finite element simulation algorithm, the real-time process parameters are optimized based on the formation stability field distribution map and the freezing defect feature set to obtain the optimal parameter set. The optimal parameter set refers to the process optimization parameter combination that achieves the best balance of freezing effect, obtained by solving the multi-objective optimization algorithm. This optimization framework that integrates data-driven and physical model optimization achieves adaptive adjustment of parameters through real-time optimization algorithms, thereby solving the problem that traditional static parameters cannot cope with freezing defects caused by geological changes. Subsequently, freezing efficiency evolution data was obtained based on the optimal parameter set and real-time state parameters. This freezing efficiency evolution data refers to the time-series data set reflecting the development process of the frozen soil wall after the implementation of the optimized parameters. A freezing efficiency evaluation feature set was then obtained based on this data. This feature set consists of a set of multi-dimensional indicators extracted from the freezing efficiency evolution data to quantitatively evaluate the freezing effect. A quality evaluation value was then calculated using a weighted summation method based on the multi-dimensional indicators and their corresponding weight coefficients. The quality evaluation value is a quantitative score representing the degree to which the freezing effect meets the design requirements. Finally, the quality evaluation value was compared with a preset threshold. The values ​​are compared, and if they do not meet the standards, the freezing defect feature set is updated based on the freezing efficiency evolution data to obtain the defect feature update set. The defect feature update set is then used as the freezing defect feature set to return to the step of optimizing the real-time process parameters based on the formation stability field distribution map and the freezing defect feature set, forming a closed-loop control. This method of achieving dynamic parameter optimization through feedback correction breaks through the rigid mode of the traditional method of "one-time optimization, unchanged throughout the process". This scheme ensures that the parameters always adapt to the dynamic geological conditions through a real-time evaluation-iterative optimization cycle, ensuring that the frozen wall is uniformly encircled and the thickness meets the standard, thereby enabling the temporary formation to achieve the best reinforcement effect.

[0023] In one embodiment, step S2, which involves obtaining the formation stability field distribution map based on the initial state parameters and the real-time state parameters, includes: S21. Obtain spatiotemporal temperature data, pore hydrodynamic parameters and mechanical property parameters of multiple monitoring points based on the real-time state parameters, and obtain the actual compressive strength field based on the mechanical property parameters; S22. Obtain formation characteristic parameters based on the initial state parameters and the spatiotemporal temperature data; S23. Obtain the thermal stress tensor field and the frost heave pressure tensor field based on the spatiotemporal temperature data and the formation characteristic parameters, and obtain the total stress tensor field based on the thermal stress tensor field and the frost heave pressure tensor field. S24. Obtain the shear failure field based on the total stress tensor field; S25. Obtain the seepage drag force field using Darcy's law based on the pore hydrodynamic parameters and the formation characteristic parameters, and obtain the combined action force field based on the seepage drag force field and the shear failure field. S26. Obtain the formation stability field distribution map based on the comprehensive force field.

[0024] As described in steps S21-S26 above, this invention arranges a temperature sensor at regular intervals along the length of the freezing pipe and acquires spatiotemporal temperature data for each monitoring point based on the temperature sensors. The spatiotemporal temperature data refers to the set of temperature data collected from multiple monitoring points at different times and spatial coordinates within a specific monitoring area, typically represented as [(x, y, z, t, T), where (x, y, z) represents the real-time spatial coordinates of the monitoring point, t represents the monitoring time of the monitoring point, and T represents the real-time monitoring temperature of the monitoring point]. For the spatiotemporal temperature data, outliers (such as temperature jumps caused by fiber optic breakage) are first removed, and then missing data points are filled using Kriging interpolation to ensure the continuity of the spatiotemporal temperature data in three-dimensional space. The real-time monitoring temperature and real-time coordinates of each monitoring point are then acquired based on the spatiotemporal temperature data. Next, the initial ground temperature, initial coordinates, and formation characteristic parameters of each monitoring point are acquired based on the initial state parameters. Finally, the linear elastic stiffness matrix and elastoplastic stiffness matrix are acquired based on the formation characteristic parameters. The formation characteristic parameters describe the inherent physical, mechanical, and thermal properties of the formation. The parameters are then calculated, followed by the difference between the real-time monitored temperature and the initial ground temperature to obtain the real-time monitored temperature difference for each monitoring point. Based on the geological characteristics and the real-time monitored temperature, the thermal expansion coefficient, frost heave potential, soil permeability coefficient, and soil porosity are obtained for each monitoring point. Next, the product of the thermal expansion coefficient and the real-time monitored temperature difference is calculated to obtain the thermal strain tensor. The corresponding displacement components are obtained based on the initial and real-time spatial coordinates of the monitoring points. The partial derivatives of the displacement components are then obtained using the finite difference method to obtain the total strain tensor. Finally, based on the thermoelastic constitutive equation and combined with linear elasticity... The initial thermal stress tensor is obtained by calculating the stiffness matrix, total strain tensor, and thermal strain tensor. Then, the initial thermal stress tensor is corrected by solving the equilibrium equations using finite element software (such as COMSOL). Finally, the thermal stress tensor that meets the static equilibrium condition at each monitoring point is obtained, and a thermal stress tensor field is constructed based on this. The thermal stress tensor field is a physical field that reflects the distribution of stress in the formation due to temperature changes in three-dimensional space. The thermal stress tensor field can clarify the magnitude and direction of thermal stress at different spatial locations, providing key mechanical parameters for subsequent analysis of formation stability. Next, a three-dimensional temperature field is constructed based on spatiotemporal temperature data. The MarchingCubes algorithm is used to extract isosurfaces where the real-time monitored temperature equals the soil freezing temperature, which are the theoretical freezing fronts. The coordinate set of the freezing fronts is then obtained. Next, the frontal normal temperature gradient at each monitoring point is obtained based on the freezing front coordinate set and the real-time monitored temperature. The time step is obtained according to the freezing stage (e.g., rapid freezing period or stable freezing period). Then, the product of the frost heave potential parameter, the frontal normal temperature gradient, and the time step is calculated to obtain the single-step frost heave strain increment at each monitoring point. The single-step frost heave strain increments of all time steps are then accumulated to obtain the total frost heave strain, forming the total frost heave strain distribution at each point in three-dimensional space, i.e., the frost heave strain field. Subsequently, the maximum value is extracted from all the total frost heave strain variables in the frost heave strain field. The free frost heave strain is used as the variable, and the difference between the total frost heave strain and the free frost heave strain at each monitoring point is calculated. This difference reflects the degree of constraint on the frost heave strain (the larger the difference, the stronger the constraint). Then, this difference is used to perform a tensor product operation with the elastoplastic stiffness matrix to obtain the frost heave pressure increment at each monitoring point. Subsequently, the frost heave pressure increment is used as the initial stress input model in finite element software (such as COMSOL) to solve the equilibrium equation and correct it. Finally, the frost heave pressure tensor field that meets the static equilibrium condition at each monitoring point is obtained. The frost heave pressure tensor field refers to the physical field that describes the distribution of pressure generated by the expansion of water during the freezing process of the stratum in three-dimensional space. The frost heave pressure tensor field can clarify the intensity and constraint difference of frost heave in three-dimensional space, and provide frost heave mechanical parameters for evaluating the stability of the frozen wall. Next, the frost heave pressure tensor field and the thermal stress tensor field are superimposed using the tensor addition rule to obtain the total stress tensor field of each monitoring point, reflecting the distribution of the total stress borne by the formation under temperature changes and frost heave in three-dimensional space. Then, the Mises equivalent stress criterion is used to calculate the shear failure force of each monitoring point based on the total stress tensor field. Based on the ordered shear failure force dataset, the shear failure forces of adjacent sampling points are connected by Gaussian fitting to obtain the shear failure field. The shear failure field refers to the physical field that reflects the shear failure risk at each point of the formation. Existing technologies usually only assess the impact of a single type of stress on the formation, resulting in an incomplete stress assessment. This scheme comprehensively considers the coupled mechanical effects of temperature changes and frost heave, avoiding the one-sidedness of assessing a single stress factor. Subsequently, hydrodynamic viscosity and pore water pressure data at each monitoring point were obtained based on pore hydrodynamic parameters. Pore hydrodynamic parameters describe the movement characteristics of fluids (mainly groundwater) in the formation pores. The water pressure gradient at each monitoring point was then obtained based on the pore water pressure data. Next, the seepage velocity at each monitoring point was calculated using Darcy's law based on the soil permeability coefficient, hydrodynamic viscosity, and water pressure gradient, thus obtaining the seepage velocity field. Finally, the seepage drag force at each monitoring point was calculated based on the soil porosity, seepage velocity, and water density. A seepage drag force field is formed, which refers to the physical field reflecting the three-dimensional distribution of the drag force generated by groundwater seepage on the formation skeleton. The cohesion field, internal friction angle field, and total stress are obtained based on mechanical property parameters. These mechanical property parameters characterize the mechanical behavior of formation materials. The pore water pressure field is obtained by calculating the product of the pore water pressure and Biot coefficient at each monitoring point. The real-time effective stress at each monitoring point is then obtained based on the difference between the pore water pressure field and the total stress, and a real-time effective stress field is constructed accordingly. Finally, based on Mohr's law... - The Coulomb strength formula obtains the actual compressive strength field based on the cohesive force field, real-time effective stress, and internal friction angle field. The actual compressive strength field refers to the three-dimensional spatial distribution field constructed based on the real-time mechanical state of the stratum, reflecting the maximum pressure (or stress) that each point of the stratum can actually withstand. It serves as a "safety threshold" for assessing stratum stability and provides a benchmark for subsequent comparison of the comprehensive force field (damage factors). Finally, the shear failure field and the seepage drag force field are vector-superimposed to obtain the comprehensive force field, which is then compared point by point with the actual compressive strength field to identify stress exceeding the limit and obtain the exceedance amplitude. Based on the spatial distribution of stress exceeding the limit area and the exceedance amplitude, a graded early warning stratum stability field distribution map is generated. Existing technologies cannot dynamically and multi-factor assess stratum stability, and relying solely on experience judgment can easily lead to missed risk assessments. However, this scheme generates a graded early warning stability field distribution map through multi-field coupling analysis, solving the limitations of static and single-factor assessment. It realizes real-time, accurate, and visual monitoring of stratum stability, providing a direct basis for dynamically adjusting construction parameters and ensuring subsequent construction safety and stratum stability.

[0025] In one embodiment, step S3, which involves obtaining the frozen defect feature set based on the real-time status parameters, includes: S31. Based on the phase change latent heat compensation algorithm, obtain the three-dimensional evolution data of the frozen front according to the real-time state parameters, and obtain the spatial variation coefficients of multiple cross sections according to the three-dimensional evolution data of the frozen front. S32. Obtain the frozen design parameters, and obtain the first threshold and the second threshold according to the frozen design parameters, and determine whether the spatial variation coefficient of each cross section is greater than the first threshold. S33. If the spatial variation coefficient of the cross section is greater than the first threshold, the cross section is determined to be an unconnected defect surface, and the unconnected defect features are obtained based on the unconnected defect surface. S34. Obtain the thickness deviation rate of each monitoring point according to the real-time status parameters, and determine whether the thickness deviation rate of each monitoring point is greater than the second threshold. S35. If the thickness deviation rate of the monitoring point is greater than the second threshold, the monitoring point is determined to be a thickness deficiency defect point, and the thickness deficiency defect characteristics are obtained based on the thickness deficiency defect point. S36. Construct a frozen defect feature set based on the thickness deficiency defect feature and the non-intersecting defect feature.

[0026] As described in steps S31-S36 above, this invention obtains the temperature time-series data of each monitoring point through real-time state parameters, and obtains the real-time temperature change rate of each monitoring point based on the temperature time-series data. Then, it determines whether each monitoring point is in the phase transition stage based on the real-time monitored temperature and the real-time temperature change rate. If a monitoring point simultaneously meets the conditions of "real-time monitored temperature close to the freezing point of water (e.g., -2℃ ≦ real-time monitored temperature ≦ 1℃)" and "real-time temperature change rate approaching 0", then it is considered that the monitoring point is in the water-ice phase transition stage. Next, the latent heat of phase change of water and the equivalent specific heat capacity of the soil are obtained, and the ratio of the latent heat of phase change of water to the equivalent specific heat capacity of the soil is calculated to obtain the latent heat compensation amount. Then, the difference between the real-time monitored temperature and the latent heat compensation amount is calculated, and... The difference is used as the final temperature value of the monitoring point. If the monitoring point does not meet any of the above conditions, it is considered that the monitoring point is not in the water-ice phase transition stage. In this case, the real-time monitoring temperature is used as the final temperature value of the monitoring point. Through this phase transition latent heat compensation algorithm, the temperature monitoring value in the water-ice phase transition stage can be accurately corrected, eliminating the interference of latent heat release during the phase transition on the temperature measurement. This ensures that the real-time temperature data can truly reflect the actual position and evolution state of the freezing front. It can also specifically correct the temperature value of the monitoring point in the phase transition stage. This can solve the problem that the existing technology relies solely on the original temperature data to calculate the freezing front, which is prone to misjudgment of the freezing front position in the phase transition stage, thus affecting the accuracy of freezing defect identification. This lays the foundation for the accurate acquisition of subsequent three-dimensional evolution data of the freezing front. Next, based on the three-dimensional temperature field, the MarchingCubes algorithm is used to extract isosurfaces whose final temperature value equals the soil freezing temperature, which are the actual freezing fronts at the current moment. Then, based on the actual freezing fronts arranged in a time series, the frontal normal displacement and adjacent time steps are obtained, and the ratio of the frontal normal displacement to the adjacent time step is calculated to obtain the frontal expansion rate. Through velocity analysis of continuous time steps, three-dimensional evolution data of the freezing front is formed. The three-dimensional evolution data of the freezing front refers to the dynamic change information of the freezing front in three-dimensional space over time, obtained through real-time monitoring and algorithm processing during the freezing process. Then, multiple cross-sections are taken along the axis of the connecting channel (e.g., every 0.5 meters), and based on the three-dimensional evolution data of the freezing front, the adjacent... If the shortest distance between all adjacent actual freezing fronts at a critical section is less than or equal to a preset tolerance, then the frozen wall at that section is considered to have formed a loop. Each spatial point within that section is marked with its looping status (looped / unlooped). Finally, the discrete looping status data is transformed into a continuous surface using a spatial interpolation algorithm and visualized using color mapping: for example, red areas represent unlooped areas and green areas represent looped areas, thus forming a frozen wall looping status surface. This enables intuitive and quantitative monitoring of the frozen wall looping status, thereby solving the problems in existing technologies where it is difficult to dynamically and accurately determine whether the frozen wall has formed a loop, and the looping status cannot be intuitively presented. This provides a clear spatial basis for the subsequent identification of unlooped defects. After obtaining multiple cross-sectional views of the connecting channels, when the distance between two freezing pipes does not exceed the design specification, they are identified as adjacent freezing pipes. Then, for each pair of identified adjacent freezing pipes, temperature data is extracted along the axial direction of the two adjacent freezing pipes at the same elevation, obtaining the first temperature value of pipe A at that elevation and the second temperature value of pipe B at the same elevation. The difference between the first and second temperature values ​​is then calculated. Next, the distance between adjacent freezing pipes is obtained based on the spatial coordinates of pipes A and B, and the ratio of this difference to the distance between adjacent freezing pipes is calculated to obtain the temperature gradient value of the adjacent freezing pipes. Finally, within the same cross-section of the connecting channel, the temperature gradient values ​​of all adjacent freezing pipe pairs are summarized to form a gradient dataset. The method obtains the standard deviation and average temperature gradient from the gradient dataset, then calculates the ratio of the standard deviation to the average temperature gradient to obtain the spatial coefficient of variation. Finally, it compares the spatial coefficient of variation with a first threshold. The spatial coefficient of variation describes the dispersion of the temperature gradient between adjacent pairs of freezing tubes within the same cross-section. The first threshold is a critical value set based on freezing design parameters to determine whether a cross-section contains unconnected defects. If the spatial coefficient of variation is greater than the first threshold, the cross-section is determined to have unconnected defects, and multiple unconnected defects constitute an unconnected defect region. This method solves the problem of relying on human experience to judge unconnected defects in existing technologies, which is subjective. To address the issues of high variability, low precision, and difficulty in standardization, this paper improves the objectivity and accuracy of identifying unconnected defects. Based on the identified unconnected defect areas, the MarchingCubes algorithm is used to extract the discontinuous parts of the actual frozen front, obtaining the spatial boundary of the unconnected defect area. This boundary consists of a series of discrete vertices. Therefore, combined with the three-dimensional temperature field, a set of coordinates for the unconnected defect boundary is obtained. The coordinates of the defect center are then calculated using an arithmetic mean based on this set. Next, the coordinates of all vertices in the unconnected defect area are projected onto a cross-section (two-dimensional plane) perpendicular to the axis of the connecting channel, resulting in a two-dimensional point set. Finally, the straight-line distance between any two points is calculated based on this two-dimensional point set. The maximum value is the maximum gap width. The defect projection area of ​​each cross section is calculated using a polygon area formula (such as the Shoelace formula). Then, based on the frozen design parameters, the theoretical closed area of ​​each cross section is obtained. Finally, the defect projection area of ​​each cross section is compared with the theoretical closed area to obtain the proportion of unclosed area of ​​each cross section. The unclosed defect features are constructed by the defect center coordinates, the maximum gap width, and the proportion of unclosed area. This method, which deepens the identification of unclosed defects from regional identification to multi-dimensional feature quantification, can solve the problem that the description of unclosed defects in the existing technology is only at the level of existence, lacking specific quantitative parameters, and making it difficult to guide the optimization of subsequent process parameters. The actual thickness distribution field of the frozen wall is obtained based on real-time state parameters using ultrasonic tomography, and the designed thickness distribution field of the frozen wall is determined according to the freezing design parameters. Then, for each monitoring point in three-dimensional space, the difference between the designed thickness and the actual thickness of the frozen wall is calculated, and the ratio of this difference to the designed thickness is calculated to obtain the thickness deviation rate for each monitoring point (a positive value indicates that the actual thickness is less than the designed thickness, and a negative value indicates that the actual thickness is greater than the designed thickness). Next, the thickness deviation rate is compared with a second threshold. If the thickness deviation rate is greater than the second threshold, which is a critical value set based on the freezing design parameters to determine whether a monitoring point has a thickness deficiency defect, then the monitoring point is determined to be a thickness deficiency defect point. If the thickness deviation rate is not greater than the second threshold, then the monitoring point is determined not to be a thickness deficiency defect point. Subsequently, a 3D connected component labeling algorithm (such as the CCL algorithm) is used to group adjacent defect points into the same region, identifying multiple unique defects. The method involves identifying areas of insufficient thickness defects, then selecting the minimum thickness of defect points within these areas. The three-dimensional space is then discretized into voxels (e.g., cubes with sides of 0.1 meters). The number of voxels marked as "insufficient thickness defect points" within the defect area is counted, and multiplied by the volume of a single voxel to obtain the defect volume. The total volume of the defect area is then calculated, and the ratio of the defect volume to the total volume is used to determine the spatial distribution density of the defect. The spatial location of the insufficient thickness defect points, the minimum thickness value, the defect volume, and the spatial distribution density constitute the characteristics of insufficient thickness defects. This method achieves quantitative and regional identification of insufficient thickness defects, solving the problems of difficulty in accurately measuring the thickness distribution of frozen walls and the lack of quantitative standards and regional division for identifying insufficient thickness defects in existing technologies. This improves the accuracy and systematic nature of thickness defect identification. Then, the maximum gap width, unclosed area ratio of non-loop defects, minimum thickness value of insufficient thickness defects, and defect spatial distribution density are normalized. The processed non-loop defect data is organized into the form of [type identifier "01", defect center coordinates, maximum gap width, maximum gap width], and the processed insufficient thickness defect data is organized into the form of [type identifier "02", minimum thickness value, defect spatial distribution density]. All data are merged to generate a frozen defect feature set containing timestamps. This solves the problem of scattered and inconsistent defect information in existing technologies, which makes it difficult for parameter optimization algorithms to effectively utilize the information. It achieves efficient connection between defect features and optimization decisions, and provides structured input for precise optimization of process parameters.

[0027] In one embodiment, step S4, which optimizes the real-time process parameters based on the formation stability field distribution map and the freezing defect feature set to obtain the optimal parameter set, includes: S41. Obtain the severity of defects in each independent defect region based on the frozen defect feature set, and obtain the geological sensitivity coefficient of each independent defect region based on the frozen defect feature set and the formation stability field distribution map. S42. Obtain the process dynamic control coefficient based on the real-time status parameters, and obtain the initial adjustment value of the process parameters based on the process dynamic control coefficient and the real-time process parameters; S43. Obtain the process parameter adjustment range based on the initial adjustment value of the process parameters and the geological sensitivity coefficient, and obtain multiple parameter optimization groups based on the process parameter adjustment range; S44. Obtain the parameter optimization objective and process optimization constraints, and filter multiple parameter optimization groups according to the parameter optimization objective and process optimization constraints to obtain the optimal parameter group.

[0028] As described in steps S41-S44 above, this invention calculates the severity of defects in each independent defect region based on the characteristics of unconnected defects on unconnected defect surfaces and the characteristics of insufficient thickness defects at insufficient thickness points using a weighted summation formula. The severity of defects refers to an index used to quantify the severity of problems in independent defect regions. Based on the distribution map of the formation stability field, the stress exceedance range of each independent defect region is obtained. Then, by weighted fusion of the stress exceedance range and the severity of defects, a geological sensitivity coefficient for each independent defect region is obtained. The geological sensitivity coefficient characterizes the risk sensitivity of a defect region under dynamic changes in geological conditions. Construction parameters are optimized according to the geological sensitivity coefficient from largest to smallest to prioritize the treatment of high-risk areas. This achieves risk-oriented optimization and can solve the problems of insufficient targeting in parameter optimization in existing technologies, untimely treatment of high-risk defect regions, and the potential for safety accidents. Next, based on real-time process parameters, the current brine temperature and flow rate are obtained. Using a three-dimensional heat conduction equation, the expansion rate of the actual freezing front is simulated, and the theoretical time required for the actual freezing front to expand from its current position to complete closure is calculated. Then, based on freezing design parameters, the design closure time is obtained, and the difference between the theoretical time and the design closure time is calculated to obtain the closure delay time. Then, based on the process dynamic control coefficients, the heat conduction compensation coefficient, formation thermal conductivity, flow rate regulation coefficient, frost heave sensitivity coefficient, and temperature adjustment coefficient are obtained. The process dynamic control coefficients refer to a set of adjustment coefficients reflecting the dynamic changes of freezing process parameters (such as brine temperature, flow rate, and liquid nitrogen quantity) with geological conditions and defect characteristics. The product of the heat conduction compensation coefficient, formation thermal conductivity, and closure delay time is calculated to obtain the temperature compensation amount. The sum of the current brine temperature and the temperature compensation amount constitutes the brine adjustment temperature. Then, the temperature is adjusted using the formula… "Calculate the brine adjustment flow rate for the cross-section with uncrossed defects, where, This indicates the adjusted flow rate of the brine at that cross-section. This indicates the current brine flow rate at that cross-section. This indicates the percentage of the unclosed area of ​​the cross-section. This represents the flow rate adjustment coefficient. The formula is based on the current brine flow rate and the proportion of unclosed area. By introducing the flow rate adjustment coefficient, the flow rate is dynamically adjusted to compensate for insufficient local heat transfer. The core idea is that a larger proportion of unclosed area indicates a delayed expansion of the freezing front in that region, requiring a significant increase in brine flow rate to enhance the heat exchange rate and accelerate freezing. Conversely, if the proportion of unclosed area is small, only a slight adjustment in flow rate is needed. The flow rate adjustment coefficient determines the sensitivity of the adjustment and is usually determined by fluid heat transfer experiments or engineering experience to ensure that the flow rate change matches the heat conduction requirements. Ultimately, the adjusted flow rate promotes the freezing process of the unclosed area by improving convective heat transfer efficiency, leading to uniform closure of the frozen wall. This method allows the initial adjustment value to accurately match the compensation requirements for unclosed defects, solving the problem in existing technologies where process parameter adjustments rely on experience and lack quantitative calculations that dynamically correlate with defect characteristics and freezing progress, resulting in poor adjustment effects. Next, the product of the frost heave sensitivity coefficient and the thickness deviation rate is calculated to obtain the pulse freezing frequency. By coupling frost heave risk and frozen wall uniformity, the freezing rhythm is dynamically adjusted to suppress frost heave while promoting uniform development of the frozen wall. The liquid nitrogen efficiency coefficient is obtained, and the product of the liquid nitrogen efficiency coefficient, defect volume, and minimum thickness deviation is calculated to obtain the liquid nitrogen compensation amount. The sum of the current liquid nitrogen amount and the liquid nitrogen compensation amount constitutes the liquid nitrogen adjustment amount. Through this method based on defect scale and local strength gap, combined with liquid nitrogen usage efficiency, the amount of additional liquid nitrogen to be injected is calculated to directionally strengthen the defect area and ensure the overall stability of the frozen wall. This method can achieve dynamic adaptation of freezing rhythm with frost heave risk and frozen wall uniformity, and make the liquid nitrogen injection amount accurately match the defect scale and strength gap, thereby improving liquid nitrogen usage efficiency. This solves the problems of fixed freezing rhythm in existing technologies, which makes it difficult to balance frost heave suppression and freezing efficiency, and the lack of targeted liquid nitrogen injection, which cannot effectively strengthen the defect area. It is beneficial to improve the dynamic control capability of freezing process and defect repair effect. Data such as brine temperature adjustment, brine flow rate adjustment, and liquid nitrogen adjustment amount together constitute the initial adjustment values ​​of the process parameters. Then, based on the geological sensitivity coefficient and temperature adjustment coefficient (determined through frost heave experiments, generally 1~3℃), the temperature adjustment range is obtained. For example, according to the formula " Calculate the minimum temperature adjustment value and apply it according to the formula. "Calculate the maximum temperature adjustment value, where, This indicates the minimum temperature adjustment value. This indicates the maximum temperature adjustment value. This indicates that the brine temperature has been adjusted. Indicates the temperature adjustment coefficient. The temperature adjustment range is defined by the geological sensitivity coefficient, the minimum temperature adjustment value, and the maximum temperature adjustment value. Similarly, the flow rate adjustment range is calculated based on the flow rate adjustment coefficient, the geological sensitivity coefficient, and the brine flow rate. The pulse frequency range is calculated based on the frequency adjustment coefficient, the geological sensitivity coefficient, and the pulse freezing frequency. The liquid nitrogen quantity adjustment range is calculated based on the liquid nitrogen adjustment coefficient, the geological sensitivity coefficient, and the liquid nitrogen quantity. These temperature, flow rate, pulse frequency, and liquid nitrogen quantity adjustment ranges constitute the process parameter adjustment range. Values ​​are then discretized within these ranges to generate candidate parameter values ​​for each optimized parameter. Multiple candidate parameter values ​​are then combined to obtain multiple parameter optimization groups. This method of generating numerous parameter optimization groups ensures the comprehensiveness and flexibility of parameter optimization and provides ample candidate solutions for multi-objective optimization. It addresses the problems in existing technologies where parameter optimization ranges are fixed, dynamic adjustments related to geological sensitivity are lacking, and candidate parameter combinations are limited, making it difficult to find the optimal solution. Subsequently, with uniform circumference of the frozen wall and achieving the required thickness as optimization objectives, a function was first constructed to quantify these two objectives: For the objective of uniform circumference of the frozen wall, the objective function was constructed by calculating the weighted sum of the areas of all non-circumference defect regions, as shown in the formula "". ",in, This represents minimizing the area of ​​the unclosed area. This indicates the number of non-intersecting defect areas. Indicates the serial number of the non-intersecting defect area. Indicates the first The area of ​​each unconnected defect region Indicates the first The geological sensitivity coefficient of unconnected defect areas; for thickness compliance targets, the objective function is constructed by calculating the weighted sum of deviations of all insufficient thickness defect areas, the formula is "". ",in, This indicates minimizing the thickness deviation. This indicates the number of areas with insufficient thickness due to defects. The serial number indicates the region with insufficient thickness defect. Indicates the first Total thickness deviation rate of each thickness-deficient defect area Indicates the first The total volume of the defect area with insufficient thickness was considered. Three types of constraints were set to ensure project safety and feasibility: Formation stability constraint: the frost heave pressure must not exceed the difference between the formation compressive strength and the reserved stress safety margin; Equipment capacity constraint: the sum of the products of the flow rate and temperature adjustment of all freezing pipes must not exceed the upper limit of the total power of the refrigeration system; Construction feasibility constraint: the minimum spacing between freezing pipes must not be less than the design value, and the daily liquid nitrogen consumption must not exceed the reserve limit. Based on these constraints, a multi-objective optimization was performed on multiple parameter groups using the Non-Dominated Sorting Genetic Algorithm (NSGA-II): 100 parameter groups containing adjustable variables such as freezing pipe temperature, flow rate, and liquid nitrogen consumption were randomly selected from the candidate parameter groups as the initial population. Each parameter group was considered an "individual." For each individual, the objective function value was calculated to minimize the unclosed area and minimize the thickness deviation. Non-dominated sorting was performed according to the "Pareto dominance" relationship. Solutions not dominated by any other individual were assigned to the first layer (Pareto front), solutions dominated only by individuals in the first layer were assigned to the second layer, and so on. Simultaneously, the crowding degree of individuals within the same layer was calculated, prioritizing... The algorithm retains the dispersed solutions and then performs genetic operations, using a roulette wheel selection method to choose parent individuals (individuals with smaller objective function values ​​are more likely to be selected). The selected parent parameter sets are then cross-combined to generate offspring parameters, and the parameter values ​​are fine-tuned with a certain probability to increase population diversity. This process of non-dominated sorting, crowding calculation, and genetic operations is repeated until the rate of change of the Pareto front is less than 5% or the maximum number of iterations (usually 50-100 generations) is reached. Finally, the algorithm outputs a set of non-dominated solutions, i.e., the Pareto optimal solution set. Then, the optimal parameter set is selected based on engineering requirements: for high-risk areas, the solution that minimizes the unclosed area is prioritized; for low-risk areas, the solution that minimizes the thickness deviation is prioritized; and for equilibrium requirements, the solution that minimizes the sum of the unclosed area and the thickness deviation is selected. This method achieves scientific decision-making under multiple objectives and constraints, ensuring that the optimal parameter set optimizes the freezing effect while meeting safety and feasibility requirements. It solves the problem in existing technologies where parameter optimization has a single objective, does not consider multi-objective balance and engineering constraints, and is difficult to obtain a globally optimal solution.

[0029] In one embodiment, step S5, which involves obtaining freezing efficiency evolution data based on the optimal parameter set and the real-time state parameters, and obtaining a freezing efficiency evaluation feature set based on the freezing efficiency evolution data, includes: S51. Obtain basic modeling data based on the real-time state parameters and the formation stability field distribution map, and construct a multiphysics coupling body based on the basic modeling data; S52. Obtain freezing efficiency evolution data based on the multiphysics coupling body and the optimal parameter set, wherein the freezing efficiency evolution data includes temperature field evolution data, formation response data and frozen wall development data; S53. Obtain the freezing design target, and obtain the project schedule compliance based on the freezing design target and the temperature field evolution data; S54. Obtain the stability coefficient based on the freezing design target and the formation response data, and obtain the thickness compliance volume ratio and the cross-ring compliance index based on the freezing design target and the frozen wall development data. S55. The intersection compliance index, construction period consistency, stability coefficient, thickness compliance volume ratio and the intersection compliance index are used as the feature set for freezing efficiency evaluation.

[0030] As described in steps S51-S55 above, this invention utilizes the freezing method principle to build a finite element model in COMSOL based on modeling foundation data. This modeling foundation data refers to the collection of various basic information required to construct a multi-physics coupled body. A three-dimensional geometric framework is constructed by importing the axial coordinates of the subway connecting passage BIM model, and geological exploration data is used to divide the geological strata into layers (such as clay layers, sand layers, etc.). Next, the actual installation positions of the freezing pipes are accurately entered into the model to ensure that the spatial distribution of the freezing pipes in the simulation is consistent with the engineering site. Simultaneously, a geological stability field distribution map is incorporated to reflect the differences in geological stability in different areas. Then, boundary conditions are set: based on the three-dimensional temperature field, it is mapped to each node of the finite element mesh using interpolation, with the current frozen wall thickness as the structural boundary. The actual freezing front is used as the initial state of the simulation, and the model is given an initial stress state by combining the stratum's self-weight and the combined force field generated by previous freezing. Finally, an "adiabatic" condition is set at the outermost boundary of the model (such as the stratum boundary far from the freezing area) to ignore the thermal interference of the distant environment on the freezing area. The model's bottom boundary is constrained to be "fixed" to prevent unreasonable deformation due to the "infinite domain." The surface boundary is not subject to displacement constraints, allowing for free settlement or uplift; this setting realistically reflects surface deformation characteristics. Temperature-dependent piecewise functions are used to characterize material parameters. For example, when the finite element mesh node temperature is higher than the freezing temperature, the thermal conductivity of the unfrozen soil is 1.2 W / (m·K), and the cohesion is c=50 kPa; when the finite element mesh node temperature is lower than the freezing temperature, the thermal conductivity of the unfrozen soil is 2.5 W / (m·K). With a cohesion of c=100kPa and a mass of m·K, this method of constructing a multiphysics coupling body achieves a high degree of simulation of the model from geometry to physical properties. The multiphysics coupling body refers to a three-dimensional simulation model constructed in finite element software (such as COMSOL) that can simultaneously simulate the interaction of multiple physical fields such as temperature field, stress field, and strain field during the freezing process. It can solve the problem that the model simplification is too high in the existing technology, and the geological stratification, the actual location of the freezing pipe, and the temperature dependence of material parameters are not considered, resulting in large deviations between the simulation results and the actual situation. The optimal parameter set is then input into a multiphysics coupling body, and an adaptive time-stepping strategy (initial step size 0.1 hours, maximum step size not exceeding 2 hours) is used to perform calculations. GPU acceleration is enabled, and the mesh is refined in key areas (such as near the actual freezing front) and coarse in non-critical areas to balance accuracy and efficiency. The system simulates the process of the freezing pipe releasing cold energy into the formation, and finally predicts the temperature field evolution data (simulating the three-dimensional temperature distribution and freezing front expansion at each moment), freezing wall development data (calculating the freezing wall thickness at different moments and comparing it with the design value), and formation response data (outputting the stress, strain, and displacement fields of the formation to analyze stability). This method can solve the problem of difficulty in comprehensively and dynamically obtaining freezing efficiency evolution data in existing technologies. Among them, freezing efficiency evolution data refers to the dataset obtained by simulation through multiphysics coupling body, which reflects the dynamic changes of freezing process effect over time and the problem of difficulty in balancing simulation efficiency and accuracy. This provides rich and accurate basic data for subsequent efficiency evaluation. The system obtains the predicted construction period based on temperature field evolution data, and simultaneously acquires data such as the planned construction period, frozen wall design thickness, and minimum design curvature based on freezing design objectives. It calculates the absolute value of the difference between the predicted and planned construction periods, divides this absolute value by the planned construction period, and finally subtracts the result from "1" to obtain the construction period consistency rate. This index directly reflects the consistency between the predicted and planned construction periods; the closer the value is to 1, the smaller the deviation and the higher the consistency rate; the lower the value, the larger the deviation and the lower the consistency rate. By using numerical values ​​to intuitively reflect the construction period deviation, the system achieves quantitative and intuitive construction period assessment, solving the problem of insufficient targeted construction period control in existing technologies and improving the scientific nature of construction progress management. Based on the formation response data, the maximum principal stress at each monitoring point is extracted. Simultaneously, the ultimate principal stress of the formation is obtained from historical data (determined through geotechnical tests; for example, 2.5 MPa is taken for frozen sand, which is a safe threshold). This is then calculated using the formula… Calculate the stability coefficient, where, Represents the stability coefficient. Indicates the maximum principal stress. Indicates the ultimate principal stress of the formation. This represents the total volume of the analysis region (the sum of all tiny volumetric units). This represents a small volumetric unit (such as 1 m³ of soil, used for integration and accumulation). Since stability cannot be directly calculated, but "instability" (stress exceeding the limit) can be quantified, the formula uses triple integration to accumulate the "proportion of exceeding the limit" of all small volumetric units in the entire analysis area to obtain the "overall influence of exceeding the limit stress". The "overall influence of exceeding the limit stress" is then transformed into the "proportion of exceeding the limit stress in the analysis area", i.e., the "proportion of exceeding the limit influence". Subtracting the "proportion of exceeding the limit influence" from 1 yields the stability coefficient, which inversely represents the overall stability. The less the exceedance, the closer the stability coefficient is to 1; the more the exceedance, the closer the coefficient is to 0. This solves the problem that existing technologies for assessing formation stability are mostly local point assessments, lacking global and quantitative indicators, and are difficult to reflect the overall stability state. Similarly, the predicted value of the frozen wall thickness is obtained based on the frozen wall development data, and then calculated using the formula " "Calculate the volume ratio that meets the thickness standard, where, Indicates the volume ratio that meets the thickness standard. This represents the predicted value of the frozen wall thickness. Indicates the design thickness of the frozen wall. Represents the total volume of the analysis region. Representing a tiny volume unit, the formula indicates the function " The function of the "thickness compliance index" is to "screen" compliant areas: the value of 1 is assigned to locations where the thickness meets the standard, and these areas are included in the volume calculation; the value of 0 is assigned to locations where the thickness does not meet the standard, and these areas are not included in the calculation. The volume of all "compliant points" (spatial points whose thickness meets the design requirements) within the analysis area is summed through triple integration. Then, the influence of the size of the analysis area is eliminated by the ratio of "compliant volume / total volume", resulting in the thickness compliance volume ratio. The higher this index, the better the overall compliance of the frozen wall thickness and the stronger the engineering applicability. This achieves a comprehensive and objective quantitative assessment of the compliance of the frozen wall thickness. Based on the development data of the frozen wall, cross-sections are taken every 0.5 meters along the tunnel axis. The length of the arc segment with a thickness not less than the minimum design arc is extracted from each cross-section. The ratio of the total closed arc length of all cross-sections to the total design perimeter is calculated to obtain the predicted intersection closure degree. Then, the ratio of the predicted intersection closure degree to the target closure degree is calculated, and this ratio is compared with "1". The minimum value of the two is selected as the intersection compliance index to achieve accurate quantification of the intersection compliance degree and accurately reflect the gap with the target closure degree.

[0031] In one embodiment, step S6, which involves obtaining a defect feature update set based on the frozen ergonomic evolution data and the frozen defect feature set, includes: S61. Obtain the updated coordinates of the defect center based on the frozen efficiency evolution data and the real-time status parameters; S62. Obtain the unconnected defect update features based on the freezing efficiency evolution data and the freezing defect feature set, wherein the unconnected defect update features include the maximum gap update width and the unclosed area update ratio. S63. Obtain the thickness growth compensation coefficient based on the freezing efficiency evolution data, and obtain the thickness deficiency defect update feature based on the thickness growth compensation coefficient and the freezing defect feature set, wherein the thickness deficiency defect update feature includes the minimum thickness update value and the defect spatial distribution update density. S64. Construct a defect feature update set based on the defect center update coordinates, the non-intersecting defect update features, and the insufficient thickness defect update features.

[0032] As described in steps S61-S64 above, this invention obtains a real-time temperature gradient field from a three-dimensional temperature field constructed based on spatiotemporal temperature data, and obtains a predicted temperature gradient field based on temperature field evolution data. Then, by comparing the spatial gradients of the predicted and real-time temperature gradient fields, the defect center coordinates of the original uncoiled defects are corrected to the extreme point of temperature anomalies, resulting in updated defect center coordinates. These updated coordinates refer to the latest spatial coordinates obtained after dynamically correcting the original center coordinates of uncoiled defects or insufficient thickness defects. This closely links defect location with the dynamic changes in the temperature field, achieving accurate tracking of defect center coordinates. Next, the actual installation coordinates of each freezing tube are obtained, and the design coordinates of each freezing tube are obtained according to the freezing design parameters. The straight-line distance between the design coordinates and the actual installation coordinates is calculated to obtain the geometric compensation amount, and then calculated according to the formula… "Calculate the maximum gap update width, where, Indicates the maximum gap update width. Indicates the maximum gap width. Indicates the achievement index of the circle. This represents the geometric compensation amount, and the "" in the formula "It corrects the original gap by adjusting the actual looping progress (the smaller the looping compliance index, the slower the actual progress, and the higher the proportion of the original gap retained)," the formula states. "This is to supplement the new gaps caused by pipeline layout errors (e.g., if the spacing between frozen pipes is 0.1m larger than the design value, the geometric compensation is positive, increasing the gap width). Ultimately, through these two dynamic corrections, the current true width of the unclosed defect is comprehensively reflected, achieving dynamic and comprehensive correction of the defect width. This solves the problem in existing technologies where the width of unclosed defects is determined only based on initial data, without considering construction errors and dynamic changes in the closure progress, leading to distorted descriptions of defect width. Next, the area of ​​new defects is obtained based on the development data of the frozen wall, and the proportion of new unclosed area is obtained based on the area of ​​new defects. Then, the proportion of unclosed area and the proportion of new unclosed area are used to replace the maximum gap width and the geometric compensation, respectively, and substituted into the above formula to calculate the updated proportion of unclosed area. This takes into account both the changes of the original defect with the actual closure progress and the new defects discovered in the prediction, ensuring that the updated proportion of unclosed area can dynamically reflect the latest defect range. This method can solve the problem of lagging defect range assessment in existing technologies." Then, the ratio of the predicted frozen wall thickness to the designed frozen wall thickness is calculated to obtain the thickness growth compensation coefficient. A modified Nixon ice lens model is used to obtain the frost heave compensation amount by real-time monitoring of the temperature difference. Next, the minimum thickness value, thickness growth compensation coefficient, and frost heave compensation amount for the insufficient thickness defect area are replaced with the maximum gap width, the intersection compliance index, and the geometric compensation amount, respectively, and substituted into the above formula to calculate the updated minimum thickness value. This formula not only adjusts the minimum thickness value according to the predicted thickness growth trend but also supplements the thickness increment caused by frost heave. Then, the density of newly added defect points is obtained based on the frozen wall development data, and the formula is used… "Calculate the spatial distribution update density of defects, where, This represents the updated density of the spatial distribution of defects. Indicates the spatial distribution density of defects. This represents the thickness growth compensation coefficient. This formula represents the density of newly added defect points. It clarifies that if the thickness growth compensation coefficient is greater than 1, the original defect density decreases proportionally (increased thickness reduces defects); conversely, if the thickness growth compensation coefficient is less than 1, the original defect density increases proportionally. Simultaneously, newly discovered defect points are included to ensure that the updated defect spatial distribution density reflects the latest defect distribution, thus improving the accuracy of defect distribution assessment. Finally, the data obtained after updating the data of non-looping defects is organized into the form of [type identifier "01", defect center update coordinates, maximum gap update width, and percentage of unclosed area update]. The data obtained after updating the data of insufficient thickness defects is organized into the form of [type identifier "02", minimum thickness update value, and defect spatial distribution update density]. All updated data are merged to generate a defect feature update set containing timestamps. This not only ensures the consistency of the data format but also allows the defect feature update set to connect with historical data and clearly reflect the temporal changes of defects, achieving standardized and dynamic management of defect features.

[0033] This application also provides a parameter optimization system for the subway connection passage freezing method, including: The first acquisition module is used to acquire multi-source real-time monitoring data, which includes formation state parameters and process real-time parameters, and to acquire initial state parameters and real-time state parameters based on the formation state parameters. The second acquisition module is used to acquire a formation stability field distribution map based on the initial state parameters and the real-time state parameters. The feature extraction module is used to obtain a set of freezing defect features based on the real-time status parameters; The parameter optimization module is used to optimize the real-time process parameters based on the formation stability field distribution map and the freezing defect feature set to obtain the optimal parameter set; The efficiency evolution module is used to obtain frozen efficiency evolution data based on the optimal parameter set and the real-time state parameters, and to obtain frozen efficiency evaluation feature set based on the frozen efficiency evolution data. The iterative optimization module is used to obtain a quality assessment value based on the frozen efficiency assessment feature set, and to determine whether the quality assessment value is greater than a preset threshold. If the quality assessment value is greater than the preset threshold, then the optimal parameter set is determined to meet the construction requirements; If the quality assessment value is not greater than the preset threshold, then the defect feature update set is obtained based on the frozen efficiency evolution data and the frozen defect feature set, and the process is returned to the step of optimizing the real-time process parameters until the optimal parameter set meets the construction requirements.

[0034] In one embodiment, the feature extraction module includes: The acquisition unit is used to acquire three-dimensional evolution data of the frozen front based on the phase change latent heat compensation algorithm according to the real-time state parameters, and to acquire the spatial variation coefficients of multiple cross sections based on the three-dimensional evolution data of the frozen front. The first judgment unit is used to obtain the frozen design parameters, obtain the first threshold and the second threshold according to the frozen design parameters, and determine whether the spatial variation coefficient of each cross section is greater than the first threshold. The first identification unit is used to determine that the cross section is an unconnected defect surface if the spatial variation coefficient of the cross section is greater than the first threshold, and to obtain the unconnected defect features based on the unconnected defect surface. The second judgment unit is used to obtain the thickness deviation rate of each monitoring point according to the real-time status parameters, and to determine whether the thickness deviation rate of each monitoring point is greater than the second threshold. The second identification unit is used to determine that the monitoring point is a thickness deficiency defect point if the thickness deviation rate of the monitoring point is greater than the second threshold, and to obtain the thickness deficiency defect features based on the thickness deficiency defect point. The feature integration unit is used to construct a frozen defect feature set based on the thickness deficiency defect feature and the non-intersecting defect feature.

[0035] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described subway connection channel freezing method parameter optimization method.

[0036] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method for optimizing parameters of the subway connection channel freezing method.

[0037] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in this application and in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0038] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0039] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for optimizing parameters of a freezing method for a subway connection passage, characterized by, The method comprises the following steps: acquiring multi-source real-time monitoring data, wherein the multi-source real-time monitoring data comprises formation state parameters and process real-time parameters, and acquiring initial state parameters and real-time state parameters according to the formation state parameters; acquiring a formation stability field distribution map according to the initial state parameters and the real-time state parameters; acquiring a freezing defect feature set according to the real-time state parameters; optimizing the process real-time parameters based on the formation stability field distribution map and the freezing defect feature set to obtain an optimal parameter group; acquiring freezing work efficiency evolution data according to the optimal parameter group and the real-time state parameters, and acquiring a freezing work efficiency evaluation feature set according to the freezing work efficiency evolution data; acquiring a quality evaluation value according to the freezing work efficiency evaluation feature set, and judging whether the quality evaluation value is greater than a preset threshold value; if the quality evaluation value is greater than the preset threshold value, it is determined that the optimal parameter group meets the construction requirements; if the quality evaluation value is not greater than the preset threshold value, acquiring a defect feature update set according to the freezing work efficiency evolution data and the freezing defect feature set, and returning to the step of optimizing the process real-time parameters until the optimal parameter group meets the construction requirements.

2. The method according to claim 1, wherein, The step of acquiring a formation stability field distribution map according to the initial state parameters and the real-time state parameters comprises the following steps: acquiring time-space temperature data, pore water dynamic parameters and mechanical property parameters of a plurality of monitoring points according to the real-time state parameters, and acquiring an actual compressive strength field according to the mechanical property parameters; acquiring formation property parameters according to the initial state parameters and the time-space temperature data; acquiring a thermal stress tensor field and a frost heaving pressure tensor field according to the time-space temperature data and the formation property parameters, and acquiring a total stress tensor field according to the thermal stress tensor field and the frost heaving pressure tensor field; acquiring a shear failure field according to the total stress tensor field; acquiring a seepage drag force field according to the pore water dynamic parameters and the formation property parameters by Darcy's law, and acquiring a comprehensive force field according to the seepage drag force field and the shear failure field; acquiring a formation stability field distribution map according to the comprehensive force field.

3. The method according to claim 1, wherein, The step of acquiring a freezing defect feature set according to the real-time state parameters comprises the following steps: acquiring freeze front three-dimensional evolution data according to the real-time state parameters based on a phase change latent heat compensation algorithm, and acquiring a spatial variation coefficient of a plurality of cross sections according to the freeze front three-dimensional evolution data; acquiring freezing design parameters, and acquiring a first threshold value and a second threshold value according to the freezing design parameters, and judging whether the spatial variation coefficient of each cross section is greater than the first threshold value; if the spatial variation coefficient of a cross section is greater than the first threshold value, it is determined that the cross section is an un-intersected defect surface, and un-intersected defect features are acquired according to the un-intersected defect surface; acquiring a thickness deviation rate of each monitoring point according to the real-time state parameters, and judging whether the thickness deviation rate of each monitoring point is greater than the second threshold value; if the thickness deviation rate of a monitoring point is greater than the second threshold value, it is determined that the monitoring point is a thickness deficiency defect point, and thickness deficiency defect features are acquired according to the thickness deficiency defect point; constructing a frozen defect feature set according to the insufficient thickness defect feature and the unlooping defect feature; 4. The method according to claim 1, wherein, the step of optimizing the process real-time parameters based on the stratum stability field distribution map and the frozen defect feature set to obtain an optimal parameter group, comprising: obtaining a defect severity of each independent defect area according to the frozen defect feature set, and obtaining a geological sensitivity coefficient of each independent defect area according to the frozen defect feature set and the stratum stability field distribution map; obtaining a process dynamic regulation coefficient according to the real-time state parameters, and obtaining a process parameter initial adjustment value according to the process dynamic regulation coefficient and the process real-time parameters; obtaining a process parameter adjustment interval according to the process parameter initial adjustment value and the geological sensitivity coefficient, and obtaining a plurality of parameter optimization groups according to the process parameter adjustment interval; obtaining a parameter optimization target and a process optimization constraint condition, and screening a plurality of the parameter optimization groups according to the parameter optimization target and the process optimization constraint condition to obtain an optimal parameter group.

5. The method according to claim 1, wherein, the step of obtaining frozen work efficiency evolution data according to the optimal parameter group and the real-time state parameters, and obtaining a frozen work efficiency evaluation feature set according to the frozen work efficiency evolution data, comprising: obtaining modeling basic data according to the real-time state parameters and the stratum stability field distribution map, and constructing a multi-physical field coupling body according to the modeling basic data; obtaining frozen work efficiency evolution data according to the multi-physical field coupling body and the optimal parameter group, wherein the frozen work efficiency evolution data comprises temperature field evolution data, stratum response data and frozen wall development data; obtaining a frozen design target, and obtaining a construction period coincidence degree according to the frozen design target and the temperature field evolution data; obtaining a stability coefficient according to the frozen design target and the stratum response data, and obtaining a thickness standard volume ratio and a loop standard index according to the frozen design target and the frozen wall development data; taking the loop standard index, the construction period coincidence degree, the stability coefficient, the thickness standard volume ratio and the loop standard index as the frozen work efficiency evaluation feature set.

6. The method according to claim 1, wherein, the step of obtaining a defect feature update set according to the frozen work efficiency evolution data and the frozen defect feature set, comprising: obtaining a defect center update coordinate according to the frozen work efficiency evolution data and the real-time state parameters; obtaining an unlooping defect update feature according to the frozen work efficiency evolution data and the frozen defect feature set, wherein the unlooping defect update feature comprises a maximum gap update width and an un-closed area update proportion; obtaining a thickness growth compensation coefficient according to the frozen work efficiency evolution data, and obtaining an insufficient thickness defect update feature according to the thickness growth compensation coefficient and the frozen defect feature set, wherein the insufficient thickness defect update feature comprises a minimum thickness update value and a defect space distribution update density; constructing a defect feature update set according to the defect center update coordinate, the unlooping defect update feature and the insufficient thickness defect update feature.

7. A system for optimizing parameters of a freezing method for a subway connection passage, characterized by, comprising: The first acquisition module is configured to acquire multi-source real-time monitoring data, the multi-source real-time monitoring data comprising formation state parameters and process real-time parameters, and to acquire initial state parameters and real-time state parameters according to the formation state parameters; The second acquisition module is configured to acquire a formation stability field distribution map according to the initial state parameters and the real-time state parameters; The feature extraction module is configured to acquire a freezing defect feature set according to the real-time state parameters; The parameter optimization module is configured to optimize the process real-time parameters based on the formation stability field distribution map and the freezing defect feature set to obtain an optimal parameter group; The work efficiency evolution module is configured to acquire freezing work efficiency evolution data according to the optimal parameter group and the real-time state parameters, and to acquire a freezing work efficiency evaluation feature set according to the freezing work efficiency evolution data; The iterative optimization module is configured to acquire a quality evaluation value according to the freezing work efficiency evaluation feature set, and to determine whether the quality evaluation value is greater than a preset threshold value; If the quality evaluation value is greater than the preset threshold value, it is determined that the optimal parameter group meets construction requirements; If the quality evaluation value is not greater than the preset threshold value, a defect feature update set is acquired according to the freezing work efficiency evolution data and the freezing defect feature set, and the step of optimizing the process real-time parameters is returned to until the optimal parameter group meets construction requirements.

8. The system for optimization of parameters of the freeze method for the construction of a metro cross-passage according to claim 7, characterized by the fact that, The feature extraction module comprises: The acquisition unit is configured to acquire freezing front three-dimensional evolution data according to the real-time state parameters based on a phase change latent heat compensation algorithm, and to acquire spatial variation coefficients of a plurality of cross sections according to the freezing front three-dimensional evolution data; The first judgment unit is configured to acquire freezing design parameters, to acquire a first threshold value and a second threshold value according to the freezing design parameters, and to determine whether the spatial variation coefficient of each cross section is greater than the first threshold value; The first identification unit is configured to determine that a cross section is an un-intersected defect surface if the spatial variation coefficient of the cross section is greater than the first threshold value, and to acquire an un-intersected defect feature according to the un-intersected defect surface; The second judgment unit is configured to acquire a thickness deviation rate of each monitoring point according to the real-time state parameters, and to determine whether the thickness deviation rate of each monitoring point is greater than the second threshold value; The second identification unit is configured to determine that a monitoring point is a thickness deficiency defect point if the thickness deviation rate of the monitoring point is greater than the second threshold value, and to acquire a thickness deficiency defect feature according to the thickness deficiency defect point; The feature integration unit is configured to construct a freezing defect feature set according to the thickness deficiency defect feature and the un-intersected defect feature. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8. The processor executes the computer program to realize the steps of the method of any one of claims 1 to 6.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the method of any one of claims 1 to 6.

Images