A method and system for predicting settlement of a tunnel base in a water-rich sandstone formation

Abstract

This invention relates to the field of geological exploration technology and discloses a method and system for predicting tunnel foundation settlement in water-rich sandy strata. The method includes: constructing a multi-source collaborative exploration system; deploying a sandification-seepage-stress coupling model at edge computing nodes and mining the correlation between data from different dimensions through a spatiotemporal attention fusion network; based on the prediction matrix and mineral mining condition parameters; triggering a high-precision re-measurement mechanism when the predicted settlement exceeds a safety threshold or the sandification expansion rate is abnormal, activating a joint verification module of cross-hole radar and sonic logging; and iteratively optimizing the coupling model parameters using a transfer learning algorithm through deviation analysis between measured data and prediction results. This invention provides a scientific basis for assessing strata stability and formulating prevention and control measures during mineral mining, effectively reducing the probability of engineering risks caused by foundation settlement in water-rich sandy strata.

Description

Technical Field

[0001] This invention relates to the field of geological exploration technology, and in particular to a method and system for predicting the settlement of tunnel foundations in water-rich sandy strata. Background Technology

[0002] In mineral resource development and underground engineering construction, the foundation settlement problem of water-rich sandy strata has long plagued engineering safety. Due to the uneven sandification of the rock mass and the significant seepage of groundwater, such strata are prone to fine particle loss and increased porosity under excavation disturbance, leading to a sharp drop in the foundation bearing capacity and causing sudden settlement or uneven deformation. Traditional prediction methods based on single geological exploration data are difficult to capture its complex dynamic evolution process.

[0003] Existing settlement prediction technologies mostly rely on empirical formulas or simplified mechanical models, neglecting the multi-field coupling effect of sandification, seepage, and stress, and failing to adequately consider the spatiotemporal variability of formation parameters under water-rich conditions. In practical engineering, insufficient prediction accuracy often leads to delayed support solutions, causing accidents such as tunnel structural cracking and equipment damage. This not only increases engineering maintenance costs but also poses a serious threat to construction and operational safety. Therefore, there is an urgent need to develop a high-precision prediction method that can integrate multi-source information. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method and system for predicting tunnel foundation settlement in water-rich sandy strata, which can effectively reduce the probability of engineering risks caused by foundation settlement in water-rich sandy strata.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for predicting the settlement of the tunnel base in water-rich sandy strata, including: constructing a multi-source collaborative exploration system, integrating a three-dimensional seismic exploration module, a distributed optical fiber sensing module and a borehole panoramic scanning module, and simultaneously collecting multi-dimensional data of water-rich sandy strata, including strata elastic wave impedance profiles, real-time strain monitoring sequences and core micro-sandification characteristic maps. A sandification-seepage-stress coupling model is deployed on edge computing nodes. The correlation between data from different dimensions is mined through a spatiotemporal attention fusion network to generate a prediction matrix that includes the settlement rate distribution, the expansion boundary of the sandification zone, and the critical instability warning time. Based on the prediction matrix and mineral mining conditions, an optimized exploration strategy set is generated using an adaptive dynamic programming algorithm. The strategy set includes a seismic wave acquisition parameter adjustment scheme, a sensor node deployment update plan, and a priority ranking of key monitoring areas. When the predicted settlement exceeds the safety threshold or the sand expansion rate is abnormal, the high-precision retesting mechanism is triggered, activating the cross-hole radar and sonic logging joint verification module. By analyzing the discrepancy between measured data and predicted results, a transfer learning algorithm is used to iteratively optimize the parameters of the coupled model.

[0007] As a preferred embodiment of the method for predicting tunnel foundation settlement in water-rich sandy strata according to the present invention, a three-dimensional seismic acquisition technology based on compressed sensing is adopted. Through multi-frequency excitation of a controllable source and reception by a geophone array, the elastic wave impedance, longitudinal and transverse wave velocity ratio and attenuation coefficient profile of the water-rich sandy strata are obtained. A distributed optical fiber sensor network is deployed in the borehole to achieve strain monitoring with centimeter-level spatial resolution by utilizing the Brillouin scattering effect. The sampling frequency is dynamically adjusted to a preset frequency according to the intensity of mining disturbance. The high-spectral camera and laser confocal lens mounted on the borehole panoramic scanning module are used to collect images of the sandification degree and pore distribution characteristics on the surface of the rock core, with an image resolution no lower than the preset resolution. A multi-source data spatiotemporal alignment protocol is designed, using a GPS synchronization clock as a reference, to correlate and map the temporal scale of seismic data, the spatial scale of fiber optic sensing, and the microscopic scale of rock core images.

[0008] As a preferred embodiment of the method for predicting the settlement of the tunnel foundation in water-rich sandy strata according to the present invention, a physical constraint framework is established with the seepage equation of porous media, the sandy damage evolution equation and the constitutive relationship of rock and soil as the core. The sandy damage evolution equation introduces the fine particle loss rate as a state variable. A two-branch spatiotemporal attention fusion network is constructed. The first branch uses the first model to extract the spatiotemporal distribution characteristics of sandy strata in the seismic profile, and the second branch processes the strain correlation map of fiber optic sensing through a graph convolutional network. The physical constraint framework is embedded into the network loss function in the form of energy conservation residual terms to simultaneously predict the base settlement, permeability coefficient changes in the sandy zone, and pore water pressure distribution at different mining stages. The coupled evolution process of sandification and subsidence was simulated using an improved cellular automata model, with the degree of sandification, groundwater head, and intensity of mining disturbance set as key parameters for the evolution rules.

[0009] As a preferred embodiment of the method for predicting tunnel foundation settlement in water-rich sandy strata according to the present invention, a multi-objective decision-making model is established, consisting of a data acquisition efficiency agent, a prediction accuracy agent, and a cost control agent. The reward function of the data acquisition efficiency agent includes the effective data volume per unit time and the equipment energy consumption ratio. The reward function of the prediction accuracy agent includes the settlement prediction error and the accuracy of sandy zone boundary identification. The reward function of the cost control agent includes the equipment operation and maintenance cost and data storage overhead. A knowledge distillation-based transfer learning strategy is adopted to transfer the exploration strategy experience of historical mining areas to the decision-making model of new mining areas, thereby shortening the strategy convergence time. A Pareto-optimal exploration strategy set was generated using a multi-objective particle swarm optimization algorithm, and the mining disturbance simulation was performed in a digital twin environment for verification within a preset time period.

[0010] As a preferred embodiment of the method for predicting the settlement of the tunnel foundation in water-rich sandy strata according to the present invention, based on the risk level classification in the prediction matrix, cross-hole radar detection is initiated for high-risk areas, and step-frequency continuous waves in a preset frequency band are used to obtain the three-dimensional distribution characteristics of the sandy area, with a spatial resolution not lower than a preset value. The P-wave and S-wave time difference data were collected using an acoustic logging tool, and the formation Poisson's ratio and dynamic elastic modulus were calculated to verify the mechanical parameter settings of the coupled model. The design incorporates a multi-method verification result fusion rule, employing evidence theory to fuse the sandification zone identification results from seismic, radar, and well logging data, thereby improving boundary positioning accuracy.

[0011] As a preferred embodiment of the method for predicting tunnel foundation settlement in water-rich sandy strata according to the present invention, a three-dimensional visualization prediction platform is constructed: integrating the stratum structure model, sandy distribution model and settlement prediction model, realizing dynamic visualization of the foundation settlement process, and supporting multi-view cross-section and parameter sensitivity analysis. An uncertainty analysis module was established: confidence intervals for settlement prediction were generated through Monte Carlo simulation, and the impact of uncertainties in sandification parameters, groundwater parameters, and mining parameters on the prediction results was quantified. Develop an interface for optimizing mining plans: Based on settlement prediction results, automatically generate adjustment suggestions for the advance speed of the mining face, support strength, and drainage plan.

[0012] As a preferred embodiment of the method for predicting the settlement of the tunnel foundation in water-rich sandy strata of the present invention, a stratigraphic structure model is constructed based on three-dimensional seismic data, and isosurface drawing technology is used to display the spatial interface between the sandy zone and the non-sandy zone, with the transparency being dynamically adjustable. The groundwater seepage path and fine particle migration trajectory are simulated by a particle system, and the pore water pressure cloud map and settlement rate vector field are superimposed and displayed. Develop a parameter sensitivity analysis tool that supports interactive adjustment of one or more parameters and updates settlement prediction results in real time.

[0013] As a preferred embodiment of the method for predicting tunnel foundation settlement in water-rich sandy strata according to the present invention, a dynamic adjustment mechanism for settlement warning threshold is established: based on the sandy strata grade, mining depth and dynamic changes in groundwater, a fuzzy logic algorithm is used to update the warning thresholds for settlement rate and cumulative settlement in real time. Real-time updates of early warning thresholds for settlement rate and cumulative settlement; Construct a historical case knowledge base: link geological conditions, mining parameters, subsidence characteristics and prevention measures of different mining areas, and use case reasoning technology to provide analogy reference for current predictions.

[0014] Secondly, the present invention provides a system for predicting the basement subsidence of water-rich sandy strata, comprising: a system construction module for constructing a multi-source collaborative exploration system, integrating a three-dimensional seismic exploration module, a distributed optical fiber sensing module and a borehole panoramic scanning module, and simultaneously acquiring multi-dimensional data of water-rich sandy strata, including strata elastic wave impedance profiles, real-time strain monitoring sequences and core micro-sandification characteristic maps. The model deployment module is used to deploy the sandification-seepage-stress coupling model on edge computing nodes. It uses a spatiotemporal attention fusion network to mine the correlation between data from different dimensions and generate a prediction matrix that includes the settlement rate distribution, the sandification zone expansion boundary, and the critical instability warning time. The strategy set generation module is used to generate an optimized exploration strategy set based on the prediction matrix and mineral mining conditions parameters using an adaptive dynamic programming algorithm. The strategy set includes a seismic wave acquisition parameter adjustment scheme, a sensor node deployment update plan, and a priority ranking of key monitoring areas. The verification module is used to trigger a high-precision retesting mechanism and activate the cross-hole radar and sonic logging joint verification module when the predicted settlement exceeds the safety threshold or the sand expansion rate is abnormal. The parameter iteration module is used to iteratively optimize the parameters of the coupled model by analyzing the deviation between measured data and prediction results and employing a transfer learning algorithm.

[0015] Thirdly, the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement a method for predicting the settlement of the tunnel foundation in water-rich sandy strata.

[0016] Compared with existing technologies, the beneficial effects of this invention are as follows: It can overcome the limitations of traditional single detection methods in fully capturing the complex evolution characteristics of water-rich sandy strata. Through a multi-source collaborative exploration system, it achieves simultaneous acquisition and fusion analysis of multi-dimensional data such as stratum elastic waves, strain, and microscopic sandification characteristics, providing comprehensive physical information support for basement settlement prediction. By leveraging the sandification-seepage-stress coupling model and spatiotemporal attention fusion network, it accurately mines data correlations, generating prediction results such as settlement rate distribution, sandification zone expansion boundary, and critical instability warning time, significantly improving the accuracy and foresight of basement settlement prediction for water-rich sandy strata. Through the adaptive dynamic programming algorithm, it generates an optimized exploration strategy set, which can be used to adjust data acquisition and monitoring schemes in a targeted manner, improving exploration efficiency and resource utilization efficiency. Combined with a high-precision retesting mechanism and model iterative optimization, it further ensures the reliability of prediction results, providing a scientific basis for stratum stability assessment and prevention measures formulation during mineral mining, and effectively reducing the probability of engineering risks caused by basement settlement in water-rich sandy strata. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram illustrating a method for predicting tunnel foundation settlement in water-rich sandy strata proposed in this invention. Figure 2 This is a schematic diagram of a basement settlement prediction system for water-rich sandy strata proposed in this invention; Figure 3 This is a schematic diagram of an electronic device proposed by the present invention. Detailed Implementation

[0019] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0020] Example 1, referring to Figure 1 This is the first embodiment of the present invention, which provides a method for predicting tunnel foundation settlement in water-rich sandy strata, comprising: S1. Construct a multi-source collaborative exploration system, integrating a 3D seismic exploration module, a distributed optical fiber sensing module, and a borehole panoramic scanning module to simultaneously collect multi-dimensional data of water-rich sandy strata. The multi-dimensional data includes strata elastic wave impedance profiles, real-time strain monitoring sequences, and core micro-sanding characteristic maps. S2. Deploy the sandification-seepage-stress coupling model on edge computing nodes, and mine the correlation of data in different dimensions through a spatiotemporal attention fusion network to generate a prediction matrix that includes the settlement rate distribution, the sandification zone expansion boundary and the critical instability warning time. S3. Based on the prediction matrix and mineral mining conditions, an optimized exploration strategy set is generated using an adaptive dynamic programming algorithm. The strategy set includes a seismic wave acquisition parameter adjustment scheme, a sensor node deployment update plan, and a priority ranking of key monitoring areas. S4. When the predicted settlement exceeds the safety threshold or the sand expansion rate is abnormal, the high-precision retesting mechanism is triggered, and the cross-hole radar and sonic logging joint verification module is activated. S5. By analyzing the deviation between the measured data and the predicted results, the parameters of the coupled model are iteratively optimized using a transfer learning algorithm.

[0021] Specifically, firstly, a multi-source collaborative exploration system enables the simultaneous acquisition and fusion analysis of multi-dimensional data such as formation elastic waves, strain, and microscopic sandification characteristics, providing comprehensive physical information support for basement settlement prediction. Secondly, by leveraging a sandification-seepage-stress coupling model and a spatiotemporal attention fusion network, data correlations are accurately mined, generating prediction results such as settlement rate distribution, sandification zone expansion boundary, and critical instability warning time, significantly improving the accuracy and foresight of basement settlement prediction for water-rich sandy strata. Thirdly, an adaptive dynamic programming algorithm generates an optimized exploration strategy set, allowing for targeted adjustments to data acquisition and monitoring schemes, improving exploration efficiency and resource utilization efficiency. Fourthly, combined with a high-precision re-measurement mechanism and model iterative optimization, the reliability of prediction results is further guaranteed, providing a scientific basis for formation stability assessment and prevention measures during mineral mining, effectively reducing the probability of engineering risks caused by basement settlement in water-rich sandy strata.

[0022] Specifically, S1 includes the following sub-steps: Using compressed sensing-based 3D seismic acquisition technology, the elastic wave impedance, P-wave velocity ratio, and attenuation coefficient profiles of water-rich sandy strata were obtained through multi-frequency excitation of a controllable source and reception by a geophone array. A distributed optical fiber sensor network is deployed in the borehole to achieve strain monitoring with centimeter-level spatial resolution by utilizing the Brillouin scattering effect. The sampling frequency is dynamically adjusted to a preset frequency according to the intensity of mining disturbance. The high-spectral camera and laser confocal lens mounted on the borehole panoramic scanning module are used to collect images of the sandification degree and pore distribution characteristics on the surface of the rock core, with an image resolution no lower than the preset resolution. A multi-source data spatiotemporal alignment protocol is designed, using a GPS synchronization clock as a reference, to correlate and map the temporal scale of seismic data, the spatial scale of fiber optic sensing, and the microscopic scale of rock core images.

[0023] Specifically, step S1 achieves comprehensive data acquisition through a multi-source collaborative exploration system. Compressed sensing 3D seismic technology is employed, with controllable source multi-frequency excitation (10-200Hz) and a detector array arranged in a quincunx pattern (5-10 meters spacing). This reduces data volume by 30% while maintaining 90% profile resolution, acquiring key parameters such as elastic wave impedance. A distributed fiber optic sensor network within the borehole, based on the Brillouin scattering effect, achieves a spatial resolution of 1cm, with the sampling frequency dynamically adjusted according to mining disturbances (1-10Hz), achieving an accuracy of ±5με. The borehole panoramic scanning module, equipped with a 12-megapixel hyperspectral camera (400-1000nm) and a laser confocal lens, acquires data every 0.5cm, identifying microcracks as small as 0.1mm. Multi-source data are synchronized with a GPS clock (±10ns) for spatiotemporal alignment, eliminating modal misalignment and improving the spatial matching degree of the sandification zone to over 95%, laying a data foundation for subsequent analysis.

[0024] The specific S2 includes the following steps: A physical constraint framework is established with the seepage equation of porous media, the evolution equation of sandification damage and the constitutive relationship of soil and rock as the core. The sandification damage evolution equation introduces the fine particle loss rate as a state variable. A two-branch spatiotemporal attention fusion network is constructed. The first branch uses the first model to extract the spatiotemporal distribution characteristics of sandy strata in the seismic profile, and the second branch processes the strain correlation map of fiber optic sensing through a graph convolutional network. The physical constraint framework is embedded into the network loss function in the form of energy conservation residual terms to simultaneously predict the base settlement, permeability coefficient changes in the sandy zone, and pore water pressure distribution at different mining stages. The coupled evolution process of sandification and subsidence was simulated using an improved cellular automata model, with the degree of sandification, groundwater head, and intensity of mining disturbance set as key parameters for the evolution rules.

[0025] Specifically, step S2 constructs a sandification-seepage-stress coupling model to achieve accurate prediction. The physical constraint framework integrates the Darcy-Fochheimer seepage equation (permeability coefficient is exponentially related to sandification rate), the sandification damage evolution equation (fine particle loss rate is a state variable), and the modified Cambridge constitutive relation (internal friction angle decreases with sandification rate). In the dual-branch spatiotemporal attention fusion network, the Transformer architecture extracts seismic wave impedance anomaly features, and the graph convolutional network processes the fiber strain correlation spectrum, fusing features across attention layers, improving accuracy by 40% compared to the single-mode model. The loss function embeds an energy conservation residual term (weight 1:1) to ensure the prediction conforms to physical laws, reducing the boundary error of the sandification zone to 0.8 meters. An improved cellular automata (5cm grid) simulates the coupled evolution, dynamically adjusting the time step to predict sandification expansion 6 months in advance, generating key results such as settlement rate and critical instability time.

[0026] The specific S3 includes the following sub-steps: A multi-objective decision-making model is established, consisting of a data acquisition efficiency agent, a prediction accuracy agent, and a cost control agent. The reward function of the data acquisition efficiency agent includes the effective data volume per unit time and the equipment energy consumption ratio. The reward function of the prediction accuracy agent includes the settlement prediction error and the accuracy of sandification zone boundary identification. The reward function of the cost control agent includes the equipment operation and maintenance cost and data storage overhead. A knowledge distillation-based transfer learning strategy is adopted to transfer the exploration strategy experience of historical mining areas to the decision-making model of new mining areas, thereby shortening the strategy convergence time. A Pareto-optimal exploration strategy set was generated using a multi-objective particle swarm optimization algorithm, and the strategy was validated through mining disturbance simulation in a digital twin environment, with the validation period not exceeding a preset time. Specifically, step S3 generates an optimized exploration strategy using intelligent algorithms. The multi-objective decision-making model comprises three agents: a data acquisition efficiency agent (combining the reward function with effective data volume and energy consumption), a prediction accuracy agent (focusing on subsidence error and boundary recognition rate), and a cost control agent (considering operation and maintenance and storage costs). Weighted voting (0.4:0.4:0.2) balances the objectives. Based on knowledge distillation-based transfer learning, experience from 50+ historical mining areas is transferred to the new mining area. The model is optimized using KL divergence (<0.1), reducing the convergence time from 14 days to 5 days. A multi-objective particle swarm optimization algorithm (population 50, 100 iterations) generates a Pareto optimal strategy set covering scenarios such as "high efficiency and low cost" and "high precision." Validation is completed within 24 hours in a digital twin environment, reducing the strategy implementation failure rate by 60%.

[0027] The specific S4 includes the following sub-steps: Based on the risk level classification in the prediction matrix, cross-aperture radar detection is initiated in high-risk areas. Stepped frequency continuous waves in a preset frequency band are used to obtain the three-dimensional distribution characteristics of the sandification zone, with a spatial resolution not lower than the preset value. The P-wave and S-wave time difference data were collected using an acoustic logging tool, and the formation Poisson's ratio and dynamic elastic modulus were calculated to verify the mechanical parameter settings of the coupled model. The design incorporates a multi-method verification result fusion rule, employing evidence theory to fuse the sandification zone identification results from seismic, radar, and well logging data, thereby improving boundary positioning accuracy.

[0028] Specifically, step S4 initiates high-precision retesting to verify the prediction results. In high-risk areas (settlement >30mm or sandification rate >5% / month), cross-hole radar (200-1000MHz stepping wave) is used to identify the three-dimensional distribution of sandification zones with a 0.5-meter resolution, achieving a 92% detection rate for low-resistivity anomalies. A sonic logging tool collects full-wavelength data, calculates Poisson's ratio and dynamic elastic modulus, and verifies the mechanical parameters of the coupled model, improving the matching degree to 90%. DS evidence theory fuses multi-source results: seismic, radar, and logging weights of 0.3:0.4:0.3. The fusion probability is calculated using an orthogonal sum formula (>0.8 indicates a sandification zone). Conflicting evidence is processed according to coefficients, achieving a boundary positioning accuracy of 0.3 meters, a 50% improvement over single methods, providing precise target areas for engineering support.

[0029] In some embodiments, a method for predicting tunnel foundation settlement in water-rich sandy strata may further include the following steps: Construct a three-dimensional visualization prediction platform: integrate stratigraphic structure model, sandification distribution model and settlement prediction model to realize dynamic visualization of the basement settlement process, and support multi-view cross-section and parameter sensitivity analysis; An uncertainty analysis module was established: confidence intervals for settlement prediction were generated through Monte Carlo simulation, and the impact of uncertainties in sandification parameters, groundwater parameters, and mining parameters on the prediction results was quantified. Develop an interface for optimizing mining plans: Based on settlement prediction results, automatically generate adjustment suggestions for the advance speed of the mining face, support strength, and drainage plan.

[0030] Specifically, when constructing the 3D visualization prediction platform, a geological structure model (based on 3D seismic data with an accuracy of 0.5 meters), a sandification distribution model (with a sandification rate of 50% as the threshold), and a settlement prediction model are integrated to dynamically display the settlement process. It supports cross-sections along the X, Y, and Z axes, intuitively presenting the spatial relationship between the sandification zone and the settlement zone. Parameter sensitivity analysis allows adjustment of parameters such as sandification rate and groundwater level (within ±30%), and real-time updates to prediction results, helping to quickly assess the impact of parameters. An uncertainty analysis module uses Monte Carlo simulation to perform 1000 samplings on sandification parameters (fine particle loss rate, coefficient of variation 15%), groundwater parameters (permeability coefficient, coefficient of variation 20%), and mining parameters (advancement speed, coefficient of variation 10%), generating 90% confidence intervals (e.g., interval [15mm, 28mm] when predicting a settlement of 20mm), quantifying the impact of parameters. The sandification parameter contributes 40%, providing a basis for key monitoring. When developing and optimizing the mining scheme interface, suggestions are automatically generated based on settlement prediction: for settlement <10mm, maintain the original scheme (advance at 3m / d); for settlement 10-30mm, reduce the speed to 1.5m / d and strengthen the support; for settlement ≥30mm, suspend mining and drain water (drainage ≥500m³ per day), and simultaneously output three-dimensional effect icons to indicate the adjusted positions, thereby improving execution efficiency.

[0031] In some embodiments, a method for predicting tunnel foundation settlement in water-rich sandy strata may further include the following steps: A stratigraphic model is constructed based on 3D seismic data, and isosurface drawing technology is used to show the spatial boundary between sandy and non-sandy areas, with dynamic adjustable transparency. The groundwater seepage path and fine particle migration trajectory are simulated by a particle system, and the pore water pressure cloud map and settlement rate vector field are superimposed and displayed. Develop a parameter sensitivity analysis tool that supports interactive adjustment of one or more parameters and updates settlement prediction results in real time.

[0032] Specifically, a stratigraphic model was constructed based on 3D seismic data. Using an elastic wave impedance of 2500 g / cm³ as a threshold, isosurface technology was used to map the interface between sandy and non-sandy zones. The transparency was adjustable from 0-100% (at 30%, the internal and external structures of the interface could be observed simultaneously), clearly showing the pocket-like and strip-like distribution of sandy zones, aiding in determining the direction of expansion. Groundwater seepage was simulated using a particle system. Particle density was positively correlated with pore water pressure (a 0.5 MPa increase in pressure corresponds to a 50% increase in density). The trajectory lines were color-coded to indicate flow velocity (blue < 0.2 m / s, red > 1 m / s). Overlaying pore water pressure cloud maps (0-3 MPa color scale) and settlement rate vector fields (arrow length indicates velocity 0-8 mm / month) visually demonstrated the correlation between faster settlement downstream of seepage. Develop a parameter sensitivity analysis tool with 12 built-in adjustable parameters (sandification rate, permeability coefficient, etc.), supporting single-parameter or multi-parameter combination adjustment (such as simultaneous increase of sandification rate and groundwater level), updating settlement prediction (rate distribution, critical time) and generating parameter influence curves within 5 seconds, and quickly locating key influencing factors.

[0033] In some embodiments, a method for predicting tunnel foundation settlement in water-rich sandy strata may further include the following sub-steps: Establish a dynamic adjustment mechanism for settlement early warning thresholds: Based on the sandification grade of the strata, the mining depth and the dynamic changes of groundwater, the early warning thresholds for settlement rate and cumulative settlement are updated in real time using a fuzzy logic algorithm; Real-time updates of early warning thresholds for settlement rate and cumulative settlement; Construct a historical case knowledge base: link geological conditions, mining parameters, subsidence characteristics and prevention measures of different mining areas, and use case reasoning technology to provide analogy reference for current predictions.

[0034] Specifically, the dynamic adjustment mechanism for settlement early warning thresholds is based on a fuzzy logic algorithm: Inputting sandification level (level 5), mining depth (0-500m), and daily groundwater variation (0-1m), it uses 20 rules for reasoning (e.g., "Level 4 sandification + variation > 0.5m → threshold reduction by 20%)" to output an adjustment coefficient (0.5-1.5), updated and pushed every 24 hours. The early warning accuracy is 95%, and the false alarm rate is reduced by 60%. The historical case knowledge base contains 200+ entries, covering geology, mining, settlement, and mitigation parameters. K-nearest neighbor algorithm matching (weights: sandification 30%, groundwater 25%, mining 20%) is used, with the top 5 cases selected for reference. Monthly incremental learning updates (5-10 entries) and BERT model extraction of text features assist in retrieval, improving the efficiency of solving similar problems by 50% and reducing the cost of repeated trials.

[0035] Example 2, refer to Figure 2This is the first embodiment of the present invention, which provides a system for predicting the basement settlement of water-rich sandy strata. The system includes: a system construction module for building a multi-source collaborative exploration system, integrating a 3D seismic exploration module, a distributed fiber optic sensing module, and a borehole panoramic scanning module, simultaneously acquiring multi-dimensional data of the water-rich sandy strata, including strata elastic wave impedance profiles, real-time strain monitoring sequences, and core microscopic sandy strata characteristic maps; and a model deployment module for deploying a sandy strata-seepage-stress coupling model at edge computing nodes, using a spatiotemporal attention fusion network to mine the correlations between different dimensions of data, generating a model including settlement rate distribution, ... The system includes: a prediction matrix for the expansion boundary of sandification zones and the early warning time of critical instability; a strategy set generation module, which generates an optimized exploration strategy set based on the prediction matrix and mineral mining parameters using an adaptive dynamic programming algorithm. The strategy set includes a seismic wave acquisition parameter adjustment scheme, a sensor node deployment update plan, and a priority ranking of key monitoring areas; a verification module, which triggers a high-precision re-measurement mechanism and activates the cross-hole radar and sonic logging joint verification module when the predicted subsidence exceeds the safety threshold or the sandification expansion rate is abnormal; and a parameter iteration module, which uses a transfer learning algorithm to iteratively optimize the coupled model parameters through deviation analysis between measured data and prediction results.

[0036] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements a method for predicting the settlement of the tunnel foundation in water-rich sandy strata.

[0037] like Figure 3 As shown, the electronic device may include a processor 610, a communication interface 620, a memory 630, and a communication bus 640. The processor 610, communication interface 620, and memory 630 communicate with each other via the communication bus 640. The processor 610 can call logical instructions from the memory 630 to execute a method for predicting tunnel foundation settlement in water-rich sandy strata.

[0038] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0039] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute a method for predicting the settlement of the tunnel foundation in water-rich sandy strata.

[0040] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform a method for predicting the settlement of the tunnel foundation in water-rich sandy strata.

[0041] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0042] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0043] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0044] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method of predicting settlement of a tunnel base in a water- richly-sandstone-ized stratum, characterized by, include: A multi-source collaborative exploration system was constructed, integrating a 3D seismic exploration module, a distributed optical fiber sensing module, and a borehole panoramic scanning module to simultaneously acquire multi-dimensional data of water-rich sandy strata. The multi-dimensional data includes strata elastic wave impedance profiles, real-time strain monitoring sequences, and core micro-sandification characteristic maps. A sandification-seepage-stress coupling model is deployed on edge computing nodes. The correlation between data from different dimensions is mined through a spatiotemporal attention fusion network to generate a prediction matrix that includes the distribution of tunnel settlement rate, the expansion boundary of sandification zone, and the critical instability warning time. The specific steps for generating the prediction matrix are as follows: A physical constraint framework is established with the seepage equation of porous media, the sandification damage evolution equation and the constitutive relationship of rock and soil as the core. The sandification damage evolution equation introduces the fine particle loss rate as a state variable. A two-branch spatiotemporal attention fusion network is constructed. The first branch uses the first model to extract the spatiotemporal distribution characteristics of sandy strata in the seismic profile, and the second branch processes the strain correlation map of fiber optic sensing through a graph convolutional network. The physical constraint framework is embedded into the network loss function in the form of energy conservation residual terms to simultaneously predict the base settlement, permeability coefficient changes in the sandy zone, and pore water pressure distribution at different mining stages. The sandification and subsidence coupled evolution process was simulated using an improved cellular automata model, with the degree of sandification, groundwater head, and mining disturbance intensity set as key parameters for the evolution rules. Based on the prediction matrix and mineral mining conditions, an optimized exploration strategy set is generated using an adaptive dynamic programming algorithm. The strategy set includes a seismic wave acquisition parameter adjustment scheme, a sensor node deployment update plan, and a priority ranking of key monitoring areas. When the predicted tunnel settlement exceeds the safety threshold or the sandification expansion rate is abnormal, a high-precision re-measurement mechanism is triggered, activating the cross-hole radar and acoustic logging joint verification module. By analyzing the discrepancy between measured data and predicted results, a transfer learning algorithm is used to iteratively optimize the parameters of the coupled model.

2. The method of predicting settlement of a tunnel base in a water- rich sandstone formation of claim 1, wherein, A multi-source collaborative exploration system was constructed, integrating a 3D seismic exploration module, a distributed fiber optic sensing module, and a borehole panoramic scanning module. This system simultaneously acquired multi-dimensional data of water-rich sandy strata, including formation elastic wave impedance profiles, real-time strain monitoring sequences, and core microscopic sandification characteristic maps. The specific steps were as follows: Using compressed sensing-based 3D seismic acquisition technology, the elastic wave impedance, P-wave velocity ratio, and attenuation coefficient profiles of water-rich sandy strata were obtained through multi-frequency excitation of a controllable source and reception by a geophone array. A distributed optical fiber sensor network is deployed in the borehole to achieve strain monitoring with centimeter-level spatial resolution by utilizing the Brillouin scattering effect. The sampling frequency is dynamically adjusted to a preset frequency according to the intensity of mining disturbance. The high-spectral camera and laser confocal lens mounted on the borehole panoramic scanning module are used to collect images of the sandification degree and pore distribution characteristics on the surface of the rock core, with an image resolution no lower than the preset resolution. A multi-source data spatiotemporal alignment protocol is designed, using a GPS synchronization clock as a reference, to correlate and map the temporal scale of seismic data, the spatial scale of fiber optic sensing, and the microscopic scale of rock core images.

3. The method for predicting tunnel foundation settlement in water-rich sandy strata as described in claim 2, characterized in that, Based on the prediction matrix and mineral mining parameters, an optimized exploration strategy set is generated using an adaptive dynamic programming algorithm. This strategy set includes steps for adjusting seismic wave acquisition parameters, updating sensor node deployment, and prioritizing key monitoring areas. Specifically: A multi-objective decision-making model is established, consisting of a data acquisition efficiency agent, a prediction accuracy agent, and a cost control agent. The reward function of the data acquisition efficiency agent includes the effective data volume per unit time and the equipment energy consumption ratio. The reward function of the prediction accuracy agent includes the settlement prediction error and the accuracy of sandification zone boundary identification. The reward function of the cost control agent includes the equipment operation and maintenance cost and data storage overhead. A knowledge distillation-based transfer learning strategy is adopted to transfer the exploration strategy experience of historical mining areas to the decision-making model of new mining areas, thereby shortening the strategy convergence time. A Pareto-optimal exploration strategy set was generated using a multi-objective particle swarm optimization algorithm, and the mining disturbance simulation was performed in a digital twin environment for verification within a preset time period.

4. The method for predicting tunnel foundation settlement in water-rich sandy strata as described in claim 3, wherein when the predicted tunnel settlement exceeds a safety threshold or the sandy expansion rate is abnormal, a high-precision re-measurement mechanism is triggered, activating the joint verification module of cross-hole radar and sonic logging, specifically as follows: Based on the risk level classification in the prediction matrix, cross-aperture radar detection is initiated in high-risk areas. Stepped frequency continuous waves in a preset frequency band are used to obtain the three-dimensional distribution characteristics of the sandification zone, with a spatial resolution not lower than the preset value. The P-wave and S-wave time difference data were collected using an acoustic logging tool, and the formation Poisson's ratio and dynamic elastic modulus were calculated to verify the mechanical parameter settings of the coupled model. The design incorporates a multi-method verification result fusion rule, employing evidence theory to fuse the sandification zone identification results from seismic, radar, and well logging data, thereby improving boundary positioning accuracy.

5. The method for predicting tunnel foundation settlement in water-rich sandy strata as described in claim 1, characterized in that, The method further includes: Construct a three-dimensional visualization prediction platform: integrate stratigraphic structure model, sandification distribution model and settlement prediction model to realize dynamic visualization of the basement settlement process, and support multi-view cross-section and parameter sensitivity analysis; An uncertainty analysis module was established: confidence intervals for settlement prediction were generated through Monte Carlo simulation, and the impact of uncertainties in sandification parameters, groundwater parameters, and mining parameters on the prediction results was quantified. Develop an interface for optimizing mining plans: Based on settlement prediction results, automatically generate adjustment suggestions for the advance speed of the mining face, support strength, and drainage plan.

6. The method for predicting tunnel foundation settlement in water-rich sandy strata as described in claim 5, characterized in that, The specific steps for building a 3D visualization prediction platform are as follows: A stratigraphic model is constructed based on 3D seismic data, and isosurface drawing technology is used to show the spatial boundary between sandy and non-sandy areas, with dynamic adjustable transparency. The groundwater seepage path and fine particle migration trajectory are simulated by a particle system, and the pore water pressure cloud map and settlement rate vector field are superimposed and displayed. Develop a parameter sensitivity analysis tool that supports interactive adjustment of one or more parameters and updates settlement prediction results in real time.

7. The method for predicting tunnel foundation settlement in water-rich sandy strata as described in claim 1, characterized in that, The method further includes: Establish a dynamic adjustment mechanism for settlement early warning thresholds: Based on the sandification grade of the strata, the mining depth and the dynamic changes of groundwater, the early warning thresholds for settlement rate and cumulative settlement are updated in real time using a fuzzy logic algorithm; Real-time updates of early warning thresholds for settlement rate and cumulative settlement; Construct a historical case knowledge base: link geological conditions, mining parameters, subsidence characteristics and prevention measures of different mining areas, and use case reasoning technology to provide analogy reference for current predictions.

8. A system for predicting tunnel foundation settlement in water-rich sandy strata, characterized in that, include: The system construction module is used to build a multi-source collaborative exploration system, integrating a 3D seismic exploration module, a distributed optical fiber sensing module, and a borehole panoramic scanning module to simultaneously collect multi-dimensional data of water-rich sandy strata. The multi-dimensional data includes strata elastic wave impedance profiles, real-time strain monitoring sequences, and core micro-sandification characteristic maps. The model deployment module is used to deploy the sandification-seepage-stress coupling model on edge computing nodes. It uses a spatiotemporal attention fusion network to mine the correlation of data in different dimensions and generate a prediction matrix including the tunnel settlement rate distribution, the sandification zone expansion boundary and the critical instability early warning time. The specific steps for generating the prediction matrix are as follows: A physical constraint framework is established with the seepage equation of porous media, the sandification damage evolution equation and the constitutive relationship of rock and soil as the core. The sandification damage evolution equation introduces the fine particle loss rate as a state variable. A two-branch spatiotemporal attention fusion network is constructed. The first branch uses the first model to extract the spatiotemporal distribution characteristics of sandy strata in the seismic profile, and the second branch processes the strain correlation map of fiber optic sensing through a graph convolutional network. The physical constraint framework is embedded into the network loss function in the form of energy conservation residual terms to simultaneously predict the base settlement, permeability coefficient changes in the sandy zone, and pore water pressure distribution at different mining stages. The sandification and subsidence coupled evolution process was simulated using an improved cellular automata model, with the degree of sandification, groundwater head, and mining disturbance intensity set as key parameters for the evolution rules. The strategy set generation module is used to generate an optimized exploration strategy set based on the prediction matrix and mineral mining conditions parameters using an adaptive dynamic programming algorithm. The strategy set includes a seismic wave acquisition parameter adjustment scheme, a sensor node deployment update plan, and a priority ranking of key monitoring areas. The verification module is used to trigger a high-precision retesting mechanism and activate the cross-hole radar and sonic logging joint verification module when the predicted tunnel settlement exceeds the safety threshold or the sandification expansion rate is abnormal. The parameter iteration module is used to iteratively optimize the parameters of the coupled model by analyzing the deviation between measured data and prediction results and employing a transfer learning algorithm.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements a method for predicting the foundation settlement of a tunnel in water-rich sandy strata as described in claims 1-7.

Images