A method and system for predicting tunnel uneven settlement tide

Abstract

This invention discloses a method and system for predicting uneven settlement tides in tunnels, belonging to the field of tunnel structural health monitoring technology. The method separates the tidal component from the current time step's surface settlement, structural settlement, and ground hydrological signal monitoring data, and uses this as an input load. Substituting this load into a constructed porosity elasticity equation, the method obtains the time-varying curves of tidal pressure and surface settlement from the simulated signal. It also extracts the phase information and corresponding time of the simulated signal, thereby determining the phase difference and time difference with the measured tidal signal. The phase offset and time delay parameters in the porosity elasticity equation are then corrected to obtain an optimized porosity elasticity equation. The continuous settlement field to be measured is input into the optimized porosity elasticity equation to obtain the pore water pressure values ​​at different locations and times, which are then input into a pre-trained prediction network model. The model predicts and outputs the differential settlement probability for future time steps. This method enables accurate assessment of tunnel settlement risk.

Description

Technical Field

[0001] This invention relates to the field of tunnel structural health monitoring technology, and more specifically to a method and system for predicting tunnel uneven settlement tides. Background Technology

[0002] In the operation and maintenance of tidal tunnels, intelligent early warning and proactive prevention and control of uneven settlement processes can effectively avoid the risk of structural instability, reduce excessive engineering intervention, thereby extending service life and optimizing operation and maintenance resources.

[0003] Currently, monitoring of uneven settlement in tunnels mainly relies on manual leveling or periodic data collection from local single-point sensors. This not only requires interrupting traffic to deploy monitoring points (e.g., drilling weakens the lining strength), significantly increasing the risk of structural damage, but also results in delayed early warnings or frequent false alarms and severe resource misallocation due to experience-driven static models. Especially in coastal tidal zones, traditional methods struggle to quantify the chain reaction mechanism of tidal load, groundwater level, and settlement, easily triggering abrupt differential settlement changes that threaten tunnel structural safety. Furthermore, data fragmentation delays proactive prevention and control.

[0004] In summary, existing technologies cannot achieve real-time perception of differential settlement, leading to inaccurate model predictions and threatening the safety of tunnel structures. Summary of the Invention

[0005] To address the problems existing in the above-mentioned fields, this invention proposes a method and system for predicting tunnel uneven settlement tides. This method can comprehensively consider the complex mechanism of tunnel settlement, avoid the limitations of single-factor analysis, and improve the comprehensiveness and accuracy of the prediction results.

[0006] To address the aforementioned technical problems, this invention discloses a method for predicting tunnel uneven settlement tides, comprising the following steps: Based on Biot's consolidation theory, a porosity elastic equation describing the transmission process of tidal pressure fluctuations is established. Surface subsidence, structural subsidence, and hydrological signal monitoring data at the current time step are acquired. Tidal components in these data are separated using fast Fourier transform. These tidal components are then used as input loads and substituted into the porosity elastic equation to obtain the time-varying curves of tidal pressure and surface subsidence in the simulated signals. The phase information of the simulated signals with the same frequency components as the monitoring data and their corresponding times are extracted. Based on the phase information of the simulated signal and its corresponding time, and the phase and time of the measured tidal signal, the phase difference and time difference are determined; based on the phase difference and time difference, the phase offset and time delay parameters in the porosity elastic equation are corrected to obtain the optimized porosity elastic equation. Historical time-step monitoring data of surface settlement, structural settlement and stratigraphic hydrological signals of the tunnel are acquired, and then fused with the acquired current time-step monitoring data of surface settlement, structural settlement and stratigraphic hydrological signals to generate a continuous settlement field. The continuous settlement field to be measured is input into the optimized pore elasticity equation to obtain the pore water pressure values ​​of tides at different locations and at different times. The pore water pressure values ​​at different times are input into the pre-trained prediction network model. After predicting and outputting the differential settlement probability curves of future time steps, the settlement envelope is simulated to assess the risk of tunnel settlement in the future.

[0007] Preferably, the establishment of the porosity elastic equation describing the tidal pressure fluctuation transmission process based on Biot's consolidation theory specifically includes: Based on Biot's consolidation theory, the equilibrium equation and seepage equation are determined. Among them, the equilibrium equation is used to describe the coupling relationship between soil skeleton stress and pore pressure, and the seepage equation is used to characterize the transmission process of tidal pressure through boundary terms in the porous medium. Based on the equilibrium equation and the seepage equation, establish the porosity elasticity equation; The porosity elastic equation is: ; In the equation: ∇ is the gradient / divergence operator in the equilibrium equation; u is the displacement vector of the solid skeleton; D D:∇u represents the elastic stiffness tensor of the solid skeleton; α is the Biot consolidation coefficient, ∇u represents the stress tensor corresponding to the strain. p This represents the pressure change of the pore fluid; in the seepage equation, M Here, represents the Biot modulus; ∇·u represents the volumetric strain of the solid skeleton. k For the transport efficiency of porous media, γ w Let be the specific weight of the fluid, ΔPtide be the change in tidal pressure, and δ(z−H) be the Dirac delta function. H For boundary depth; Based on the established pore elasticity equation, by introducing tidal periodic boundary conditions, the pressure change of the pore fluid is transformed into the pore water pressure change in the porous medium. Among them, the transport efficiency of porous media k The amount of pressure change in the pore fluid is determined by ∇ P The degree of attenuation, elastic stiffness tensor D Affects the energy conversion ratio; By solving the porosity elasticity equation, the quantification unit ∇ is determined. P The resulting change in effective formation stress characterizes the transmission efficiency. k The value of .

[0008] Preferably, the step of acquiring surface subsidence, structural subsidence, and formation hydrological signal monitoring data at the current time step, separating the tidal component from the surface subsidence, structural subsidence, and formation hydrological signal monitoring data using Fast Fourier Transform, substituting the tidal component as an input load into the porosity elasticity equation, obtaining the time-varying curves of tidal pressure and surface subsidence of the simulated signal, and extracting the phase information of the simulated signal with the same frequency components as the monitoring data and its corresponding time, specifically includes: Acquire monitoring data on surface subsidence, structural subsidence, and stratigraphic hydrological signals at the current time step; By using Fast Fourier Transform, the monitoring data of surface subsidence, structural subsidence and stratigraphic hydrological signals are transformed from the time domain to the frequency domain, and the periodic-frequency components corresponding to the astronomical tide table are identified as tidal components, while random load signals including high-frequency noise and low-frequency drift are filtered out. The denoised tidal components are used as input loads and substituted into the porosity elastic equation for numerical simulation to obtain the tidal pressure and surface subsidence variation curves of the simulated signals over time. The FFT transformation is performed on the time-varying curves of tidal pressure and surface subsidence of the acquired simulated signals to extract the phase information of the components with the same frequency as the actual tidal components, and the time when the simulated signal reaches the phase information is determined. The phase and time of the analog signal are compared with the phase and time of the actual tidal signal to determine the phase difference and time difference; Based on the phase difference and time difference, correct the phase shift and time delay parameters in the porosity elasticity equation; Regarding phase offset, when the phase of the analog signal lags behind the actual tidal signal, decrease the phase offset parameter; when the phase of the analog signal leads the actual tidal signal, increase the phase offset parameter. Regarding the time delay parameter, when the time of the analog signal lags behind the actual tidal signal, decrease the time delay parameter; when the time of the analog signal leads the actual tidal signal, increase the time delay parameter. The porosity elastic equation is updated using the corrected phase offset and time delay parameters. Numerical simulation is then performed again using the updated porosity elastic equation until the phase difference and time difference between the simulated signal and the actual tidal signal meet the preset threshold requirements, thus obtaining the optimized porosity elastic equation.

[0009] Preferably, the step of inputting the continuous settlement field to be measured into the optimized pore elasticity equation to obtain the pore water pressure values ​​of the tides at different locations at different times; and inputting the pore water pressure values ​​at different times into a pre-trained prediction network model to predict and output the differential settlement probability of future time steps, specifically includes: The continuous settlement field to be measured is input into the optimized pore elastic equation, and the pore water pressure values ​​of the tide at different locations at different times are output. The pore water pressure values ​​of tides at different locations at different times are input into the Transformer multi-head attention mechanism time series model, which outputs the differential settlement probability and settlement rate curves for the next few hours.

[0010] Preferably, the step of predicting and outputting the differential settlement probability at future time steps further includes: Based on the differential settlement probability and settlement rate curves over the next few hours, Monte Carlo simulation is used to generate multiple settlement sample paths that conform to the set distribution. Statistical analysis is performed on the generated settlement sample paths that conform to the set distribution to determine the statistical value of settlement and draw the settlement envelope. Based on the output differential settlement probability and settlement rate curves and settlement envelope for the next few hours, assess the risk of tunnel settlement in the future.

[0011] Preferably, it also includes, based on the risk of tunnel settlement over a future period, identifying high-risk sections and obtaining settlement monitoring data after grouting, specifically including: Multiple warning events are obtained by determining multiple dynamic warning thresholds of the same level; 3D grouting scheme generation: Based on the BIM model, areas with excessive settlement risk in the tunnel over a period of time are marked. By identifying high-risk sections, a 3D grouting scheme is generated, including: grouting hole locations, grout type suitable for geological characteristics, and grouting volume that meets the optimized transfer efficiency. Multiple early warning events, high-risk sections, and 3D grouting solutions are synchronized to the digital twin platform, and corresponding reference solutions are pushed by associating them with historical handling cases. Based on the corresponding reference scheme, the settlement control effect under different grouting hole positions and grouting volumes was simulated using a reinforcement effect simulator, and settlement monitoring data after grouting was obtained.

[0012] Preferably, the method further includes feeding back the acquired post-grouting settlement monitoring data to the Transformer multi-head attention mechanism time series model and the porosity elasticity equation, and dynamically correcting the weights of the model and optimizing the porosity elasticity equation, respectively.

[0013] Preferably, the step of acquiring historical time-step monitoring data of surface settlement, structural settlement, and stratigraphic hydrological signals of the tunnel, and then merging these data with the acquired current time-step monitoring data of surface settlement, structural settlement, and stratigraphic hydrological signals to generate a continuous settlement field specifically includes: By using distributed optical fibers, InSAR satellite imagery, and pore water pressure sensors that are equidistantly deployed along the tunnel axis, historical time step monitoring data of the tunnel's structural settlement, surface settlement, and stratum hydrological signals are collected. The historical time step monitoring data of the collected tunnel structural settlement, surface settlement and stratum hydrological signals and the current time step monitoring data of surface settlement, structural settlement and stratum hydrological signals are spatiotemporally aligned. After achieving millimeter-level time synchronization based on BeiDou time synchronization, the IDW algorithm is used to fuse the discrete monitoring point data of time-synchronized structural settlement, surface settlement and stratum hydrological signals into a millimeter-level continuous settlement field. The confidence scores of discrete monitoring point data in a continuous settlement field are obtained. Anomalies in discrete monitoring point data in the continuous settlement field are identified using the isolated forest algorithm, and the filtered continuous settlement field is obtained as the input to the prediction network model.

[0014] Preferably, it also includes a prediction system for tunnel uneven settlement tides, comprising: The porosity elasticity equation construction module is used to establish a porosity elasticity equation describing the tidal pressure fluctuation transmission process based on Biot's consolidation theory; acquire surface subsidence, structural subsidence, and formation hydrological signal monitoring data at the current time step; separate the tidal components in the surface subsidence, structural subsidence, and formation hydrological signal monitoring data through fast Fourier transform; use the tidal components as input loads, substitute them into the porosity elasticity equation, obtain the tidal pressure and surface subsidence variation curves of the simulated signals over time, and extract the phase information of the simulated signals with the same frequency components as the monitoring data and their corresponding times; The porosity elasticity equation optimization module is used to determine the phase difference and time difference based on the phase information of the simulated signal and its corresponding time and the phase and time of the measured tidal signal; based on the phase difference and time difference, the phase offset and time delay parameters in the porosity elasticity equation are corrected to obtain the optimized porosity elasticity equation; The multi-sensor fusion module is used to acquire historical time step monitoring data of surface settlement, structural settlement and stratigraphic hydrological signals of the tunnel, and then fuse them with the acquired current time step monitoring data of surface settlement, structural settlement and stratigraphic hydrological signals to generate a continuous settlement field. The time-series prediction engine module is used to input the continuous settlement field to be measured into the optimized pore elasticity equation to obtain the pore water pressure values ​​of tides at different locations at different times; input the pore water pressure values ​​at different times into the pre-trained prediction network model, and output the differential settlement probability of future time steps through prediction.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The proposed method for predicting tunnel uneven settlement tides constructs a pore elasticity equation that accurately describes the variation of pore pressure in the stratum under tidal loads from a physical perspective. By separating the tidal component from the surface settlement, structural settlement, and hydrological signal monitoring data of the current time step, and correcting the phase shift and time delay parameters in the pore elasticity equation, the transmission efficiency of tidal pressure fluctuations caused by tidal loads in the tunnel stratum can be precisely quantified. This clarifies the specific impact of tidal factors on tunnel settlement and provides key parameters for accurate tunnel settlement prediction. Spatiotemporal alignment and fusion of monitoring data from the current and historical time steps fully consider the temporal and spatial correlations of different data types, facilitating a more accurate analysis of the complex relationship between tunnel settlement and hydrological changes. This provides a high-quality, high-resolution data foundation for subsequent analysis and prediction, and more comprehensively and realistically reflects the actual situation of tunnel settlement. The method inputs the continuously measured settlement field into the optimized pore elasticity equation to obtain the pore water pressure values ​​at different locations and times. These pore water pressure values ​​are then input into a pre-trained prediction network model. The model predicts and outputs the differential settlement probability for future time steps, reflecting the range and trend of tunnel settlement over a future period. This method comprehensively considers the complex mechanisms of tunnel settlement, avoids the limitations of single-factor analysis, and improves the comprehensiveness and accuracy of the prediction results. Attached Figure Description

[0016] Figure 1 This is a flowchart of the method for predicting tunnel uneven settlement tides proposed in this invention. Figure 2 A schematic diagram of the components of the tunnel uneven settlement tidal prediction system provided in an embodiment of the present invention; Figure 3 A schematic diagram of the specific modules of the tunnel uneven settlement tidal prediction system provided in this embodiment of the invention; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0017] The following will refer to the appendices in the embodiments of the present invention. Figures 1-4 The technical solutions in the embodiments of the present invention will be clearly and completely described. It should be understood that the terminology used in the present invention is only for describing particular implementation methods and is not intended to limit the present invention.

[0018] Example like Figure 1 As shown, this invention proposes a method for predicting tunnel uneven settlement tides, comprising the following steps: S1: Based on Biot's consolidation theory, a porosity elastic equation describing the transmission process of tidal pressure fluctuations is established; surface subsidence, structural subsidence, and formation hydrological signal monitoring data at the current time step are acquired, and the tidal component in the surface subsidence, structural subsidence, and formation hydrological signal monitoring data is separated by fast Fourier transform; the tidal component is used as the input load and substituted into the porosity elastic equation to obtain the tidal pressure and surface subsidence variation curves of the simulated signal over time, and the phase information of the simulated signal with the same frequency component as the monitoring data and its corresponding time are extracted; S2: Based on the phase information of the simulated signal and its corresponding time, and the phase and time of the measured tidal signal, determine the phase difference and time difference; based on the phase difference and time difference, correct the phase offset and time delay parameters in the porosity elastic equation to obtain the optimized porosity elastic equation. S3: Acquire historical time-step monitoring data of surface settlement, structural settlement and stratum hydrological signals of the tunnel, and fuse them with the acquired current time-step monitoring data of surface settlement, structural settlement and stratum hydrological signals to generate a millimeter-level continuous settlement field. S4: Input the continuous settlement field to be measured into the optimized pore elasticity equation to obtain the pore water pressure values ​​of the tide at different locations and at different times; input the pore water pressure values ​​at different times into the pre-trained prediction network model, and output the differential settlement probability of future time steps through prediction.

[0019] like Figure 2 As shown, this invention also proposes a prediction system for tunnel uneven settlement tides, comprising: The multi-sensor fusion module 100 is used to acquire historical time-step monitoring data of surface settlement, structural settlement, and ground hydrological signals of the tunnel, and then fuse these data with the acquired current time-step monitoring data of surface settlement, structural settlement, and ground hydrological signals after spatiotemporal alignment to generate a millimeter-level continuous settlement field. The multi-sensor fusion module includes a heterogeneous data access module 101, a spatiotemporal alignment module 102, and a quality verification module 103, such as... Figure 3 As shown.

[0020] The tidal-driven modeling module 200 is used to construct a porosity elastic equation describing the transmission process of tidal pressure fluctuations based on Biot's consolidation theory, in order to quantify the transmission efficiency of tidal pressure fluctuations caused by tidal loads. It acquires monitoring data of surface subsidence, structural subsidence, and ground hydrological signals at the current time step, and separates the tidal components from these data using Fast Fourier Transform. The tidal components are used as input loads and substituted into the porosity elastic equation to obtain the time-varying curves of tidal pressure and surface subsidence of the simulated signals, and to extract the phase information and corresponding time of the simulated signals with the same frequency components as the monitoring data. Based on the phase information and corresponding time of the simulated signals and the phase and time of the measured tidal signals, the phase difference and time difference are determined. Based on the phase difference and time difference, the phase offset and time delay parameters in the porosity elastic equation are corrected to obtain the optimized porosity elastic equation. The tidal-driven modeling module includes a fluid-structure interaction module 201, a period correction module 202, and a parameter inversion module 203, as shown below. Figure 3 As shown.

[0021] The time-series prediction engine module 300 is used to input the continuous settlement field to be measured into the optimized pore elasticity equation to obtain the pore water pressure values ​​of tides at different locations at different times; it inputs the pore water pressure values ​​at different times into a pre-trained prediction network model to predict and output the differential settlement probability of future time steps. The time-series prediction engine module includes a Transformer core module 301, a settlement inference module 302, and a model optimization module 303, such as... Figure 3 As shown.

[0022] In step S1, the equilibrium equation and the seepage equation are determined by the fluid-structure interaction module 201 based on Biot consolidation theory. The equilibrium equation is used to describe the coupling relationship between soil skeleton stress and pore pressure, and the seepage equation is used to characterize the transmission process of tidal pressure through boundary terms in the porous medium. The two are bidirectionally coupled through the Biot consolidation coefficient.

[0023] The constructed porosity elastic equation is as follows: ; In the equation: ∇ is the gradient / divergence operator in the equilibrium equation; u is the displacement vector of the solid skeleton; D D:∇u represents the elastic stiffness tensor of the solid skeleton; α is the Biot consolidation coefficient, ∇u represents the stress tensor corresponding to the strain. p This represents the pressure change of the pore fluid; in the seepage equation, M Here, represents the Biot modulus; ∇·u represents the volumetric strain of the solid skeleton. k For the transport efficiency of porous media, γ wLet be the specific weight of the fluid, ΔPtide be the change in tidal pressure, δ(z−H) be the Dirac delta function, and H be the boundary depth. Based on the equilibrium equation and the seepage equation, a pore elasticity equation is established. The established pore elasticity equation, by introducing tidal periodic boundary conditions, transforms the pressure change of the pore fluid into the pore water pressure change within the porous medium.

[0024] Among them, the transport efficiency of porous media k The amount of pressure change in the pore fluid is determined by ∇ P The degree of attenuation, elastic stiffness tensor D Affects the energy conversion ratio; By solving the porosity elasticity equation, the quantification unit ∇ is determined. P The resulting change in effective formation stress characterizes the transmission efficiency. k The value of .

[0025] The transfer efficiency of the porosity elasticity equation through porous media k With elastic stiffness tensor D The efficiency of tidal load transfer from the seawater boundary to the surrounding soil strata was jointly quantified. Among these, the transfer efficiency of porous media was considered. k The amount of pressure change Δ that determines the pore fluid P The degree of attenuation, elastic stiffness tensor D It affects the energy conversion ratio.

[0026] Its constitutive relation is specifically manifested as: the change in pressure of the pore fluid under tidal load ∇ p The fluid is transferred to the formation through the seepage of pore fluids, and its transfer efficiency is jointly controlled by Darcy's law and soil compressibility.

[0027] By solving the pore elasticity equation, the change in effective stress in the formation caused by the change in pressure of a unit pore fluid is quantified, thereby characterizing the transmission efficiency of tidal pressure to the porous media around the tunnel. Finally, the correlation between the change in effective stress in the formation and the settlement is obtained by solving the pore elasticity equation.

[0028] Biot consolidation theory is a classic theory for studying the deformation and consolidation of soil under the coupled effects of seepage and stress fields, and it can well describe the transmission process of tidal pressure fluctuations in soil. This invention establishes a porosity elasticity equation based on this theory, providing a solid theoretical foundation for subsequent simulations and analyses, and considering the relationship between tidal pressure and soil deformation from a theoretical perspective.

[0029] The periodic correction module 202 uses Fast Fourier Transform to convert the acquired surface subsidence, structural subsidence, and stratigraphic hydrological signal monitoring data of the current time step from the time domain to the frequency domain, identifies the periodic-frequency component (12.4-hour dominant frequency component) corresponding to the astronomical tide table as the tidal component, and filters out random load signals including high-frequency noise and low-frequency drift. Based on the denoised tidal component, the phase shift and time delay parameters in the porosity elastic equation are calibrated to ensure that the porosity elastic equation focuses only on the core mechanism of tidal driving, thereby ensuring that the tidal driving mechanism accurately reflects the actual hydrological dynamics.

[0030] After acquiring the monitoring data of surface subsidence, structural subsidence, and stratigraphic hydrological signals at the current time step, the tidal component is separated using Fast Fourier Transform (FFT). FFT is a highly efficient signal processing tool that can convert time-domain signals into frequency-domain signals, thereby accurately extracting tidal components with specific frequency characteristics, providing accurate data input for subsequent simulations and analyses.

[0031] The parameter inversion module 203 uses the adjoint state method to invert the permeability of porous media in the porosity elastic equation. Combined with the surface subsidence, structural subsidence and formation hydrological signal monitoring data of the current time step obtained by Ensemble Kalman filtering assimilation, the porosity elastic equation is gradually optimized to ensure the accurate characterization of the tidal driving mechanism.

[0032] Based on the phase shift and time delay parameters in the calibrated pore elastic equation, the pore elastic equation is optimized so that the optimized pore elastic equation can adapt to changes in formation physical properties in real time.

[0033] Based on Biot's consolidation theory, the constructed porosity elasticity equation accurately describes the variation of pore pressure in formations under tidal loads from a physical perspective. This theoretically constructed model has a solid scientific foundation, reflects actual conditions more realistically, and has higher reliability and accuracy compared to empirical or simplified models.

[0034] The separated tidal component was substituted into the porosity elasticity equation as the input load to obtain the time-varying curves of tidal pressure and surface subsidence from the simulated signals. The phase information and corresponding time of the simulated signals with the same frequency components as the monitoring data were also extracted. This step utilized the previously established porosity elasticity equation to obtain simulated signals relevant to actual conditions through numerical simulation, and further extracted key information, providing a basis for subsequent parameter correction.

[0035] By acquiring the tidal period coefficient and using Fast Fourier Transform to separate the tidal component from surface settlement, structural settlement, and stratum hydrological signal monitoring data, the phase shift and time delay parameters in the porosity elastic equation are corrected, and finally, the optimized porosity elastic equation is obtained through inversion. This series of operations can accurately quantify the transmission efficiency of tidal pressure fluctuations caused by tidal loads in the tunnel strata, clarify the specific influence of tidal factors on tunnel settlement, and provide key parameters for accurately predicting tunnel settlement.

[0036] In step S2, the phase difference and time difference are determined based on the phase information of the simulated signal and its corresponding time, and the phase and time of the measured tidal signal. This step quantifies the difference between the simulation results and the measured data by comparing the two, providing a clear target for parameter correction.

[0037] Based on the phase difference and time difference, the phase shift and time delay parameters in the porosity elasticity equation are corrected to obtain the optimized porosity elasticity equation. By correcting the parameters, the simulation results are made closer to reality, improving the accuracy and reliability of the model.

[0038] Specifically, in step S3, distributed optical fibers, InSAR satellite images, and pore water pressure sensors, which are equidistantly arranged along the tunnel axis, collect monitoring data on the tunnel's structural settlement, surface settlement, and stratum hydrological signals received by the distributed optical fibers, InSAR satellite images, and pore water pressure sensors respectively through the heterogeneous data access module 101.

[0039] It can access data through eight industrial protocols such as Modbus / TCP and MQTT, and can cover the monitoring needs of "surface-structure-stratum".

[0040] The spatiotemporal alignment module 102 performs spatiotemporal alignment processing on the collected monitoring data of structural settlement, surface settlement, and stratigraphic hydrological signals of the tunnel. After achieving millimeter-level (10ns) time synchronization based on BeiDou time synchronization, it uses the IDW algorithm (R²≥0.95) to fuse the discrete monitoring point data of structural settlement, surface settlement, and stratigraphic hydrological signals into a millimeter-level continuous settlement field. The spatiotemporal alignment module 102 can achieve millimeter-level continuous settlement field fusion with a spatiotemporal synchronization accuracy of 10ns.

[0041] The confidence level of each discrete monitoring point in the millimeter-level continuous settlement field is obtained; the quality verification module 103 identifies abnormal discrete monitoring point data in the millimeter-level continuous settlement field through the isolated forest algorithm, and obtains the filtered millimeter-level continuous settlement field as the input of the prediction network model.

[0042] When the proportion of abnormal discrete monitoring point data / low-confidence discrete monitoring point data exceeds 15%, a data alarm is triggered to ensure the reliability of subsequent modeling and model prediction.

[0043] Millimeter-level precision can more accurately capture subtle changes in tunnel settlement, providing a high-quality, high-resolution data foundation for subsequent analysis and prediction. Compared with discrete or imprecise data obtained by traditional methods, it can more comprehensively and realistically reflect the actual situation of tunnel settlement.

[0044] The spatiotemporal alignment and fusion process fully considers the temporal and spatial correlations of different types of data, organically combining the originally independent surface subsidence, structural subsidence, and stratigraphic hydrological signals into a unified continuous subsidence field. This correlation helps to more accurately analyze the complex relationship between tunnel subsidence and stratigraphic hydrological changes, providing strong support for a deeper understanding of tunnel subsidence mechanisms.

[0045] In step S4, the Transformer core module 301 is used to construct an 8-head attention mechanism time series model, inputting the millimeter-level continuous settlement field to be measured into the optimized pore elastic equation to obtain the pore water pressure values ​​of tides at different locations at different times.

[0046] The pore water pressure values ​​of tides at different locations at different times were preprocessed, and the preprocessed pore water pressure values ​​of tides at different locations at different times were input into an 8-head attention mechanism time series model. The model generates differential settlement probability and settlement rate curves for the next 72 hours based on the input sequence.

[0047] The differential settlement probability curve reflects the likelihood of differential settlement occurring at different locations at different times in the future.

[0048] The settlement rate curve describes how the settlement velocity changes over time.

[0049] By using the tidal phase embedding layer of the 8-head attention mechanism time series model, the tidal periodic features are extracted and transformed into a 128-dimensional feature vector. This vector is then input into the Transformer multi-head attention mechanism time series model to predict differential settlement risk. The settlement inference module 302 outputs the differential settlement probability (0-1 risk value) and settlement rate curve for the next 72 hours. Combined with Monte Carlo simulation, a settlement envelope is generated to clarify "when, where, and how much settlement may occur".

[0050] Based on the output differential settlement probability and settlement rate curves and settlement envelope for the next 72 hours, the risk of tunnel settlement in the future is assessed. When the predicted tunnel settlement is ≥3mm / 10m, high-risk sections are marked, and a settlement envelope is generated in conjunction with Monte Carlo simulation.

[0051] Specifically, based on the differential settlement probability and settlement rate curves for the next 72 hours, Monte Carlo simulation is used to generate multiple settlement sample paths that conform to the set distribution. Statistical analysis is performed on the generated settlement sample paths that conform to the set distribution to determine the statistical value of settlement and draw the settlement envelope.

[0052] Monte Carlo simulation is a probabilistic and statistical stochastic simulation method that approximates the solution of complex problems by generating a large number of random samples and calculating their statistical characteristics. In settlement prediction, Monte Carlo simulation can be used to generate a large number of possible settlement paths, thereby assessing the uncertainty of settlement.

[0053] This invention quantifies the risks of extreme working conditions using the Monte Carlo settlement envelope, achieving an accuracy rate of over 92% in identifying high-risk sections.

[0054] Model optimization module 303 deploys an online learning framework to perform incremental training based on the deviation between the daily added monitoring data (true values) and the prediction results. Through knowledge distillation technology, it compresses the volume of the Transformer multi-head attention mechanism time series model to 30% of its original size, ensuring prediction accuracy while adapting to edge deployment requirements.

[0055] By inputting a millimeter-scale continuous settlement field and an optimized porosity elasticity equation into the prediction model, the differential settlement probability curves for future time steps are output. This prediction method, based on high-precision data and accurate parameters, can more accurately predict the settlement of tunnels at different future time points, significantly improving the accuracy and reliability of prediction results compared to traditional prediction methods.

[0056] By simulating and generating a settlement envelope, the extent and trend of tunnel settlement over a future period can be visually displayed, allowing engineers to clearly understand the risk level and potential impact of tunnel settlement. This intuitive risk assessment method helps engineers take timely and effective countermeasures, such as strengthening monitoring, adjusting construction plans, or carrying out reinforcement treatments, thereby reducing the impact of tunnel settlement on project safety and operation.

[0057] like Figure 2 As shown, the tunnel uneven settlement tidal prediction system proposed in this invention also includes an edge response gateway module 400 and an active prevention and control decision module 500.

[0058] like Figure 3 As shown, the edge response gateway module 400 includes a lightweight inference module 401, a protocol conversion module 402, and a containerized management module 403, wherein: The lightweight inference module 401 is used to develop model slicing technology to distribute the computational load to multi-core processors, enabling a single device to process 200 tunnel cross-section data points in parallel. It supports wide temperature range operation from -25℃ to 70℃, ensuring millimeter-level risk assessment efficiency. The lightweight inference module 401 distributes the multi-core computing load to process 200 tunnel cross-section data points. The protocol conversion module 402 enables automatic switching between 5G / LoRa dual-mode communication, and the containerized management module supports hot-swapping of new sensor drivers. The protocol conversion module 402 integrates 12 industrial protocol converters, including Modbus / OPC UA, and automatically switches transmission modes in signal dead zones using 5G+LoRa dual-mode communication. The containerized management module 403 is used for Docker containerized deployment of functional modules, providing a hot-swappable interface to support the loading of new sensor drivers.

[0059] By integrating 12 industrial protocol converters such as Modbus / OPC UA, it supports automatic switching between 5G (high-speed, long-distance) and LoRa (low-power, wide-coverage) dual modes, adapts to the complex communication environment in tunnels, and enables real-time transmission of prediction results to decision-making modules and digital twin platforms.

[0060] The proactive prevention and control decision-making module 500 includes a threshold early warning module 501, a reinforcement scheme module 502, and a digital twin linkage module 503.

[0061] The method for predicting uneven settlement tides in tunnels proposed in this invention also includes, based on the risk of tunnel settlement over a future period, identifying high-risk sections and obtaining settlement monitoring data after grouting, specifically including: The threshold warning module 501 determines multiple dynamic warning thresholds of the same level and obtains multiple warning events, including: under normal working conditions, differential settlement ≥3mm / 10m triggers a level one alarm; during heavy rain / high tide (decrease in ground stability) ≥2mm / 10m triggers a level one alarm; when the predicted limit is exceeded for 3 consecutive times, an audible and visual alarm is activated to avoid false alarms / missed alarms caused by a single threshold.

[0062] The reinforcement scheme module 502 generates a three-dimensional grouting scheme: Based on the BIM model, the settlement exceeding the standard area of ​​the tunnel settlement risk in the future period is marked. By identifying high-risk sections, a three-dimensional grouting scheme is generated, including: grouting hole positions, grout type adapted to the stratum characteristics, and grouting volume that meets the optimized transfer efficiency. The digital twin linkage module 503 synchronizes multiple early warning events, high-risk sections, and three-dimensional grouting solutions to the digital twin platform, and pushes corresponding reference solutions by associating them with historical handling cases from the past 5 years. Based on the corresponding reference scheme, the settlement control effect under different grouting hole positions and grouting volume is simulated using a reinforcement effect simulator. Settlement monitoring data after grouting is obtained to assist in optimizing and generating a precise prevention and control scheme.

[0063] The acquired settlement monitoring data after grouting is fed back to the Transformer multi-head attention mechanism time series model and the porosity elasticity equation. The weights of the model and the porosity elasticity equation are dynamically corrected and optimized, forming a closed loop of "monitoring → modeling → prediction → prevention and control → feedback → optimization", which continuously improves the accuracy of early warning and the effectiveness of prevention and control.

[0064] like Figure 4 The diagram shown is a structural schematic of an electronic device provided in an embodiment of the present invention. The electronic device may include: a memory 1, a processor 2, and a computer program stored on the memory 1 and capable of running on the processor 2.

[0065] When processor 2 executes the program, it implements the method and system for predicting tunnel uneven settlement tides provided in the above embodiments.

[0066] Furthermore, electronic devices also include: Communication interface 3 is used for communication between memory 1 and processor 2.

[0067] Memory 1 is used to store computer programs that can run on processor 2.

[0068] Memory 1 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0069] If memory 1, processor 2, and communication interface 3 are implemented independently, then communication interface 3, memory 1, and processor 2 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 4 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0070] In practical implementation, if memory 1, processor 2 and communication interface 3 are integrated on a single chip, then memory 1, processor 2 and communication interface 3 can communicate with each other through the internal interface.

[0071] Processor 2 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0072] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described digital twin-based intelligent early warning method for tidal-driven uneven settlement of tunnels.

[0073] The proposed method for predicting tunnel uneven settlement tides comprehensively considers various influencing factors of tunnel settlement, including surface settlement, structural settlement, changes in ground hydrology, and tidal loads. Through a series of steps, these factors are organically combined to form a complete prediction system. This systematic and comprehensive analysis method can more fully consider the complex mechanisms of tunnel settlement, avoid the limitations of single-factor analysis, and improve the comprehensiveness and accuracy of the prediction results.

[0074] The proposed tunnel uneven settlement tidal prediction system is based on a multi-sensor fusion module 100. It collects multi-dimensional surface-structure-stratum data from distributed optical fibers, InSAR satellite imagery, and pore water pressure sensors, and after spatiotemporal alignment and quality verification, forms a millimeter-level continuous settlement field. A tidal-driven modeling module 200 then uses this data to construct a pore elasticity equation, quantifying the chain reaction mechanism of tidal load, groundwater level, and tunnel settlement, and dynamically optimizing the pore elasticity equation (quantifying the transmission efficiency of seawater pressure caused by tidal load). A time-series prediction engine module 300 outputs the probability and envelope of differential settlement risk for the next 72 hours based on this prediction model. An edge response gateway module 400 processes the prediction results in real time, achieving millisecond-level risk assessment and protocol conversion. Finally, an active prevention and control decision module 500 dynamically adjusts thresholds and generates grouting schemes based on early warning information, and simulates the reinforcement effect through a digital twin platform, forming a closed-loop optimization process from data perception to risk warning and then to prevention and control execution.

[0075] The spatiotemporal alignment module 102 can achieve millimeter-level continuous settlement field fusion with a spatiotemporal synchronization accuracy of 10ns. The 10ns-level time synchronization achieved by the spatiotemporal alignment module refers to using BeiDou time service to provide a unified timestamp reference for heterogeneous data sources such as distributed optical fibers, InSAR, and pore water pressure sensors, so as to eliminate device clock drift and ensure extremely high time consistency of multi-source data when fused into a continuous settlement field.

[0076] Unlike the spatiotemporal alignment module, the edge response gateway module 400 achieves millisecond-level risk assessment and protocol conversion. Millisecond-level risk assessment means that after receiving prediction results, the module rapidly calculates the risk level from tunnel cross-section data using model slicing and lightweight inference technology, while simultaneously completing the entire industrial protocol conversion process in milliseconds. While the spatiotemporal alignment module focuses on the time alignment accuracy during data acquisition, the edge response gateway module 400 emphasizes the real-time processing latency during the decision-making and response phase. Both address different performance metrics—data fusion accuracy and early warning response speed—to jointly ensure the closed-loop efficiency of the system from high-precision perception to real-time prevention and control.

[0077] The method provided by the embodiments of the present invention enables tidal tunnel operation and maintenance to leapfrog towards real-time adaptive development, providing a technical paradigm for improving the resilience of coastal infrastructure.

[0078] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

[0079] Furthermore, unless otherwise stated, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. All references to this specification are incorporated by way of citation to disclose and describe methods relating to those references. In the event of any conflict with any incorporated reference, the content of this specification shall prevail.

Claims

1. A method for predicting tunnel uneven settlement tides, characterized in that, Includes the following steps: Based on Biot's consolidation theory, a porosity elastic equation describing the transmission process of tidal pressure fluctuations is established. Surface subsidence, structural subsidence, and hydrological signal monitoring data at the current time step are acquired. Tidal components in these data are separated using fast Fourier transform. These tidal components are then used as input loads and substituted into the porosity elastic equation to obtain the time-varying curves of tidal pressure and surface subsidence in the simulated signals. The phase information of the simulated signals with the same frequency components as the monitoring data and their corresponding times are extracted. Based on the phase information of the simulated signal and its corresponding time, and the phase and time of the measured tidal signal, the phase difference and time difference are determined; based on the phase difference and time difference, the phase offset and time delay parameters in the porosity elastic equation are corrected to obtain the optimized porosity elastic equation. Historical time-step monitoring data of surface settlement, structural settlement and stratigraphic hydrological signals of the tunnel are acquired, and then fused with the acquired current time-step monitoring data of surface settlement, structural settlement and stratigraphic hydrological signals to generate a continuous settlement field. The continuous settlement field to be measured is input into the optimized pore elasticity equation to obtain the pore water pressure values ​​of tides at different locations and at different times; the pore water pressure values ​​at different times are input into the pre-trained prediction network model, and the differential settlement probability of future time steps is output through prediction.

2. The method for predicting tunnel uneven settlement tides according to claim 1, characterized in that, The porosity elastic equation describing the tidal pressure fluctuation transmission process, based on Biot's consolidation theory, specifically includes: Based on Biot's consolidation theory, the equilibrium equation and seepage equation are determined. Among them, the equilibrium equation is used to describe the coupling relationship between soil skeleton stress and pore pressure, and the seepage equation is used to characterize the transmission process of tidal pressure through boundary terms in the porous medium. Based on the equilibrium equation and the seepage equation, establish the porosity elasticity equation; The porosity elastic equation is: ； In the equation: ∇ is the gradient / divergence operator in the equilibrium equation; u is the displacement vector of the solid skeleton; D D:∇u represents the elastic stiffness tensor of the solid skeleton; α is the Biot consolidation coefficient, ∇u represents the stress tensor corresponding to the strain. p This represents the pressure change of the pore fluid; in the seepage equation, M Here, represents the Biot modulus; ∇·u represents the volumetric strain of the solid skeleton. k For the transport efficiency of porous media, γ w Let be the specific weight of the fluid, ΔPtide be the change in tidal pressure, and δ(z−H) be the Dirac delta function. H For boundary depth; Based on the established pore elasticity equation, by introducing tidal periodic boundary conditions, the pressure change of the pore fluid is transformed into the pore water pressure change in the porous medium. Among them, the transport efficiency of porous media k The amount of pressure change in the pore fluid is determined by ∇ P The degree of attenuation, elastic stiffness tensor D Affects the energy conversion ratio; By solving the porosity elasticity equation, the quantification unit ∇ is determined. P The resulting change in effective formation stress characterizes the transmission efficiency. k The value of .

3. The method for predicting tunnel uneven settlement tides according to claim 1, characterized in that, The process involves acquiring surface subsidence, structural subsidence, and hydrological signal monitoring data at the current time step; separating the tidal component from these data using Fast Fourier Transform; substituting the tidal component as an input load into the porosity elasticity equation to obtain the time-varying curves of tidal pressure and surface subsidence in the simulated signals; and extracting the phase information and corresponding time of the simulated signals with the same frequency components as the monitoring data. Specifically, this includes: Acquire monitoring data on surface subsidence, structural subsidence, and stratigraphic hydrological signals at the current time step; By using Fast Fourier Transform, the monitoring data of surface subsidence, structural subsidence and stratigraphic hydrological signals are transformed from the time domain to the frequency domain, and the periodic-frequency components corresponding to the astronomical tide table are identified as tidal components, while random load signals including high-frequency noise and low-frequency drift are filtered out. The denoised tidal components are used as input loads and substituted into the porosity elastic equation for numerical simulation to obtain the tidal pressure and surface subsidence variation curves of the simulated signals over time. The FFT transformation is performed on the time-varying curves of tidal pressure and surface subsidence of the acquired simulated signals to extract the phase information of the components with the same frequency as the actual tidal components, and the time when the simulated signal reaches the phase information is determined. The phase and time of the analog signal are compared with the phase and time of the actual tidal signal to determine the phase difference and time difference; Based on the phase difference and time difference, correct the phase shift and time delay parameters in the porosity elasticity equation; Regarding phase offset, when the phase of the analog signal lags behind the actual tidal signal, decrease the phase offset parameter; when the phase of the analog signal leads the actual tidal signal, increase the phase offset parameter. Regarding the time delay parameter, when the time of the analog signal lags behind the actual tidal signal, decrease the time delay parameter; when the time of the analog signal leads the actual tidal signal, increase the time delay parameter. The porosity elastic equation is updated using the corrected phase offset and time delay parameters. Numerical simulation is then performed again using the updated porosity elastic equation until the phase difference and time difference between the simulated signal and the actual tidal signal meet the preset threshold requirements, thus obtaining the optimized porosity elastic equation.

4. The method for predicting tunnel uneven settlement tides according to claim 3, characterized in that, The process involves inputting the continuous settlement field to be measured into the optimized pore elasticity equation to obtain the pore water pressure values ​​at different locations and times; inputting the pore water pressure values ​​at different times into a pre-trained prediction network model to predict and output the differential settlement probability at future time steps, specifically including: The continuous settlement field to be measured is input into the optimized pore elastic equation, and the pore water pressure values ​​of the tide at different locations at different times are output. The pore water pressure values ​​of tides at different locations at different times are input into the Transformer multi-head attention mechanism time series model, which outputs the differential settlement probability and settlement rate curves for the next few hours.

5. The method for predicting tunnel uneven settlement tides according to claim 4, characterized in that, The method of predicting and outputting the differential settlement probability at future time steps also includes: Based on the differential settlement probability and settlement rate curves over the next few hours, Monte Carlo simulation is used to generate multiple settlement sample paths that conform to the set distribution. Statistical analysis is performed on the generated settlement sample paths that conform to the set distribution to determine the statistical value of settlement and draw the settlement envelope. Based on the output differential settlement probability and settlement rate curves and settlement envelope for the next few hours, assess the risk of tunnel settlement in the future.

6. The method for predicting tunnel uneven settlement tides according to claim 4, characterized in that, This also includes identifying high-risk sections based on the risk of tunnel settlement over a future period, and obtaining settlement monitoring data after grouting. Specifically, this includes: Multiple warning events are obtained by determining multiple dynamic warning thresholds of the same level; 3D grouting scheme generation: Based on the BIM model, areas with excessive settlement risk in the tunnel over a period of time are marked. By identifying high-risk sections, a 3D grouting scheme is generated, including: grouting hole locations, grout type suitable for geological characteristics, and grouting volume that meets the optimized transfer efficiency. Multiple early warning events, high-risk sections, and 3D grouting solutions are synchronized to the digital twin platform, and corresponding reference solutions are pushed by associating them with historical handling cases. Based on the corresponding reference scheme, the settlement control effect under different grouting hole positions and grouting volumes was simulated using a reinforcement effect simulator, and settlement monitoring data after grouting was obtained.

7. The method for predicting tunnel uneven settlement tides according to claim 6, characterized in that, It also includes feeding back the acquired settlement monitoring data after grouting to the Transformer multi-head attention mechanism time series model and the porosity elasticity equation, and dynamically correcting the weights of the model and optimizing the porosity elasticity equation, respectively.

8. The method for predicting tunnel uneven settlement tides according to claim 1, characterized in that, The process involves acquiring historical time-step monitoring data of surface settlement, structural settlement, and stratigraphic hydrological signals of the tunnel, aligning these data with the current time-step monitoring data, and then fusing them to generate a continuous settlement field. Specifically, this includes: By using distributed optical fibers, InSAR satellite imagery, and pore water pressure sensors that are equidistantly deployed along the tunnel axis, historical time step monitoring data of the tunnel's structural settlement, surface settlement, and stratum hydrological signals are collected. The historical time step monitoring data of the collected tunnel structural settlement, surface settlement and stratum hydrological signals and the current time step monitoring data of surface settlement, structural settlement and stratum hydrological signals are spatiotemporally aligned. After achieving millimeter-level time synchronization based on BeiDou time synchronization, the IDW algorithm is used to fuse the discrete monitoring point data of time-synchronized structural settlement, surface settlement and stratum hydrological signals into a millimeter-level continuous settlement field. The confidence scores of discrete monitoring point data in a continuous settlement field are obtained. Anomalies in discrete monitoring point data in the continuous settlement field are identified using the isolated forest algorithm, and the filtered continuous settlement field is obtained as the input to the prediction network model.

9. A prediction system for tunnel uneven settlement tides, characterized in that, include: The pore elasticity equation construction module is used to establish pore elasticity equations describing the tidal pressure fluctuation transmission process based on Biot's consolidation theory. Acquire surface subsidence, structural subsidence, and hydrological signal monitoring data at the current time step. Separate the tidal components from the surface subsidence, structural subsidence, and hydrological signal monitoring data using Fast Fourier Transform. Substitute the tidal components as input loads into the porosity elasticity equation to obtain the time-varying curves of tidal pressure and surface subsidence of the simulated signals. Extract the phase information of the simulated signals with the same frequency components as the monitoring data and their corresponding times. The porosity elasticity equation optimization module is used to determine the phase difference and time difference based on the phase information of the simulated signal and its corresponding time and the phase and time of the measured tidal signal; based on the phase difference and time difference, the phase offset and time delay parameters in the porosity elasticity equation are corrected to obtain the optimized porosity elasticity equation; The multi-sensor fusion module is used to acquire historical time step monitoring data of surface settlement, structural settlement and stratigraphic hydrological signals of the tunnel, and then fuse them with the acquired current time step monitoring data of surface settlement, structural settlement and stratigraphic hydrological signals to generate a continuous settlement field. The time-series prediction engine module is used to input the continuous settlement field to be measured into the optimized pore elasticity equation to obtain the pore water pressure values ​​of tides at different locations at different times; input the pore water pressure values ​​at different times into the pre-trained prediction network model, and output the differential settlement probability of future time steps through prediction.

Images