A tunnel full-life-cycle intelligent operation management and control system based on digital twinning

Abstract

This invention relates to the field of intelligent operation and maintenance in tunnel engineering, and discloses a smart operation and maintenance management system for the entire life cycle of tunnels based on digital twins. The system divides tunnel segments, joints, backfill grouting zones, and drainage branches into tunnel physical units, establishes the state inheritance relationship of design, construction, completion, operation, and maintenance data, and determines the range of structural and hydraulic parameters for each physical unit. It constructs a dynamic hydraulic-to-structural coupling propagation diagram through parameter perturbation, generates abnormal event sequences based on structural and hydraulic monitoring data, and reversely determines the initial unit of the defect and the effective propagation path. Candidate maintenance measures are converted into changes in physical parameters for maintenance simulation, and maintenance intervention combinations that can block the effective propagation path without causing risk transfer to adjacent units are selected. After maintenance, the propagation path blocking status is verified through boundary condition pairing or hydraulic perturbation, and local correction is performed on the propagation edge where the predicted response differs from the measured response for the first time.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance of tunnel engineering, specifically to a smart operation and maintenance management system for the entire life cycle of tunnels based on digital twins. Background Technology

[0002] With the continuous expansion of urban rail transit, highway tunnels, and underground infrastructure construction, shield tunnels are gradually entering the long-term operation phase. During construction and operation, shield tunnels are prone to defects such as segment misalignment, joint opening, backfill voids, localized leakage, and abnormal drainage due to factors such as uneven ground deformation, defects in backfill grouting, reduced joint stiffness, changes in groundwater level, and decreased drainage capacity. These defects are usually not isolated but rather propagate through structural and hydraulic responses along segment joints, backfill voids, and drainage channels, leading to discrepancies between the actual leakage location and the initial defect location, increasing the difficulty of defect localization and maintenance decisions.

[0003] Existing intelligent tunnel operation and maintenance systems typically store design data, construction records, monitoring data, and maintenance records separately, and use 3D models, monitoring thresholds, or historical defect cases to display the tunnel's operational status. Some systems use digital twin models for structural response calculations or defect trend predictions, but the correspondence between data from different stages and specific segments, joints, backfill grouting areas, and drainage branches is not clear enough. Model parameters often use design values, empirical values, or uniform values ​​for different zones, making it difficult to reflect local state changes caused by construction differences and long-term operation. Furthermore, existing defect diagnoses often rely on the location of abnormal monitoring points, the correlation of monitoring quantities, or pre-established knowledge relationships between defects and causes. This makes it difficult to distinguish the continuous coupled processes such as backfill voids causing joint opening, joint opening causing changes in permeability, and water pressure changes affecting the joint's stress state. This easily leads to the direct misinterpretation of defect locations as defect source locations.

[0004] Current maintenance decisions typically select grouting, sealing, dredging, or structural reinforcement measures based on the type and location of the defect. However, there is a lack of verifiable criteria to determine whether these measures can effectively block the spread of the defect. For example, while sealing a single water-producing joint may temporarily reduce surface water flow, it could cause water pressure to shift to adjacent joints or areas behind the wall, creating new leakage points. After maintenance, existing systems often evaluate the effectiveness based on a decrease in water flow or a recovery in monitoring values. This makes it difficult to eliminate the influence of changes in rainfall, groundwater level, temperature, and operational loads on monitoring results, and also makes it difficult to accurately determine the local parameters requiring correction when model predictions differ from post-maintenance measurements. Therefore, it is necessary to establish an intelligent operation and maintenance management system that can integrate tunnel lifecycle data, identify the propagation relationship between structural and hydraulic responses, invert the defect initiation location and effective propagation path, and verify the blocking status of the propagation path after maintenance. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a smart operation and maintenance management system for the entire lifecycle of tunnels based on digital twins, in order to solve the technical problems existing in the prior art.

[0006] The above-mentioned technical objective of the present invention is achieved through the following technical solution: A digital twin-based intelligent operation and maintenance management system for the entire lifecycle of tunnels, applied to shield tunnels, includes: The physical unit construction module is used to divide the shield tunnel into at least two tunnel physical units and map design data, construction data, completion acceptance data and operation monitoring data to each tunnel physical unit. The life cycle status construction module is used to determine the initial value range of joint stiffness, wall-back constraint stiffness, wall-back void rate, joint permeability coefficient, and hydraulic conduction capacity of drainage branches for each tunnel physical unit based on design data, construction data, and completion acceptance data. The coupled propagation graph construction module is used to apply perturbations to the physical parameters corresponding to the initial value range in the digital twin model, determine the structural propagation relationship based on the structural response of other tunnel physical units, determine the hydraulic propagation relationship based on the head difference and hydraulic response of adjacent tunnel physical units, and construct a dynamic hydraulic-to-structure coupled propagation graph with tunnel physical units as nodes and the structural propagation relationship, hydraulic propagation relationship and bidirectional conversion relationship between structural response and hydraulic response as propagation edges. The anomaly sequence generation module is used to compare the structural and hydraulic monitoring data of each tunnel physical unit with the operational baseline state, determine the abnormal unit and the start time of the anomaly, and generate an anomaly event sequence. The disease inversion module is used to trace candidate starting units backward from the abnormal unit along the dynamic hydraulic-to-structure coupling propagation diagram. Within the initial value range of each candidate starting unit, hydraulic-to-structure coupling propagation simulation is performed. Based on the differences between the simulation results and the abnormal event sequence in terms of abnormal range, occurrence order, propagation delay and response direction, the disease starting unit with the smallest comprehensive difference and the effective propagation path are determined from the candidate starting units that meet the structural equilibrium condition and water conservation condition. The maintenance simulation module is used to convert candidate maintenance measures into changes in physical parameters and perform maintenance simulations. It determines the maintenance intervention combination that ensures that the propagation coefficient of at least one propagation edge in each effective propagation path is lower than the corresponding activation threshold and that the response of adjacent tunnel physical units does not exceed the limit value, and generates maintenance tasks. The maintenance verification and correction module is used to obtain the post-maintenance response sequence after the maintenance task is carried out. When the post-maintenance response sequence no longer satisfies the propagation timing relationship of the effective propagation path, it is determined that the effective propagation path is blocked, and the physical parameters of the propagation edge where the predicted recovery sequence and the post-maintenance response sequence first show a difference are corrected.

[0007] Preferably, the physical unit construction module determines the spatial range of the segment block according to the tunnel axial mileage and the circumferential position inside the segment ring, and associates the circumferential joint located at the axial boundary of the spatial range, the longitudinal joint located at the circumferential boundary of the spatial range, the wall grouting zone covering the outer projection area of ​​the spatial range, and the drainage branch section receiving drainage within the spatial range to the same tunnel physical unit, and configures a unique unit identifier for each tunnel physical unit. The lifecycle state construction module writes design data, construction data, completion acceptance data, operation monitoring data, and historical maintenance data into the design state, construction formation state, completion benchmark state, operation evolution state, and maintenance intervention state respectively, according to the unique unit identifier. It also establishes a state inheritance chain according to the formation time of each state, and the subsequent state is associated with the previous state by adding a new state record without overwriting the data of the previous state.

[0008] Preferably, the life cycle status construction module uses the joint design stiffness, wall back grouting design parameters, waterproofing design parameters and drainage design capacity in the design data as physical parameter benchmarks; Based on the joint opening, joint misalignment, and joint sealing test results, the parameter test values ​​of joint stiffness and joint permeability coefficient are determined using the calibration relationship between the test values ​​and physical parameters in the digital twin model; the parameter test values ​​of back wall constraint stiffness and back wall void rate are determined based on the actual measurement of back wall grouting, grouting pressure, and back wall void test results; the parameter test values ​​of hydraulic conduction capacity of drainage branches are determined based on the flow rate and pressure loss test results of drainage branches. For each physical parameter, the larger of the parameter detection value minus the converted measurement uncertainty and the design allowable lower limit is used as the lower limit of the initial value range. The smaller of the parameter detection value plus the converted measurement uncertainty and the design allowable upper limit is used as the upper limit of the initial value range.

[0009] Preferably, the coupled propagation graph construction module sequentially applies a perturbation to one physical parameter of a tunnel physical unit, and obtains the changes in displacement, strain, joint opening, or internal force of other tunnel physical units while keeping other physical parameters and boundary conditions unchanged. The ratio of the disturbance amount of the physical parameter to the value of the physical parameter before the disturbance is determined as the relative disturbance amount of the physical parameter. The ratio of the change in the structural response to the normalized reference value of the structural response is determined as the relative change in the structural response. The normalized reference value of the structural response is the larger value between the absolute value of the structural response before the disturbance and the upper limit of the structural response noise. For any two tunnel physical units, the ratio of the relative change in the structural response of the downstream tunnel physical unit to the relative disturbance of the physical parameters of the upstream tunnel physical unit is determined as the structural propagation coefficient. When the structural propagation coefficient is not lower than the structural propagation activation threshold and the direction of structural response change conforms to the structural equilibrium calculation result of the digital twin model, a structural propagation edge is established between the upstream tunnel physical unit and the downstream tunnel physical unit. The structural propagation activation threshold is determined based on the larger of the normalized upper limit of the structural monitoring quantity under the operating baseline state and the normalized measurement uncertainty of the monitoring device.

[0010] Preferably, the coupled propagation graph construction module determines the hydraulic propagation direction based on the head difference between adjacent tunnel physical units, determines the ratio of the relative change in water pressure or drainage flow of the downstream tunnel physical unit to the relative change in head difference of the upstream tunnel physical unit as the hydraulic propagation coefficient, and establishes a hydraulic propagation edge when the hydraulic propagation coefficient is not lower than the hydraulic propagation activation threshold. The ratio of the relative change in the joint permeability coefficient to the relative change in the joint opening is determined as the structure-to-hydraulic conversion coefficient, and a structure-to-hydraulic conversion edge is established when the structure-to-hydraulic conversion coefficient is not lower than the structure-to-hydraulic conversion activation threshold. The ratio of the relative change in the normal force of the joint or the relative change in the joint opening to the relative change in the water pressure is determined as the hydraulic-to-structure conversion coefficient, and a hydraulic-to-structure conversion edge is established when the hydraulic-to-structure conversion coefficient is not lower than the hydraulic-to-structure conversion activation threshold. The activation thresholds for hydraulic propagation, structural-to-hydraulic conversion, and hydraulic-to-structural conversion are determined based on the larger of the normalized upper limit of the corresponding response quantity under the operating baseline state and the normalized measurement uncertainty of the corresponding monitoring device, respectively. The coupled propagation graph construction module recalculates each propagation coefficient as the groundwater level, joint opening, and hydraulic conduction capacity of the drainage branch change. Propagation edges with propagation coefficients not lower than the corresponding activation threshold are marked as active, while propagation edges with propagation coefficients lower than the corresponding activation threshold are marked as inactive.

[0011] Preferably, the abnormal sequence generation module extracts historical periods from the state inheritance chain where there is no maintenance work and the structural monitoring and hydraulic monitoring values ​​do not exceed the operation monitoring control values. The historical periods are grouped according to the numerical range of rainfall, groundwater level, temperature and tunnel operation load, and an operation benchmark state is established based on the structural monitoring and hydraulic monitoring values ​​in each group. The width of the numerical interval is not less than the measurement resolution of the corresponding boundary condition monitoring quantity, and the difference between the maximum and minimum values ​​of the corresponding boundary condition monitoring quantity within the same numerical interval does not exceed the upper limit of the fluctuation of the boundary condition monitoring quantity during the stable operating period. The abnormal sequence generation module selects the operating benchmark state based on the current rainfall, groundwater level, temperature and tunnel operating load values, and compares the current structural monitoring and hydraulic monitoring values ​​with the selected operating benchmark state. When the deviation between the displacement, strain, or joint opening and the operational baseline state exceeds the structural anomaly identification threshold, and the duration of the continuous exceedance of the structural anomaly identification threshold is greater than the longest continuous duration of occasional overruns of structural monitoring quantities under the operational baseline state, it is determined that the corresponding tunnel physical unit has a structural response anomaly. When the deviation between water pressure, moisture content, or drainage flow rate and the operating baseline state exceeds the hydraulic anomaly identification threshold, and the duration of continuous exceedance of the hydraulic anomaly identification threshold is greater than the longest continuous duration of occasional exceedance of hydraulic monitoring data under the operating baseline state, the corresponding tunnel physical unit is determined to have a hydraulic response anomaly. The structural anomaly identification threshold and the hydraulic anomaly identification threshold are determined based on the larger of the upper limit of the fluctuation of the corresponding monitored quantity under the operating baseline state and the measurement uncertainty of the corresponding monitoring device, respectively; the abnormal event sequence is arranged according to the start time of the anomaly, and the abnormal unit, anomaly type, response direction, response amplitude and duration are recorded.

[0012] Preferably, the disease inversion module traces each abnormal unit in the abnormal event sequence in reverse along the active propagation edge in the dynamic hydraulic to structural coupling propagation diagram, and determines the node that can reach at least one abnormal unit through the propagation edge as the candidate starting unit. For each candidate starting unit, the disease inversion module changes at least one physical parameter within the initial value range corresponding to the candidate starting unit, and loads the rainfall, groundwater level, temperature and tunnel operation load within the time period corresponding to the abnormal event sequence into the digital twin model to obtain the simulated anomaly range, simulated anomaly start time, simulated propagation delay and simulated response direction. The differences in the number of units between the simulated anomaly range and the measured anomaly range, the time difference between the simulated anomaly start time and the measured anomaly start time, the time delay difference between the simulated propagation delay and the measured propagation delay, and the directional difference between the simulated response direction and the measured response direction are determined respectively. Each difference is normalized by dividing it by the maximum value of each difference among all candidate starting units. The four normalized differences are multiplied by one-quarter and then summed to determine the comprehensive difference. The disease inversion module excludes candidate starting units that do not meet the structural equilibrium condition or water conservation condition, and determines the candidate starting unit with the smallest comprehensive difference as the disease starting unit; when at least two candidate starting units have the same comprehensive difference, the candidate starting units with fewer activation state propagation edges and smaller relative physical parameter adjustments are selected as the disease starting units in sequence. An effective propagation path consists of a propagation edge between the disease initiation unit and the abnormal unit that is in an active state, whose response direction is consistent with the abnormal event sequence, and whose cumulative propagation delay and measured propagation delay do not exceed the allowable deviation of propagation delay; the allowable deviation of propagation delay is determined based on the sum of the sampling interval and the time synchronization error at the start of the abnormality.

[0013] Preferably, the maintenance simulation module converts the back wall grouting measures into a reduction in the back wall void rate, an increase in the back wall constraint stiffness, and a reduction in the back wall grouting zone permeability; it converts the joint sealing measures into a reduction in the joint permeability coefficient; it converts the drainage branch dredging measures into an increase in the hydraulic conduction capacity of the drainage branch; and it converts the joint structure reinforcement measures into an increase in the joint stiffness. The maintenance simulation module loads the physical parameter changes corresponding to different combinations of candidate maintenance measures into the digital twin model, recalculates the propagation coefficient of the propagation edge in each effective propagation path, excludes candidate maintenance measure combinations that do not make the propagation coefficient of at least one propagation edge in each effective propagation path lower than the corresponding activation threshold, and excludes candidate maintenance measure combinations that make the stress, joint opening, or water pressure of adjacent tunnel physical units exceed the limit value. The limit value is the smaller of the design allowable value and the operation monitoring and control value for the corresponding parameter. The maintenance simulation module sorts the remaining candidate maintenance measures combinations in ascending order according to the number of physical units of the tunnel to be maintained. When the number of physical units of the tunnel to be maintained is the same, it sorts them in ascending order according to the number of propagation edges to be processed. When both of the above two numbers are the same, they sort them in ascending order according to the sum of the relative adjustment ranges of each physical parameter. The candidate maintenance measures combination ranked first is determined as the maintenance intervention combination. The maintenance task includes the physical unit of the tunnel to be maintained, the propagation edge to be processed, the maintenance range, the target physical parameters, the maintenance construction parameters corresponding to the target physical parameters, and the post-maintenance verification location.

[0014] Preferably, the maintenance verification and correction module uses boundary condition pairing verification or hydraulic disturbance verification to obtain the post-maintenance response sequence; When using boundary condition pairing verification, a reference period is selected from the historical monitoring data before maintenance, in which the corresponding differences between rainfall, groundwater level, temperature, and tunnel operating load and the verification period after maintenance do not exceed the allowable deviation of their respective boundary conditions. The response sequence of the reference period is then compared with the response sequence after maintenance. The allowable deviation of the boundary conditions is determined based on the larger of the measurement uncertainty of the corresponding boundary condition monitoring quantity and the upper limit of fluctuation during the stable operating period. When using hydraulic disturbance verification, before the maintenance task is carried out, the target drainage branch associated with the hydraulic propagation edge to be treated is controlled to generate hydraulic disturbance and the disturbance response sequence before maintenance is obtained. After the maintenance task is carried out, the target drainage branch is controlled again with the same valve opening change and holding time and the disturbance response sequence after maintenance is obtained. The valve opening change and holding time are limited to ensure that the water pressure at the monitoring point does not exceed the design allowable value and the drainage flow does not exceed the rated capacity of the drainage branch. When the fault initiation unit generates a structural or hydraulic response exceeding the upper limit of the response noise during the post-repair verification period, while the tunnel physical unit downstream of the propagation edge to be treated does not generate a response exceeding the corresponding upper limit of the response noise, or when the difference between the abnormal start times of adjacent tunnel physical units in the post-repair response sequence no longer meets the allowable deviation of the propagation delay of the corresponding propagation edge, it is determined that the effective propagation path is blocked. The upper limit of the response noise is determined based on the larger of the upper limit of the fluctuation of the corresponding monitored quantity and the measurement uncertainty of the corresponding monitoring device during a stable operating period without maintenance or hydraulic disturbance.

[0015] Preferably, the maintenance verification and correction module compares the recovery start time, recovery direction and recovery magnitude of each tunnel physical unit in the predicted recovery sequence and the post-maintenance response sequence in the direction of propagation from the initial defect unit to the abnormal unit. When the difference between the predicted response and the measured response of the tunnel physical unit located upstream in the propagation direction does not exceed the response identification threshold, while the difference between the predicted response and the measured response of the tunnel physical unit located downstream in the propagation direction exceeds the response identification threshold, the propagation edge connecting the two tunnel physical units is determined as the propagation edge where the difference first appears. The maintenance verification and correction module determines the correction range of the physical parameters associated with the propagation edge where the first difference occurs, based on the target physical parameters in the maintenance task, the actual execution values ​​of the maintenance construction parameters, and the post-maintenance test results. Under the condition of keeping the physical parameters of other propagation edges in the dynamic hydraulic-to-structural coupling propagation diagram unchanged, the module adjusts the joint stiffness, joint permeability coefficient, hydraulic conduction capacity of the drainage branch, or propagation delay of the propagation edge within the correction range until the difference between the recalculated predicted response and the post-maintenance response does not exceed the response identification threshold. The response identification threshold is determined based on the larger of the upper limit of the fluctuation of the corresponding monitored quantity during the stable period after maintenance and the measurement uncertainty of the corresponding monitoring device. The maintenance verification and correction module writes the maintenance task, the actual execution value of the maintenance construction parameters, the post-maintenance response sequence, the blocking result of the effective propagation path, and the corrected propagation edge physical parameters into the maintenance intervention status of the corresponding tunnel physical unit.

[0016] In summary, the present invention has the following main beneficial effects: By dividing tunnel segments, circumferential joints, longitudinal joints, grouting zones behind the tunnel walls, and drainage branch sections into uniquely identified tunnel physical units, and establishing a state inheritance chain in chronological order based on design data, construction data, completion acceptance data, operation monitoring data, and historical maintenance data, the joint stiffness, back-wall constraint stiffness, back-wall void rate, joint permeability coefficient, and hydraulic conduction capacity of drainage branches for each tunnel physical unit can be determined in conjunction with the actual construction and operation conditions. This achieves the effect of avoiding the use of uniform empirical parameters to describe the entire tunnel and improving the consistency between the digital twin model and the actual structural state of the tunnel.

[0017] By perturbing the structural and hydraulic parameters of the tunnel physical unit, and based on the structural response, head change, and bidirectional conversion relationship between the structural and hydraulic responses of adjacent tunnel physical units, a dynamic hydraulic-to-structural coupling propagation diagram is constructed. Combined with the range of abnormal units, the order of abnormal occurrence, propagation delay, and response direction, the starting unit and effective propagation path of the defect are determined in reverse. This achieves the goal of restoring the coupling relationship between surface leakage, joint opening, and wall voids to a propagation chain with actual structural connection and hydraulic connectivity, thereby reducing misjudgments caused by judging the cause of defects solely based on the location of abnormalities or historical experience.

[0018] By converting wall grouting, joint sealing, drainage branch dredging, and joint structure reinforcement into changes in physical parameters in a digital twin model, and using the inactivation of at least one propagation edge in each effective propagation path and the absence of stress, water pressure, or joint opening risk transfer in adjacent tunnel physical units as the screening criteria for maintenance schemes, the propagation paths after maintenance are verified by boundary condition pairing or hydraulic disturbance, and only the propagation edges where the predicted response and measured response differ for the first time are locally corrected. This achieves the effect of controlling the maintenance scope while blocking the spread of defects, verifying the maintenance effect, and reducing indiscriminate parameter adjustments in the digital twin model. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Example 1 refer to Figure 1 A digital twin-based intelligent operation and maintenance management system for the entire lifecycle of tunnels, applied to shield tunnels, includes: The physical unit construction module is used to divide the shield tunnel into at least two tunnel physical units and map design data, construction data, completion acceptance data and operation monitoring data to each tunnel physical unit. The life cycle status construction module is used to determine the initial value range of joint stiffness, wall-back constraint stiffness, wall-back void rate, joint permeability coefficient, and hydraulic conduction capacity of drainage branches for each tunnel physical unit based on design data, construction data, and completion acceptance data. The coupled propagation graph construction module is used to apply perturbations to the physical parameters corresponding to the initial value range in the digital twin model, determine the structural propagation relationship based on the structural response of other tunnel physical units, determine the hydraulic propagation relationship based on the head difference and hydraulic response of adjacent tunnel physical units, and construct a dynamic hydraulic-to-structure coupled propagation graph with tunnel physical units as nodes and the structural propagation relationship, hydraulic propagation relationship and bidirectional conversion relationship between structural response and hydraulic response as propagation edges. The anomaly sequence generation module is used to compare the structural and hydraulic monitoring data of each tunnel physical unit with the operational baseline state, determine the abnormal unit and the start time of the anomaly, and generate an anomaly event sequence. The disease inversion module is used to trace candidate starting units backward from the abnormal unit along the dynamic hydraulic-to-structure coupling propagation diagram. Within the initial value range of each candidate starting unit, hydraulic-to-structure coupling propagation simulation is performed. Based on the differences between the simulation results and the abnormal event sequence in terms of abnormal range, occurrence order, propagation delay and response direction, the disease starting unit with the smallest comprehensive difference and the effective propagation path are determined from the candidate starting units that meet the structural equilibrium condition and water conservation condition. The maintenance simulation module is used to convert candidate maintenance measures into changes in physical parameters and perform maintenance simulations. It determines the maintenance intervention combination that ensures that the propagation coefficient of at least one propagation edge in each effective propagation path is lower than the corresponding activation threshold and that the response of adjacent tunnel physical units does not exceed the limit value, and generates maintenance tasks. The maintenance verification and correction module is used to obtain the post-maintenance response sequence after the maintenance task is carried out. When the post-maintenance response sequence no longer satisfies the propagation timing relationship of the effective propagation path, it is determined that the effective propagation path is blocked, and the physical parameters of the propagation edge where the predicted recovery sequence and the post-maintenance response sequence first show a difference are corrected.

[0022] The on-site data acquisition device is used to collect structural monitoring data, hydraulic monitoring data, and boundary condition data for the shield tunnel. Structural monitoring data include displacement, strain, and joint opening; hydraulic monitoring data include water pressure, water content, and drainage flow; boundary condition data include rainfall, groundwater level, temperature, and tunnel operating load.

[0023] Displacement can be obtained using a total station, hydrostatic level, displacement gauge, or laser measuring device; strain can be obtained using a strain sensor or fiber optic monitoring device; joint opening can be obtained using a crack gauge or joint displacement gauge; water pressure can be obtained using a piezometer; moisture content can be obtained using a moisture content sensor placed near the joint or on the lining surface; and drainage flow rate can be obtained using a flow meter placed in the drainage branch.

[0024] The tunnel operating load is determined based on the tunnel's purpose. For rail transit tunnels, it can be characterized by train passing times, axle loads, operating speeds, and frequency of passage; for road tunnels, it can be characterized by traffic flow, vehicle type, and vehicle load. The same operating load characterization method is used for the same shield tunnel during the construction of operational baseline conditions, anomaly identification, defect inversion, and maintenance verification.

[0025] The data processing device includes a processor, a memory, and a communication interface. The processor executes the program in the memory to implement the physical unit construction module, the life cycle state construction module, the coupling propagation graph construction module, the anomaly sequence generation module, the disease inversion module, the maintenance simulation module, and the maintenance verification and correction module.

[0026] All monitoring devices use a unified time reference. Monitoring devices that can access the unified time signal directly use the unified time signal; monitoring devices that cannot directly access the unified time signal correct the sampling time based on the deviation between the clock of the acquisition host and the reference clock of the data processing device when data is received. The sampling interval, time synchronization error, measurement resolution, and measurement uncertainty of each monitoring device are all recorded in the monitoring device configuration record.

[0027] The physical unit construction module obtains the design layout of the shield tunnel, tunnel axis mileage, segment ring number, segment block number, ring joint location, longitudinal joint location, grouting zone behind the wall, and drainage pipeline layout relationship.

[0028] The spatial range of each segment is determined according to the tunnel axial mileage and the circumferential position within the segment ring. The circumferential joints on both sides of the spatial range, the longitudinal joints on both sides of the spatial range, the wall grouting zone covering the projected area outside the spatial range, and the drainage branch section receiving seepage and drainage within the spatial range are associated with the same tunnel physical unit.

[0029] Each tunnel physical unit is assigned a unique unit identifier. This unique unit identifier consists of at least the tunnel section identifier, the segment ring number, and the segment block number. Design data, construction data, completion and acceptance data, operation monitoring data, and historical maintenance data are all mapped to their corresponding tunnel physical units using this unique unit identifier.

[0030] When a backfill grouting zone covers more than two tunnel physical units, the overlapping area between the backfill grouting zone and the projected area of ​​each tunnel physical unit is recorded, and the ratio of the overlapping area to the area of ​​the backfill grouting zone is used as the coverage ratio. The backfill grouting detection results are mapped to the corresponding tunnel physical units according to the coverage ratio. When the detection device can directly identify the void area behind each tunnel segment, the direct detection results shall prevail.

[0031] When a drainage branch receives seepage and drainage from two or more tunnel physical units, a flow convergence relationship is established based on the branch connection direction, drainage slope, confluence location, and pipeline elevation. If separate branch inlet flow monitoring points are not set up, the total flow of the drainage branch is used as a common hydraulic observation for all connected tunnel physical units, and no unfounded average distribution of the total flow is performed.

[0032] When the effective monitoring range of a monitoring point covers more than two tunnel physical units, a one-to-many mapping is established between the monitoring point and the tunnel physical units. If the contribution ratio of each tunnel physical unit to the monitoring results cannot be distinguished, the monitoring result is used as a common observation of all associated tunnel physical units.

[0033] The above division ensures that each tunnel physical unit simultaneously corresponds to the actual segment structure, actual joint relationship, actual backfill grouting area, and actual drainage path, which is different from the data management method of dividing the tunnel into equidistant segments according to a fixed length.

[0034] The lifecycle state construction module writes the data of each stage into the corresponding tunnel physical unit according to the unique unit identifier, forming a state inheritance chain.

[0035] The state inheritance chain includes the design state, construction formation state, as-built benchmark state, operation evolution state, and maintenance intervention state.

[0036] The design parameters include the segment structure dimensions, concrete design parameters, joint design stiffness, waterproofing structure, backfill grouting design volume, grouting design pressure, design drainage capacity, and design allowable values ​​for structural and hydraulic responses.

[0037] The construction status includes joint opening, joint misalignment, connection fastening records, actual measurements of grouting behind the wall, grouting pressure, grouting duration, grouting interruption records, and construction anomaly records.

[0038] The completion benchmark status includes the results of segment location detection, joint condition detection, wall void detection, waterproof sealing detection, and drainage branch flow and pressure loss detection.

[0039] The operational evolution status is recorded according to the structural monitoring data, hydraulic monitoring data, and boundary condition data at the time of data collection.

[0040] The maintenance intervention status record includes the maintenance object, the propagation edge to be processed, the maintenance scope, the target physical parameters, the maintenance construction parameters, the actual execution value of the maintenance construction parameters, the post-maintenance test results, the post-maintenance response sequence, the effective propagation path blocking results, and the corrected propagation edge physical parameters.

[0041] The subsequent state is associated with the previous state by adding a new state record, without overwriting the data of the previous state. This allows for the differentiation of the design state, actual construction state, operational deterioration state, and post-maintenance state of the same tunnel physical unit.

[0042] The life cycle status construction module determines the initial value range of joint stiffness, wall-back constraint stiffness, wall-back void ratio, joint permeability coefficient, and hydraulic conduction capacity of drainage branches for each tunnel physical unit based on design data, construction data, and completion acceptance data.

[0043] Joint stiffness is used to characterize the ability of circumferential or longitudinal joints to resist relative opening, relative slippage, or relative rotation. Based on the structural response type corresponding to the current defect, normal stiffness, tangential stiffness, or rotational stiffness is selected as the joint stiffness used in the calculation. When the defect involves both joint opening and misalignment, the normal stiffness and tangential stiffness are calibrated separately.

[0044] Using the joint design stiffness in the design documents as the parameter benchmark, the corresponding boundary conditions are loaded into the digital twin structural model by selecting identifiable load periods caused by train passage, temperature changes, or groundwater pressure changes.

[0045] Within the design allowable range, adjust the joint stiffness sequentially, calculate the joint opening or relative displacement corresponding to each candidate joint stiffness, and determine the candidate value with the smallest overall difference between the simulated response and the measured response as the joint stiffness test value.

[0046] When the joint force and relative displacement can be directly obtained on site, the joint stiffness test value can be obtained based on the relationship between the force increment and the relative displacement increment, and the digital twin model can be used to verify whether it can reproduce the actual measured structural response.

[0047] The back-wall constraint stiffness is used to characterize the ability of the back-wall grouting body and the surrounding strata to constrain the radial displacement of the tunnel segments.

[0048] First, the candidate range of the back wall constraint stiffness is determined based on the actual measurement of the back wall grouting, the grouting pressure, and the back wall void detection results. Then, a monitoring period with clear operational loads is selected, and the radial displacement or strain of the tunnel segment is used as the verification quantity.

[0049] Adjust the back wall constraint stiffness within the candidate range, compare the radial displacement or strain of the tunnel segment calculated by the digital twin model with the measured results, and determine the parameter value with the smallest difference as the back wall constraint stiffness detection value.

[0050] The grouting volume and grouting pressure behind the wall are used to limit the candidate range of the back wall constraint stiffness, and are not directly converted into back wall constraint stiffness according to a fixed ratio.

[0051] The back wall voiding rate is determined based on the ratio between the area of ​​the back wall voiding detection area and the area of ​​the corresponding back wall grouting zone.

[0052] When the detection results clearly define the void boundary, the void area obtained from the detection is used to calculate the void rate behind the wall. When the detection results can only provide the possible boundaries of the void area, the lower limit of the void rate behind the wall is formed by determining the area of ​​the inner boundary of the void area, and the upper limit of the void rate behind the wall is formed by determining the area of ​​the outer boundary that includes the possible void area.

[0053] The joint permeability coefficient in this application is an equivalent physical parameter used in the digital twin hydraulic model to characterize the water flow capacity of the joint.

[0054] For joints where the water head on both sides and the flow rate through the joint can be obtained, the time period in which the joint flow rate and the water head on both sides are within a stable fluctuation range during the continuous sampling period is selected as the parameter determination time period. The pressure water head and the position water head on both sides of the joint are added to obtain the total water head on both sides, and the absolute value of the difference between the total water heads on both sides is determined as the joint water head difference.

[0055] The joint is divided into sub-segments along its length based on the measurement points of the opening between adjacent joints. For the first... For each joint segment, the residual effective opening is determined based on the average opening amount of the joints at both ends of the segment, the compression state of the sealing structure, and the extent of sealing failure. The product of the residual effective opening amount and the length of the segment is determined as the effective water-passing area of ​​the segment. The sum of the effective water-passing areas of all joint segments constitutes the effective water-passing area of ​​the joint. .

[0056] When the test results of the sealing structure indicate that the joint only forms a continuous water passage within a local area, only the joint sub-segment corresponding to the continuous water passage is included in the effective water passage area of ​​the joint; when the boundary of the continuous water passage cannot be accurately determined, the lower limit and upper limit of the effective water passage area are calculated by determining the water passage area and the area containing the possible water passage area, thereby forming the range of values ​​for the joint permeability coefficient.

[0057] Equivalent seepage length The equivalent seepage length is determined based on the joint waterproofing structure design section, as-built construction records, and on-site sealing condition testing results. Specifically, it is the length of the water flow path measured along a continuous water passage from the water-facing side of the joint to the water outlet on the inner side. When there are two or more continuous water passages, the path length corresponding to the continuous water passage that connects to the measured water outlet and has the lowest hydraulic resistance is selected as the equivalent seepage length. Joint permeability coefficient. Determined according to the following formula:

[0058] In the formula, The seam permeability coefficient; To determine the stable flow rate through the joint within a specified time period for the parameters; This is the equivalent seepage length; The effective water-passing area of ​​the joint; This represents the total head difference across the joint. For joints where no measurable flow rate is observed, the detection limit of the flow monitoring device is set to... As the upper limit of seam flow, and Substituting into the above formula, we obtain the upper limit of the joint permeability coefficient. When the total head difference on both sides of the joint is not greater than the head measurement uncertainty, the joint permeability coefficient is not converted based on the current test results. Instead, the joint is marked as a joint to be corrected, and its initial value range is taken from the design allowable range.

[0059] When the drainage branch is in a stable flow state, the hydraulic conduction capacity of the drainage branch is determined based on the flow rate of the drainage branch and the head difference between the inlet and outlet.

[0060] When the drainage flow rate and head difference have a nonlinear relationship, the local hydraulic conduction capacity is characterized by the ratio between the change in flow rate and the change in head difference near the current operating state.

[0061] The measurement uncertainty of the monitoring device comes from the device verification or calibration records.

[0062] For the physical parameters obtained through model inversion, the measurement uncertainty of each detection quantity involved in the calibration is increased and decreased sequentially, and the physical parameters are recalibrated while keeping other detection quantities unchanged. The maximum absolute difference between each calibration result and the original parameter detection value is taken as the converted measurement uncertainty of that physical parameter.

[0063] The larger of the parameter detection value minus the converted measurement uncertainty and the design allowable lower limit is determined as the lower limit of the initial value range; the smaller of the parameter detection value plus the converted measurement uncertainty and the design allowable upper limit is determined as the upper limit of the initial value range.

[0064] When on-site testing data is lacking, the design allowable range should be used as the initial value range, and the physical parameter should be marked as a parameter to be calibrated. The parameter to be calibrated can participate in the selection of candidate starting units, but it should be verified through supplementary testing or pre-maintenance disturbance verification before the maintenance plan is determined.

[0065] The digital twin model includes a structural calculation layer and a hydraulic calculation layer.

[0066] The structural calculation layer establishes the structural calculation relationships between segments, circumferential joints, longitudinal joints, backfill grouting body, and surrounding strata based on the shield tunnel geometric model. Segments are constructed using structural elements capable of characterizing axial, radial, and bending responses; joints are constructed using connection elements capable of characterizing normal opening, tangential sliding, and rotation; and backfill constraints are applied to the outside of the segments based on the backfill constraint stiffness.

[0067] The hydraulic calculation layer establishes water flow migration relationships based on joints, grouting zones behind the wall, void areas behind the wall, and drainage branches. The joint permeability coefficient controls the water flow capacity of the joints, the void rate behind the wall and the permeability of the grouting zones behind the wall control the water migration behind the wall, and the hydraulic conduction capacity of the drainage branches controls the water discharge capacity.

[0068] The structural calculation layer outputs the joint opening amount, and the hydraulic calculation layer updates the joint water passage capacity based on the joint opening amount; the hydraulic calculation layer outputs the water pressure at the joint location, and the structural calculation layer updates the joint normal force and segment stress state based on the water pressure.

[0069] Each computational unit and physical parameter in the digital twin model is associated with a unique unit identifier of the tunnel physical unit.

[0070] The abnormal sequence generation module extracts historical periods from the state inheritance chain where there is no maintenance work, the monitoring device is operating normally, and the structural monitoring and hydraulic monitoring values ​​do not exceed the operational monitoring control values.

[0071] Historical data are grouped according to the numerical intervals of rainfall, groundwater level, temperature, and tunnel operating load. The width of the numerical interval is not less than the measurement resolution of the corresponding monitoring device, and the difference between the maximum and minimum values ​​of the corresponding boundary condition quantity within the same interval does not exceed the upper limit of the fluctuation of the boundary condition quantity during the stable operating period.

[0072] For each monitoring quantity within the same boundary condition group, the median of the effective samples is used as the operational baseline value.

[0073] If the monitoring device is disconnected, the monitoring value exceeds the device's range, or the monitoring value undergoes a sudden change relative to consecutive sampling points that cannot be explained by changes in boundary conditions, the sample will be excluded as a monitoring fault sample.

[0074] The larger of the maximum absolute deviation of the valid samples within a group relative to the operating baseline and the measurement uncertainty of the monitoring device is determined as the upper limit of the response noise for the corresponding monitored quantity.

[0075] The coupling propagation graph construction module sequentially applies a perturbation to one physical parameter of a selected tunnel physical unit. Only one physical parameter is changed each time, while other physical parameters and boundary conditions remain unchanged.

[0076] The disturbance is initially determined by the conversion measurement uncertainty of the physical parameter. As long as the responses of other tunnel physical units do not exceed the corresponding response noise limit, the disturbance is gradually increased until a identifiable response is generated or the boundary of the initial value range is reached.

[0077] To eliminate the dimensional differences between different physical parameters and different monitoring quantities, the normalized propagation coefficient is used to represent the structural propagation, hydraulic propagation, and the bidirectional conversion relationship between structural response and hydraulic response.

[0078] From the upstream tunnel physical unit Pointing to downstream tunnel physical unit Normalized propagation coefficient Determined according to the following formula: ; In the formula: Indicates the physical unit of the upstream tunnel Pointing to downstream tunnel physical unit The normalized propagation coefficient; This indicates the change in upstream input physical parameters or upstream input response; This represents the upstream input value before the disturbance; This represents the minimum identifiable change in the upstream input physical parameters, or the upper limit of the response noise of the upstream input response; Indicates the change in downstream response; This represents the downstream response value before the disturbance; This indicates the upper limit of the response noise in the downstream response.

[0079] When upstream input When the response is to water head, water pressure, joint opening, or other inputs that can be directly monitored, This is the upper limit of the response noise for the input response. Specifically, valid samples are extracted from the operating reference state, which belongs to the same numerical range as the current boundary condition. The median of the valid samples is used as the input response reference value. The larger of the maximum absolute deviation of each valid sample relative to the input response reference value and the corresponding measurement uncertainty of the monitoring device is determined as... .

[0080] When upstream input When considering joint stiffness, wall-backed constraint stiffness, wall-backed void ratio, joint permeability coefficient, or hydraulic conduction capacity of drainage branches, This refers to the minimum identifiable change in the corresponding physical parameter. The minimum identifiable change is determined as follows: using the converted measurement uncertainty of the physical parameter as the initial perturbation, the perturbation is gradually increased within its initial value range, and digital twin calculations are performed while keeping other physical parameters and boundary conditions unchanged after each perturbation; the minimum perturbation that causes the change in at least one downstream response to exceed the upper limit of the downstream response noise, and for which the digital twin model satisfies the convergence condition, is determined as the minimum identifiable change in the physical parameter. If changes in the physical parameter within its initial value range still cannot produce a response exceeding the upper limit of the downstream response noise, a corresponding propagation edge is not established based on that physical parameter.

[0081] When the upstream input is physical parameters At that time, the propagation coefficient is: ; When the upstream input is a monitorable response At that time, the propagation coefficient is: ; In the formula, The propagation coefficient corresponding to the physical parameter disturbance; The input response perturbation corresponds to the propagation coefficient; This refers to the disturbance of physical parameters; These are the physical parameter values ​​before the disturbance.

[0082] When establishing the structure propagation edge, in the formula This refers to joint stiffness, back-wall constraint stiffness, or back-wall void ratio. This refers to displacement, strain, joint opening, or internal force.

[0083] When establishing a hydraulic propagation edge, This refers to the upstream water head, joint permeability coefficient, or hydraulic conduction capacity of the drainage branch. This refers to the downstream water pressure or drainage flow rate.

[0084] When establishing the structural-to-hydraulic transition edge, For the seam opening amount, This refers to the joint permeability coefficient, water pressure, or drainage flow rate.

[0085] When establishing the hydraulic-to-structural transition edge, For water pressure, This refers to the normal force on the joint, the joint opening, displacement, or strain.

[0086] For a single upstream input disturbance, first calculate the relative change in the upstream input based on the actual change in the upstream input. The normalized noise of the downstream response is calculated based on the upper limit of the downstream response noise. The normalized noise value of the downstream response is determined according to the following formula:

[0087] ; Activation threshold of propagation edge Determined according to the following formula: ; In the formula, For the upstream tunnel physical unit Pointing to downstream tunnel physical unit The propagation edge activation threshold; This represents the normalized noise value of the downstream response; This represents the relative change in upstream input.

[0088] When the upstream input is a physical parameter This is the ratio of the physical parameter disturbance to the original physical parameter value; when the upstream input is a monitorable response. It is the ratio of the change in input response to the normalized baseline value of the input response.

[0089] When the change in upstream input does not exceed its minimum identifiable change or the upper limit of response noise, this disturbance will not be used for propagation edge activation determination. After the upstream input reaches the identifiable condition, if the propagation coefficient is not lower than the determined activation threshold and the direction of response change conforms to the structural connection relationship, head direction, or bidirectional conversion relationship, the corresponding propagation edge will be marked as active; otherwise, the corresponding propagation edge will be marked as inactive.

[0090] The activation thresholds for structural propagation edges, hydraulic propagation edges, structural-to-hydraulic conversion edges, and hydraulic-to-structural conversion edges are all determined using the above method, requiring only the substitution of the corresponding upstream input and downstream response quantities.

[0091] When the propagation coefficient is not lower than the corresponding activation threshold and the direction of response change conforms to the structural connection relationship or the direction of water flow, the corresponding propagation edge is established and marked as active; when the propagation coefficient is lower than the corresponding activation threshold, the corresponding propagation edge is marked as inactive.

[0092] The direction of the propagation edge of the structure should also conform to the structural equilibrium calculation results of the digital twin structural model. The direction of the hydraulic propagation edge is determined based on the total head difference, which includes pressure head and position head.

[0093] When the groundwater level, joint opening, or hydraulic conduction capacity of the drainage branch changes, the propagation coefficient and activation state of the relevant propagation edge are recalculated.

[0094] This results in a dynamic hydraulic-to-structure coupling propagation diagram that includes structural propagation edges, hydraulic propagation edges, structural-to-hydraulic conversion edges, and hydraulic-to-structural conversion edges.

[0095] The propagation edges in the dynamic hydraulic-to-structure coupling propagation diagram are jointly determined by the actual structural connections, hydraulic connectivity, and physical parameter disturbance results, and are not semantic relationships between diseases and causes directly established based on disease names, expert experience, or historical cases.

[0096] The abnormal sequence generation module selects the corresponding operational baseline state based on the current rainfall, groundwater level, temperature, and the numerical range of tunnel operating load.

[0097] The threshold for structural anomaly identification is the upper limit of the response noise of the corresponding structural monitoring quantity, and the threshold for hydraulic anomaly identification is the upper limit of the response noise of the corresponding hydraulic monitoring quantity.

[0098] When the deviation of displacement, strain, or joint opening from the operating baseline value exceeds the structural anomaly identification threshold, and the duration of continuous exceedance of the threshold is greater than the longest continuous duration of occasional overruns of structural monitoring data under historical operating baseline conditions, the corresponding tunnel physical unit is determined to have a structural response anomaly.

[0099] When the deviation of water pressure, moisture content, or drainage flow rate from the operating baseline value exceeds the hydraulic anomaly identification threshold, and the duration of continuous exceedance of the threshold is greater than the longest continuous duration of occasional exceedance of hydraulic monitoring data under historical operating baseline conditions, the corresponding tunnel physical unit is determined to have a hydraulic response anomaly.

[0100] The longest continuous duration of occasional exceedances is determined based on the sample segment in which all short-term exceedances of the normal fluctuation range under the operating baseline state do not exceed the operating monitoring control value and subsequently recover on their own.

[0101] If there are no occasional out-of-limit sample segments in the historical data, the duration of the occasional out-of-limit is taken as one complete sampling cycle of the corresponding monitoring device.

[0102] When consecutive sampling points meet the anomaly identification conditions, the first sampling time in the consecutive sampling segment that exceeds the anomaly identification threshold is determined as the anomaly start time.

[0103] Each abnormal event record includes the abnormal unit, abnormal type, abnormal start time, response direction, response magnitude, and duration. All abnormal events are arranged according to their abnormal start time to form an abnormal event sequence.

[0104] The disease inversion module starts from the anomalous element and traces backward along the active propagation edges in the dynamic hydraulic-to-structural coupling propagation graph. Nodes that can reach at least one anomalous element through the active propagation edges are identified as candidate starting elements.

[0105] For each candidate starting unit, at least one physical parameter is changed within its initial value range, and the rainfall, groundwater level, temperature and tunnel operation load during the period of the abnormal event are loaded into the digital twin model.

[0106] Candidate disease parameters are selected based on the life cycle status.

[0107] If construction records or completion inspections show insufficient grouting behind the wall, interrupted grouting, or voids behind the wall, priority should be given to changing the void rate and the constraint stiffness behind the wall.

[0108] When the misalignment or opening of the joint is too large, the joint stiffness should be changed first.

[0109] If the waterproof seal test is abnormal or there is a history of leakage at the joint, the joint permeability coefficient should be changed first.

[0110] When there is historical siltation or flow attenuation in the drainage branch, the hydraulic conduction capacity of the drainage branch should be changed first.

[0111] The digital twin model outputs the simulated anomaly range, simulated anomaly start time, simulated propagation delay, and simulated response direction.

[0112] The simulated propagation delay of a single propagation edge is determined by the time difference between the moment when the downstream node's response first continuously exceeds the upper limit of the response noise after applying parameter disturbance or hydraulic disturbance to the upstream node and the moment when the upstream disturbance begins.

[0113] When the downstream response exceeds the upper limit of the response noise at the first sampling time, the propagation delay is recorded as a delay interval not greater than the sum of a sampling interval and the time synchronization error, and is not directly recorded as zero.

[0114] The cumulative propagation delay of a propagation path is the sum of the propagation delays of each propagation edge within that path. The allowable deviation of the propagation delay is the sum of the maximum sampling interval and the maximum time synchronization error of the relevant monitoring devices.

[0115] For each candidate starting unit, the range difference between the simulated anomaly range and the measured anomaly range, the time difference between the simulated anomaly start time and the measured anomaly start time, the time delay difference between the simulated propagation delay and the measured propagation delay, and the directional difference between the simulated response direction and the measured response direction are determined respectively.

[0116] The simulated anomaly range, anomaly start time, propagation delay, and response direction characterize the spatial coverage, temporal start point, propagation speed, and response trend of disease propagation, respectively. Since all four differences have been normalized, and this embodiment does not pre-increase the priority of any difference item, the four normalized differences are treated with equal weights, with each item having a weight of one-quarter.

[0117] Candidate start unit Comprehensive differences Determined according to the following formula: In the formula, Candidate starting unit The overall differences; To normalize the difference in the range of outliers; To normalize the difference at the start time of anomalies; To normalize the differences in propagation delay; The difference in the direction of the normalized response.

[0118] When a certain difference is zero in all candidate starting units, the normalized difference for that difference is set to zero, and its weight is not redistributed to other difference items. This ensures that the same comprehensive difference judgment scale is used in different disease inversion processes.

[0119] The difference in the anomaly range is determined based on the number of non-overlapping tunnel physical units in the simulated anomaly unit set and the measured anomaly unit set.

[0120] The difference in the anomaly start time is determined based on the difference in the anomaly start time of tunnel physical units with the same unique unit identifier in the simulated anomaly unit set and the measured anomaly unit set. Measured anomaly units that did not appear in the simulation results have already been included in the anomaly range difference and will not be included again in the anomaly start time difference.

[0121] The propagation delay difference is determined based on the time interval between the cumulative propagation delay of the simulated propagation path and the corresponding measured abnormal event.

[0122] The difference in response direction is determined based on the number of anomalous units whose simulated response direction is inconsistent with the measured response direction.

[0123] The disease inversion module also performs structural balance and water conservation checks on the simulation results.

[0124] The disease inversion module performs structural balance checks and water conservation checks on the simulation results corresponding to each candidate starting unit.

[0125] The nodal internal forces used in structural equilibrium checks are obtained by summing the unit internal forces and constraint reactions of the segment structural units, joint connection units, and wall-back constraint units from the digital twin structural model. External forces acting on the nodes consist of design dead load, measured or converted operating loads based on train operation data, water loads converted from measured water pressure, and other recorded external loads. The conversion error of the operating load is determined based on the measurement uncertainty of the load monitoring device or the verification error of the operating load conversion model.

[0126] Normalized structural equilibrium residuals of candidate propagation regions Determined according to the following formula: ; In the formula, For normalized structural equilibrium residuals; It is a vector composed of the internal forces and constraint reactions of each computation node in the candidate propagation region; The external load vector acting on the same computation node; As the load noise baseline, the larger of the external load conversion uncertainty and the absolute force residual limit of the structural solver is taken. The structural equilibrium allowable residual is the larger of the normalized convergence residual recorded by the structural solver in the model calculation configuration and the normalized uncertainty of the external load. If the residual does not exceed the allowable residual for structural equilibrium, the simulation result corresponding to the candidate starting unit is determined to satisfy the structural equilibrium condition.

[0127] The water volume entering the candidate propagation area used for water conservation checks is determined by the boundary inflow rate output by the digital twin hydraulic model under measured groundwater level and boundary head conditions; the water volume flowing out through the drainage branch is obtained by the drainage branch flow meter; the water volume flowing out through the visible leakage location is obtained by the leakage collection device, water measuring device, or the leakage flow detection method specified on site; the change in water storage within the candidate propagation area is transiently calculated by the hydraulic model based on the void area behind the wall, the joint area, and its water storage capacity, and verified using water pressure or moisture content monitoring results.

[0128] Normalized water conservation residual of candidate propagation region Determined according to the following formula: ; In the formula, The normalized water quantity conservation residual; The water flow rate entering the candidate propagation area; This represents the total flow rate through drainage branches and leakage points; Time interval Changes in water storage in the candidate propagation region; As the flow noise benchmark, the larger value among the combined measurement uncertainty of each flow monitoring device involved in the water volume calculation, the detection limit of the flow monitoring device, and the absolute flow residual limit of the hydraulic solver is taken.

[0129] Water conservation allows for the larger of the normalized convergence residual recorded by the hydraulic solver in the calculation configuration and the normalized measurement uncertainty of the flow rate involved in the water quantity calculation. If the residual does not exceed the allowable residual for water conservation, the simulation result corresponding to the candidate starting unit is determined to satisfy the water conservation condition.

[0130] Candidate starting units that do not meet the structural equilibrium condition or water conservation condition are excluded, and the candidate starting unit with the smallest comprehensive difference is determined as the disease starting unit.

[0131] When two or more candidate starting units have the same overall difference, the one with fewer active state propagation edges to the measured abnormal unit is selected first; if the number of propagation edges is still the same, the one with the smaller sum of relative adjustments to the candidate physical parameters is selected.

[0132] An effective propagation path is formed when the propagation from the disease initiation unit to the abnormal unit is in an active state, the response direction is consistent with the abnormal event sequence, and the difference between the cumulative propagation delay and the measured propagation delay does not exceed the allowable deviation of the propagation delay.

[0133] When no single candidate starting unit can satisfy the requirements of structural balance, water conservation and propagation timing, the abnormal event sequence is divided into at least two abnormal event subsequences based on whether the abnormal units have a common activation propagation edge and whether the abnormal start time is continuous, and the disease starting unit inversion is performed separately.

[0134] The maintenance simulation module converts candidate maintenance measures into changes in physical parameters in a digital twin model.

[0135] The back wall grouting measures have been transformed into measures that reduce the back wall void rate, increase the back wall constraint stiffness, and reduce the permeability of the back wall grouting zones.

[0136] Joint sealing measures translate into a reduction in the joint permeability coefficient. When using sealing or reinforcing structures with structural restraint, this also translates into an increase in joint stiffness.

[0137] The measures to clear drainage branches have been transformed into measures to improve the hydraulic conduction capacity of drainage branches.

[0138] The measures to strengthen the joint structure are transformed into measures to increase the joint stiffness.

[0139] The correspondence between maintenance construction parameters and changes in physical parameters comes from maintenance material testing reports, maintenance process tests, results of field test sections, or maintenance records that have been verified in similar structures.

[0140] For the same maintenance process, establish a correspondence between the maintenance construction parameter samples and the physical parameter change samples before and after maintenance. When the target physical parameter is between two verified samples, piecewise interpolation is used to determine the maintenance construction parameters; when the target physical parameter exceeds the range of verified samples, extrapolation is not performed, and a small-scale process test is conducted first.

[0141] Candidate maintenance measures are only deployed on active propagation edges in the effective propagation path, or on tunnel physical units directly associated with such propagation edges.

[0142] Generate candidate maintenance measure combinations that can cause parameter changes in at least one propagation edge in each effective propagation path, without adding maintenance locations unrelated to the effective propagation path to the candidate combinations.

[0143] The physical parameter changes corresponding to the candidate maintenance measure combinations are loaded into the digital twin model, and the propagation coefficients of each propagation edge are recalculated.

[0144] Candidate maintenance measure combinations that do not reduce the propagation coefficient of at least one propagation edge in each effective propagation path to below the corresponding activation threshold are excluded.

[0145] For the remaining candidate maintenance measure combinations, examine the stress, joint opening, and water pressure of adjacent tunnel physical units. Use the smaller of the design allowable value and the operational monitoring control value as the corresponding limit. Candidate maintenance measure combinations that cause the response of any adjacent tunnel physical unit to exceed the limit value are excluded.

[0146] For the remaining candidate combinations of maintenance measures, they are first arranged in order of increasing number of physical units of the tunnel to be maintained; if the number is the same, they are arranged in order of increasing number of propagation edges to be treated; if the first two are the same, they are arranged in order of increasing sum of adjustment ratios of each physical parameter relative to the parameters before maintenance.

[0147] The top-ranked candidate maintenance measure combination was determined as the maintenance intervention combination.

[0148] The maintenance simulation module generates maintenance tasks based on maintenance intervention combinations. Each maintenance task includes the physical unit of the tunnel to be maintained, the propagation edge to be addressed, the maintenance scope, target physical parameters, maintenance construction parameters, and post-maintenance verification location.

[0149] After determining the maintenance intervention combination, the target physical parameters corresponding to the maintenance intervention combination are loaded into the digital twin model, and the time history is calculated using the same rainfall, groundwater level, temperature and tunnel operation load as the time period during which the abnormal event occurred.

[0150] For each tunnel physical unit in the effective propagation path, the moment when the simulated response first continuously returns to the range formed by the operation baseline plus or minus the upper limit of the response noise is determined as the predicted recovery start time.

[0151] The predicted recovery direction is determined based on the direction of change of the pre-maintenance response value toward the operational baseline value. The predicted recovery magnitude is the difference between the pre-maintenance response value and the post-maintenance stable response value.

[0152] Based on the propagation direction of the effective propagation path, the predicted recovery start time, predicted recovery direction, and predicted recovery magnitude of each tunnel physical unit are arranged to form a predicted recovery sequence.

[0153] When performing backfill grouting, record the location of the grouting holes, the actual grouting pressure, the actual grouting volume, the grouting duration, and the grout return status. After maintenance, determine the actual backfill void rate and backfill constraint stiffness by detecting backfill voids and segment radial response.

[0154] When sealing joints, record the actual sealing location, sealing length, material batch, and curing status; after repair, determine the joint permeability coefficient through water tightness testing or seepage testing under the same working conditions.

[0155] When dredging drainage branches, record the drainage flow and pressure loss before and after dredging to determine the hydraulic transmission capacity of the drainage branches.

[0156] When reinforcing the joint structure, the joint stiffness is determined based on the relative displacement response of the joint before and after reinforcement.

[0157] The actual execution values ​​of maintenance construction parameters and the post-maintenance test results are written into the maintenance intervention status and used to determine the correction range of physical parameters after maintenance.

[0158] After the backfilling or joint sealing is completed, the moment when the repair materials reach the detectable state specified in the construction technical documents and the continuous structural monitoring and hydraulic monitoring quantities enter a stable fluctuation range shall be taken as the starting point for post-repair verification.

[0159] After the drainage branch is cleared, the moment when the drainage flow and pressure loss enter a stable fluctuation range is taken as the starting point for post-maintenance verification.

[0160] Do not conclude that the effective transmission path has been blocked until the post-repair monitoring response has stabilized.

[0161] When using boundary condition pairing verification, a reference period is selected from historical monitoring data prior to maintenance.

[0162] The corresponding differences in rainfall, groundwater level, temperature, and tunnel operating load between the reference period and the post-maintenance verification period shall not exceed the allowable deviations of their respective boundary conditions.

[0163] The permissible deviation of the boundary condition is determined based on the larger of the upper limit of the fluctuation of the corresponding boundary condition quantity during the stable operating period and the measurement uncertainty of the monitoring device.

[0164] Historical periods involving maintenance work, temporary adjustments to the drainage system, monitoring device malfunctions, or external engineering disturbances are not considered as reference periods.

[0165] When the response sequence after maintenance no longer satisfies the propagation timing relationship between adjacent tunnel physical units in the original effective propagation path, and the response of the downstream tunnel physical unit of the propagation edge to be processed does not exceed the corresponding anomaly identification threshold, it is determined that the effective propagation path has been blocked.

[0166] If there is no reference period that meets all the allowable deviations of the boundary conditions before maintenance, the mismatched historical data will not be used to make a judgment, but hydraulic disturbance verification will be used instead.

[0167] The target drainage branch is directly connected to the hydraulic propagation edge to be treated, and its adjustment can form a drainage branch with measurable head changes in the disease initiation unit.

[0168] Before the maintenance task is carried out, hydraulic disturbances are generated in the drainage branch of the control target, and the changes in valve opening, holding time, and disturbance response sequence before maintenance are recorded.

[0169] The valve opening change and holding time are determined based on the rated capacity of the drainage branch, the design allowable water pressure at the monitoring point, and the on-site drainage operation procedures, so that the water pressure at the monitoring point does not exceed the design allowable value and the drainage flow does not exceed the rated capacity of the drainage branch during the disturbance.

[0170] After the maintenance task is carried out, the target drainage branch is controlled again using the same valve opening change and holding time to obtain the post-maintenance disturbance response sequence.

[0171] The response of the downstream tunnel physical unit on the propagation side to be treated is evaluated only when the disease initiation unit generates a water pressure or drainage flow response exceeding the upper limit of the response noise.

[0172] If the initial unit of the hazard does not produce an identifiable response, the current disturbance verification is invalid. Re-verify by adjusting the valve opening change or the duration of the adjustment, provided that the design allowable water pressure and the rated capacity of the drainage branch are not exceeded.

[0173] When the disease initiation unit generates a response exceeding the upper limit of the response noise, while the downstream tunnel physical unit of the propagation edge to be treated does not generate a response exceeding the corresponding upper limit of the response noise, the propagation edge is determined to be inactive.

[0174] When each pathway from the disease initiation unit to the abnormal unit contains at least one verified inactive propagation edge, the effective propagation path is determined to be blocked.

[0175] The maintenance verification and correction module compares the recovery start time, recovery direction and recovery magnitude of each tunnel physical unit in the predicted recovery sequence and the post-maintenance response sequence in the direction of propagation from the initial defect unit to the abnormal unit.

[0176] The response identification threshold is the upper limit of the response noise of the corresponding monitored quantity during the stable period after maintenance.

[0177] When the difference between the predicted response and the measured response of the tunnel physical unit located upstream in the propagation direction does not exceed the response identification threshold, while the difference between the predicted response and the measured response of the tunnel physical unit located downstream in the propagation direction exceeds the response identification threshold, the propagation edge connecting the two tunnel physical units is determined as the propagation edge where the difference first appears.

[0178] When a propagation edge is associated with more than two physical parameters, the physical parameters directly affected by the maintenance measures are corrected first. When the maintenance measures do not directly affect the propagation edge, the parameters to be corrected are determined according to their sensitivity from high to low based on the changes in the propagation coefficient caused by disturbances in each physical parameter during the construction of the propagation edge.

[0179] The calibration range is determined based on the actual executed values ​​of the target physical parameters and maintenance construction parameters in the maintenance task, the post-maintenance test values, and their converted measurement uncertainties, and the initial range before maintenance is not used.

[0180] Using the post-repair test value as the initial value for correction and the converted measurement uncertainty as the initial adjustment step size, the predicted recovery sequence is recalculated along the direction of parameter increase and the direction of parameter decrease, respectively. The direction that reduces the difference between the predicted response and the measured response is selected for further adjustment.

[0181] If the difference no longer decreases after one adjustment, the adjustment step size is halved. Correction is stopped when the difference between the predicted response and the measured response does not exceed the response identification threshold.

[0182] When the adjustment step size is already less than the minimum identifiable change of the parameter, or when the parameter reaches the boundary of the correction value range, and the difference between the predicted response and the measured response still exceeds the response identification threshold, the propagation edge is marked as a propagation edge to be supplemented for detection, and a joint detection, wall void detection, or drainage branch detection task is generated.

[0183] The correction range for propagation delay is determined based on the sampling interval, time synchronization error, and the simulated propagation delay before the propagation edge is repaired. The corrected propagation delay must not be less than zero.

[0184] During local calibration, the physical parameters of other propagation edges are kept unchanged to avoid model parameter drift unrelated to the repair area caused by local repair.

[0185] When a certain monitoring quantity is missing, other types of monitoring quantities are not directly used as substitutes.

[0186] When adjacent monitoring points of the same type exist, and these adjacent monitoring points have a stable response relationship during historical stable periods, a value to be verified can be formed based on the historical response relationship. The value to be verified is only used for screening candidate starting units and is not used as the final basis for blocking effective propagation paths.

[0187] When the missing monitoring data is located at the starting unit of the disease, on both sides of the spread edge to be treated, or at the verification position after maintenance, temporary monitoring points should be set up before maintenance verification, or supplementary detection methods that can obtain the same physical response should be adopted.

[0188] Before obtaining valid verification data, we should not conclude that the effective transmission path has been blocked.

[0189] During operation, the abnormal sequence generation module detected that the water pressure of an upstream tunnel physical unit was continuously higher than the operating benchmark value under the same boundary conditions. Subsequently, it detected that the joint opening of the adjacent ring joint exceeded the structural anomaly identification threshold. Then, it detected that the water content and drainage flow of the downstream tunnel physical unit exceeded the hydraulic anomaly identification threshold.

[0190] The system generates upstream water pressure anomaly events, joint opening anomaly events, and downstream hydraulic response anomaly events sequentially based on the sampling time when each monitored quantity first exceeds the identification threshold.

[0191] The disease inversion module traces backward from the downstream abnormal unit along the active hydraulic propagation edge, structural propagation edge, and structural-to-hydraulic conversion edge to obtain two candidate starting units.

[0192] The construction formation status of the first candidate starting unit shows that the actual grouting behind the wall is lower than the design amount, and the as-built benchmark status shows that there is void behind the wall; the second candidate starting unit only has historical surface sealing records.

[0193] For the first candidate starting unit, adjust the parameters within the initial range of the back wall void ratio and the back wall constraint stiffness; for the second candidate starting unit, adjust the parameters within the initial range of the joint permeability coefficient.

[0194] After loading the groundwater level, temperature and operational load during the period of the abnormal event, the simulation results of the first candidate starting unit can sequentially produce an increase in upstream water pressure, an increase in the opening of adjacent joints and an increase in downstream water content, and satisfy the structural equilibrium and water conservation conditions.

[0195] Although the second candidate starting unit can simulate downstream leakage, it cannot reproduce the measured upstream water pressure rise in advance, and the water conservation residual exceeds the allowable range.

[0196] Therefore, the first candidate starting unit was determined as the disease initiation unit, and the propagation path consisting of the back wall grouting zone, adjacent joints and downstream leakage units was determined as the effective propagation path.

[0197] The maintenance simulation module simulates maintenance for individual sealing of downstream joints, grouting behind the wall of the initial unit of the defect, dredging of drainage branches, and combinations of the above measures.

[0198] While sealing the downstream joint separately can reduce the surface water discharge, it causes the upstream water pressure to exceed the smaller of the design allowable value and the operation monitoring control value, so it is excluded.

[0199] Dredging only the drainage branches can reduce the upstream water pressure, but the structural propagation edges remain active and cannot block all effective propagation paths.

[0200] After grouting behind the wall of the initial unit of the disease and dredging of the associated drainage branch, the void rate behind the wall is reduced, the constraint stiffness behind the wall is increased, the hydraulic conduction capacity of the drainage branch is improved, and at least one propagation edge in each effective propagation path has a propagation coefficient lower than the corresponding activation threshold, and the stress, joint opening and water pressure of the adjacent tunnel physical unit do not exceed the limit value.

[0201] Based on this, the system determines the maintenance intervention combination, generates maintenance tasks and predicted recovery sequences.

[0202] After the repair is completed, the physical parameters and correction ranges are determined based on the actual grouting volume, grouting pressure, retest results of voids behind the wall, and flow test results of drainage branches.

[0203] Hydraulic disturbance was implemented using the same valve opening change and holding duration as before maintenance. The disease initiation unit generated a water pressure response exceeding the upper limit of the response noise, while the downstream tunnel physical unit of the propagation side to be treated did not generate a change in water content or drainage flow exceeding the upper limit of the response noise, thus determining that the effective propagation path was blocked.

[0204] If the actual recovery start time of the downstream tunnel physical unit is later than the predicted recovery start time, and the difference between the measured response and the predicted response of the disease initiation unit and the intermediate tunnel physical unit does not exceed the response identification threshold, then the propagation edge between the intermediate tunnel physical unit and the downstream tunnel physical unit is determined as the propagation edge where the difference first appears, and only the joint permeability coefficient or propagation delay of the propagation edge is corrected.

[0205] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A digital twin-based intelligent operation and maintenance management system for the entire lifecycle of tunnels, applied to shield tunnels, characterized in that: include: The physical unit construction module is used to divide the shield tunnel into at least two tunnel physical units and map design data, construction data, completion acceptance data and operation monitoring data to each tunnel physical unit. The life cycle status construction module is used to determine the initial value range of joint stiffness, wall-back constraint stiffness, wall-back void rate, joint permeability coefficient, and hydraulic conduction capacity of drainage branches for each tunnel physical unit based on design data, construction data, and completion acceptance data. The coupled propagation graph construction module is used to apply perturbations to the physical parameters corresponding to the initial value range in the digital twin model, determine the structural propagation relationship based on the structural response of other tunnel physical units, determine the hydraulic propagation relationship based on the head difference and hydraulic response of adjacent tunnel physical units, and construct a dynamic hydraulic-to-structure coupled propagation graph with tunnel physical units as nodes and the structural propagation relationship, hydraulic propagation relationship and bidirectional conversion relationship between structural response and hydraulic response as propagation edges. The anomaly sequence generation module is used to compare the structural and hydraulic monitoring data of each tunnel physical unit with the operational baseline state, determine the abnormal unit and the start time of the anomaly, and generate an anomaly event sequence. The disease inversion module is used to trace candidate starting units backward from the abnormal unit along the dynamic hydraulic-to-structure coupling propagation diagram. Within the initial value range of each candidate starting unit, hydraulic-to-structure coupling propagation simulation is performed. Based on the differences between the simulation results and the abnormal event sequence in terms of abnormal range, occurrence order, propagation delay and response direction, the disease starting unit with the smallest comprehensive difference and the effective propagation path are determined from the candidate starting units that meet the structural equilibrium condition and water conservation condition. The maintenance simulation module is used to convert candidate maintenance measures into changes in physical parameters and perform maintenance simulations. It determines the maintenance intervention combination that ensures that the propagation coefficient of at least one propagation edge in each effective propagation path is lower than the corresponding activation threshold and that the response of adjacent tunnel physical units does not exceed the limit value, and generates maintenance tasks. The maintenance verification and correction module is used to obtain the post-maintenance response sequence after the maintenance task is carried out. When the post-maintenance response sequence no longer satisfies the propagation timing relationship of the effective propagation path, it is determined that the effective propagation path is blocked, and the physical parameters of the propagation edge where the predicted recovery sequence and the post-maintenance response sequence first show a difference are corrected.

2. The intelligent operation and maintenance management system for the entire lifecycle of a tunnel based on digital twins as described in claim 1, characterized in that, The physical unit construction module determines the spatial range of the segment block according to the tunnel axial mileage and the circumferential position inside the segment ring. It associates the circumferential joint located at the axial boundary of the spatial range, the longitudinal joint located at the circumferential boundary of the spatial range, the wall grouting zone covering the outer projection area of ​​the spatial range, and the drainage branch section receiving drainage within the spatial range to the same tunnel physical unit, and configures a unique unit identifier for each tunnel physical unit. The lifecycle state construction module writes design data, construction data, completion acceptance data, operation monitoring data, and historical maintenance data into the design state, construction formation state, completion benchmark state, operation evolution state, and maintenance intervention state respectively, according to the unique unit identifier. It also establishes a state inheritance chain according to the formation time of each state, and the subsequent state is associated with the previous state by adding a new state record without overwriting the data of the previous state.

3. The intelligent operation and maintenance management system for the entire lifecycle of a tunnel based on digital twins as described in claim 2, characterized in that, The life cycle status construction module uses the joint design stiffness, wall back grouting design parameters, waterproofing design parameters, and drainage design capacity in the design data as physical parameter benchmarks. Based on the joint opening, joint misalignment, and joint sealing test results, the parameter test values ​​of joint stiffness and joint permeability coefficient are determined using the calibration relationship between the test values ​​and physical parameters in the digital twin model. The parameter test values ​​for the back wall constraint stiffness and back wall void rate are determined based on the actual measurement of back wall grouting, grouting pressure, and back wall void detection results; the parameter test values ​​for the hydraulic conduction capacity of the drainage branch are determined based on the flow rate and pressure loss detection results of the drainage branch. For each physical parameter, the larger of the parameter detection value minus the converted measurement uncertainty and the design allowable lower limit is used as the lower limit of the initial value range. The smaller of the parameter detection value plus the converted measurement uncertainty and the design allowable upper limit is used as the upper limit of the initial value range.

4. The intelligent operation and maintenance management system for the entire lifecycle of a tunnel based on digital twins as described in claim 3, characterized in that, The coupled propagation graph construction module sequentially applies a perturbation to one physical parameter of a tunnel physical unit, and obtains the changes in displacement, strain, joint opening, or internal force of other tunnel physical units while keeping other physical parameters and boundary conditions unchanged. The ratio of the disturbance amount of the physical parameter to the value of the physical parameter before the disturbance is determined as the relative disturbance amount of the physical parameter. The ratio of the change in the structural response to the normalized reference value of the structural response is determined as the relative change in the structural response. The normalized reference value of the structural response is the larger value between the absolute value of the structural response before the disturbance and the upper limit of the structural response noise. For any two tunnel physical units, the ratio of the relative change in the structural response of the downstream tunnel physical unit to the relative disturbance of the physical parameters of the upstream tunnel physical unit is determined as the structural propagation coefficient. When the structural propagation coefficient is not lower than the structural propagation activation threshold and the direction of structural response change conforms to the structural equilibrium calculation result of the digital twin model, a structural propagation edge is established between the upstream tunnel physical unit and the downstream tunnel physical unit. The structural propagation activation threshold is determined based on the larger of the normalized upper limit of the structural monitoring quantity under the operating baseline state and the normalized measurement uncertainty of the monitoring device.

5. A tunnel full lifecycle intelligent operation and maintenance management system based on digital twins as described in claim 4, characterized in that, The coupled propagation graph construction module determines the hydraulic propagation direction based on the head difference between adjacent tunnel physical units, and determines the hydraulic propagation coefficient by the ratio of the relative change in water pressure or drainage flow of the downstream tunnel physical unit to the relative change in head difference of the upstream tunnel physical unit. Hydraulic propagation edges are established when the hydraulic propagation coefficient is not lower than the hydraulic propagation activation threshold. The ratio of the relative change in the joint permeability coefficient to the relative change in the joint opening is determined as the structure-to-hydraulic conversion coefficient, and a structure-to-hydraulic conversion edge is established when the structure-to-hydraulic conversion coefficient is not lower than the structure-to-hydraulic conversion activation threshold. The ratio of the relative change in the normal force of the joint or the relative change in the joint opening to the relative change in the water pressure is determined as the hydraulic-to-structure conversion coefficient, and a hydraulic-to-structure conversion edge is established when the hydraulic-to-structure conversion coefficient is not lower than the hydraulic-to-structure conversion activation threshold. The activation thresholds for hydraulic propagation, structural-to-hydraulic conversion, and hydraulic-to-structural conversion are determined based on the larger of the normalized upper limit of the corresponding response quantity under the operating baseline state and the normalized measurement uncertainty of the corresponding monitoring device, respectively. The coupled propagation graph construction module recalculates each propagation coefficient as the groundwater level, joint opening, and hydraulic conduction capacity of the drainage branch change. Propagation edges with propagation coefficients not lower than the corresponding activation threshold are marked as active, while propagation edges with propagation coefficients lower than the corresponding activation threshold are marked as inactive.

6. A tunnel full lifecycle intelligent operation and maintenance management system based on digital twins as described in claim 5, characterized in that, The abnormal sequence generation module extracts historical periods from the state inheritance chain where there is no maintenance work and the structural and hydraulic monitoring values ​​do not exceed the operational monitoring control values. It groups the historical periods according to the numerical range of rainfall, groundwater level, temperature and tunnel operation load, and establishes the operational baseline state based on the structural and hydraulic monitoring values ​​within each group. The width of the numerical interval is not less than the measurement resolution of the corresponding boundary condition monitoring quantity, and the difference between the maximum and minimum values ​​of the corresponding boundary condition monitoring quantity within the same numerical interval does not exceed the upper limit of the fluctuation of the boundary condition monitoring quantity during the stable operating period. The abnormal sequence generation module selects the operating benchmark state based on the current rainfall, groundwater level, temperature and tunnel operating load values, and compares the current structural monitoring and hydraulic monitoring values ​​with the selected operating benchmark state. When the deviation between the displacement, strain, or joint opening and the operational baseline state exceeds the structural anomaly identification threshold, and the duration of the continuous exceedance of the structural anomaly identification threshold is greater than the longest continuous duration of occasional overruns of structural monitoring quantities under the operational baseline state, it is determined that the corresponding tunnel physical unit has a structural response anomaly. When the deviation between water pressure, moisture content, or drainage flow rate and the operating baseline state exceeds the hydraulic anomaly identification threshold, and the duration of continuous exceedance of the hydraulic anomaly identification threshold is greater than the longest continuous duration of occasional exceedance of hydraulic monitoring data under the operating baseline state, the corresponding tunnel physical unit is determined to have a hydraulic response anomaly. The structural anomaly identification threshold and the hydraulic anomaly identification threshold are determined based on the larger of the upper limit of the fluctuation of the corresponding monitored quantity under the operating baseline state and the measurement uncertainty of the corresponding monitoring device, respectively; the abnormal event sequence is arranged according to the start time of the anomaly, and the abnormal unit, anomaly type, response direction, response amplitude and duration are recorded.

7. A tunnel full lifecycle intelligent operation and maintenance management system based on digital twins as described in claim 6, characterized in that, The disease inversion module traces each abnormal unit in the abnormal event sequence in reverse along the active propagation edge in the dynamic hydraulic to structural coupling propagation diagram, and determines the node that can reach at least one abnormal unit through the propagation edge as the candidate starting unit. For each candidate starting unit, the disease inversion module changes at least one physical parameter within the initial value range corresponding to the candidate starting unit, and loads the rainfall, groundwater level, temperature and tunnel operation load within the time period corresponding to the abnormal event sequence into the digital twin model to obtain the simulated anomaly range, simulated anomaly start time, simulated propagation delay and simulated response direction. The differences in the number of units between the simulated anomaly range and the measured anomaly range, the time difference between the simulated anomaly start time and the measured anomaly start time, the time delay difference between the simulated propagation delay and the measured propagation delay, and the directional difference between the simulated response direction and the measured response direction are determined respectively. Each difference is normalized by dividing it by the maximum value of each difference among all candidate starting units. The four normalized differences are multiplied by one-quarter and then summed to determine the comprehensive difference. The disease inversion module excludes candidate starting units that do not meet the structural equilibrium condition or water conservation condition, and determines the candidate starting unit with the smallest comprehensive difference as the disease starting unit; when at least two candidate starting units have the same comprehensive difference, the candidate starting units with fewer activation state propagation edges and smaller relative physical parameter adjustments are selected as the disease starting units in sequence. An effective propagation path consists of a propagation edge between the disease initiation unit and the abnormal unit that is in an active state, whose response direction is consistent with the abnormal event sequence, and whose cumulative propagation delay and measured propagation delay do not exceed the allowable deviation of propagation delay; the allowable deviation of propagation delay is determined based on the sum of the sampling interval and the time synchronization error at the start of the abnormality.

8. A tunnel full lifecycle intelligent operation and maintenance management system based on digital twins as described in claim 7, characterized in that, The maintenance simulation module transforms the back wall grouting measures into reducing the back wall void rate, increasing the back wall constraint stiffness, and reducing the back wall grouting zone permeability; transforms the joint sealing measures into reducing the joint permeability coefficient; transforms the drainage branch dredging measures into increasing the drainage branch hydraulic conduction capacity; and transforms the joint structure reinforcement measures into increasing the joint stiffness. The maintenance simulation module loads the physical parameter changes corresponding to different combinations of candidate maintenance measures into the digital twin model, recalculates the propagation coefficient of the propagation edge in each effective propagation path, excludes candidate maintenance measure combinations that do not make the propagation coefficient of at least one propagation edge in each effective propagation path lower than the corresponding activation threshold, and excludes candidate maintenance measure combinations that make the stress, joint opening, or water pressure of adjacent tunnel physical units exceed the limit value. The limit value is the smaller of the design allowable value and the operation monitoring and control value for the corresponding parameter. The maintenance simulation module sorts the remaining candidate maintenance measures combinations in ascending order according to the number of physical units of the tunnel to be maintained. When the number of physical units of the tunnel to be maintained is the same, it sorts them in ascending order according to the number of propagation edges to be processed. When both of the above two numbers are the same, they sort them in ascending order according to the sum of the relative adjustment ranges of each physical parameter. The candidate maintenance measures combination ranked first is determined as the maintenance intervention combination. The maintenance task includes the physical unit of the tunnel to be maintained, the propagation edge to be processed, the maintenance range, the target physical parameters, the maintenance construction parameters corresponding to the target physical parameters, and the post-maintenance verification location.

9. A tunnel full lifecycle intelligent operation and maintenance management system based on digital twins as described in claim 8, characterized in that, The maintenance verification and correction module uses boundary condition pairing verification or hydraulic disturbance verification to obtain the post-maintenance response sequence. When using boundary condition pairing verification, a reference period is selected from the historical monitoring data before maintenance, in which the corresponding differences between rainfall, groundwater level, temperature, and tunnel operating load and the verification period after maintenance do not exceed the allowable deviation of their respective boundary conditions. The response sequence of the reference period is then compared with the response sequence after maintenance. The allowable deviation of the boundary conditions is determined based on the larger of the measurement uncertainty of the corresponding boundary condition monitoring quantity and the upper limit of fluctuation during the stable operating period. When using hydraulic disturbance verification, before the maintenance task is carried out, the target drainage branch associated with the hydraulic propagation edge to be treated is controlled to generate hydraulic disturbance and the disturbance response sequence before maintenance is obtained. After the maintenance task is carried out, the target drainage branch is controlled again with the same valve opening change and holding time and the disturbance response sequence after maintenance is obtained. The valve opening change and holding time are limited to ensure that the water pressure at the monitoring point does not exceed the design allowable value and the drainage flow does not exceed the rated capacity of the drainage branch. When the fault initiation unit generates a structural or hydraulic response exceeding the upper limit of the response noise during the post-repair verification period, while the tunnel physical unit downstream of the propagation edge to be treated does not generate a response exceeding the corresponding upper limit of the response noise, or when the difference between the abnormal start times of adjacent tunnel physical units in the post-repair response sequence no longer meets the allowable deviation of the propagation delay of the corresponding propagation edge, it is determined that the effective propagation path is blocked. The upper limit of the response noise is determined based on the larger of the upper limit of the fluctuation of the corresponding monitored quantity and the measurement uncertainty of the corresponding monitoring device during a stable operating period without maintenance or hydraulic disturbance.

10. A tunnel full lifecycle intelligent operation and maintenance management system based on digital twins as described in claim 9, characterized in that, The maintenance verification and correction module compares the recovery start time, recovery direction and recovery magnitude of each tunnel physical unit in the predicted recovery sequence and the post-maintenance response sequence in the direction of propagation from the initial defect unit to the abnormal unit. When the difference between the predicted response and the measured response of the tunnel physical unit located upstream in the propagation direction does not exceed the response identification threshold, while the difference between the predicted response and the measured response of the tunnel physical unit located downstream in the propagation direction exceeds the response identification threshold, the propagation edge connecting the two tunnel physical units is determined as the propagation edge where the difference first appears. The maintenance verification and correction module determines the correction range of the physical parameters associated with the propagation edge where the first difference occurs, based on the target physical parameters in the maintenance task, the actual execution values ​​of the maintenance construction parameters, and the post-maintenance test results. Under the condition of keeping the physical parameters of other propagation edges in the dynamic hydraulic-to-structural coupling propagation diagram unchanged, the module adjusts the joint stiffness, joint permeability coefficient, hydraulic conduction capacity of the drainage branch, or propagation delay of the propagation edge within the correction range until the difference between the recalculated predicted response and the post-maintenance response does not exceed the response identification threshold. The response identification threshold is determined based on the larger of the upper limit of the fluctuation of the corresponding monitored quantity during the stable period after maintenance and the measurement uncertainty of the corresponding monitoring device. The maintenance verification and correction module writes the maintenance task, the actual execution value of the maintenance construction parameters, the post-maintenance response sequence, the blocking result of the effective propagation path, and the corrected propagation edge physical parameters into the maintenance intervention status of the corresponding tunnel physical unit.

Images