A method for analyzing a composite stratum shield tunnel face stability support

Abstract

The application discloses a kind of composite stratum shield tunnel face stability support analysis methods, it is related to supporting analysis technical field, and the regional division is carried out to face and its front area, the region with similar deformation trend or instability tendency is classified into the same kind, and the different supporting control target is set for each region, the balanced scheduling framework of support and settlement is constructed, before entering risk condition, relevant risk characteristics are extracted, and input safety evaluation model, according to the evaluation result, the preventive correction of face pressure regulation amplitude, advancing rate change slope and grouting strategy is carried out, in the whole analysis and control process, cross-checking is carried out using different dimensional numerical method and field measurement, when difference is found, feedback to balanced scheduling framework, and support strategy is revised again.The analysis method constructs scheduling framework by real-time data and partition classification, combined with safety evaluation and cross-checking, overcomes the deficiency of composite stratum analysis, improves the stability of face and construction safety.

Description

Technical Field

[0001] This invention relates to the field of support analysis technology, specifically to a method for analyzing the stability support of a shield tunnel face in composite strata. Background Technology

[0002] In the construction of shield tunnels in complex strata, the stability control of the tunnel face is particularly critical due to the differences in the physical and mechanical properties of different soil and rock layers. Shield construction not only requires precise control of earth pressure or slurry pressure, but also requires the design of appropriate support measures for the characteristics of complex strata. Through support analysis, the stability and safety of different construction methods under complex strata conditions can be effectively evaluated, ensuring the smooth progress of tunnel construction and the safety of the ground environment.

[0003] The existing technology has the following drawbacks:

[0004] Existing analytical methods cannot accurately capture the stress release and seepage effects in the three-dimensional area in front of the tunnel face during tunnel excavation. This results in significant deviations in the predicted settlement trough width, volume loss rate, and deformation patterns. Furthermore, the methods cannot quickly correct support strategies when deviations or anomalies occur, which can easily lead to risks such as local instability of the tunnel face, excessive surface settlement, or sudden water and sand inrush. In particular, under the complex strata conditions where water-rich sand layers and silty mudstones coexist, these methods lack specificity and adaptability, making it difficult to effectively address the stability control requirements of the tunnel face under complex working conditions.

[0005] Based on this, the present invention proposes a method for analyzing the stability support of the tunnel face in composite strata shield tunnels. By constructing a scheduling framework through real-time data and zoning classification, and combining safety assessment and cross-verification, the method overcomes the shortcomings of composite strata analysis and improves the stability of the tunnel face and construction safety. Summary of the Invention

[0006] The purpose of this invention is to provide a method for analyzing the stability support of shield tunnel faces in composite strata, so as to solve the problems in the background technology.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for analyzing the stability support of a shield tunnel face in composite strata, the method comprising the following steps:

[0008] S1: During the tunnel boring machine (TBM) excavation process, multi-source data are collected in real time to form a dynamic information set reflecting the stress and seepage characteristics of the tunnel face;

[0009] S2: Based on the dynamic information set, the working face and the area in front of it are divided into regions. Regions with similar deformation trends or instability tendencies are classified into the same category, and different support control targets are set for each region.

[0010] S3: Combining the target requirements, construction plans and stratum deformation patterns of each region, construct a balanced scheduling framework for support and settlement to coordinate the matching relationship between face pressure, advance rate and grouting strategy;

[0011] S4: Before entering a risky working condition, extract relevant risk characteristics and input them into the safety assessment model. Based on the assessment results, make preventive corrections to the working face pressure adjustment range, the slope of the advance rate change, and the grouting strategy.

[0012] S5: Throughout the analysis and control process, numerical methods of different dimensions are used to cross-check with field measurements. When discrepancies are found, they are fed back to the balanced scheduling framework to revise the support strategy.

[0013] In a preferred embodiment, step S3: Combining the target requirements, construction plan, and stratum deformation patterns of each area, a balanced scheduling framework for support and settlement is constructed to coordinate the matching relationship between face pressure, advance rate, and grouting strategy, including the following steps:

[0014] The system acquires the categories of each area and their corresponding differentiated support control targets, real-time geological and construction parameters under the current construction status, and imports the current construction plan information to form the control input space of the balanced scheduling framework.

[0015] Analyze the statistical correlation or response mapping relationship between different combinations of construction parameters and formation response, and identify parameter combination patterns under different formation conditions;

[0016] Based on the current regional objectives, a multi-objective trade-off logic is adopted to coordinate the adjustment of face pressure, advance rate, and grouting strategy, ultimately generating support and construction parameter control instructions and completing the construction of a balanced scheduling framework.

[0017] For example, a tunnel section is divided into area A (silty sand layer), and the support control target is a maximum settlement ≤12mm and a volume loss rate ≤0.8%. Real-time construction parameters: current face pressure = 150kPa; advance rate = 20mm / min; grouting volume = 2.5m³ / m; design advance rate range 15–25mm / min; planned construction period requirement is 1.2m per hour. This constitutes the input space: Input space = {face pressure, advance rate, grouting volume, target settlement, target loss rate, planned rate} = {150, 20, 2.5, 12, 0.8%, 1.2}.

[0018] Based on numerical simulation data (such as finite element calculation output), the response relationships between face pressure P, advance rate v, grouting volume Q, settlement rate S, and volume loss rate V are shown in Table 1.

[0019] Table 1 Response Relationship Table:

[0020]

[0021] The comparison shows that when P=150kPa, v=20mm / min, and Q=2.5m³ / m, S=11.8mm and VL=0.78%, which just meets the control target. If the propulsion rate is increased to 25mm / min, the settlement exceeds the standard (13.5mm). If it is reduced to 15mm / min, the settlement is even smaller (10.6mm), but the propulsion efficiency is insufficient (lower than the planned 1.2m / h).

[0022] The objective function is set as follows: ensure S≤12mm; ensure VL≤0.8%; the advance rate v should be as close as possible to the planned value (20mm / min≈1.2m / h). Through constraint screening, the optimal solution is: face pressure PkPa; advance rate v=20mm / min; grouting volume Q=2.5m³ / m;

[0023] The final output control command is: control command = (P, v, Q) = (150 kPa, 20 mm / min, 2.5 m³ / m). This parameter combination satisfies the settlement and loss rate constraints while ensuring the progress requirements of the construction plan, thus completing the construction of the balanced scheduling framework.

[0024] In a preferred embodiment, support and construction parameter control instructions are generated, including the face pressure control range or target set value, advance rate adjustment value, grouting volume, grouting pressure, and grouting timing strategy.

[0025] In a preferred embodiment, step S2: Based on the dynamic information set, the working face and the area in front of it are divided into regions, and regions with similar deformation trends or instability tendencies are grouped into the same category. Differentiated support control targets are set for each region, including the following steps:

[0026] Multidimensional feature parameters covering the working face and the area in front are extracted from the dynamic information set at each monitoring moment to form a feature matrix with time as the sequence and spatial grid as the unit;

[0027] The face and the area in front are gridded or segmented continuously in the spatial dimension. Each segmentation unit is a control sub-region, and the set of feature parameters corresponding to the control sub-region is mapped.

[0028] Based on the feature matrix, clustering analysis logic is used to group each control sub-region;

[0029] Assign a label to each region category and summarize the common characteristics and response patterns of that category;

[0030] Based on the results of the control sub-region division, differentiated support control objectives are set for each region category.

[0031] Assume that at a certain monitoring moment, data was collected covering the tunnel face and a 5m × 5m grid area in front of it, with the grid subdivided into 4 × 4 = 16 cells. For each cell, the following parameters were collected: settlement rate S (mm / h); pore water pressure U (kPa); and horizontal displacement rate H (mm / h). The resulting feature matrix is ​​shown in Table 2.

[0032] Table 2 Feature Matrix Table:

[0033]

[0034] The area in front of the face of the tunnel boring machine was divided into 16 control sub-regions based on the gridding result, with each region corresponding to the aforementioned parameter set. For example, sub-region 3 → (S=3.5, U=160, H=1.8). The feature vectors (S, U, H) of all sub-regions were clustered (e.g., K-means, K=3). The results are as follows:

[0035] Category A (Low Risk Zone): Settlement rate <1.5mm / h, pore pressure <130kPa, displacement rate <0.6mm / h → Includes sub-regions {1,2,5,...};

[0036] Category B (Medium Risk Zone): Settlement rate of 2–3.5 mm / h, pore pressure of 140–155 kPa, displacement rate of 1–1.8 mm / h → includes sub-regions {6,...};

[0037] Category C (High-risk area): Settlement rate > 3.5 mm / h, pore pressure > 155 kPa, displacement rate > 1.8 mm / h → includes sub-regions {3,4,...}.

[0038] Category A → Labeled "Stable Zone": Characterized by low settlement and low displacement, with a stable response mode. Category B → Labeled "Transition Zone": Characterized by moderate settlement and displacement, requiring close monitoring. Category C → Labeled "Instability-Sensitive Zone": Characterized by rapid settlement and high pore pressure, posing a significant risk of instability.

[0039] Category A (Stable Zone): Maintain a conventional face pressure of 140–150 kPa and a grouting volume of 2.0 m³ / m.

[0040] Category B (Transition Zone): Moderately increase the working face pressure to 155–160 kPa and increase the grouting volume to 2.8 m³ / m.

[0041] Category C (High-risk area): Significantly increase the working face pressure to ≥165kPa, grouting volume to ≥3.5m³ / m, and reduce the advance rate.

[0042] Through the above example: extract data (settlement, pore pressure, displacement) from the feature matrix → divide the space into control sub-regions → cluster to obtain low / medium / high risk zones → assign differentiated support targets to different regions.

[0043] In a preferred embodiment, multidimensional feature parameters covering the working face and the area in front are extracted from the dynamic information set at each monitoring moment, including:

[0044] Stratigraphic types and their spatial distribution;

[0045] Surface subsidence rate, subsidence rate, and subsidence gradient of the working face and the area in front at each time point;

[0046] Characteristics of pressure fluctuation at the working face, trend of advance rate variation, and response of grouting volume and grouting pressure;

[0047] Changes in groundwater pressure, increases in pore water pressure, and permeability indicators;

[0048] The usage of support parameters in historical construction sections and the corresponding stratum response feedback;

[0049] Deformation pattern features were extracted, including gradual settlement, abrupt settlement, and uneven settlement.

[0050] In a preferred embodiment, step S4: Before entering a risky working condition, relevant risk characteristics are extracted and input into a safety assessment model. Based on the assessment results, preventative corrections are made to the working face pressure adjustment range, the slope of the advance rate change, and the grouting strategy, including the following steps:

[0051] Based on geological survey reports, geological feedback from previous construction sections, three-dimensional geological models, shield tunneling path planning, and real-time dynamic information sets, we can identify whether there are risky working conditions in the construction section.

[0052] For each type of risky working condition, based on the historical engineering case library, the geology-construction-response correlation database, and the current real-time monitoring data, characteristic parameters related to the risky working condition are extracted;

[0053] The feature parameters are input into a pre-trained safety assessment model to assess the instability modes that the current or upcoming section will face during future construction.

[0054] Based on the instability mode prediction results output by the safety assessment model, the logic for preventive adjustment of support parameters is triggered.

[0055] Input data source:

[0056] Geological survey report: The layer ahead is a medium-rich sandy layer 20m ahead, and the groundwater level is 3.5m below the surface;

[0057] Feedback from the early construction phase: Similar strata have previously experienced excessive settlement;

[0058] Three-dimensional geological model: There is a region with a high permeability coefficient at a depth of 10m along the tunnel axis. );

[0059] Path planning: The tunnel boring machine will enter this section in 5 hours;

[0060] Dynamic information set: Current working face pressure P = 150 kPa, settlement rate 1.0 mm / h. → Determined that this section has a risk condition of "high permeability sand layer".

[0061] Relevant parameters were extracted from the database and monitoring data: permeability coefficient. Pore ​​water pressure U = 120 kPa; Current settlement rate S = 1.0 mm / h; Stratum thickness H = 8 m; Tunnel boring machine advance rate v = 20 mm / min;

[0062] Constructing a risk feature vector: ;

[0063] Input the feature vectors into the security assessment model trained based on rules and a database to obtain the output:

[0064] Predicted instability mode: Seepage ahead of the tunnel face leads to "funnel-shaped" instability; the system automatically adjusts parameters based on model output.

[0065] Original settings: face pressure P = 150 kPa; advance rate v = 20 mm / min; grouting volume Q = 2.5 m³ / m;

[0066] Corrected: Working face pressure increased to The propulsion rate was reduced to The grouting volume was increased to → Generate a new control command: Control command = (165 kPa, 17 mm / min, 3.0 m³ / m).

[0067] In this example: the system identifies the risk of "water-rich sand layers" through geological models and monitoring data; extracts the feature vector X=(k,U,S,H,v); the safety assessment model predicts "funnel-shaped instability"; and automatically triggers the support parameter correction logic to increase the face pressure, slow down the advance rate, and enhance grouting. This ensures that the tunnel boring machine adjusts its strategy in advance before entering the risky working condition, avoiding instability accidents caused by passive responses.

[0068] In a preferred embodiment, the preventive adjustment logic for the support parameters includes control of the face pressure adjustment amplitude, control of the slope of the advance rate change, and optimization of the grouting strategy.

[0069] In a preferred embodiment, S5: Throughout the analysis and control process, cross-verification is performed using numerical methods of different dimensions and field measurements. When discrepancies are found, feedback is sent to the balanced scheduling framework to revise the support strategy, including the following steps:

[0070] Real-time acquisition of measured data from the construction site; mapping and alignment of the numerical method's prediction output with the measured data at the corresponding time and in the corresponding section using a unified timestamp and spatial location index logic.

[0071] The numerical prediction results are compared with the field measured data. The difference between the numerical prediction and the measured data is calculated, and the presence of deviation is determined based on the pre-set tolerance range.

[0072] If a deviation exists, the feedback control logic is triggered, and the deviation information is fed back into the balanced scheduling framework to dynamically re-evaluate and re-optimize the existing support strategy.

[0073] Cross-checking and feedback correction process: Time T=100min, shield tunneling to tunnel mileage K0+320m. Measured data (monitoring point located directly above the tunnel face for surface settlement monitoring): Measured settlement Smeas=11.5mm, numerical method prediction results (PLAXIS3D simulation output):

[0074] The predicted settlement Spred = 9.8 mm was mapped by aligning the two using timestamps (T = 100 min) and spatial indices (K0 + 320 m).

[0075] Calculate the deviation between the prediction and the actual measurement: ;

[0076] Tolerance range: ±1.0mm.

[0077] The current deviation ΔS = 1.7 mm > tolerance (1.0 mm), indicating that a deviation exists.

[0078] Because the deviation exceeds the tolerance range, the system triggers a feedback mechanism: the deviation information is sent to... Input the balanced scheduling framework; the balanced scheduling framework re-evaluates the current control parameters (face pressure, advance rate, grouting volume) and optimizes them.

[0079] Existing control parameters: face pressure P=150kPa, advance rate v=20v=20v=20mm / min, grouting volume Q=2.5m³ / m.

[0080] After optimization (recalculated by the system): the working face pressure is increased to P=158kPa, the advance rate is maintained at v=20mm / min, and the grouting volume is increased to Q=3.0m³ / m.

[0081] The new predicted settlement is updated to: Spred,new=11.2mm; the deviation from the measured value of 11.5mm is only 0.3mm, which falls within the tolerance range, and the closed-loop verification is completed.

[0082] In a preferred embodiment, real-time acquisition of measured data from the construction site includes the measured surface settlement rate and its rate of change at each monitoring point, the actual face pressure, the advance rate, the grouting volume and grouting pressure, tunnel convergence, surrounding displacement, groundwater pressure, and pore water pressure.

[0083] In a preferred embodiment, the multi-source data includes geological and hydrological parameters, shield tunneling parameters, grouting parameters, and surface and structural response monitoring data;

[0084] The pre-processed geological and hydrological parameters, shield tunneling parameters, grouting parameters, and surface and structural response monitoring data are integrated according to time nodes to construct a multi-dimensional dynamic data set.

[0085] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0086] This invention constructs a dynamic information set through multi-source real-time data acquisition, partitions and categorizes the tunnel face area, and establishes a balanced scheduling framework based on differentiated objectives to achieve coordinated control of tunnel face pressure, advance rate, and grouting strategy. A safety assessment model is introduced before high-risk conditions for preventative correction, avoiding instability accidents caused by reactive responses. Simultaneously, cross-verification using numerical simulation and field measurements forms a closed-loop feedback loop, ensuring a high degree of consistency between predictions and reality. This method effectively solves the problems of existing technologies failing to reflect the complexity of complex geological formations, the disconnect between predictions and reality, and the lag in countermeasures. It significantly improves the accuracy and adaptability of tunnel face stability analysis, reduces tunnel construction risks, and enhances the overall safety and reliability of construction. Attached Figure Description

[0087] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0088] Figure 1 This is a flowchart of the analysis method of the present invention.

[0089] Figure 2 This is a timing diagram of the analysis method of the present invention. Detailed Implementation

[0090] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0091] Example 1: Please refer to Figure 1 - Figure 2 As shown in the figure, this embodiment provides a method for analyzing the stability support of the tunnel face in composite strata shield tunnels. The analysis method includes the following steps:

[0092] S1: During the tunnel boring machine (TBM) excavation process, multi-source data, including geological composition, groundwater level, surface settlement monitoring, TBM thrust, grouting volume, and tunnel face pressure, are collected in real time. These data are then organized according to time nodes to form a dynamic information set that reflects the stress and seepage characteristics of the tunnel face, laying the foundation for subsequent analysis.

[0093] S2: Based on the dynamic information set, the working face and the area in front of it are divided into regions. Regions with similar deformation trends or instability tendencies are grouped into the same category, and differentiated support control targets are set for each region to make the working face control more targeted.

[0094] S3: Combining the target requirements, construction plans, and stratum deformation patterns of each region, a balanced scheduling framework for support and settlement is constructed. This framework coordinates the matching relationship between face pressure, advance rate, and grouting strategy to ensure face stability while controlling settlement of the surrounding environment. Upon completion, the adjustment results are written back to the dataset to form a new state update.

[0095] S4: Before entering high-permeability strata, shallow buried sections, or other high-risk working conditions, relevant risk characteristics are extracted in advance and input into the safety assessment model. Based on the assessment results, preventive corrections are made to the working face pressure adjustment range, advance rate change slope, and grouting strategy to ensure that the working face remains stable under special working conditions.

[0096] S5: Throughout the analysis and control process, numerical methods from different dimensions are used to cross-check with on-site measurements to ensure the reliability of the results. When discrepancies are found, they are immediately fed back to the balanced scheduling framework to revise the support strategy, forming a complete closed loop of "collection—analysis—adjustment—re-verification" to ensure the continuous stability of the tunnel face.

[0097] This application constructs a dynamic information set through multi-source real-time data acquisition, divides and classifies the tunnel face area, and establishes a balanced scheduling framework based on differentiated objectives to achieve coordinated control of tunnel face pressure, advance rate, and grouting strategy. A safety assessment model is introduced before risky conditions to perform preventative corrections, avoiding instability accidents caused by reactive responses. Simultaneously, a closed-loop feedback loop is formed by cross-verification of numerical simulation and field measurements to ensure a high degree of consistency between predictions and reality. This method effectively solves the problems of existing technologies that struggle to reflect the complexity of complex strata, the disconnect between predictions and reality, and the lag in countermeasures. It significantly improves the accuracy and adaptability of tunnel face stability analysis, reduces tunnel construction risks, and enhances the overall safety and reliability of construction.

[0098] Example 2:

[0099] S1: During the tunnel boring machine (TBM) excavation process, multi-source data, including geological composition, groundwater level, surface settlement monitoring, TBM thrust, grouting volume, and tunnel face pressure, are collected in real time. These data are then organized according to time nodes to form a dynamic information set that reflects the stress and seepage characteristics of the tunnel face, laying the foundation for subsequent analysis.

[0100] During shield tunnel construction, especially when traversing complex strata (such as alternating layers of sand, clay, gravel, and rock), high-water-pressure areas, shallow-buried sections, or sensitive sections near important buildings or structures, the tunnel face, as the leading edge of the shield advance, is directly affected by the coupled effects of multiple factors, including the physical and mechanical properties of the strata, groundwater seepage, construction loads, and support parameters. To accurately grasp the response state of the rock and soil in front of and around the tunnel face, and to effectively predict and control the stability of the tunnel face and stratum deformation, this method first constructs a multi-source heterogeneous data real-time acquisition and fusion processing mechanism. Through continuous monitoring and systematic integration of key construction and geological parameters, a high spatiotemporal resolution dynamic information set is formed that comprehensively characterizes the dynamic evolution of the stress field and seepage field at the tunnel face. This provides accurate and reliable data support for subsequent regional division, support target setting, and dynamic control.

[0101] Specifically, this step utilizes a multi-type sensor monitoring network and engineering data acquisition system deployed at the construction site and on the tunnel boring machine to acquire the following key categories of engineering and geological parameters in real time. All data is tagged and synchronously collected according to a unified timestamp to ensure strict alignment of data from different sources in the time dimension, facilitating subsequent multi-dimensional correlation analysis and modeling:

[0102] Geological and hydrological parameters include:

[0103] Information on stratigraphic distribution (such as soil / rock layer type, layer thickness, burial depth, interface location, etc.) obtained through real-time detection or preliminary geological exploration can be obtained through ground-penetrating radar, geological prediction systems carried by TBMs (such as BEAM, geological advanced prediction instruments), or geological borehole data before construction.

[0104] Groundwater level (or pore water pressure) monitoring data is usually collected in real time by pore water pressure gauges and water level gauges deployed in the strata in front of or around the tunnel face, in order to reflect the occurrence state of water in the strata and the seepage driving conditions.

[0105] If conditions permit, soil physical parameters (such as natural water content, void ratio, permeability coefficient, etc.) may also be included. Although these parameters may be provided by previous tests or geological reports, they are also included in the dynamic information set in this system as the basic input for stratigraphic classification and seepage analysis.

[0106] Shield tunneling construction parameters include:

[0107] The total thrust of the tunnel boring machine (TBM) is directly output by the TBM's main control system. It reflects the total contact load between the cutterhead and the soil during the advancement process and is an important indicator for evaluating the face compression effect and the response of the ground ahead.

[0108] The cutterhead torque reflects the energy consumption and ground hardness feedback during the shield cutting process, and indirectly reflects the shear strength characteristics of the ground.

[0109] Advance-Rate, which is the distance the tunnel boring machine moves forward per unit time, is a key variable for construction progress control and is also closely related to geological adaptability.

[0110] Auxiliary parameters such as cutterhead rotation speed and penetration-per-revolution can be incorporated into the information set as needed to further refine the formation cutting response analysis.

[0111] Grouting parameters include:

[0112] The synchronous grouting volume is the volume of grout injected through the grouting system behind the tunnel segments during the tunnel boring machine's advance. It is usually counted by ring (per ring advance) or per unit length.

[0113] Grouting pressure reflects the diffusion capacity of grout during the filling of formation voids and the additional stress it exerts on the surrounding soil.

[0114] Auxiliary information such as grout type, mix ratio, and setting time is used to evaluate the grout diffusion effect and long-term reinforcement performance, and should also be included in the information set if detailed records are available.

[0115] Surface and structural response monitoring data include:

[0116] Ground-Surface-Settlement (GSD) data, which is the vertical displacement data of the ground surface collected periodically or in real time by settlement monitoring points (such as leveling instruments, InSAR, and total station monitoring points) deployed on the ground, is a core indicator for evaluating the impact of tunnel boring machine (TBM) construction on the surrounding environment.

[0117] If monitoring conditions permit, data on secondary impacts such as the tilting of surrounding buildings, crack development, and deformation of underground pipelines can also be included for a more comprehensive environmental safety assessment.

[0118] All the aforementioned monitoring data are collected in real time through distributed data acquisition systems (such as SCADA systems, IoT sensor platforms, BIM data platforms, etc.) and transmitted to the central data processing platform via standardized data interfaces and communication protocols (such as MQTT, OPCUA, Modbus, etc.). During the data acquisition process, to ensure the timeliness and consistency of the information set, all data are sampled and aligned at a unified time node (e.g., every minute, every 5 minutes, or every cycle), forming a time-series indexed, multi-dimensional, multi-source synchronized data matrix.

[0119] Subsequently, the system performs preprocessing and fusion processing on the collected raw data. This processing logic mainly includes the following key steps:

[0120] The system identifies and removes obvious outliers (such as sudden zero values, maximum / minimum values, and missing data segments) caused by sensor malfunctions, communication interruptions, or environmental interference. Common logic includes threshold judgment and sliding window statistical filtering (such as the 3σ criterion and moving average method). Since different types of sensors may have different sampling frequencies, the system uses timestamp matching and linear / spline interpolation to unify all parameters to the same time point, ensuring that variables are comparable at the same moment. Statistical processing (such as mean) is performed on continuously collected raw data according to predefined analysis periods (such as each cycle, each minute, or a custom time window). The data includes maximum, minimum, and cumulative values, and key characteristic parameters for subsequent analysis are extracted, such as "grouting volume per unit advance length," "thrust-to-advance rate ratio," and "face water pressure gradient." Preprocessed geological and hydrological parameters, shield tunneling parameters, grouting parameters, and surface and structural response monitoring data are integrated according to time nodes to construct a structured, multi-dimensional dynamic data set. This set includes not only the original monitoring values ​​but also calculated derived features, thus comprehensively and accurately depicting the coupled state of the "geology-construction-support-environment" at each moment of the tunnel face. The resulting dynamic information set is a high-dimensional data volume indexed by time and using multi-source parameters as variables. Its core function is to reflect in real-time and systematically the stress distribution characteristics, seepage field trends, construction load response status, and surrounding environmental deformation of the tunnel face and its surrounding strata, providing a complete, accurate, and dynamic data foundation for subsequent tunnel face area division, support target setting, parameter coordinated scheduling, and risk early warning.

[0121] S2: Based on the dynamic information set, the working face and the area in front of it are divided into regions. Regions with similar deformation trends or instability tendencies are grouped into the same category, and differentiated support control targets are set for each region to make the working face control more targeted.

[0122] Based on the constructed multi-source dynamic information set, this method further focuses on the tunnel face and the adjacent geological space within a certain range. This area is a high-risk zone and key control domain for stress release, ground disturbance, changes in groundwater seepage paths, and potential instability events (such as tunnel face collapse, water inrush, mudslide, and excessive surface subsidence) during tunnel boring machine (TBM) advancement. To achieve more precise and adaptive support control in this area, this method introduces a data-driven regional division and differentiated target setting mechanism. The core of this mechanism is that instead of treating the area in front of the tunnel face as a homogeneous and undifferentiated single control object, it further subdivides the tunnel face and the area in front into multiple "control sub-regions" with similar engineering behavior characteristics (such as deformation trends, stress response, seepage sensitivity, and instability tendency) based on ground response characteristics, deformation development trends, and potential instability modes. For each sub-region, a personalized support control target is set that matches its risk level, geological conditions, and construction response, thereby significantly improving the targeting, effectiveness, and construction safety of the support strategy.

[0123] In the specific implementation process, this step relies on the dynamic information set constructed in S1 to extract key feature parameters related to formation response, construction disturbance, and environmental deformation, including but not limited to:

[0124] Stratigraphic types and their spatial distribution (such as sand layers, clay layers, interbedded structures, etc.).

[0125] Surface subsidence rate, subsidence rate, and subsidence gradient of the working face and the area in front at each time point;

[0126] Characteristics of pressure fluctuation at the working face, trend of advance rate variation, and response of grouting volume and grouting pressure;

[0127] Changes in groundwater pressure, increments in pore water pressure, and permeability indicators (such as inferred or inverted permeability coefficients).

[0128] The usage of support parameters in historical construction sections and the corresponding stratum response feedback (such as whether a certain type of stratum experiences increased settlement or face instability under specific thrust and grouting conditions).

[0129] Extract deformation pattern features (such as gradual settlement, abrupt settlement, uneven settlement, etc.).

[0130] Based on the aforementioned multidimensional feature parameters, the system employs data clustering and pattern recognition logic to dynamically divide the spatial area of ​​the tunnel face and a certain range ahead (usually several times the tunnel diameter, such as 3D to 5D, where D is the excavation diameter of the tunnel boring machine). This processing logic mainly includes the following technical steps:

[0131] First, multidimensional feature parameters corresponding to each monitoring moment and covering the working face and the area in front are extracted from the dynamic information set to form a high-dimensional feature matrix with time as the sequence and spatial grids (or continuous regions) as the units. To eliminate the influence of the difference in dimensions between different parameters on the subsequent clustering analysis, the system adopts standardization processing logic (such as Z-score standardization or Min-Max normalization) to perform dimensionless transformation on various features, ensuring that all variables participate in the subsequent analysis at the same scale.

[0132] The tunnel face and the area in front of it are divided into grids or continuous segments in the spatial dimension (for example, along the tunnel axis, each ring is advanced in units of length, or several control segments are divided at fixed intervals; in the radial or circumferential dimension, they are further subdivided according to stratigraphic interfaces, hydrological units, etc.). Each segmented unit is a "control sub-region" and is mapped to a set of corresponding characteristic parameters (such as the average settlement rate, grouting response, stratigraphic type, etc. in the region over the past several time periods).

[0133] Based on the constructed feature matrix, the system employs clustering analysis logic to group each control sub-region, identifying sets of regions with high similarity in deformation trends, stress responses, seepage behavior, or instability tendencies. This processing logic does not rely on pre-defined classification labels; instead, it automatically groups regions with similar feature patterns into one category by calculating the similarity of feature vectors between different regions (such as Euclidean distance, cosine similarity, Mahalanobis distance, etc., or based on density, hierarchy, cluster center, and other algorithmic logic).

[0134] For example, if several areas exhibit characteristics such as rapid increase in surface subsidence rate, sudden drop in tunnel face pressure, and significant increase in grouting volume but limited subsidence control effect during continuous advancement, the system will classify them as "soft and water-rich strata control areas with high instability tendency"; conversely, if an area only experiences slight subsidence and stable tunnel face pressure under similar construction parameters, it may be classified as "low-sensitivity hard strata control areas".

[0135] After clustering, the system assigns an identifier label to each region category (such as category A, category B, etc., or classifies regions into categories I, II, III, etc. according to risk level), and summarizes the common characteristics and typical response patterns of the category, for example:

[0136] Category A (High-permeability sand layer area): High sensitivity to surface subsidence, poor working face stability, and strong response to changes in grouting pressure and advance rate;

[0137] Category B (clay strata): Deformation develops slowly but with a significant lag effect, and is more sensitive to changes in working face pressure;

[0138] Category C (Hard Rock Interbedded Zone): Overall stability is good, but excavation disturbance may cause local rockfall or rock burst risk.

[0139] Based on the above regional division results, this method further sets differentiated support control targets for each regional category. The logic behind these targets fully considers the geological characteristics, historical response behavior, current construction parameters, and future progress trends of the region, aiming to achieve precise control through a "one-category-one-policy" or "one-section-one-target" approach. For example:

[0140] For areas with a high tendency to instability, set more conservative support targets, such as strictly controlling the increase in surface settlement (e.g., not exceeding 10 mm), maintaining the working face pressure above the safety threshold, and increasing the grouting volume and grouting pressure to strengthen the stratum reinforcement.

[0141] For low-sensitivity areas, control standards can be appropriately relaxed, the advance rate and support parameter configuration can be optimized to improve construction efficiency while ensuring overall stability.

[0142] For transitional areas or areas with unclear characteristics, the system can adopt a dynamic observation and step-by-step correction logic. Based on the initial conservative target, the target value is gradually optimized as the construction progresses and feedback information is obtained.

[0143] S3: Combining the target requirements, construction plans, and stratum deformation patterns of each region, a balanced scheduling framework for support and settlement is constructed. This framework coordinates the matching relationship between face pressure, advance rate, and grouting strategy to ensure face stability while controlling settlement of the surrounding environment. Upon completion, the adjustment results are written back to the dataset to form a new state update.

[0144] Based on the dynamic information set that enables the division of the working face and the area in front, and the setting of differentiated support control targets for each area, this step further focuses on how to coordinate various key support parameters and construction operation variables in the specific construction process to achieve the dual goals of working face stability and surrounding environment settlement control. To this end, this method constructs a "support and settlement equilibrium scheduling framework." This framework is essentially a multi-objective, multi-parameter, dynamically adaptable intelligent decision-making and coordination control mechanism. Its core function is to dynamically optimize and match key construction control parameters such as face pressure, shield tunneling speed, and synchronous grouting (or secondary grouting) strategies based on the clearly defined support objectives for different areas, the established construction organization plan (such as daily progress indicators, ring assembly arrangements, equipment operation and maintenance windows, etc.), and the real-time response law of stratum deformation. This enables a coordinated and consistent matching relationship among the parameters, thereby ensuring that the face does not become unstable (such as collapse, water inrush, mudslide, pressure imbalance, etc.) while controlling the disturbances to the surrounding strata and surface environment (such as settlement, tilting, cracking, etc.) within the design or safety allowable range.

[0145] Specifically, the construction and operation of this balanced scheduling framework mainly includes the following key technical aspects:

[0146] First, the system obtains the pre-defined regional categories and their corresponding differentiated support control objectives (such as maximum allowable settlement rate, upper and lower limits of face pressure, grouting volume range, etc.) from step S2, and extracts real-time geological and construction parameters under the current construction status (such as current stratum type, face pressure, advance rate, grouting volume and pressure, surface settlement monitoring values ​​and their rate of change, etc.) from the dynamic information set in step S1. Simultaneously, the system also imports current construction plan information, including: planned advance length (or number of loops), construction period requirements, equipment capacity limitations (such as maximum / minimum advance speed, grouting pumping capacity, etc.), assembly operation window, shutdown and maintenance plan, and other actual engineering constraints. These inputs collectively constitute the "control input space" of the balanced scheduling framework, clarifying the multi-objective constraints that must be met during the scheduling process, such as:

[0147] Stability objective: The pressure at the working face should be maintained within a predetermined range to avoid ground collapse or pressure imbalance ahead;

[0148] Settlement control target: The settlement rate of the ground surface or buildings and the cumulative settlement rate shall not exceed the design limit;

[0149] Construction feasibility objectives: The advancement rate and grouting strategy should be matched with the equipment capacity and the assembly rhythm of the ring pieces to avoid construction interruption or efficiency reduction due to parameter conflicts.

[0150] Based on the above input conditions, the system adopts multi-parameter collaborative control logic to dynamically match and jointly optimize the three core control variables: face pressure, propulsion rate, and grouting strategy. Its processing logic mainly includes the following key technical points:

[0151] The system first analyzes the statistical correlation or response mapping relationship between different combinations of construction parameters (such as the face pressure and grouting volume corresponding to a certain advance rate) and the stratum response (such as settlement rate, face pressure fluctuation, grouting pressure feedback, etc.) based on historical data and real-time monitoring information accumulated in S1 and S2.

[0152] Based on a clear understanding of the correlation patterns among parameters, the system further employs a multi-objective trade-off logic to coordinately adjust the face pressure, advance rate, and grouting strategy according to the target constraints of the current region (e.g., if a region is a high settlement-sensitive area, the settlement control weight is higher; if it is a high instability risk area, the face pressure stability weight is higher). This logic does not pursue the ultimate optimization of a single parameter, but rather seeks a set of parameters that are "overall optimal or risk-controllable" through dynamic trade-offs (e.g., appropriately reducing the advance rate to achieve a lower settlement rate, or slightly increasing the face pressure to reduce the grouting volume requirement), thus achieving a reasonable compromise between the contradictions of various control objectives.

[0153] The scheduling framework does not set fixed parameters all at once. Instead, it continuously receives real-time feedback information from a dynamic information set (such as the latest settlement monitoring values, face pressure fluctuations, and grouting effect feedback) as construction progresses and the strata response changes. Based on this feedback, it dynamically re-evaluates and readjusts the existing parameter combinations. For example, if a sudden increase in the settlement rate occurs in a certain area at the current advance rate, the system may automatically trigger parameter adjustment logic, suggesting a reduction in advance speed or an increase in grouting volume to suppress settlement development. Conversely, if the face pressure remains low and the strata are stable, a moderate increase in speed may be allowed to improve efficiency.

[0154] After the above-mentioned collaborative optimization and dynamic matching processing, the system finally generates a set of coordinated and optimized support and construction parameter control instructions, including: recommended face pressure control range or target set value; suggested advance rate adjustment value (or upper / lower limit constraint); optimized grouting volume, grouting pressure and grouting sequence strategy (such as whether supplementary grouting is needed, grouting in stages, etc.).

[0155] These adjustments serve two purposes. First, they act as construction guidance parameters, which can be output to the tunnel boring machine control system, grouting management system, or construction scheduling platform to assist on-site engineers in setting parameters and making operational decisions. Second, the system writes these adjusted parameter values, along with their corresponding time nodes, regional categories, construction status, and other information, back to the dynamic information set in step S1, forming new status data records. These records are used to update the current construction status and parameter feedback, thereby providing the latest input for subsequent continuous monitoring, regional re-division, target re-evaluation, and iterative optimization of scheduling strategies.

[0156] The significant features of this balanced scheduling framework are its dynamism, synergy, and adaptability: it is not a static parameter template, nor does it adjust a single variable in isolation. Instead, it organically combines support objectives, construction plans, and geological response patterns to construct a dynamic feedback loop between parameters, responses, and objectives. This allows the support strategy to be continuously optimized as construction progresses, geological changes occur, and risk conditions evolve. Thus, while ensuring the stability of the working face and the safety of the surrounding environment, it achieves a comprehensive balance between construction efficiency, support costs, and environmental impact.

[0157] In summary, by constructing and implementing this support and settlement equilibrium scheduling framework, this method achieves refined, dynamic, and collaborative control under multiple parameters, objectives, and constraints during shield tunneling construction. This lays a solid foundation for strategy execution and status feedback for the prediction and preventive adjustment of high-risk conditions in the subsequent S4 step, as well as the closed-loop verification and continuous optimization in the S5 step. It is one of the key links to ensure the stability of the shield tunnel face throughout the entire process.

[0158] S4: Before entering high-permeability strata, shallow buried sections, or other high-risk working conditions, relevant risk characteristics are extracted in advance and input into the safety assessment model. Based on the assessment results, preventive corrections are made to the working face pressure adjustment range, advance rate change slope, and grouting strategy to ensure that the working face remains stable under special working conditions.

[0159] During the construction of shield tunnels, when the tunnel boring machine is about to enter high-permeability strata (such as coarse sand, gravel, and pebble layers with high porosity and strong water permeability), shallow buried sections (i.e., shallow tunnel depth, insufficient overburden thickness, and weak stratum self-stabilization capacity), uneven interbedded soft and hard layers, karst development areas, adjacent important buildings or underground pipelines, and areas sensitive to historical settlement, the tunnel face and surrounding soil and rock are extremely prone to instability, water inrush, mudslides, excessive settlement, or collapse under the disturbance of the tunnel boring machine. These areas generally have characteristics such as low stratum bearing capacity, strong groundwater activity, sensitive stress release, and strict surrounding environmental constraints. This poses a serious threat to construction safety, structural stability, and environmental safety.

[0160] To effectively address the aforementioned high-risk construction scenarios and avoid the potential for uncontrolled consequences due to reactive adjustments after risks actually occur, this method innovatively introduces an integrated mechanism of "risk prediction—model evaluation—preventive parameter correction." Before the tunnel boring machine (TBM) enters the high-risk section, these characteristics are quantified and input into a pre-constructed safety assessment model. Through model analysis and evaluation, the potential face instability modes, ground deformation trends, and environmental response risks of that section are predicted. Based on this, the key support parameters to be implemented—including the face pressure adjustment range, the slope of the advance rate change (i.e., the control of the advance rate change trend over time), and synchronous or secondary grouting strategies—are proactively and preventively optimized and adjusted. This results in a "risk-adaptive" support parameter configuration before entering the high-risk zone, ensuring that the TBM remains in a controllable and stable technical state throughout the tunneling process. The specific implementation process mainly includes the following technical aspects:

[0161] The system first identifies whether there are known high-risk geological or environmental conditions in the upcoming construction section based on geological survey reports, geological feedback from previous construction sections, 3D geological models, shield tunneling path planning, and real-time dynamic information sets. For example:

[0162] High-permeability strata: Sand, gravel or fractured zones with high permeability coefficients, abundant groundwater, and prone to water inrush are identified through geological forecasting (such as TBM geological advance forecasting system, pore water pressure monitoring, resistivity inversion, etc.).

[0163] Shallow buried section: Based on the tunnel longitudinal section design and surface elevation data, determine whether the current or upcoming section is in a shallow buried condition with a small overlying soil layer (such as the tunnel arch depth being less than 2 to 3 times the diameter of the shield machine). In such sections, the strata have poor self-stabilizing ability and the surface settlement control requirements are extremely high.

[0164] Interlayers of soft and hard or abrupt changes: such as interlayers of soft clay and hard rock, fault fracture zones, and deep weathering troughs, are prone to stress concentration or uneven deformation induced by shield tunneling disturbance.

[0165] Environmentally sensitive areas: such as areas with important buildings, rail transit lines, underground pipelines, historical sites, etc. above or to the side of the construction area, have strict requirements for deformation indicators such as settlement, tilting, and vibration.

[0166] For each of the aforementioned high-risk conditions, the system extracts key characteristic parameters directly related to that type of risk based on a historical engineering case library, a geology-construction-response correlation database, and current real-time monitoring data. These parameters include: stratum permeability indicators (such as inferred permeability coefficient, groundwater level depth, and hydraulic head); overlying soil thickness and soil physical and mechanical parameters (such as unit weight, internal friction angle, and cohesion, which can be empirical values ​​or previously inverted values); current face pressure level, historical trends in advance rate, grouting volume, and pressure feedback effects; settlement response, face instability records, and parameter adjustment history of past construction sections under similar geological conditions. These risk characteristics are summarized in structured data to form a "risk feature vector" or "risk description data package" for specific high-risk sections, serving as the input basis for subsequent safety assessment models.

[0167] The system inputs the extracted risk feature data into a pre-trained security assessment model. This model is a comprehensive assessment tool built on a Bayesian network, and its core function is:

[0168] Based on the input risk characteristics, the model assesses the main instability modes (such as face extrusion, water and mud inrush, and excessive settlement) that the current or upcoming section may face during future construction. This assessment logic does not rely on a single fixed algorithm but rather uses a multi-dimensional risk factor comprehensive analysis to identify the key control points most likely to affect construction safety. For example, if the input characteristics show a high-permeability sand layer + shallow burial + high advance rate, the model may determine "high risk of water inrush + high sensitivity to surface settlement" and list "insufficient face pressure maintenance" and "untimely grouting reinforcement" as key risk drivers. If it is a soft-hard interlayer + sharp bend section, the model may indicate that "uneven stress at the face" and "abnormal tool wear or collapse risk" require special attention.

[0169] The pre-training steps for the security assessment model are as follows:

[0170] We collected a large amount of construction data from historical tunnel / underground engineering projects, including:

[0171] Engineering geological and hydrogeological data: soil and rock type, permeability, stratigraphic structure (such as alternating layers of soft and hard soil), burial depth, groundwater conditions, etc.

[0172] Construction environment and geometric parameters: tunnel depth, cross-sectional shape, curve radius (whether it is a sharp bend), surrounding environment (such as whether it is adjacent to buildings);

[0173] Construction process parameters: advance rate, excavation method (such as shield tunneling, TBM, drill and blast method), support type and parameters, grouting conditions, etc.

[0174] Risk events and instability patterns: actual instability phenomena (such as water inrush and mudslide, face extrusion, surface subsidence, collapse, abnormal tool wear, etc.);

[0175] Risk drivers include: abnormal face pressure, delayed support, untimely grouting, and excessive disturbance of the surrounding rock.

[0176] Monitoring data (optional): Real-time or historical monitoring information such as surface subsidence, surrounding rock deformation, internal forces of support structures, and pore water pressure.

[0177] Bayesian networks are a type of probabilistic graphical model that can express the conditional dependencies between variables, making them suitable for uncertainty reasoning and multi-factor risk analysis.

[0178] Input risk characteristics (such as soil type, burial depth, permeability, construction rate, support parameters, etc.) are defined as parent nodes or intermediate variables in the network; possible instability modes (such as water inrush, settlement, collapse, etc.) and key risk drivers are defined as child nodes or output nodes.

[0179] In the absence of strong priors, the following algorithm can be used to automatically learn the dependencies between variables from the data:

[0180] Score-based methods, such as the K2 algorithm, Hill-Climbing, and BIC / MDL scoring, seek the optimal network structure to maximize posterior probability or minimize information loss.

[0181] Constraint-based methods, such as the PC algorithm, infer whether there is a direct connection between variables through conditional independence tests.

[0182] Redundant connections are eliminated to ensure a simple network structure with engineering interpretability; the logical coherence of key risk transmission paths (e.g., geological conditions → support response → instability mode) is strengthened; and the network is ensured to map typical risk scenarios (e.g., "shallow burial + high water pressure + rapid advancement → water inrush + settlement"). After determining the network structure, the conditional probability tables (CPTs) of each node need to be estimated, i.e., the probability of a child node taking different states given the state of the parent node.

[0183] Historical data-based statistics: Using labeled historical data, statistically analyze the conditional probability distribution of various variable combinations;

[0184] Maximum likelihood estimation (MLE): Used when there is sufficient data;

[0185] Bayesian estimation (introducing prior): When the data for certain risk events is sparse, a Dirichlet prior can be introduced to smooth the data and avoid the zero probability problem.

[0186] Use cross-validation (such as k-fold cross-validation) or hold-out method to divide the dataset into training and test sets; evaluation metrics include: classification accuracy / recall / F1-score (for unstable pattern classification tasks); correct association of risk drivers; probability prediction calibration (whether the predicted probability of risk occurrence is consistent with the actual frequency).

[0187] Adjust the network structure (e.g., add / remove edges) to improve prediction performance and interpretability; optimize CPT parameters, especially for low-probability, high-impact events; if data is insufficient, transfer learning or semi-supervised learning methods can be used to supplement training with similar engineering case data.

[0188] Industry standards and design specifications are embedded as supplementary rules into the model logic to enhance the model's robustness in situations with missing data or boundary conditions; for example, "if the burial depth is <10m and it is a water-rich sand layer, the default water inrush risk level is increased." The instability modes and risk drivers output by the model are connected to the subsequent preventive adjustment module for support parameters; for example, if the model predicts "insufficient pressure maintenance at the working face," the system automatically suggests increasing the support strength or grouting in advance.

[0189] Based on the instability mode prediction results output by the safety assessment model, the system further triggers the preventive adjustment logic of the support parameters, and performs forward-looking optimization settings for key construction parameters before and after entering the high-risk section. This mainly includes the adjustment of the following three types of core parameters:

[0190] Face pressure adjustment range control: If the assessment shows that there is a risk of instability or collapse at the face (such as too low pressure or loose strata), the system will suggest increasing the face pressure setting value or reducing the pressure reduction range to ensure that the soil in front maintains the necessary support force and prevent collapse or water inrush; conversely, if there is a risk of excessive compression, the pressure rise slope will be appropriately controlled to avoid face extrusion.

[0191] Advance rate variation slope control: For high-risk sections, the system generally recommends reducing the advance rate or controlling its variation slope (i.e., the rate increase / decrease per unit time) to a low level to reduce the formation disturbance rate and avoid stress abrupt changes or face instability caused by rapid advance. For example, before entering high-permeability sand layers, the advance rate can be reduced from the normal value of 8-10 mm / min to 4-6 mm / min, and the acceleration process should be controlled to be gradual.

[0192] Grouting strategy optimization: Based on the formation permeability and settlement control requirements, the system will adjust grouting parameters, such as increasing the synchronous grouting volume and grouting pressure to fill the shield tail voids in a timely manner and control formation relaxation and settlement development; for high-permeability formations, it may be recommended to grout in advance or use grouts with higher early strength and water-stopping effect, such as two-liquid grout or quick-setting grout; adjust the grouting sequence logic, such as increasing the grouting frequency, segmented grouting, or supplementary grouting strategies, to ensure that the reinforcement effect covers the entire risk section.

[0193] S5: Throughout the analysis and control process, numerical methods from different dimensions are used to cross-check with on-site measurements to ensure the reliability of the results. When discrepancies are found, they are immediately fed back to the balanced scheduling framework to revise the support strategy, forming a complete closed loop of "collection—analysis—adjustment—re-verification" to ensure the continuous stability of the tunnel face.

[0194] Throughout the entire process of shield tunnel construction, in order to ensure the high reliability and engineering applicability of the established support strategies, parameter adjustment schemes, and analysis conclusions on the stability of the tunnel face and the settlement control of the surrounding environment, this method constructs and implements a closed-loop control mechanism of "cross-verification of multi-dimensional numerical simulation and field measured data, deviation identification, feedback correction, and strategy iteration". This mechanism runs through all the analysis and control stages from S1 to S4 mentioned above. Its core objective is to systematically compare and verify the analysis results obtained based on theoretical models, empirical rules, or data-driven methods (i.e., "numerical method output") with the actual engineering response data collected in real time by various sensors deployed at the construction site (i.e., "on-site measured data") to identify possible deviations or inconsistencies between the two. Once a significant difference is found, the feedback control logic is immediately triggered to input the deviation information back into the equilibrium scheduling framework, driving the dynamic re-optimization and readjustment of support parameters (such as face pressure, advance rate, grouting strategy, etc.), thereby forming a complete closed-loop process of "data acquisition → analysis and decision-making → parameter adjustment → effect verification" to achieve continuous and adaptive control of shield tunnel face stability and construction safety.

[0195] At the same time, the system acquires "measured data" from the construction site in real time through the multi-source dynamic information set established in step S1, including: the measured surface settlement rate and its rate of change at each monitoring point; the actual face pressure, advance rate, grouting volume and grouting pressure; stratum deformation monitoring data (such as tunnel convergence, surrounding displacement, etc., if applicable); and measured values ​​of groundwater pressure or pore water pressure.

[0196] The system uses a unified timestamp and spatial location indexing logic to accurately map and align the predicted output of numerical methods with the measured data at the corresponding time and in the corresponding section, ensuring that the two are comparable in terms of time and engineering location, thus laying the foundation for subsequent cross-verification.

[0197] After completing the data mapping, the system employs a multi-dimensional comparative analysis process to systematically compare the prediction results of the numerical methods with the field measured data. This analysis process mainly includes the following key points:

[0198] Core indicators that are crucial to construction safety and control effectiveness, such as surface settlement rate, actual and predicted values ​​of face pressure, relationship between advance rate and stratum response, and grouting volume and settlement inhibition effect, were selected as the main objects of comparative analysis.

[0199] By calculating the difference between numerical predictions and measured data (such as absolute difference, relative error, deviation of trend, etc.), and based on a pre-set reasonable tolerance range (this threshold can be dynamically adjusted according to different geological conditions, construction stages, or risk levels), the system determines whether there is a "significant deviation" between the two. For example, if the surface settlement rate predicted by the numerical method at a certain monitoring point is 12 mm, while the measured value has reached 22 mm and exceeds the preset settlement warning value (such as 20 mm), the system determines that there is a significant deviation in the settlement control of this section, and the feedback mechanism needs to be triggered.

[0200] Once the system identifies a significant deviation or trend inconsistency between numerical predictions and measured data through the aforementioned comparative analysis logic, indicating that the current support strategy may be ineffective in responding to the actual formation response, the system will immediately trigger feedback control logic. This logic will input the deviation information (including deviation type, magnitude, location, time point, and relevant parameter configurations) back into the equilibrium scheduling framework in step S3, driving the system to dynamically re-evaluate and re-optimize the existing support strategy. This feedback correction logic mainly includes the following processing steps:

[0201] Identify the specific section where the deviation occurred, the corresponding construction parameters (such as excessive advance rate, insufficient grouting, etc.), and the potential risks (such as excessive settlement, face instability, etc.). Based on the cause of the deviation, the balanced scheduling framework recalculates and optimizes the support parameters. For example, if the measured settlement is too large, it may be recommended to further reduce the advance rate, increase the grouting volume, or appropriately increase the face pressure. The corrected parameter scheme is output as a new control command to the construction site and is simultaneously updated to the dynamic information set of step S1 to form a new status record for guiding the next stage of construction.

[0202] Through the aforementioned cyclical process of "collecting (real-time and predicted data) → analyzing (comparing numerical simulation and actual measurements) → adjusting (optimizing support parameters) → re-verifying (a new round of monitoring and comparison)," this method constructs a highly adaptive, self-learning, and self-correcting closed-loop control system. This system can continuously optimize support strategies and control parameters based on actual feedback during construction, ensuring that the tunnel face maintains a stable structure, controllable deformation, predictable risks, and optimizable parameters when facing complex and variable geological conditions, construction disturbances, and environmental constraints.

[0203] This closed-loop mechanism not only improves the reliability and robustness of the support analysis method, but also achieves a technological leap from "static pre-setting" to "dynamic adaptation" and from "experience-driven" to "data and model-driven". It is one of the core technical links to ensure the safety, efficiency and controllability of the entire shield tunnel construction process.

[0204] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0205] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for analyzing the stability support of a shield tunnel face in composite strata, characterized in that: The analytical method includes the following steps: S1: During the tunnel boring machine (TBM) excavation process, multi-source data are collected in real time to form a dynamic information set reflecting the stress and seepage characteristics of the tunnel face; S2: Based on the dynamic information set, the working face and the area in front of it are divided into regions. Regions with similar deformation trends or instability tendencies are classified into the same category, and different support control targets are set for each region. S3: Combining the target requirements, construction plans and stratum deformation patterns of each region, construct a balanced scheduling framework for support and settlement to coordinate the matching relationship between face pressure, advance rate and grouting strategy; S4: Before entering a risky working condition, extract relevant risk characteristics and input them into the safety assessment model. Based on the assessment results, make preventive corrections to the working face pressure adjustment range, the slope of the advance rate change, and the grouting strategy. S5: Throughout the analysis and control process, cross-checking is performed using numerical methods of different dimensions and on-site measurements. When discrepancies are found, feedback is sent to the balanced scheduling framework to revise the support strategy. Step S2: Based on the dynamic information set, the working face and the area in front of it are divided into regions. Regions with similar deformation trends or instability tendencies are grouped into the same category, and differentiated support control targets are set for each region, including the following steps: Multidimensional feature parameters covering the working face and the area in front are extracted from the dynamic information set at each monitoring moment to form a feature matrix with time as the sequence and spatial grid as the unit; The face and the area in front are gridded or segmented continuously in the spatial dimension. Each segmentation unit is a control sub-region, and the set of feature parameters corresponding to the control sub-region is mapped. Based on the feature matrix, clustering analysis logic is used to group each control sub-region; Assign a label to each region category and summarize the common characteristics and response patterns of that category; Based on the results of the control sub-region division, differentiated support control objectives are set for each region category.

2. The method for analyzing the stability support of a shield tunnel face in composite strata according to claim 1, characterized in that: Step S3: Combining the target requirements, construction plan, and stratum deformation patterns of each area, construct a balanced scheduling framework for support and settlement to coordinate the matching relationship between face pressure, advance rate, and grouting strategy. This includes the following steps: The system acquires the categories of each area and their corresponding differentiated support control targets, real-time geological and construction parameters under the current construction status, and imports the current construction plan information to form the control input space of the balanced scheduling framework. Analyze the statistical correlation or response mapping relationship between different combinations of construction parameters and formation response, and identify parameter combination patterns under different formation conditions; Based on the current regional objectives, a multi-objective trade-off logic is adopted to coordinate the adjustment of face pressure, advance rate, and grouting strategy, ultimately generating support and construction parameter control instructions and completing the construction of a balanced scheduling framework.

3. The method for analyzing the stability support of a shield tunnel face in composite strata according to claim 2, characterized in that: Generate support and construction parameter control instructions, including the face pressure control range or target setting value, advance rate adjustment value, grouting volume, grouting pressure, and grouting sequence strategy.

4. The method for analyzing the stability support of a shield tunnel face in composite strata according to claim 3, characterized in that: Multidimensional feature parameters covering the working face and the area in front are extracted from the dynamic information set at each monitoring moment, including: Stratigraphic types and their spatial distribution; Surface subsidence rate, subsidence rate, and subsidence gradient of the working face and the area in front at each time point; Characteristics of pressure fluctuation at the working face, trend of advance rate variation, and response of grouting volume and grouting pressure; Changes in groundwater pressure, increases in pore water pressure, and permeability indicators; The usage of support parameters in historical construction sections and the corresponding stratum response feedback; Deformation pattern features were extracted, including gradual settlement, abrupt settlement, and uneven settlement.

5. A method for analyzing the stability support of a shield tunnel face in composite strata according to any one of claims 2-4, characterized in that: Step S4: Before entering the risky working condition, extract relevant risk characteristics and input them into the safety assessment model. Based on the assessment results, make preventive corrections to the working face pressure adjustment range, the slope of the advance rate change, and the grouting strategy, including the following steps: Based on geological survey reports, geological feedback from previous construction sections, three-dimensional geological models, shield tunneling path planning, and real-time dynamic information sets, we can identify whether there are risky working conditions in the construction section. For each type of risky working condition, based on the historical engineering case library, the geology-construction-response correlation database, and the current real-time monitoring data, characteristic parameters related to the risky working condition are extracted; The feature parameters are input into a pre-trained safety assessment model to assess the instability modes that the current or upcoming section will face during future construction. Based on the instability mode prediction results output by the safety assessment model, the logic for preventive adjustment of support parameters is triggered.

6. The method for analyzing the stability support of a shield tunnel face in composite strata according to claim 5, characterized in that: The preventive adjustment logic for the support parameters includes control of the working face pressure adjustment range, control of the slope of the advance rate change, and optimization of the grouting strategy.

7. The method for analyzing the stability support of a shield tunnel face in composite strata as described in claim 6. Its features include: S5: Throughout the analysis and control process, numerical methods of different dimensions are used for cross-verification with field measurements. When discrepancies are found, feedback is given to the balanced scheduling framework to revise the support strategy, including the following steps: Real-time acquisition of measured data from the construction site; mapping and alignment of the numerical method's prediction output with the measured data at the corresponding time and in the corresponding section using a unified timestamp and spatial location index logic. The numerical prediction results are compared with the field measured data. The difference between the numerical prediction and the measured data is calculated, and the presence of deviation is determined based on the pre-set tolerance range. If a deviation exists, the feedback control logic is triggered, and the deviation information is fed back into the balanced scheduling framework to dynamically re-evaluate and re-optimize the existing support strategy.

8. The method for analyzing the stability support of a shield tunnel face in composite strata according to claim 7, characterized in that: Real-time acquisition of measured data from the construction site includes the measured surface settlement rate and its rate of change at each monitoring point, actual face pressure, advance rate, grouting volume and grouting pressure, tunnel convergence, surrounding displacement, groundwater pressure, and measured values ​​of pore water pressure.

9. The method for analyzing the stability support of a shield tunnel face in composite strata according to claim 1, characterized in that: The multi-source data includes geological and hydrological parameters, shield tunneling construction parameters, grouting parameters, and surface and structural response monitoring data; The pre-processed geological and hydrological parameters, shield tunneling parameters, grouting parameters, and surface and structural response monitoring data are integrated according to time nodes to construct a multi-dimensional dynamic data set.

Images