A soil pressure risk early warning decision method for subway shield construction

Abstract

The application discloses a soil pressure risk early warning decision method for subway shield construction, and relates to the technical field of shield construction. The application discloses a soil pressure risk early warning decision method for subway shield construction, and relates to the technical field of shield construction. The application discloses a soil pressure risk early warning decision method for subway shield construction, and relates to the technical field of shield construction. The application discloses a soil pressure risk early warning decision method for subway shield construction, and relates to the technical field of shield construction. The application discloses a soil pressure risk early warning decision method for subway shield construction, and relates to the technical field of shield construction. The application discloses a soil pressure risk early warning decision method for subway shield construction, and relates to the technical field of shield construction. The application discloses a soil pressure risk early warning decision method for subway shield construction, and relates to the technical field of shield construction. The application discloses a soil pressure risk early warning decision method for subway shield construction, and relates to the technical field of shield construction. The application discloses a soil pressure risk early warning decision method for subway shield construction, and relates to the technical field of shield construction. The application discloses a soil pressure risk early warning decision method for subway shield construction, and relates to the technical field of shield construction. The application discloses a method for early warning and decision of soil pressure risk of subway shield construction, which simplifies the emergency disposal process of soil pressure risk and improves the accuracy and response speed of soil pressure risk control.

Description

Technical Field

[0001] This invention relates to the field of tunnel boring machine (TBM) construction technology, and more specifically, to a method for early warning and decision-making regarding earth pressure risks in subway TBM construction. Background Technology

[0002] Earth pressure risk early warning and decision-making methods used in subway tunnel boring machine (TBM) construction serve as the core support for the construction of a digital and intelligent construction safety management system for urban rail transit projects. The accuracy of risk identification, the timeliness of early warning response, the scientific nature of decision-making schemes, and the adaptability of the system directly determine the safety level of subway TBM construction, the progress of project construction, and the economic benefits of the project.

[0003] The geological conditions of subway tunnel boring machine (TBM) construction are complex and variable, with numerous factors affecting earth pressure, rapid risk evolution, significant differences in construction scenarios, and high requirements for emergency response. As a core constraint affecting the efficient application and stable implementation of earth pressure risk management technology, it is dynamically influenced by multiple factors such as differences in geological conditions (e.g., uneven strata, composite strata, water-rich strata), fluctuations in TBM equipment performance, dynamic changes in construction parameters, and varying sensitivities of the surrounding environment. Meanwhile, intelligent early warning and decision-making technology for earth pressure risk, specifically designed for subway TBM construction scenarios, is a key means to solve the problems of delayed early warning, blind decision-making, and poor adaptability of existing earth pressure management methods, directly ensuring the safety and stability of the TBM tunneling process and the smooth progress of the project.

[0004] However, existing methods for early warning and decision-making regarding earth pressure risks in subway tunnel construction often rely on single threshold warnings and human experience in practical applications. They lack comprehensive assessment mechanisms in both spatiotemporal dimensions, scenario-adaptive multi-objective decision-making mechanisms, standardized control measures mapping mechanisms, and closed-loop dynamic iteration mechanisms. These methods fail to cover the complex needs of different geological conditions, different tunneling equipment, different construction stages, and different surrounding environments. Consequently, existing control methods tend to be one-sided, emphasizing post-event handling over pre-event warning. Furthermore, existing methods often adopt static control models, failing to address issues such as single-dimensional risk assessment, inaccurate anomaly identification, highly subjective decision-making schemes, and lagging system updates. This results in high rates of missed or misjudged earth pressure risks, untimely emergency response, significant construction safety hazards, and a heavy workload for management personnel.

[0005] No effective solutions have yet been proposed to address the problems in the relevant technologies. Summary of the Invention

[0006] To address the problems in related technologies, this invention proposes an earth pressure risk early warning and decision-making method for subway tunnel construction, in order to overcome the aforementioned technical problems existing in the existing related technologies.

[0007] To achieve the above objectives, the specific technical solution adopted by the present invention is as follows: A method for early warning and decision-making regarding earth pressure risk in subway tunnel construction includes the following steps: S1. Obtain geological survey data, shield equipment parameters, and historical earth pressure risk management case data of the area to be constructed, construct an earth pressure risk management benchmark system, and generate a risk management priority weight vector. S2. During the tunnel boring machine excavation process, real-time soil pressure parameters, equipment operation data and underground environment monitoring data are collected. After data preprocessing, a standardized monitoring dataset is obtained. A multi-dimensional comprehensive evaluation of the standardized monitoring dataset is carried out to obtain the real-time comprehensive soil pressure status index. Then, it is benchmarked against the soil pressure risk control benchmark system to identify the types of soil pressure anomalies, determine the risk control level, and match alternative control measures. As a preferred option, the process of collecting earth pressure parameters, equipment operation data, and underground environment monitoring data in real time during shield tunneling, preprocessing the data to obtain a standardized monitoring dataset, conducting a multi-dimensional comprehensive evaluation of the standardized monitoring dataset to obtain a real-time comprehensive earth pressure status index, and then benchmarking it against the earth pressure risk management benchmark system to identify the type of earth pressure anomaly, determine the risk management level, and match alternative management measures includes the following steps: S21. Preset spatiotemporal synchronous sampling rules, collect soil pressure parameters, equipment operation data and underground environment monitoring data in real time, and perform outlier removal and missing value completion processing to obtain a standardized monitoring dataset; S22. Perform spatiotemporal registration and dimensional normalization on the standardized monitoring dataset, extract the temporal change indicators and spatial correlation indicators of each monitoring dimension, and obtain the real-time comprehensive earth pressure state index through multi-dimensional weighted comprehensive calculation. As a preferred embodiment, the steps of performing spatiotemporal registration and dimensional normalization on the standardized monitoring dataset, extracting time-series change indicators and spatial correlation indicators for each monitoring dimension, and obtaining the real-time comprehensive earth pressure state index through multi-dimensional weighted comprehensive calculation include the following steps: S221. Set a unified data collection time benchmark and construction area spatial coordinate benchmark, and perform time sequence alignment and point matching for various types of data in the standardized monitoring dataset; S222. Standardize the data units and value ranges of the standardized monitoring dataset after time-series alignment and location matching. S223. Perform statistical analysis on the standardized monitoring data after unifying the data units and numerical value ranges, obtain the numerical fluctuation law of each monitoring parameter as the tunneling mileage changes, extract the time series change index of each monitoring dimension, and analyze the correlation of monitoring data at different points in combination with the distribution relationship of construction points, and extract spatial correlation indexes. As a preferred embodiment, the statistical analysis of standardized monitoring data after unifying data units and numerical value ranges to obtain the numerical fluctuation patterns of each monitoring parameter with tunneling mileage, extracting time-series change indicators for each monitoring dimension, and analyzing the correlation of monitoring data at different locations in conjunction with the distribution relationship of construction points to extract spatial correlation indicators includes the following steps: S2231. Perform sliding window statistical analysis on the normalized monitoring data to obtain the numerical fluctuation pattern of each monitoring parameter as the tunneling mileage changes, and extract trend change indicators, abrupt change indicators and periodic indicators of each monitoring dimension based on the numerical fluctuation pattern, and integrate them to form a time series change indicator set. S2232. Organize the spatial coordinate relationships of each construction point in the standardized monitoring dataset, construct a spatial correlation network of points, analyze the Pearson correlation coefficient of monitoring data of spatially adjacent points, extract spatial correlation indicators between different points, and integrate them to form a set of spatial correlation indicators.

[0008] S224. Preset the weight allocation of indicators and the reference rules for standard state of earth pressure, assign weights to time-series change indicators and spatially related indicators respectively, and integrate them according to the logic of superimposing the indicator components one by one and complementing the related indicators to obtain a comprehensive evaluation dataset. S225. The comprehensive assessment dataset is compared and fitted with the standard state reference rule for earth pressure item by item to calculate the real-time comprehensive state index of earth pressure.

[0009] S23. Compare the real-time comprehensive state index of earth pressure with the standard state index in the earth pressure risk management benchmark system, and identify the type of earth pressure anomaly and determine the risk management level based on the similarity comparison results. S24. Set the mapping relationship between earth pressure anomaly types, risk control levels and control measures, and obtain the corresponding set of alternative control measures from the earth pressure risk control benchmark system.

[0010] S3. Based on the risk control level and the type of earth pressure anomaly, construct a multi-objective optimization decision-making model for earth pressure risk, and input the alternative control measures and risk control priority weight vector into the model to calculate the cost-risk balance coefficient of each alternative control measure. As a preferred option, the construction of a multi-objective optimization decision-making model for earth pressure risk based on risk management level and earth pressure anomaly type, and the input of alternative management measures and risk management priority weight vectors into the model to calculate the cost-risk balance coefficient of each alternative management measure, includes the following steps: S31. Set three core optimization objectives: cost control, risk reduction, and construction progress assurance, and define the initial constraint range for each core optimization objective; S32. Preset optimization and adjustment rules, and adjust the constraint boundaries and relative weights of the core optimization objectives according to the type of earth pressure anomaly and the risk control level, and construct a multi-objective optimization decision-making model for earth pressure risk. As a preferred embodiment, the preset optimization adjustment rules, and the adjustment of the constraint boundaries and relative weights of the core optimization objectives based on the type of earth pressure anomaly and the risk control level, to construct a multi-objective optimization decision model for earth pressure risk, include the following steps: S321. Pre-set multi-objective optimization and adjustment rules to classify construction control scenarios into different types of earth pressure anomalies and different risk control levels. S322. Adjust the constraint boundaries of the core optimization objectives according to the construction control scenario classification standards and the types of abnormal earth pressure. S323. Adjust the relative weight of core optimization objectives according to the construction management scenario classification standard and the risk management level. S324. Set the adjusted constraint boundary as the model solution constraint condition, and embed the adjusted relative weight ratio to build a multi-objective optimization decision model for earth pressure risk.

[0011] S33. Input the weight vectors of each alternative control measure and risk control priority into the multi-objective optimization decision-making model for earth pressure risk, calculate the multi-objective comprehensive score of each alternative control measure, and use it as the cost-risk balance coefficient corresponding to the control measure.

[0012] S4. Prioritize each alternative control measure according to the cost-risk balance coefficient, generate graded early warning signals and executable control decision plans, and collect decision execution effect data simultaneously. As a preferred option, the process of prioritizing each alternative control measure based on the cost-risk balance coefficient, generating tiered early warning signals and executable control decision plans, and simultaneously collecting data on the effectiveness of decision implementation includes the following steps: S41. Establish a four-level early warning threshold system for earth pressure risk and rules for generating control decision-making schemes, and clarify the signal types, release scope, response time limits and responsible entities corresponding to different early warning levels; S42. Prioritize each alternative control measure according to the cost-risk balance coefficient from largest to smallest, and match the ranking results with the four-level early warning threshold system for earth pressure risk to generate a graded early warning signal corresponding to the risk control level. S43. Analyze the signal type, release scope, response time limit and responsible entity in the graded early warning signal, and generate an executable control decision plan in combination with the control decision plan generation rules; As a preferred embodiment, the process of analyzing the signal type, release scope, response time limit, and responsible entity in the graded early warning signal, and generating an executable control decision scheme in conjunction with the control decision scheme generation rules, includes the following steps: S431. Pre-set a hierarchical control measure template library, analyze and extract three core control parameters from the hierarchical early warning signals: signal type, release scope, and response time limit, and match the corresponding level of standardized handling process template from the control measure template library according to the risk control level. S432. Embed the three types of core control parameters into the standardized handling process template, supplement and clarify the person in charge of handling, the equipment allocation list, the material demand plan and the key points of on-site operation, and standardize the format according to the control decision plan generation rules to generate an executable control decision plan.

[0013] S44. Distribute the graded early warning signals and executable control decision schemes to the construction control terminal for execution, and simultaneously collect data on earth pressure changes, construction progress, and resource consumption during the decision execution process, and integrate them to form decision execution effect data.

[0014] S5. Based on the decision-making execution effect data, dynamically iterate and update the earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision-making model.

[0015] As a preferred option, the dynamic iterative update of the earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision model based on decision execution effect data includes the following steps: S51. Pre-set the construction control effect evaluation index system and evaluation judgment threshold, and extract quantitative evaluation indicators from the decision execution effect data; S52. Compare and analyze the quantitative evaluation indicators with the evaluation judgment thresholds to determine the degree of adaptability of the earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision-making model to the construction scenario. As a preferred embodiment, the step of comparing and analyzing the quantitative evaluation indicators with the evaluation judgment thresholds to determine the adaptability of the earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision-making model to the construction scenario includes the following steps: S521. Preset indicator comparison rules and adaptation level grading standards, compare the quantitative evaluation indicators with the evaluation judgment thresholds one by one, and obtain the set of indicator compliance status. S522. Based on the set of indicator compliance, analyze the overall adaptability of the earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision-making model to the construction scenario, and determine the degree of adaptability based on the overall adaptability.

[0016] S53. Based on the results of the adaptation degree judgment, correct the standard state indicators and mapping correlations within the earth pressure risk management benchmark system, and simultaneously optimize the constraint boundary and weight allocation parameters of the multi-objective optimization decision model for earth pressure risk. S54. The revised earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision-making model are solidified, stored and dynamically updated.

[0017] The beneficial effects of this invention are as follows: 1. This invention constructs a method for early warning and decision-making regarding earth pressure risks in subway shield tunneling, integrating multi-source data benchmark construction, spatiotemporal dual-dimensional comprehensive evaluation, scenario-adaptive multi-objective decision-making, hierarchical early warning scheme generation, and closed-loop dynamic iteration. Combined with risk control priority weight generation, real-time earth pressure comprehensive state calculation, cost-risk balance coefficient deduction, and anomaly type mapping mechanism, it achieves intelligent early warning and decision-making for earth pressure risks that are precisely matched to the construction scenario. Simultaneously, through the configuration of a standardized control measure template library, dynamic adjustment of multi-objective constraints, and closed-loop feedback of execution effects, it simplifies the emergency response process for earth pressure risks, reduces the decision-making burden and professional threshold for construction management personnel, and improves the accuracy and response speed of earth pressure risk control in subway shield tunneling.

[0018] 2. This invention achieves continuous quantitative assessment of earth pressure risk status through real-time acquisition of earth pressure parameters, synchronous monitoring of equipment operation data, and real-time monitoring of underground environmental parameters. Simultaneously, it ensures the timeliness and standardization of risk management by hierarchically pushing early warning signals and automatically generating executable control plans. Furthermore, it improves the safety and economy of subway tunnel construction through spatiotemporal feature fusion, scenario-adaptive multi-objective optimization, and closed-loop feedback of execution effects. This avoids the problems of missed or misjudged assessments that easily occur with single-threshold early warnings, the poor adaptability of static decision-making schemes to complex geology, and the high subjectivity and low efficiency of manual decision-making. Thus, it realizes the intelligent, precise, and closed-loop development of earth pressure risk management in subway tunnel construction. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of a method for early warning and decision-making on earth pressure risk in subway tunnel construction according to an embodiment of the present invention. Detailed Implementation

[0021] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0022] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0023] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments, such as... Figure 1 As shown, the earth pressure risk early warning and decision-making method for subway shield tunneling according to an embodiment of the present invention includes the following steps: S1. Obtain geological survey data, shield equipment parameters, and historical earth pressure risk management case data of the area to be constructed, construct an earth pressure risk management benchmark system, and generate a risk management priority weight vector. Specifically, the method for constructing the earth pressure risk management benchmark system and generating the risk management priority weight vector is as follows: The acquired geological survey data, shield tunneling equipment parameters, and historical earth pressure risk management case data are classified, cleaned, formatted, and redundancy is removed to establish a structured historical database of earth pressure risk management. Earth pressure risk management is divided into basic scenarios based on four dimensions: geological conditions, tunnel boring machine type, construction stage and surrounding environment. A standard state index library for normal, early warning and dangerous earth pressure states under each scenario is established. By analyzing the risk evolution patterns and treatment effects of different types of earth pressure anomalies in historical cases, a three-level mapping relationship database of anomaly type, risk control level, and alternative control measures is constructed. Integrate the standard state index library and the three-level mapping relationship library to form a complete earth pressure risk management benchmark system; Based on successful handling data from historical risk management cases, an improved entropy weight method is used to calculate the objective weights of each risk management dimension. The analytic hierarchy process is combined with expert experience to determine the subjective weights. The final risk management priority weight vector is obtained by integrating the objective and subjective weights.

[0024] It should be explained that the principles for constructing the earth pressure risk management benchmark system are as follows: Geological adaptability principle: Different geological conditions correspond to independent benchmark index systems. For water-rich strata, the focus is on strengthening the seepage pressure and soil chamber pressure fluctuation index. For soft and hard uneven strata, the focus is on strengthening the cutterhead torque and thrust change index. Equipment matching principle: The benchmark thresholds are set differently based on the model, service life and performance status of the tunnel boring machine. The thresholds for normal state are appropriately relaxed for older equipment, while the thresholds for warning state are tightened. Historical verification principle: All benchmark indicators and mapping relationships must be verified by historical data from at least three similar construction projects. Indicators with a verification pass rate of less than 80% need to be readjusted.

[0025] The principle for allocating risk management priority weights is as follows: The weight of the safety risk dimension is ≥0.5, the weight of the cost control dimension is 0.2~0.3, and the weight of the construction progress dimension is 0.1~0.2; Level I risk management scenario: weight of security risk dimension ≥ 0.7, weight of other dimensions ≤ 0.15; Level IV risk management scenario: Cost control dimension weight ≥ 0.4, construction progress dimension weight 0.3~0.4, safety risk dimension weight 0.2~0.3.

[0026] The parameter settings for the benchmark system and weight vector are as follows: Historical database update cycle: Automatic update once every 100 rings of tunneling completed; mandatory update within 24 hours after a major risk event. The range of the subjective and objective weighting fusion coefficient is 0.4 to 0.6, with a default value of 0.5. The weight vector can take values ​​from 0.1 to 0.8 for a single dimension, and the sum of the weights for all dimensions must be equal to 1. Weight adjustment step size: 0.05, to ensure adjustment accuracy.

[0027] A dynamic calibration mechanism based on the benchmark system is adopted. The benchmark index threshold is locally calibrated every 50 rings of tunneling, with the calibration range not exceeding ±10% of the current threshold. A comprehensive calibration is performed after each construction section is completed, with the comprehensive calibration range not exceeding ±20% of the current threshold, so as to avoid the benchmark system from deviating from the actual construction scenario.

[0028] S2. During the tunnel boring machine excavation process, real-time soil pressure parameters, equipment operation data and underground environment monitoring data are collected. After data preprocessing, a standardized monitoring dataset is obtained. A multi-dimensional comprehensive evaluation of the standardized monitoring dataset is carried out to obtain the real-time comprehensive soil pressure status index. Then, it is benchmarked against the soil pressure risk control benchmark system to identify the types of soil pressure anomalies, determine the risk control level, and match alternative control measures. In this embodiment of the invention, the real-time acquisition of earth pressure parameters, equipment operation data, and underground environment monitoring data during shield tunneling, followed by data preprocessing to obtain a standardized monitoring dataset, and a multi-dimensional comprehensive evaluation of the standardized monitoring dataset to obtain a real-time comprehensive earth pressure status index, is then benchmarked against the earth pressure risk management benchmark system to identify earth pressure anomaly types, determine risk management levels, and match alternative management measures. The process includes the following steps: S21. Preset spatiotemporal synchronous sampling rules, collect soil pressure parameters, equipment operation data and underground environment monitoring data in real time, and perform outlier removal and missing value completion processing to obtain a standardized monitoring dataset; Specifically, the spatiotemporal synchronous sampling and data preprocessing method is as follows: A unified UTC time reference and a WGS84 construction area spatial coordinate reference are preset to establish a spatiotemporal synchronous mapping relationship for multi-source data; differentiated sampling frequencies are set according to the variation characteristics of different monitoring parameters, and earth chamber pressure parameters, shield tunneling equipment operation data, and underground environment monitoring data are collected synchronously; the collected raw data are timestamped and matched with spatial coordinates to generate a raw monitoring dataset with spatiotemporal labels; outlier identification and removal are performed on the raw monitoring dataset using the 3σ principle combined with the DBSCAN clustering algorithm to eliminate invalid data caused by sensor drift, electromagnetic interference, and human error; spatiotemporal interpolation is used to complete the missing data generated after outlier removal, generating a complete intermediate monitoring dataset; finally, the intermediate monitoring dataset is formatted and redundant data is removed to obtain the final standardized monitoring dataset.

[0029] It should be explained that spatiotemporal synchronous sampling follows three principles: time synchronization priority, spatial matching, and frequency adaptation. The timestamp error of all monitoring data must not exceed 100ms to ensure the comparability of different parameters at the same time point. All monitoring data must be associated with the corresponding tunnel boring machine mileage and cutterhead spatial coordinates to achieve a one-to-one correspondence between data and construction location. Parameters with rapid changes use a high sampling frequency, while parameters with slow changes use a low sampling frequency, balancing data accuracy and transmission efficiency. Data preprocessing follows the principles of outlier classification, spatiotemporal priority for missing values, and data integrity. Identified outliers are classified, isolated outliers are directly removed, and continuous outlier segments are marked and processed separately. Data from adjacent time points and adjacent spatial points are used for interpolation to ensure data continuity and relevance. The data integrity rate after preprocessing must not be lower than 95%; time periods below this threshold must be marked and trigger manual review.

[0030] The core parameters for spatiotemporal synchronous sampling and data preprocessing are set as follows: Earth chamber pressure parameter sampling frequency 10–50 Hz (default 20 Hz); shield tunneling equipment operation data sampling frequency 1–10 Hz (default 5 Hz); underground environment monitoring data sampling frequency 0.1–1 Hz (default 0.5 Hz); time synchronization error threshold ≤100 ms; spatial synchronization error threshold ≤0.1 m; DBSCAN clustering algorithm neighborhood radius 0.2; minimum number of samples 5; sliding window length 5–20 sampling points (default 10); spatiotemporal interpolation method time weight coefficient 0.6–0.8; spatial weight coefficient 0.2–0.4. Simultaneously, a dynamic sampling frequency adjustment mechanism is adopted, automatically adjusting the sampling frequency according to changes in geological conditions. In complex geological areas such as uneven strata and water-rich strata, the sampling frequency is doubled; in areas with stable geological conditions, the sampling frequency is reduced by 50%, avoiding data redundancy or insufficient monitoring.

[0031] S22. Perform spatiotemporal registration and dimensional normalization on the standardized monitoring dataset, extract the temporal change indicators and spatial correlation indicators of each monitoring dimension, and obtain the real-time comprehensive earth pressure state index through multi-dimensional weighted comprehensive calculation. In this embodiment of the invention, the steps of performing spatiotemporal registration and dimensional normalization on the standardized monitoring dataset, extracting time-series change indicators and spatial correlation indicators for each monitoring dimension, and obtaining the real-time comprehensive earth pressure state index through multi-dimensional weighted comprehensive calculation include the following steps: S221. Set a unified data collection time benchmark and construction area spatial coordinate benchmark, and perform time sequence alignment and point matching for various types of data in the standardized monitoring dataset; Specifically, the method for unifying spatiotemporal references and registering data is as follows: The entire process adopts the same UTC time reference as S21 and the WGS84 construction area spatial coordinate reference, establishing a global reference parameter sharing mechanism to ensure that all equipment and terminals use the same reference system; Secondary timestamp calibration is performed on data from different sources and with different sampling frequencies to correct equipment clock drift deviations, and all data are uniformly mapped to the same time axis; The cutterhead center coordinates, tunneling mileage, and ring number information are extracted from the real-time shield attitude data, and spatial location and construction progress labels are associated with each monitoring data point; The three types of data—soil chamber pressure, equipment operation, and underground environment—are matched point by point according to spatiotemporal correspondence, and finally, data with conflicting spatiotemporal labels are eliminated to complete time sequence alignment and point matching.

[0032] It should be explained that this step follows three core principles: the principle of uniqueness of the benchmark, which means that the benchmark remains constant throughout the construction period, and historical data must be uniformly back-calibrated when adjustments are necessary; the principle of temporal continuity, which means that the data after alignment has no breaks or overlaps, and the time interval is consistent with the sampling frequency; and the principle of spatial accuracy, which means that the point matching is accurate to the specific cutterhead position within a single ring tunneling mileage.

[0033] The core parameters are set as follows: timestamp secondary calibration accuracy ≤ 10ms, spatial coordinate transformation error ≤ 0.05m; maximum allowable deviation of time sequence alignment ≤ 1 sampling cycle; point matching distance threshold ≤ 0.2m; shield attitude data update frequency is consistent with soil chamber pressure sampling frequency. Simultaneously, a benchmark dynamic verification mechanism is adopted, automatically performing global benchmark verification every 10 rings of tunneling. When a deviation exceeds the threshold, calibration is automatically triggered and the registered data labels are synchronously corrected.

[0034] S222. Standardize the data units and value ranges of the standardized monitoring dataset after time-series alignment and location matching. Specifically, the method for unifying the data dimensions and value ranges is as follows: The min-max standardization method is used to normalize the monitoring data of different physical dimensions. First, the parameters are divided into five categories: pressure, torque, thrust, displacement, and environment. The extreme value ranges of historical normal operation data for each type of parameter are extracted. Positive indicators (the larger the value, the better the condition) and negative indicators (the larger the value, the worse the condition) are distinguished. The negative indicators are converted in reverse and then uniformly mapped to the value range of [0, 1]. The extreme data that exceeds the extreme value range are truncated at the boundary. Finally, a standardized dataset with unified dimensions and consistent values ​​is obtained.

[0035] It should be explained that this step follows three core principles: the principle of dimensionlessness, which completely eliminates the influence of different physical dimensions on the subsequent comprehensive evaluation; the principle of indicator symmetry, which converts all evaluation indicators into positive indicators to ensure the consistency of the comprehensive calculation logic; and the principle of extreme value robustness, which uses historical data quantiles rather than absolute extreme values ​​as the standardization benchmark to avoid individual outliers from causing overall data distortion.

[0036] The core parameters are set as follows: the extreme value range is taken from the 5th and 95th percentiles of historical normal data; the standardized values ​​are uniformly retained to 4 decimal places; the extreme value range is automatically updated every 50 rings of tunneling, and immediately updated in areas of abrupt changes in geological conditions; data exceeding the extreme value range is truncated to 0 or 1. Simultaneously, a segmented standardization mechanism is adopted, using the corresponding extreme value range for different geological sections to further improve the accuracy of data standardization.

[0037] S223. Perform statistical analysis on the standardized monitoring data after unifying the data units and numerical value ranges, obtain the numerical fluctuation law of each monitoring parameter as the tunneling mileage changes, extract the time series change index of each monitoring dimension, and analyze the correlation of monitoring data at different points in combination with the distribution relationship of construction points, and extract spatial correlation indexes. In this embodiment of the invention, the statistical analysis of standardized monitoring data after unifying data units and numerical value ranges to obtain the numerical fluctuation patterns of each monitoring parameter with tunneling mileage, the extraction of time-series change indicators for each monitoring dimension, and the analysis of the correlation between monitoring data at different locations in conjunction with the distribution relationship of construction points to extract spatial correlation indicators include the following steps: S2231. Perform sliding window statistical analysis on the normalized monitoring data to obtain the numerical fluctuation pattern of each monitoring parameter as the tunneling mileage changes, and extract trend change indicators, abrupt change indicators and periodic indicators of each monitoring dimension based on the numerical fluctuation pattern, and integrate them to form a time series change indicator set. Specifically, the extraction method for the time-series change index set is as follows: A fixed-length sliding window is used to traverse the normalized monitoring data segment by segment along the tunneling mileage direction. The mean, slope, and variance of the monitoring parameters within the window are statistically analyzed. The evolution characteristics of the parameters as tunneling progresses are fitted, and the trend change index of continuous rise and fall of the data is extracted. A 3σ anomaly discrimination criterion is introduced to screen abruptly deviating data points from the normal range within the window, identifying abrupt changes such as sudden pressure increases and torque jumps, and generating abrupt change indexes. Spectral analysis is performed on the stable fluctuation data within the window to uncover the periodic oscillation patterns of the parameters, extract periodic fluctuation characteristics, and generate periodic indexes. The three types of indexes are categorized and integrated according to dimensions to construct a complete time-series change index set.

[0038] It should be explained that the indicator extraction follows three basic principles: the principle of trend continuity, which relies on continuous fitting of multiple windows to ensure the authenticity and reliability of trend changes; the principle of abrupt change sensitivity, which prioritizes the identification of data with sudden and large jumps to enhance the ability to capture risk precursors; and the principle of periodic stability, which only identifies data with repeated oscillations and constant fluctuation amplitude as periodic indicators.

[0039] The core parameters are set as follows: the sliding window length is set to 8–15 rings, and the sliding step size is fixed at 2 rings; the confidence level for abrupt change is 99.7%, and the 3σ criterion is strictly implemented; the oscillation frequency range of the periodic index is controlled within 0.05–0.20 Hz. Simultaneously, an adaptive window adjustment mechanism is adopted, automatically reducing the window length in complex geological sections to improve the sensitivity of abrupt change identification and ensure that the time-series indicators closely match the actual tunneling conditions of the shield machine.

[0040] S2232. Organize the spatial coordinate relationships of each construction point in the standardized monitoring dataset, construct a spatial correlation network of points, analyze the Pearson correlation coefficient of monitoring data of spatially adjacent points, extract spatial correlation indicators between different points, and integrate them to form a set of spatial correlation indicators.

[0041] Specifically, the method for extracting the spatial correlation index set is as follows: Based on the spatial coordinate information of each construction point, monitoring points around the cutterhead and within the radial range of tunneling are selected to build a spatial correlation network of shield tunneling construction points; using adjacent points as analysis units, the Pearson correlation coefficient of different monitoring parameter sequences is calculated to quantify the degree of data linkage change between points; combined with the characteristics of soil and rock transmission, three spatial correlation levels are divided into strong correlation, medium correlation, and weak correlation to capture spatial evolution characteristics such as stress transmission and pressure superposition in the strata; the correlation characteristics of different parameters at each point are summarized, classified and integrated to construct a complete spatial correlation index set.

[0042] It should be explained that the index extraction follows three basic principles: the principle of spatial proximity, which prioritizes calculating points that are physically closer to each other to ensure that the spatial transmission logic conforms to the stress law of the strata; the principle of objective correlation, which relies on the Pearson coefficient to quantify the degree of correlation and avoid index deviation caused by subjective judgment; and the principle of adaptability to working conditions, which appropriately expands the correlation range of points in complex strata to enhance the ability to capture spatial coupling features.

[0043] The core parameters are set as follows: the search radius for construction points is set to 1.5–3.0 m; the Pearson correlation coefficient is used to determine the correlation: greater than 0.7 indicates strong correlation, 0.4–0.7 indicates medium correlation, and less than 0.4 indicates weak correlation; the number of nodes included in the spatial network at a time does not exceed 8. A dynamic spatial network update mechanism is also adopted, refreshing the point topology every 5 rings of tunneling to dynamically correct the spatial correlation network and ensure that the spatial correlation indicators match the real-time geological changes.

[0044] S224. Preset the weight allocation of indicators and the reference rules for standard state of earth pressure, assign weights to time-series change indicators and spatially related indicators respectively, and integrate them according to the logic of superimposing the indicator components one by one and complementing the related indicators to obtain a comprehensive evaluation dataset. Specifically, the method for assigning weights to indicators and integrating data is as follows: First, establish a reference rule for the standard state of earth pressure, clarifying the benchmark value ranges of various temporal and spatial indicators under normal tunneling conditions. Second, assign differentiated weights to temporal variation indicators and spatially correlated indicators based on the geological conditions of shield tunneling construction, determining the basic weights for both types of indicators. Third, use an indicator component superposition algorithm to complete the basic data fusion, supplementing and correcting indicators with weak spatial correlation and abnormal temporal fluctuations to compensate for the data deficiencies of single indicators. Fourth, eliminate redundant indicators with low weight and no reference value, unify the indicator data format, and finally integrate to generate a comprehensive earth pressure assessment dataset.

[0045] It should be explained that the integration of indicators follows three basic principles: the principle of weight differentiation, which increases the weight of spatial correlation indicators in complex geology and emphasizes the weight of temporal variation indicators in stable strata; the principle of complementarity and reinforcement, which uses spatial correlation characteristics to correct the lag defects of temporal data and uses temporal fluctuation characteristics to supplement local blind spots in space; and the principle of benchmark alignment, which relies on the earth pressure standard state reference rules to constrain the amplitude of indicators throughout the process to ensure that the integrated data closely matches the actual construction conditions.

[0046] The core parameters are set as follows: the basic weight of the time-series index is set to 0.5–0.7, and the basic weight of the spatial correlation index is set to 0.3–0.5; the weight adjustment step size is 0.05 to ensure the accuracy of weight allocation; the index reinforcement correction range does not exceed ±15%. At the same time, a dynamic fine-tuning mechanism for weights is adopted, and the weights of the two types of indicators are finely adjusted every 10 rings of tunneling based on geological changes to ensure that the index integration results can accurately reflect the true evolution of earth pressure.

[0047] S225. The comprehensive assessment dataset is compared and fitted with the standard state reference rule for earth pressure item by item to calculate the real-time comprehensive state index of earth pressure.

[0048] Specifically, the calculation method for the comprehensive earth pressure state index is as follows: The preset earth pressure standard state reference rules are retrieved to clarify the standard reference intervals for each evaluation index under normal tunneling conditions; an Euclidean distance fitting algorithm is used to compare the time-series and spatial indicators in the comprehensive evaluation dataset with the standard reference intervals item by item, quantifying the degree of deviation of each individual indicator relative to the standard working condition; the individual deviations are weighted and fused together with the aforementioned indicator allocation weights to calculate the overall earth pressure deviation fitting value; the fitting results are normalized and compressed to eliminate calculation biases caused by differences in indicator dimensions, ultimately generating a real-time comprehensive earth pressure state index with a fixed value range that can be compared horizontally.

[0049] It should be explained that the index fitting calculation follows three basic principles: the principle of benchmark solidification, which keeps the standard state reference rules constant within the same geological interval to ensure the uniformity of the evaluation criteria; the principle of deviation quantification, which judges the degree of anomaly by the deviation magnitude rather than the simple numerical value, in line with the gradual evolution characteristics of earth pressure; and the principle of state uniqueness, which outputs only one comprehensive state index at any given time, simplifying the subsequent risk level determination logic.

[0050] The core parameters are set as follows: the comprehensive state index value range is fixed at [0, 1], and the closer the value is to 1, the higher the degree of earth pressure anomaly; the deviation calculation accuracy of individual indexes is retained to 4 decimal places; the Euclidean distance sensitivity coefficient is set to 1.0~1.3; and the index weighted fusion deviation correction threshold is 5%. At the same time, a benchmark dynamic adaptation mechanism is adopted, which automatically updates the standard state reference interval when the geological layer changes, to ensure that the fitting calculation results fit the real stress characteristics of the current stratum.

[0051] S23. Compare the real-time comprehensive state index of earth pressure with the standard state index in the earth pressure risk management benchmark system, and identify the type of earth pressure anomaly and determine the risk management level based on the similarity comparison results. Specifically, the method for anomaly identification and risk level determination is as follows: Samples of four standard state indicators—normal, high, low, and disordered—from the earth pressure risk management benchmark system are retrieved. A cosine similarity algorithm is used to compare the real-time comprehensive earth pressure state indicators with each of the standard samples to quantify the degree of fit between the real-time working conditions and the standard working conditions. The standard sample with the highest similarity is selected as the basis for judgment, distinguishing between three types of earth pressure anomalies: high pressure, low pressure, and disordered pressure fluctuations. A preset similarity grading threshold is used, and four risk management levels are defined based on the deviation of the indicators. The anomaly evolution trend is simultaneously marked, ultimately completing the anomaly type identification and risk level determination.

[0052] It should be explained that the rating determination follows three basic principles: the optimal matching principle, which uses the sample with the highest similarity as the basis for judgment to ensure the accuracy of anomaly classification; the rating progression principle, which increases the risk level step by step as the indicator deviates, and prevents blind rating across levels; and the trend linkage principle, which takes into account both the instantaneous state and the continuous trend of change in the rating process to avoid misjudgment caused by a single fluctuation.

[0053] The core parameters are set as follows: cosine similarity calculation precision is retained to 4 decimal places; the similarity judgment threshold is set as follows: greater than 0.8 for high fit, 0.4-0.8 for moderate deviation, and less than 0.4 for severe anomaly; correspondingly, four risk levels are divided: Level IV, Level III, Level II, and Level I; the anomaly type judgment difference threshold is controlled within ±0.15. A graded anti-shake correction mechanism is also adopted, and the risk level is finally locked only when three consecutive sampling periods determine the same type of anomaly, avoiding misidentification problems caused by instantaneous data jitter.

[0054] S24. Set the mapping relationship between earth pressure anomaly types, risk control levels and control measures, and obtain the corresponding set of alternative control measures from the earth pressure risk control benchmark system.

[0055] Specifically, the method for mapping and screening control measures is as follows: A three-dimensional mapping relationship table is built based on historical construction cases, identifying the routine handling methods corresponding to different abnormal conditions. Combining three abnormal types—high pressure, low pressure, and erratic pressure fluctuations—the following treatment categories are horizontally divided: tunneling parameter adjustment, grouting reinforcement, soil improvement, and monitoring reinforcement. Then, based on the four-level risk control level, the treatment intensity level is vertically divided. Using the currently identified abnormal type and risk level as search tags, suitable control measures are initially screened in the benchmark system. Incompatible measures are eliminated based on construction geological conditions and equipment operating conditions. Treatment plans with construction conflicts or geological incompatibilities are also eliminated. The most suitable control measures are optimized and retained, ultimately generating a standardized set of alternative control measures.

[0056] It should be explained that the measure mapping follows three basic principles: the principle of hierarchical adaptation, where low-risk areas adopt fine-tuning measures and high-risk areas adopt strong intervention measures to avoid over-control or under-control; the principle of scenario exclusivity, where conflicting control measures should not be matched within the same working condition to ensure smooth construction logic; and the principle of safety priority, where reinforcement measures are prioritized for water-rich, soft, and dangerous strata to enhance construction safety.

[0057] The core parameters are set as follows: the three-dimensional mapping table has no fewer than 28 built-in control measures; the single-scenario alternative measure set retains 3 to 6 optimal solutions; low-risk measures prioritize adjusting tunneling parameters, and high-risk measures are forcibly matched with stratum reinforcement solutions; the threshold for determining conflict of measures is a working condition overlap of ≥60%.

[0058] S3. Based on the risk control level and the type of earth pressure anomaly, construct a multi-objective optimization decision-making model for earth pressure risk, and input the alternative control measures and risk control priority weight vector into the model to calculate the cost-risk balance coefficient of each alternative control measure. In this embodiment of the invention, the step of constructing a multi-objective optimization decision-making model for earth pressure risk based on risk control level and earth pressure anomaly type, and inputting alternative control measures and risk control priority weight vectors into the model to calculate the cost-risk balance coefficient of each alternative control measure includes the following steps: S31. Set three core optimization objectives: cost control, risk reduction, and construction progress assurance, and define the initial constraint range for each core optimization objective; Specifically, the method for setting initial constraints for multiple objectives is as follows: Based on the needs of shield tunneling construction site management, three core optimization objectives are established: cost control, risk reduction, and construction progress assurance. The cost control objective focuses on construction material consumption, equipment energy consumption, and labor management costs; the risk reduction objective focuses on the abnormal earth pressure rate, the degree of ground disturbance, and construction safety hazards; and the construction progress assurance objective focuses on the number of tunneling loops per day, the duration of process connections, and downtime. Using historical construction data under normal geological conditions, the normal fluctuation range of the three objectives is statistically analyzed, and the initial constraint thresholds for each optimization objective are calibrated. The constraints are then transformed into quantitative boundaries, establishing a set of initial constraint equations for multiple objectives, thus completing the initial range setting for the optimization objectives and providing a fixed benchmark for subsequent model solving.

[0059] It should be explained that the optimization target setting follows three basic principles: the principle of safety priority, where the risk reduction target has the most stringent constraints, prioritizing construction safety; the principle of mutual checks and balances, where the three major targets are mutually exclusive, taking into account cost, safety and schedule, and avoiding extreme optimization of a single indicator; and the principle of universal benchmarks, where the initial constraint range is adapted to conventional strata, and can be dynamically fine-tuned later for complex strata.

[0060] The core parameters are set as follows: the initial constraint for cost control is an increase in overall construction cost ≤ 12%; the initial constraint for risk reduction is an abnormal earth pressure occurrence rate ≤ 5% and ground disturbance settlement ≤ 8mm; the initial constraint for construction progress is an average of ≥ 6 rings excavated per day; all constraint boundaries are retained to two decimal places. Simultaneously, a constraint boundary self-checking mechanism is adopted, reviewing the target achievement rate every 20 rings excavated and automatically determining the adaptability of the initial constraints.

[0061] S32. Preset optimization and adjustment rules, and adjust the constraint boundaries and relative weights of the core optimization objectives according to the type of earth pressure anomaly and the risk control level, and construct a multi-objective optimization decision-making model for earth pressure risk. In this embodiment of the invention, the preset optimization adjustment rules, and the adjustment of the constraint boundaries and relative weights of the core optimization objectives based on the type of earth pressure anomaly and the risk control level, to construct a multi-objective optimization decision model for earth pressure risk, include the following steps: S321. Pre-set multi-objective optimization and adjustment rules to classify construction control scenarios into different types of earth pressure anomalies and different risk control levels. Specifically, the classification method for construction control scenarios is as follows: A two-dimensional scenario judgment matrix is ​​constructed using pre-defined multi-objective optimization and adjustment rules, with soil pressure anomaly type and risk control level as the two dimensions. Three types of anomalies—high pressure, low pressure, and erratic pressure fluctuations—are used as the horizontal classification criteria, and risk control levels I to IV are used as the vertical classification criteria. Construction control scenarios are divided into four categories—ordinary control, routine control, enhanced control, and emergency control—based on the degree of ground disturbance, construction hazard coefficient, and urgency of response. The geological sensitivity, construction restrictions, and response priorities corresponding to each scenario are clearly defined, and the scenario judgment boundaries and classification standards are solidified to form a standardized scenario classification system, providing a basis for subsequent dynamic adjustment of model constraint boundaries and weight parameters.

[0062] It should be explained that the scenario classification follows three basic principles: the two-dimensional coupling principle, which combines the anomaly type and risk level to comprehensively determine the scenario, avoiding the one-sidedness of a single dimension; the principle of increasing danger, which increases the level of scenario control with the risk level, and compresses and optimizes the adjustable space for high-risk scenarios; and the principle of type adaptation, which prioritizes increasing safety constraints for scenarios with higher pressure under the same risk, and tightens fluctuation constraints for scenarios with disordered pressure.

[0063] The core parameters are set as follows: Level IV risk corresponds to ordinary control scenarios, Level III risk corresponds to routine control scenarios, Level II risk corresponds to enhanced control scenarios, and Level I risk corresponds to emergency control scenarios; the scenario judgment matrix contains a total of 12 working condition combinations; the scenario switching judgment latency does not exceed 2 sampling periods. Simultaneously, a scenario anti-jitter judgment mechanism is adopted, ensuring that the same control scenario is continuously judged before a level switch is completed, preventing frequent scenario jumps caused by instantaneous oscillations in earth pressure and guaranteeing the stability of model optimization.

[0064] S322. Adjust the constraint boundaries of the core optimization objectives according to the construction control scenario classification standards and the types of abnormal earth pressure. Specifically, the method for adjusting the target constraint boundaries is as follows: Based on the aforementioned construction control scenario classification standard, the constraint boundaries of the three core optimization targets are differentiated for three anomaly types: high pressure, low pressure, and disordered pressure fluctuations. For high pressure anomalies, the risk of formation instability is high, so the risk reduction target constraint boundaries are tightened to strictly control the range of formation subsidence and disturbance, and the cost control constraints are appropriately relaxed. For low pressure anomalies, the degree of construction risk is low, so the construction progress guarantee constraint boundaries are relaxed to improve tunneling efficiency, and the cost control boundaries are slightly tightened. For disordered pressure fluctuation anomalies, the working condition is unstable, so the cost fluctuation and risk disturbance constraint boundaries are tightened simultaneously to limit large parameter fluctuations. The initial constraint range is scaled and corrected by preset adjustment coefficients to generate differentiated constraint boundaries adapted to different anomaly types.

[0065] It should be explained that the adjustment of constraint boundaries follows three basic principles: the anomaly adaptation principle, which tightens or relaxes constraints according to the degree of anomaly severity to align with the engineering handling logic; the boundary rigidity principle, which ensures that the constraint boundaries of high-risk anomalies cannot be relaxed, thus guaranteeing construction safety; and the interval independence principle, which ensures that the constraint intervals corresponding to different anomaly types do not overlap, thus avoiding confusion in model solutions.

[0066] The core parameters are set as follows: Scaling factor for high pressure risk constraint: 0.6–0.8; Scaling factor for low pressure schedule constraint: 1.1–1.3; Scaling factor for comprehensive fluctuation and disorder constraint: 0.9–1.1; Constraint boundary correction accuracy is retained to two decimal places. A boundary locking mechanism is also adopted, fixing the constraint boundaries during abnormal working conditions and prohibiting arbitrary changes until the earth pressure index returns to the normal range, ensuring the stability of the model's optimization solution.

[0067] S323. Adjust the relative weight of core optimization objectives according to the construction management scenario classification standard and the risk management level. Specifically, the method for adjusting the weight ratio of the optimized objectives is as follows: Based on the preset four-level risk control levels, the weights of the three core objectives—cost control, risk reduction, and construction progress assurance—are redistributed in a differentiated manner; for high-risk scenarios at levels I and II, with construction safety as the core orientation, the weight of the risk reduction objective is significantly increased, while the weight ratios of construction progress and cost control are weakened, prioritizing the avoidance of geological instability risks; for level III conventional risk scenarios, a balanced allocation model is adopted, taking into account safety, cost, and tunneling progress to achieve multi-dimensional balanced control; for level IV low-risk scenarios, the safety weight restrictions are relaxed, the weights of cost control and construction progress are increased, and construction and maintenance costs are reduced; based on the fixed weight adjustment step size, the weight allocation is finely adjusted level by level, and after completion, normalization is performed to ensure that the sum of the weights of the three objectives is always 1, generating an optimized weight combination suitable for the current risk level.

[0068] It should be explained that the weight adjustment follows three basic principles: the principle of grade progression, where the weight ratio transitions linearly and smoothly with the risk level without abrupt changes; the principle of safety net, where the target weight for risk reduction at any risk level must not be lower than 0.2 to maintain a safety baseline; and the principle of unique allocation, where a fixed and unique weight allocation is used for the same risk level to ensure consistency in model solutions.

[0069] The core parameters are set as follows: Level I risk reduction weight ≥ 0.7; Level II risk reduction weight 0.5–0.6; Level III target weights maintained in a balanced range of 0.3–0.4; Level IV cost control weight ≥ 0.5; minimum weight adjustment step is 0.05. A weight stabilization mechanism is also adopted, ensuring that the weights remain fixed within a single risk assessment period, without changing the allocation based on small fluctuations in instantaneous data, thus guaranteeing the stability of the decision model output.

[0070] S324. Set the adjusted constraint boundary as the model solution constraint condition, and embed the adjusted relative weight ratio to build a multi-objective optimization decision model for earth pressure risk.

[0071] Specifically, the method for building the multi-objective optimization decision-making model is as follows: the previously adjusted constraint boundary is transformed into inequality constraints, the legal feasible region for model solution is defined, and invalid solution schemes that exceed construction limits or violate geological conditions are eliminated; normalized relative weights of optimization objectives are embedded, and weighted values ​​are assigned to each sub-objective function to distinguish the optimization emphasis under different control scenarios; cost loss sub-functions, risk disturbance sub-functions, and schedule delay sub-functions are constructed separately and integrated into a comprehensive objective optimization function through weighted superposition; a genetic algorithm is selected as the model solving algorithm, and algorithm iteration rules and convergence judgment conditions are set to search for the comprehensive optimal solution within the feasible region; the constraints, weight ratios, objective functions, and solving algorithms are solidified to complete the construction of the multi-objective optimization decision-making model for earth pressure risk.

[0072] It should be explained that the model construction follows three basic principles: the principle of hard boundary constraints, which means that all solution results must not exceed the constraint boundaries and prevent illegal construction schemes; the principle of dynamic coupling of weights, which means that weights directly participate in the function weighting operation and accurately map the scenario control bias; and the principle of comprehensive optimization, which means that the model does not take the extreme value of a single indicator as the optimization goal, but pursues the optimal solution that comprehensively balances cost, risk and schedule.

[0073] The core parameters are set as follows: the feasible region solved by the model strictly fits the corrected constraint boundary; the genetic algorithm iterates 150–400 times, with a convergence accuracy set to 1e-5; the penalty coefficient for the sub-objective function is 0.1–0.3; and the calculation result of the comprehensive optimization function is retained to four decimal places. Simultaneously, a scheme filtering mechanism is embedded to automatically remove extreme solutions with abrupt parameter changes or those that do not conform to construction specifications, further improving the engineering practicality of the model's output control scheme.

[0074] S33. Input the weight vectors of each alternative control measure and risk control priority into the multi-objective optimization decision-making model for earth pressure risk, calculate the multi-objective comprehensive score of each alternative control measure, and use it as the cost-risk balance coefficient corresponding to the control measure.

[0075] Specifically, the calculation method for the cost-risk balance coefficient is as follows: the selected set of alternative control measures and the risk control priority weight vector are simultaneously imported into the multi-objective optimization decision model; for each alternative measure, the cost increment, formation disturbance magnitude, and schedule delay duration after the measure is implemented are calculated, and the three categories of cost, risk, and schedule are mapped to obtain three individual quantitative scores; the individual scores are weighted and integrated with the priority weight vector to eliminate the difference in target dimensions, and the multi-objective comprehensive score of the control measures is calculated; this comprehensive score is defined as the cost-risk balance coefficient, and the lower the coefficient value, the better the comprehensive control effect of the measures, thus completing the coefficient quantification calculation for all alternative measures.

[0076] It should be explained that the coefficient calculation follows three basic principles: the principle of objective quantification, in which all evaluation indicators are calculated based on actual construction measurement data to avoid bias from subjective scoring by humans; the principle of differentiated evaluation, in which different control measures are calculated using the same caliber to ensure fairness in horizontal comparison; and the principle of balanced optimization, which takes into account safety, cost and schedule, and does not unilaterally pursue the optimality of a single indicator.

[0077] The core parameters are set as follows: the balance coefficient is fixed in the range of [0, 1] and retained to four decimal places; the priority weight vector is normalized when participating in the weighted calculation; the risk disturbance penalty coefficient for high-risk working conditions is increased by 0.2 to 0.4; and the full score for each individual indicator is 100 points. At the same time, a coefficient correction mechanism is adopted to apply an additional attenuation coefficient to measures with high construction difficulty and poor geological adaptability, further conforming to the actual on-site construction conditions and providing a precise quantitative basis for the subsequent priority ranking of measures.

[0078] S4. Prioritize each alternative control measure according to the cost-risk balance coefficient, generate graded early warning signals and executable control decision plans, and collect decision execution effect data simultaneously. In this embodiment of the invention, the steps of prioritizing the alternative control measures based on the cost-risk balance coefficient, generating tiered early warning signals and executable control decision schemes, and simultaneously collecting decision execution effect data include the following steps: S41. Establish a four-level early warning threshold system for earth pressure risk and rules for generating control decision-making schemes, and clarify the signal types, release scope, response time limits and responsible entities corresponding to different early warning levels; Specifically, the method for setting up the early warning system and decision-making rules is as follows: Based on the risk control level classification standards mentioned above, a four-level earth pressure risk early warning threshold system (blue, yellow, orange, and red) is established, and critical judgment thresholds are defined for different risk levels; exclusive signal types are assigned to each level of early warning, and the information release scope is differentiated according to the degree of risk hazard, distinguishing the release levels of on-site work teams, project departments, and general contractor management; graded response time limits are set according to the urgency of the work situation, and the corresponding responsible entities for construction operators, technical control personnel, and safety management personnel are simultaneously clarified; the rules for generating control decision-making plans are solidified, binding early warning levels with handling procedures, reporting procedures, and review procedures to form a standardized early warning response chain, providing a standardized basis for subsequent early warning pushes and plan generation.

[0079] It should be explained that the early warning system is set up according to three basic principles: the principle of progressive levels, where the early warning threshold increases with the risk level and there are no overlapping or blank intervals between levels; the principle of matching rights and responsibilities, where higher-level early warnings correspond to higher-level responsible entities, and the principle of timeliness control, where the higher the risk level, the shorter the response time, and the shorter the time required to handle high-risk situations.

[0080] The core parameters are set as follows: Level IV blue alerts are broadcast on-site only, with a response time of ≤30 minutes; Level III yellow alerts are pushed to the project department, with a response time of ≤15 minutes; Level II orange alerts are simultaneously reported to the management department, with a response time of ≤8 minutes; Level I red alerts are pushed urgently across the entire region, with a response time of ≤3 minutes. A warning anti-shake mechanism is also employed, requiring three consecutive sets of stable sampling data to trigger a threshold before an alert is issued, preventing false alarms caused by instantaneous data fluctuations and ensuring the reliability of the warning system.

[0081] S42. Prioritize each alternative control measure according to the cost-risk balance coefficient from largest to smallest, and match the ranking results with the four-level early warning threshold system for earth pressure risk to generate a graded early warning signal corresponding to the risk control level. Specifically, the method for prioritizing measures and generating graded early warnings is as follows: The cost-risk balance coefficients corresponding to each alternative control measure are retrieved, and priority is assigned in descending order; a larger coefficient indicates better overall suitability. Combining this with a four-level early warning threshold system, the ranking results are bidirectionally matched with real-time risk control levels. High-risk levels are linked to higher-priority control measures, while low-risk levels are matched with lower-priority conventional control measures. Based on preset early warning generation rules, and combined with the current type of soil pressure anomaly and the state of ground disturbance, four-level graded early warning signals (blue, yellow, orange, and red) are compiled and generated. The early warning signals simultaneously carry a measure ranking table, anomaly causes, and handling suggestions, completing standardized early warning output.

[0082] It should be explained that the sorting and early warning generation follow three basic principles: the principle of optimal policy allocation, which prioritizes the allocation of the best control measures to high-risk working conditions to minimize construction hazards; the principle of hierarchical isolation, which strictly distinguishes the push links of different levels of early warning signals to avoid signal confusion; and the principle of fixed order, which fixes the sorting results of measures within a single early warning cycle and does not allow temporary changes to the order of action.

[0083] The core parameters are set as follows: the balance coefficient sorting precision is retained to four decimal places; the sorting interval for measures within the same level is not less than 0.02; the top two optimal measures are locked for red alerts, the top three measures are selected for orange alerts, and the measures for yellow and blue alerts are matched sequentially; the alert signal generation delay is ≤1 sampling period. A sorting locking mechanism is also adopted to fix the sorting results of measures until the alert is lifted, preventing frequent re-sorting from causing command confusion and ensuring a smooth and stable on-site handling process.

[0084] S43. Analyze the signal type, release scope, response time limit and responsible entity in the graded early warning signal, and generate an executable control decision plan in combination with the control decision plan generation rules; In this embodiment of the invention, the process of analyzing the signal type, release scope, response time limit, and responsible entity in the graded early warning signal, and generating an executable control decision scheme in conjunction with the control decision scheme generation rules, includes the following steps: S431. Pre-set a hierarchical control measure template library, analyze and extract three core control parameters from the hierarchical early warning signals: signal type, release scope, and response time limit, and match the corresponding level of standardized handling process template from the control measure template library according to the risk control level. Specifically, the method for matching the handling process templates is as follows: A tiered control measure template library adapted to the four-level early warning system is pre-built, categorized into four types of standardized process templates based on risk strength: routine fine-tuning, standard handling, enhanced control, and emergency rescue. The templates contain basic content such as tunneling parameters, operating procedures, and construction specifications. The tiered early warning signals are structurally analyzed to accurately extract three core control parameters: signal type, release range, and response time limit, clarifying the control level and response requirements for this risk handling. Based on the real-time risk control level, a one-to-one matching rule is used to retrieve the corresponding handling template, and the template parameters are slightly adapted and modified according to the current geological conditions and tunnel boring machine operating conditions. Redundant procedures that conflict with the on-site conditions are eliminated, and a highly adaptable and standardized handling process is generated.

[0085] It should be explained that template matching follows three basic principles: the principle of hierarchical exclusivity, where different risk levels correspond to exclusive templates, and cross-level mixed handling procedures are prohibited; the principle of parameter traceability, where the core parameters of the early warning signal are used as the basis for matching judgment to ensure that the control intensity is consistent with the early warning level; and the principle of flexible adaptation, where the basic template structure is kept unchanged, and only local construction parameters are slightly adjusted to balance standardization and on-site adaptability.

[0086] The core parameters are set as follows: the template library contains no fewer than 16 process templates, with 4 sets of differentiated work condition templates configured for each of the four levels; the template parameter correction range does not exceed 15%; the process matching completion delay is ≤2 sampling cycles; the high-level emergency template includes emergency shutdown reporting procedures. Simultaneously, a template iteration mechanism is adopted, recording the process implementation effect after each handling, and optimizing the template procedure parameters every 100 tunneling cycles to continuously improve the fit of the handling templates.

[0087] S432. Embed the three types of core control parameters into the standardized handling process template, supplement and clarify the person in charge of handling, the equipment allocation list, the material demand plan and the key points of on-site operation, and standardize the format according to the control decision plan generation rules to generate an executable control decision plan.

[0088] Specifically, the method for generating control and management decision-making plans is as follows: Three core control parameters—signal type, release scope, and response time limit—obtained from the analysis of early warning signals are embedded into the handling process template to limit the control intensity and execution time of this handling action; the responsibilities of each level of personnel in the construction team are clearly defined, and the responsibilities of technical, safety, and operational positions are divided; a list of equipment to be allocated, such as shield tunneling adjustment equipment and monitoring instruments, is matched according to the handling process, and the required quantities of grouting materials and improved slag are simultaneously verified; key on-site operational points such as construction speed limits, grouting pressure, and adjustment range are marked based on the geological characteristics; following the preset control and management decision-making plan generation rules, the sequence of procedures, text format, and execution standards are uniformly standardized, redundant information is eliminated, and finally, a clear, directly implementable, and executable control and management decision-making plan is generated.

[0089] It should be explained that the plan was developed following three basic principles: the principle of parameter constraint, which uses the core parameters of the early warning to lock in the handling standards and ensure that the control intensity matches the risk level; the principle of clear authority and responsibility, which assigns specific positions and personnel to avoid gaps in responsibility during the handling process; and the principle of practicality, which simplifies professional terminology, clarifies operational limits, and adapts to the direct execution of on-site construction personnel.

[0090] The core parameters are set as follows: the time to complete the plan preparation is ≤5 minutes; the responsibility for handling the situation by a single person is accurately assigned to the work team and position; the error in the equipment allocation list does not exceed 1 unit; and a 10% reserve is reserved for planned material reserves. At the same time, a plan verification mechanism is adopted to automatically check whether the parameters in the plan exceed the construction safety threshold, and to provide pop-up prompts for correction of unreasonable operating parameters.

[0091] S44. Distribute the graded early warning signals and executable control decision schemes to the construction control terminal for execution, and simultaneously collect data on earth pressure changes, construction progress, and resource consumption during the decision execution process, and integrate them to form decision execution effect data.

[0092] Specifically, the methods for issuing the plan and collecting data are as follows: The generated tiered early warning signals and standardized control decision-making plans are pushed to the on-site operation terminals and project management terminals according to the preset release scope, and tiered execution instructions are issued; Construction personnel strictly follow the operational points in the plan to complete control operations such as adjusting tunneling parameters and reinforcing the strata, and the background control system monitors the on-site execution process in real time; During the plan execution period, high-frequency data on real-time fluctuations in earth pressure, progress data of tunneling ring numbers, and data on the consumption of consumable materials, energy, and resources are collected to fully record the changes in working conditions throughout the entire disposal process; The multi-source collected data is cleaned, deduplicated, classified, and integrated, and data on abnormal equipment interference is removed to generate a dataset that can intuitively reflect the quality of the disposal and the effectiveness of the decision-making execution.

[0093] It should be explained that the issuance and collection follow three basic principles: the principle of hierarchical push, with high-level warnings being issued at encrypted levels to ensure that instructions quickly reach management personnel; the principle of full traceability, with a complete record kept from the issuance of instructions to the completion of the work to achieve traceability of the construction process; and the principle of truthfulness and objectivity, with the collected data not being tampered with and truthfully reflecting the implementation effect of the management plan.

[0094] The core parameters are set as follows: command issuance and transmission latency ≤ 2s; earth pressure sampling frequency increased to 2Hz during the process; progress data recorded every 5 minutes, and consumable consumption data statistically analyzed on a single-loop basis; the execution effect dataset retains complete timestamps. Simultaneously, a data linkage archiving mechanism is adopted. After a single control process is completed, the early warning signal, decision plan, and execution data are bound and archived, providing reliable engineering sample data for subsequent model iterations and work condition reviews.

[0095] S5. Based on the decision-making execution effect data, dynamically iterate and update the earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision-making model.

[0096] In this embodiment of the invention, the dynamic iterative update of the earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision model based on decision execution effect data includes the following steps: S51. Pre-set the construction control effect evaluation index system and evaluation judgment threshold, and extract quantitative evaluation indicators from the decision execution effect data; Specifically, the evaluation system and indicator extraction methods are as follows: A construction control effectiveness evaluation indicator system is established by combining three major control dimensions: construction safety, economic cost, and tunneling progress. For the safety dimension, earth pressure fluctuation amplitude and ground settlement deformation are selected as core indicators; for the progress dimension, delay time and process recovery efficiency are selected as evaluation indicators; and for the cost dimension, incremental material consumption and energy consumption increase are selected as consumption indicators. Based on historical project samples and construction specifications, thresholds for judging the merits of each indicator are preset, defining three levels of evaluation: excellent, qualified, and unqualified. Valid monitoring samples are screened from the data on the execution effects of prior decisions, and abnormal, dirty data caused by equipment malfunctions and signal interference is cleaned and removed. The screened compliant data is converted and processed to unify the indicator dimensions, extracting standardized quantitative indicators that can be directly used for evaluation, thus completing the pre-evaluation preparation work.

[0097] It should be explained that the indicator evaluation follows three basic principles: the principle of full coverage, which covers safety, progress, and cost to comprehensively reflect the quality of management and control; the principle of threshold adaptation, which sets different judgment thresholds for conventional and complex strata; and the principle of objective quantification, which generates all indicators based on measured data without any subjective human intervention.

[0098] The core parameters are set as follows: earth pressure fluctuation threshold ≤ 8%, ground settlement threshold ≤ 6mm; handling delay time threshold ≤ 12min; material energy consumption increase threshold ≤ 10%; all assessment indicators are uniformly normalized to the [0, 1] range. Simultaneously, a dynamic threshold calibration mechanism is adopted. After each control and handling review, the indicator judgment thresholds are fine-tuned to continuously optimize the adaptability of the assessment system and improve the accuracy of subsequent assessments.

[0099] S52. Compare and analyze the quantitative evaluation indicators with the evaluation judgment thresholds to determine the degree of adaptability of the earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision-making model to the construction scenario. In this embodiment of the invention, the step of comparing and analyzing the quantitative evaluation indicators with the evaluation judgment threshold to determine the adaptability of the earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision-making model to the construction scenario includes the following steps: S521. Preset indicator comparison rules and adaptation level grading standards, compare the quantitative evaluation indicators with the evaluation judgment thresholds one by one, and obtain the set of indicator compliance status. Specifically, the method for indicator comparison and situation aggregation is as follows: Standardized indicator comparison rules are pre-defined to distinguish between positive benefit indicators and negative loss indicators, clarifying the judgment logic and deviation calculation method for different indicators; the extracted three categories of quantitative evaluation indicators—safety, progress, and cost—are compared one by one against the preset evaluation judgment thresholds to calculate the actual deviation of each indicator; based on the adaptation degree grading standard, three judgment levels are divided: excellent, qualified, and abnormal, with the compliance status, deviation direction, and deviation magnitude of each indicator marked; the comparison results are categorized and stored one by one, invalid and duplicate data are removed, and an indicator compliance situation set containing all evaluation dimensions is generated, providing the original judgment basis for subsequent comprehensive rating.

[0100] It should be explained that the indicator comparison follows three basic principles: the principle of logical differentiation, which differentiates positive and negative indicators to prevent the results from being distorted due to uniform evaluation logic; the principle of deviation quantification, which avoids making a rough judgment based solely on whether an indicator is qualified or not, and simultaneously retains the deviation range for detailed rating; and the principle of independent comparison, which ensures that each indicator does not interfere with each other and is judged separately to guarantee the objectivity of the judgment of each indicator.

[0101] The core parameters are set as follows: the accuracy of the indicator deviation calculation is retained to four decimal places; a deviation of 0-5% is judged as excellent, 5%-15% as qualified, and greater than 15% as abnormal; critically exceeding indicators are marked with a separate yellow warning label; the time for a single comparison does not exceed 1 minute. At the same time, a comparison and error correction mechanism is adopted to automatically correct the temporary comparison benchmark for sudden geological changes, weaken the interference of extreme geological conditions on indicator judgment, and ensure that the comparison results are consistent with the actual control level.

[0102] S522. Based on the set of indicator compliance, analyze the overall adaptability of the earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision-making model to the construction scenario, and determine the degree of adaptability based on the overall adaptability.

[0103] Specifically, the overall adaptability analysis and judgment method is as follows: integrate the comparison results of all individual indicators to determine the compliance status of indicators, statistically analyze the overall compliance rate of indicators across all dimensions of safety control, construction progress, and resource costs, and summarize the actual implementation effectiveness of the control plan; combine the handling performance under different geological conditions and risk levels to comprehensively assess the rationality of the threshold setting of the earth pressure risk control benchmark system, as well as the degree of fit between the weight allocation and constraint adjustment logic of the multi-objective optimization decision-making model and the actual construction scenario on site; identify the working scenarios where indicators frequently exceed the standards and the control effect is poor, and accurately locate the compatibility shortcomings between the system standards and the optimization model; according to the preset overall adaptability level grading standard, combined with the comprehensive compliance rate and the actual control effect, uniformly divide the overall adaptability level and complete the overall adaptability level judgment.

[0104] It should be explained that the suitability assessment follows three basic principles: the principle of comprehensive planning, which integrates multi-dimensional evaluation data to conduct overall judgment and avoids one-sided evaluation based on a single indicator; the principle of prioritizing weaknesses, which focuses on high-frequency abnormal indicators and prioritizes the investigation of core suitability issues between the system and the model; and the principle of scenario differentiation, which separately assesses the suitability effect of stable strata and complex and weak strata to improve the accuracy of the assessment.

[0105] The core parameters are set as follows: the overall fit range is [0, 1], with higher values ​​indicating better fit; a comprehensive compliance rate of 90% or above is defined as high fit, 70% to 90% as medium fit, and below 70% as low fit; a list of deviation indicators is retained concurrently with the fit assessment; the completion time for a single fit assessment does not exceed one data statistical cycle. Simultaneously, a fit ledger recording mechanism is established to archive and retain each assessment result, providing reliable data for subsequent benchmark system correction and model parameter optimization iterations.

[0106] S53. Based on the results of the adaptation degree judgment, correct the standard state indicators and mapping correlations within the earth pressure risk management benchmark system, and simultaneously optimize the constraint boundary and weight allocation parameters of the multi-objective optimization decision model for earth pressure risk. Specifically, the iterative optimization method for the system and model is as follows: The results of the adaptation level determination are retrieved, and a graded correction strategy is adopted to complete the parameter iterative optimization; for highly adapted working conditions, only minor adjustments are made to the standard state indicators with slight deviations within the benchmark system, and the model weight allocation is slightly optimized; for moderately adapted working conditions, the mapping relationship between anomaly types and control measures is corrected, and the model constraint boundaries under conventional strata are optimized; for low-adapted working conditions, the benchmark determination thresholds for sensitive strata are reconstructed on a large scale, and the multi-objective weight allocation ratio is recalculated to compensate for the model's adaptation shortcomings; combined with historical out-of-standard indicators, the causes of deviations are traced, and the normal fluctuation range of earth pressure and the critical value for risk determination are corrected in a targeted manner, while simultaneously updating the model scaling factor and adjustment step size to ensure that the corrected benchmark system and decision model conform to the actual construction conditions, forming a closed-loop optimization process.

[0107] It should be explained that the iterative correction follows three basic principles: the principle of hierarchical optimization, which controls the correction range according to the adaptation level to avoid excessive changes that could damage the original stable parameters; the principle of source tracing and rectification, which targets high-frequency out-of-range points for targeted optimization to accurately address system shortcomings; and the principle of smooth transition, which continuously fine-tunes all parameters to prevent numerical jumps from affecting the stability of model operations.

[0108] The core parameters are set as follows: the correction range for high-degree adaptation is ≤5%, the correction range for medium-degree adaptation is controlled between 5% and 15%, and the correction range for low-degree adaptation is 15% to 25%; the minimum adjustment step size for weights is maintained at 0.05; the corrected parameters are locked for at least 5 tunneling rings to prevent frequent changes.

[0109] S54. The revised earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision-making model are solidified, stored and dynamically updated.

[0110] Specifically, the method for storing and iterating the model and system is as follows: all optimized parameters, including the standard state indicators, measure mapping relationships, constraint boundaries, and weight ratios, that have been corrected in this revision, are entered into the construction database according to a classified archiving format; a dual-solid storage mode of local terminal + cloud server is adopted to retain the running version and backup version respectively to prevent parameter loss or accidental tampering; static basic parameters and dynamic optimization parameters are divided, with static parameters remaining fixed in the long term and dynamic optimization parameters having reserved iterative update interfaces; automatic iteration trigger conditions are preset, and the system automatically starts the parameter update process when the stratum changes, the cumulative tunneling reaches a specified number of rings, or the adaptation level is continuously determined to be medium or below; old and failed working condition samples in the database are cleaned up simultaneously, high-quality training data is retained, the model's generalization ability is continuously strengthened, and finally, the long-term dynamic iteration of the control system and decision-making model is achieved.

[0111] It should be explained that the storage iteration follows three basic principles: the dual backup security principle, which ensures parameter security through dual storage and avoids data loss due to the failure of a single storage medium; the dynamic and static separation principle, which fixes the basic underlying parameters and optimizes only the dynamic calculation parameters to ensure the stability of the underlying logic of the model; and the low-speed iteration principle, which uses a smooth replacement method for parameter updates and prohibits instantaneous forced refreshes to ensure the continuous operation of the system during construction.

[0112] The core parameters are set as follows: the database is divided into three categories: basic database, runtime database, and archive database; the cloud backup refresh cycle is 24 hours; the automatic iteration trigger condition is a cumulative tunneling of 50 rings or 3 consecutive low-degree adaptations; the historical data retention period is no less than 180 days. It also features an access control encryption mechanism, allowing only technical management personnel to manually modify core parameters; all other states are automatically iterated by the system, ensuring the intelligent, closed-loop, and long-term operation of the entire earth pressure control system.

[0113] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for early warning and decision-making regarding earth pressure risk in subway shield tunneling, characterized in that, Includes the following steps: S1. Obtain geological survey data, shield tunneling equipment parameters, and historical earth pressure risk management case data for the area to be constructed, construct an earth pressure risk management benchmark system, and generate a risk management priority weight vector. S2. During the tunnel boring machine excavation process, real-time soil pressure parameters, equipment operation data and underground environment monitoring data are collected. After data preprocessing, a standardized monitoring dataset is obtained. A multi-dimensional comprehensive evaluation of the standardized monitoring dataset is carried out to obtain the real-time comprehensive soil pressure status index. Then, it is benchmarked against the soil pressure risk control benchmark system to identify the types of soil pressure anomalies, determine the risk control level, and match alternative control measures. S3. Based on the risk control level and the type of earth pressure anomaly, construct a multi-objective optimization decision-making model for earth pressure risk, and input the alternative control measures and risk control priority weight vector into the model to calculate the cost-risk balance coefficient of each alternative control measure. S4. Prioritize each alternative control measure according to the cost-risk balance coefficient, generate graded early warning signals and executable control decision plans, and collect decision execution effect data simultaneously. S5. Based on the decision-making execution effect data, dynamically iterate and update the earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision-making model. The process of collecting earth pressure parameters, equipment operation data, and underground environment monitoring data in real time during shield tunneling, preprocessing the data to obtain a standardized monitoring dataset, and conducting a multi-dimensional comprehensive evaluation of the standardized monitoring dataset to obtain a real-time comprehensive earth pressure status index, then benchmarking it against the earth pressure risk management benchmark system to identify the type of earth pressure anomaly, determine the risk management level, and match alternative management measures includes the following steps: S21. Preset spatiotemporal synchronous sampling rules, collect soil pressure parameters, equipment operation data and underground environment monitoring data in real time, and perform outlier removal and missing value completion to obtain a standardized monitoring dataset; S22. Perform spatiotemporal registration and dimensional normalization on the standardized monitoring dataset, extract the temporal change indicators and spatial correlation indicators of each monitoring dimension, and obtain the real-time comprehensive earth pressure state index through multi-dimensional weighted comprehensive calculation. S23. Compare the real-time comprehensive state index of earth pressure with the standard state index in the earth pressure risk management benchmark system, and identify the type of earth pressure anomaly and determine the risk management level based on the similarity comparison results. S24. Set up the mapping relationship between earth pressure anomaly types, risk control levels and control measures, and obtain the corresponding set of alternative control measures from the earth pressure risk control benchmark system. The process of constructing a multi-objective optimization decision-making model for earth pressure risk based on risk management level and earth pressure anomaly type, and inputting alternative management measures and risk management priority weight vectors into the model to calculate the cost-risk balance coefficient of each alternative management measure includes the following steps: S31. Set three core optimization objectives: cost control, risk reduction, and construction progress assurance, and define the initial constraint range for each core optimization objective; S32. Preset optimization and adjustment rules, and adjust the constraint boundaries and relative weights of the core optimization objectives according to the type of earth pressure anomaly and the risk control level, and construct a multi-objective optimization decision-making model for earth pressure risk. S33. Input the weight vectors of each alternative control measure and risk control priority into the multi-objective optimization decision-making model for earth pressure risk, calculate the multi-objective comprehensive score of each alternative control measure, and use it as the cost-risk balance coefficient corresponding to the control measure. The process of prioritizing alternative control measures based on the cost-risk balance coefficient, generating tiered early warning signals and executable control decision schemes, and simultaneously collecting data on the effectiveness of decision implementation includes the following steps: S41. Establish a four-level early warning threshold system for earth pressure risk and rules for generating control decision-making schemes, and clarify the signal types, release scope, response time limits and responsible entities corresponding to different early warning levels; S42. Prioritize each alternative control measure according to the cost-risk balance coefficient from largest to smallest, and match the ranking results with the four-level early warning threshold system for earth pressure risk to generate a graded early warning signal corresponding to the risk control level. S43. Analyze the signal type, release scope, response time limit and responsible entity in the graded early warning signal, and generate an executable control decision plan in combination with the control decision plan generation rules; S44. Distribute the graded early warning signals and executable control decision schemes to the construction control terminal for execution, and simultaneously collect data on earth pressure changes, construction progress, and resource consumption during the decision execution process, and integrate them to form decision execution effect data; The dynamic iterative update of the earth pressure risk management benchmark system and the multi-objective optimization decision-making model for earth pressure risk based on decision execution effect data includes the following steps: S51. Pre-set the construction control effect evaluation index system and evaluation judgment threshold, and extract quantitative evaluation indicators from the decision execution effect data; S52. Compare and analyze the quantitative evaluation indicators with the evaluation judgment thresholds to determine the degree of adaptability of the earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision-making model to the construction scenario. S53. Based on the results of the adaptation degree judgment, correct the standard state indicators and mapping correlations within the earth pressure risk management benchmark system, and simultaneously optimize the constraint boundary and weight allocation parameters of the multi-objective optimization decision model for earth pressure risk. S54. The revised earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision-making model are solidified, stored and dynamically updated.

2. The earth pressure risk early warning and decision-making method for subway shield tunneling construction according to claim 1, characterized in that, The process of performing spatiotemporal registration and dimensional normalization on the standardized monitoring dataset, extracting time-series change indicators and spatial correlation indicators for each monitoring dimension, and obtaining the real-time comprehensive earth pressure state index through multi-dimensional weighted comprehensive calculation includes the following steps: S221. Set a unified data collection time benchmark and construction area spatial coordinate benchmark, and perform time sequence alignment and point matching for various types of data in the standardized monitoring dataset; S222. Standardize the data units and value ranges of the standardized monitoring dataset after time-series alignment and location matching. S223. Perform statistical analysis on the standardized monitoring data after unifying the data units and numerical value ranges, obtain the numerical fluctuation law of each monitoring parameter as the tunneling mileage changes, extract the time series change index of each monitoring dimension, and analyze the correlation of monitoring data at different points in combination with the distribution relationship of construction points, and extract spatial correlation indexes. S224. Preset the weight allocation of indicators and the reference rules for standard state of earth pressure, assign weights to time-series change indicators and spatially related indicators respectively, and integrate them according to the logic of superimposing the indicator components one by one and complementing the related indicators to obtain a comprehensive evaluation dataset. S225. The comprehensive assessment dataset is compared and fitted with the standard state reference rule for earth pressure item by item to calculate the real-time comprehensive state index of earth pressure.

3. The method for early warning and decision-making regarding earth pressure risk in subway shield tunneling as described in claim 1, characterized in that, The preset optimization adjustment rules, based on the type of earth pressure anomaly and the risk control level, adjust the constraint boundaries and relative weights of the core optimization objectives to construct a multi-objective optimization decision-making model for earth pressure risk, including the following steps: S321. Pre-set multi-objective optimization and adjustment rules to classify construction control scenarios into different types of earth pressure anomalies and different risk control levels. S322. Adjust the constraint boundaries of the core optimization objectives according to the construction control scenario classification standards and the types of abnormal earth pressure. S323. Adjust the relative weight of core optimization objectives according to the construction management scenario classification standard and the risk management level. S324. Set the adjusted constraint boundary as the model solution constraint condition, and embed the adjusted relative weight ratio to build a multi-objective optimization decision model for earth pressure risk.

4. The earth pressure risk early warning and decision-making method for subway shield tunneling as described in claim 1, characterized in that, The process of analyzing the signal type, release scope, response time limit, and responsible entity in the graded early warning signal, and generating an executable control decision plan in conjunction with the control decision plan generation rules, includes the following steps: S431. Pre-set a hierarchical control measure template library, analyze and extract three core control parameters from the hierarchical early warning signals: signal type, release scope, and response time limit, and match the corresponding level of standardized handling process template from the control measure template library according to the risk control level. S432. Embed the three types of core control parameters into the standardized handling process template, supplement and clarify the person in charge of handling, the equipment allocation list, the material demand plan and the key points of on-site operation, and standardize the format according to the control decision plan generation rules to generate an executable control decision plan.

5. The earth pressure risk early warning and decision-making method for subway shield tunneling as described in claim 1, characterized in that, The process of comparing and analyzing quantitative evaluation indicators with evaluation thresholds to determine the suitability of the earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision-making model for the construction scenario includes the following steps: S521. Preset indicator comparison rules and adaptation level grading standards, compare the quantitative evaluation indicators with the evaluation judgment thresholds one by one, and obtain the set of indicator compliance status. S522. Based on the set of indicator compliance, analyze the overall adaptability of the earth pressure risk management benchmark system and the earth pressure risk multi-objective optimization decision-making model to the construction scenario, and determine the degree of adaptability based on the overall adaptability.

6. The earth pressure risk early warning and decision-making method for subway shield tunneling as described in claim 2, characterized in that, The statistical analysis of standardized monitoring data after unifying data units and numerical value ranges, to obtain the numerical fluctuation patterns of each monitoring parameter with tunneling mileage, to extract time-series variation indicators for each monitoring dimension, and to analyze the correlation of monitoring data at different locations in conjunction with the distribution relationship of construction points, and to extract spatial correlation indicators, includes the following steps: S2231. Perform sliding window statistical analysis on the normalized and standardized monitoring data to obtain the numerical fluctuation pattern of each monitoring parameter as the tunneling mileage changes, and extract trend change indicators, abrupt change indicators and periodic indicators of each monitoring dimension based on the numerical fluctuation pattern, and integrate them to form a time series change indicator set. S2232. Organize the spatial coordinate relationships of each construction point in the standardized monitoring dataset, construct a spatial correlation network of points, analyze the Pearson correlation coefficient of monitoring data of spatially adjacent points, extract spatial correlation indicators between different points, and integrate them to form a set of spatial correlation indicators.

Images