Construction settlement monitoring method and device, equipment and storage medium

Abstract

The application provides a construction settlement monitoring method and device, equipment and storage medium, and belongs to the field of tunnel construction monitoring. The method comprises the following steps: acquiring planar settlement data and construction activity data carrying a time stamp; dividing a monitoring grid and marking a spatial coordinate, matching the two types of data to generate a construction settlement data set; identifying a settlement change rate abnormal section and a differential settlement zone; extracting relevant time sequence features and obtaining a correlation coefficient of construction behavior and settlement anomaly; calculating an influence weight based on the correlation coefficient to determine the construction behavior causing the settlement anomaly. The construction settlement monitoring method and device, equipment and storage medium provided by the application can reduce the probability of construction settlement accidents.

Description

Technical Field

[0001] This application belongs to the field of tunnel construction monitoring technology, and more specifically, relates to a construction settlement monitoring method, device, equipment, and storage medium. Background Technology

[0002] During subway construction, settlement monitoring is a crucial aspect of controlling construction safety risks. Its core requirement is to accurately pinpoint the causes of settlement anomalies by correlating construction activities with the anomalies, thereby enabling targeted control. Currently, existing construction settlement monitoring technologies lack quantitative analysis to support the correlation between construction activities and settlement anomalies, relying primarily on qualitative judgments based on human experience. This makes it difficult to accurately define the degree of impact of each construction activity on settlement anomalies, and consequently, to identify the dominant construction activity causing the anomalies. This judgment method is prone to bias, resulting in a lack of targeted settlement anomaly response measures, insufficient monitoring and control effectiveness, and an inability to promptly curb the further development of settlement anomalies, significantly increasing the risk of construction settlement accidents. Therefore, it is urgent to address these issues to reduce the probability of construction settlement accidents. Summary of the Invention

[0003] The purpose of this application is to provide a construction settlement monitoring method, device, equipment, and storage medium that can reduce the probability of construction settlement accidents. To achieve the above objective, the technical solution provided by this application is as follows:

[0004] Firstly, a construction settlement monitoring method is provided, including:

[0005] Acquire surface settlement data and construction activity data in the construction tunnel area. Both surface settlement data and construction activity data are time-stamped. Surface settlement data is data that characterizes the overall settlement distribution in the construction tunnel area and the surrounding pre-defined area. Construction activity data is data that reflects the construction status and behavioral parameters during the subway construction process.

[0006] The construction tunnel area was divided into multiple monitoring grids, and the corresponding spatial coordinates were labeled for each monitoring grid. Construction activity data and areal settlement data were matched with the spatial coordinates of the corresponding monitoring grids to generate a construction settlement dataset.

[0007] Based on the construction settlement dataset, the continuous areas corresponding to the monitoring grids whose settlement change rate exceeds the first preset threshold are identified as settlement change rate abnormal segments; and based on the settlement difference between each monitoring grid and its adjacent monitoring grids, cluster analysis is performed on all monitoring grids to identify continuous areas whose settlement difference exceeds the second preset threshold as differential settlement zones.

[0008] Settlement time sequence features and construction procedure time sequence features associated with settlement change rate anomaly segments and differential settlement zones are extracted; and based on the settlement time sequence features and construction procedure time sequence features, the correlation coefficients between each construction behavior and settlement anomaly are obtained;

[0009] Based on the correlation coefficient, the influence weight of each construction behavior on differential settlement is obtained, and the construction behavior that leads to abnormal settlement is determined based on the influence weight.

[0010] Secondly, a construction settlement monitoring device is provided, comprising:

[0011] The data acquisition module is used to acquire surface settlement data and construction activity data in the construction tunnel area. Both surface settlement data and construction activity data are timestamped. Surface settlement data is data that characterizes the overall settlement distribution in the construction tunnel area and the surrounding preset area, while construction activity data is data that reflects the construction status and behavioral parameters during the subway construction process.

[0012] The dataset generation module is used to divide the construction tunnel area into multiple monitoring grids and label the corresponding spatial coordinates of each monitoring grid; it matches the construction activity data and areal settlement data with the spatial coordinates of the corresponding monitoring grids to generate a construction settlement dataset.

[0013] The settlement analysis module is used to identify, based on the construction settlement dataset, the continuous area corresponding to the monitoring grid where the settlement change rate exceeds the first preset threshold, as the settlement change rate abnormal segment; and to perform cluster analysis on all monitoring grids based on the settlement difference between each monitoring grid and adjacent monitoring grids, to identify the continuous area where the settlement difference exceeds the second preset threshold, as the differential settlement zone.

[0014] The correlation calculation module is used to extract the settlement time sequence features and construction procedure time sequence features associated with the settlement change rate anomaly segment and differential settlement zone; and based on the settlement time sequence features and construction procedure time sequence features, obtain the correlation coefficient between each construction behavior and settlement anomaly.

[0015] The settlement monitoring module is used to obtain the influence weight of each construction behavior on differential settlement based on the correlation coefficient, and to determine the construction behavior that leads to settlement anomalies based on the influence weight.

[0016] Thirdly, embodiments of this application also provide an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the construction settlement monitoring method provided by any possible implementation of the first aspect.

[0017] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the construction settlement monitoring method provided by any possible implementation of the first aspect.

[0018] The beneficial effects of the technical solution provided in this application are as follows:

[0019] Compared with related technologies, the construction settlement monitoring method, device, equipment, and storage medium provided in this application embodiment overcome the limitations of traditional point-based monitoring by acquiring isometric settlement data and construction activity data with timestamps. This allows for a comprehensive reflection of the spatial distribution characteristics of settlement in the construction area. Simultaneously, the timestamps enable spatiotemporal synchronization of the two types of data, avoiding distortion in correlation analysis caused by time deviations. The construction area is divided into monitoring grids and labeled with coordinates. A construction settlement dataset is generated through data matching, providing a unified spatiotemporal carrier for subsequent analysis. By dual-identifying abnormal settlement rate segments and differential settlement zones, abnormal settlement areas are accurately located, reducing invalid analysis. Temporal features are extracted and correlation coefficients are obtained, quantifying the fuzzy correlation between construction and settlement. The dominant inducing factors are then determined by calculating influence weights, eliminating reliance on manual experience. This embodiment achieves precision and scientific management throughout the entire process of construction settlement monitoring, from data acquisition to causation diagnosis, effectively improving the accuracy and targeting of settlement monitoring, providing a reliable basis for handling abnormal settlement, and reducing the probability of construction settlement accidents. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below.

[0021] Figure 1 A schematic flowchart illustrating the construction settlement monitoring method provided in this application embodiment;

[0022] Figure 2 A structural block diagram of the construction settlement monitoring device provided in the embodiments of this application;

[0023] Figure 3 A schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0024] The embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the embodiments described below with reference to the accompanying drawings are exemplary descriptions for explaining the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions of the embodiments of this application.

[0025] Those skilled in the art will understand that, unless otherwise stated, the singular forms “a,” “an,” and “the” used herein may also include the plural forms. It should be further understood that the terms “comprising” and “including” as used in embodiments of this application mean that the corresponding feature can be implemented as the presented feature, information, data, step, operation, element, and / or component, but do not exclude implementation as other features, information, data, step, operation, element, component, and / or combinations thereof supported by the art. It should be understood that when we say that an element is “connected” or “coupled” to another element, the one element can be directly connected or coupled to the other element, or it can mean that the one element and the other element establish a connection relationship through an intermediate element. Furthermore, “connected” or “coupled” as used herein can include wireless connection or wireless coupling. The term “and / or” as used herein indicates at least one of the items defined by the term; for example, “A and / or B” can be implemented as “A,” or as “B,” or as “A and B.” When describing multiple (two or more) items, if the relationship between the multiple items is not explicitly defined, the multiple items can refer to one, several or all of the multiple items. For example, the description of "parameter A includes A1, A2, A3" can be implemented as parameter A includes A1 or A2 or A3, or it can be implemented as parameter A includes at least two of the three items A1, A2 and A3.

[0026] To make the objectives, technical solutions, and advantages of this application clearer, the following description will be provided in conjunction with the accompanying drawings and specific embodiments.

[0027] This application provides a construction settlement monitoring method, which can be executed by electronic equipment, such as... Figure 1 As shown, the method may include:

[0028] S101: Obtain surface settlement data and construction activity data for the construction tunnel area. Both surface settlement data and construction activity data are time-stamped. Surface settlement data is data that characterizes the overall settlement distribution of the construction tunnel area and the surrounding preset area. Construction activity data is data that reflects the construction status and behavioral parameters during the subway construction process.

[0029] In this embodiment, the areal settlement data characterizes the overall settlement distribution of the construction tunnel area and its surrounding pre-defined area. Unlike point settlement data, which only reflects the settlement state of a single point, areal settlement data comprehensively presents the settlement differences and spatial distribution characteristics at different locations within the construction area, making it crucial data reflecting the spatial state of settlement. This embodiment can acquire the data using a combination of a small baseline synthetic aperture radar interferometry device and a ground-based gridded settlement monitoring instrument. The pre-defined surrounding area can be set to 50m on each side of the construction tunnel centerline, covering key areas potentially affected by construction activities and avoiding omissions of settlement-affected areas. Specific indicators of the areal settlement data can include the cumulative settlement value of each monitoring grid, the settlement change rate, and the settlement difference between adjacent grids. The acquisition frequency can be set to once per day for the synthetic aperture radar interferometry device and once every two hours for the ground-based gridded settlement monitoring instrument, ensuring that the data reflects settlement dynamics in real time.

[0030] In this embodiment, construction activity data refers to raw data that directly reflects the construction operation process during subway construction, and can reflect the specific operational status and parameter changes of subway construction in real time. This embodiment can collect data through multiple types of sensors, specifically including shield tunneling speed sensors, segment assembly parameter sensors, dewatering well water level sensors, and retaining structure support force sensors. Specific indicators of construction activity data include shield tunneling speed (mm / min), segment assembly axial pressure (MPa), segment assembly circumferential gap (mm), dewatering well water level depth (m), and retaining structure support axial force (kN). The acquisition frequency is set to 1 time / second for shield tunneling parameter sensors and 1 time / 30 minutes for dewatering well and retaining structure sensors to ensure accurate capture of real-time parameter changes in construction behavior.

[0031] In this embodiment, the timestamp is a time identifier information assigned to the areal settlement data and construction activity data. It is used to mark the data collection time, realize the synchronization and alignment of settlement data and construction data in the time dimension, and avoid distortion of subsequent correlation analysis due to time deviation.

[0032] In this embodiment, monitoring equipment and sensors are used to collect areal settlement data that reflects the overall spatial distribution of settlement and construction activity data that reflects real-time operational parameters of construction. All data are assigned a timestamp of uniform precision to achieve time synchronization between settlement data and construction data, avoiding distortion of subsequent correlation analysis due to time deviation. At the same time, areal settlement data breaks through the limitations of traditional point settlement monitoring and can fully reflect the spatial distribution characteristics of settlement in the construction area, laying a data foundation for subsequent anomaly identification in the spatial dimension.

[0033] S102: Divide the construction tunnel area into multiple monitoring grids and label the corresponding spatial coordinates for each monitoring grid; match the construction activity data and areal settlement data with the spatial coordinates of the corresponding monitoring grids to generate a construction settlement dataset.

[0034] In this embodiment, the monitoring grid is a standardized spatial unit formed by dividing the construction tunnel area according to a fixed scale. It serves as a unified spatial carrier for settlement data and construction data, enabling spatial dimensional correlation and matching of dispersed multi-source data. This embodiment can employ a square grid division method, dividing the construction tunnel area and its surrounding preset area into multiple monitoring grids at a scale of 10m × 10m. This scale ensures monitoring accuracy while avoiding data processing redundancy caused by an excessive number of grids. A Gaussian Cartesian coordinate system can be used to label the spatial coordinates of each monitoring grid, assigning each grid a unique coordinate code. The coding format can be "X coordinate - Y coordinate," ensuring that the spatial position of each monitoring grid can be uniquely determined.

[0035] The construction settlement dataset is a structured and spatiotemporally fused dataset formed by matching construction activity data and areal settlement data with the spatial coordinates of the corresponding monitoring grids. In this embodiment, areal settlement data and construction activity data can be transmitted to a cloud server in real time via industrial Ethernet. The server uses a coordinate matching algorithm to match the collection location coordinates of the construction activity data with the spatial coordinates of the corresponding monitoring grids, ensuring that each monitoring grid corresponds to unique areal settlement data and construction activity data. At the same time, the matched data is deduplicated and completed to generate a structured construction settlement dataset.

[0036] In this embodiment, by dividing the construction tunnel area into standardized monitoring grids and marking them with unique spatial coordinates, the spatial distribution characteristics of the areal settlement data and the collection location characteristics of the construction activity data are both associated with a unified gridded spatial unit. Then, through coordinate matching, the areal settlement data and construction activity data corresponding to the same monitoring grid are integrated to generate a structured construction settlement dataset. This achieves spatial binding between settlement data and construction data, allowing the two originally independent types of data to form a four-dimensional correlation relationship of spatial location, time, settlement state, and construction state, providing a unified analysis carrier for subsequent anomaly identification and correlation analysis.

[0037] S103: Based on the construction settlement dataset, identify the continuous area corresponding to the monitoring grid where the settlement change rate exceeds the first preset threshold, as the settlement change rate abnormal segment; and based on the settlement difference between each monitoring grid and adjacent monitoring grids, perform cluster analysis on all monitoring grids to identify the continuous area where the settlement difference exceeds the second preset threshold, as the differential settlement zone.

[0038] In this embodiment, the first preset threshold is a critical value for settlement change rate set in conjunction with subway construction specifications, engineering geological conditions, and construction experience. It serves as a standard to distinguish between normal and abnormal settlement change rates. If the settlement change rate exceeds this value, it is determined to be an abnormal settlement rate. The second preset threshold is a critical value for settlement difference set in conjunction with subway construction specifications and engineering experience. It serves as a standard to distinguish between normal and abnormal spatial non-uniformity of settlement. If the settlement difference exceeds this value, it is determined to be an abnormal local settlement distribution. The first and second preset thresholds can be set in conjunction with subway construction specifications and actual engineering construction experience.

[0039] In this embodiment, the settlement change rate is the cumulative change in settlement at a certain location in the construction tunnel area per unit time, and it is an indicator reflecting the dynamic trend of settlement change. This embodiment can use the slope threshold method to analyze the construction settlement dataset, calculate the slope of the settlement change rate time series for each monitoring grid, and determine whether the settlement change rate of three consecutive collection cycles exceeds a first preset threshold. If the condition is met, and the monitoring grid forms a continuous linear area with the surrounding over-limit monitoring grids (the number of grids in the continuous area is ≥3, and the spatial distance between adjacent grids is ≤10m), then the continuous area is determined as an abnormal settlement change rate segment, and the spatial coordinates, average settlement change rate, and abnormal duration of the abnormal segment are recorded.

[0040] In this embodiment, the settlement difference value is the absolute value of the difference between the cumulative settlement values ​​of a certain monitoring grid and its adjacent monitoring grids at the same acquisition time. It is an indicator reflecting the spatial non-uniformity of settlement in the construction area; the larger the value, the more significant the spatial difference in settlement in a local area. This embodiment uses the density clustering algorithm (DBSCAN) to perform cluster analysis on all monitoring grids, with the algorithm parameter set to the neighborhood radius. =5m, minimum number of cluster points MinPts=8. This parameter setting can accurately cluster continuous settlement difference areas and avoid clustering errors. First, calculate the settlement difference between each monitoring grid and all its adjacent monitoring grids (the settlement difference is the absolute value of the cumulative settlement of two monitoring grids). Then, using the settlement difference as the clustering feature, the monitoring grids are divided into multiple clusters through a clustering algorithm. Clusters with a mean settlement difference exceeding the second preset threshold and where the monitoring grids within the cluster form a continuous planar area (the area of ​​the continuous planar area ≥ 300㎡) are selected. The area corresponding to this cluster is determined as the differential settlement zone.

[0041] In this embodiment, from a temporal perspective, by determining whether the settlement change rate exceeds a first preset threshold, monitoring grids with continuously abnormal settlement rates are identified and aggregated into continuous abnormal settlement change rate segments, reflecting the abnormal trend of settlement dynamic changes. From a spatial perspective, based on the settlement difference between each monitoring grid and its adjacent grids, cluster analysis is used to aggregate monitoring grids with excessive spatial non-uniformity of settlement into continuous differential settlement zones, reflecting the spatial anomaly characteristics of settlement distribution. The identification of these two types of abnormal regions achieves a dual temporal and spatial definition of settlement anomalies, accurately locking the target area for subsequent analysis and avoiding the inefficiency and result distortion caused by indiscriminate global analysis.

[0042] S104: Extract the settlement time sequence characteristics and construction procedure time sequence characteristics associated with the settlement change rate anomaly segment and differential settlement zone; and based on the settlement time sequence characteristics and construction procedure time sequence characteristics, obtain the correlation coefficient between each construction behavior and settlement anomaly.

[0043] In this embodiment, the settlement time-series feature is the monitoring grid corresponding to the settlement change rate anomaly segment and differential settlement zone. It represents the characteristic sequence of settlement data changing over time during the occurrence or development of the settlement anomaly, fully reflecting the time-varying pattern of the settlement anomaly from its appearance to its development. The construction procedure time-series feature is the characteristic sequence of construction activity data changing over time during the occurrence or development of the settlement anomaly, corresponding to the spatial location of the settlement change rate anomaly segment and differential settlement zone. It fully reflects the time-varying pattern of construction operation parameters in that area when the settlement anomaly occurs. This embodiment can extract the relevant data from the monitoring grid corresponding to the settlement change rate anomaly segment and differential settlement zone 72 hours before the anomaly occurs as the time-series feature. The settlement time-series feature can include the time series of cumulative settlement value, settlement change rate, and settlement difference between adjacent grids. The construction procedure time-series feature can include the time series of shield tunneling speed, segment assembly parameters, dewatering well water level, and retaining structure support force. Both types of time-series features are converted into equal time interval feature sequences of length N, where N is the number of effective acquisitions within 72 hours, ensuring complete alignment of the time steps of the time-series features.

[0044] In this embodiment, construction behavior refers to specific operational actions or parameter adjustments implemented based on construction procedures during subway construction. It reflects the actual construction operations behind the construction activity data, such as adjusting the tunnel boring machine's advance speed, pumping water from dewatering wells, and adjusting the pressure during segment assembly. Anomaly settlement refers to a settlement state within the construction tunnel area where the settlement change rate exceeds a first preset threshold or the settlement difference exceeds a second preset threshold. It can include two manifestations: abnormal settlement rate and abnormal spatial distribution of settlement.

[0045] In this embodiment, the correlation coefficient is a value characterizing the degree of temporal correlation between a certain construction behavior and settlement anomaly, with a value range of [0,1]. The larger the value, the stronger the correlation between the temporal changes of the construction behavior and the temporal changes of settlement anomaly. This embodiment can use a temporal correlation model to process the extracted temporal features. This model is trained based on the correspondence between historical settlement temporal features and historical construction procedure temporal features. The training set accounts for 80% of the total sample size, and the validation set accounts for 20%. The mean squared error is used as the loss function, and the model parameters are optimized through the backpropagation algorithm until the prediction accuracy of the model validation set reaches more than 95%. The aligned settlement temporal features and construction procedure temporal features are input into the trained model. The model is processed by temporal encoding, linear transformation, and nonlinear normalization to output the correlation coefficient between each construction behavior and settlement anomaly, with the correlation coefficient value range of [0,1].

[0046] In this embodiment, for the identified abnormal settlement rate segments and differential settlement zones, the settlement time sequence features during the occurrence of the anomalies and the construction procedure time sequence features at the corresponding spatial locations are extracted to ensure that the extracted features are directly related to the settlement anomalies. Then, the two types of time sequence features are deeply mined through the trained time sequence analysis model to capture the time correlation between changes in construction procedures and changes in settlement anomalies. This correlation is then transformed into a correlation coefficient with a fixed range of values, realizing the transformation from qualitative judgment of correlation based on human experience to quantitative calculation of correlation by algorithms, providing a quantitative basis for subsequent accurate diagnosis of the causes of settlement anomalies.

[0047] S105: Based on the correlation coefficient, obtain the influence weight of each construction behavior on differential settlement, and determine the construction behavior that leads to abnormal settlement based on the influence weight.

[0048] In this embodiment, the influence weight is a value that characterizes the degree of contribution or dominance of a certain construction behavior to the occurrence of differential settlement. It is a relative value calculated based on the correlation coefficient through the feature importance analysis algorithm. The sum of the influence weights of all construction behaviors is 1. The larger the value, the higher the probability that the construction behavior is the dominant factor causing the settlement anomaly.

[0049] This embodiment employs a random forest feature importance analysis algorithm, using the correlation coefficient as the input feature to calculate the impact weight of each construction activity on differential settlement. The algorithm calculates the reduction in node impurity for each construction activity through ensemble learning of multiple independent decision trees, and then obtains the impact weight of each activity through proportion normalization. The sum of the impact weights of all construction activities is 1. The top three construction activities with the highest impact weights are selected as candidate construction activities. These candidate activities are then subjected to time-series matching verification. If a candidate construction activity occurs before the settlement anomaly occurs, and the parameter change of this trigger exceeds 1.5 times its normal construction threshold, then this candidate construction activity is determined to be the dominant construction activity causing the settlement anomaly. If multiple candidate construction activities satisfy the time-series matching verification, the one with the highest impact weight is selected as the dominant construction activity. Among them, the normal construction thresholds can be preset according to the subway construction specifications, such as: shield tunneling speed 30-50mm / min, axial pressure of segment assembly 2-3MPa, circumferential gap of segment assembly 2-5mm, daily water level variation of dewatering well ≤0.5m, and axial force of retaining structure support ≥80% of the design value.

[0050] In this embodiment, a feature importance analysis algorithm is used to conduct in-depth analysis of the correlation coefficients of various construction behaviors, calculate the quantitative value (influence weight) of the contribution of each construction behavior to the occurrence of differential settlement, and rank the contribution of each construction behavior based on the relative magnitude of the influence weights. Finally, the core construction behavior causing the settlement anomaly is determined based on the weight magnitude. This embodiment further transforms the quantitative correlation between construction and settlement into the precise determination of specific construction causes, overcoming the drawbacks of traditional monitoring methods that rely on manual experience to judge causes. This provides scientific quantitative support for the diagnosis of settlement anomalies and offers a clear basis for targeted treatment at the construction site.

[0051] As can be seen from the above, this embodiment, by acquiring time-stamped areal settlement data and construction activity data, overcomes the limitations of traditional point-based monitoring, comprehensively reflecting the spatial distribution characteristics of settlement in the construction area. Simultaneously, relying on timestamps achieves spatiotemporal synchronization of the two types of data, avoiding distortion in correlation analysis caused by time deviations. The construction area is divided into monitoring grids and labeled with coordinates; a construction settlement dataset is generated through data matching, providing a unified spatiotemporal carrier for subsequent analysis. By dual-identifying abnormal settlement rate segments and differential settlement zones, abnormal settlement areas are accurately located, reducing invalid analysis. Temporal features are extracted and correlation coefficients are obtained, quantifying the fuzzy correlation between construction and settlement. Then, by calculating influence weights, the dominant inducing factors are determined, eliminating reliance on manual experience. This embodiment achieves precision and scientific rigor throughout the entire process of construction settlement monitoring, from data acquisition to causation diagnosis, effectively improving the accuracy and targeting of settlement monitoring, providing a reliable basis for handling abnormal settlement, and reducing the probability of construction settlement accidents.

[0052] In one embodiment of this application, based on a construction settlement dataset, continuous regions corresponding to monitoring grids where the settlement change rate exceeds a first preset threshold are identified as abnormal settlement change rate segments, including:

[0053] Based on the construction settlement dataset, the settlement change rate of each monitoring grid is calculated over time, and a time series of the settlement change rate of each monitoring grid is generated.

[0054] In the time series of settlement change rate, identify the time interval in which the settlement change rate continuously exceeds the first preset threshold within the preset collection period;

[0055] Spatially aggregate the monitoring grids corresponding to the time series intervals to obtain the abnormal settlement change rate segments.

[0056] In this embodiment, based on the cumulative settlement value of each monitoring grid in the construction settlement dataset, linear interpolation can be used to complete the missing cumulative settlement value data to ensure data continuity. Then, the settlement change rate of each monitoring grid over time is calculated using the difference method. The specific calculation formula is: Settlement Change Rate ,in This represents the cumulative settlement value from the previous collection period. This represents the cumulative settlement value for the current collection period. This refers to the time of the previous data collection cycle. The current collection period is represented by the time unit, which is uniformly converted to days (d). The unit of the settlement change rate is mm / d to ensure the consistency and comparability of the calculation results.

[0057] In this embodiment, the settlement change rate calculated for each monitoring grid in different acquisition cycles is arranged in chronological order of acquisition time to generate a time series of the settlement change rate of the monitoring grid. The length of the time series is consistent with the effective acquisition number of the monitoring grid, and each data point corresponds to a unique acquisition timestamp, which facilitates the identification of subsequent continuous over-limit intervals.

[0058] In this embodiment, the preset acquisition cycle can be set to 3 consecutive acquisition cycles. Combined with the acquisition frequency of the areal settlement data (1 time / 2 hours for the ground grid settlement monitoring instrument), the 3 acquisition cycles correspond to 6 hours. This duration can avoid misjudgment caused by fluctuations in single-cycle data, and can also capture continuous anomalies in the settlement change rate in a timely manner, ensuring the timeliness and accuracy of anomaly identification. The preset acquisition cycle can be fine-tuned according to the monitoring accuracy requirements, and the fine-tuning range is 2-4 acquisition cycles.

[0059] In this embodiment, a sliding window method can be used to analyze the time series of settlement change rate. The length of the sliding window is set to a preset acquisition period (3 acquisition periods). The window is slid along the time series sequentially, and it is determined whether all settlement change rate data within each window exceeds a first preset threshold (0.5 mm / d). If they all exceed the threshold, the time interval corresponding to the window is determined as a continuous excess time series interval. The start time, end time, and average settlement change rate of this time series interval are recorded to provide a basis for subsequent spatial aggregation. All monitoring grids with continuous excess time series intervals are extracted, and the spatial coordinates of these monitoring grids are obtained. A spatial proximity analysis algorithm can be used to determine whether the spatial distance between adjacent monitoring grids is ≤10m (the side length of the monitoring grid). If it is satisfied, these monitoring grids are grouped into an aggregation area until all monitoring grids with continuous excess time series intervals are aggregated. A continuous area with ≥3 monitoring grids within the aggregation area is selected and determined as an abnormal settlement change rate segment. The spatial range, average settlement change rate, and duration of the abnormal segment are recorded.

[0060] In this embodiment, firstly, based on the cumulative settlement value in the construction settlement dataset, the settlement change rate of each monitoring grid is calculated using the differential method to generate a time series of settlement change rates, transforming the static data of cumulative settlement values ​​into time series data reflecting dynamic changes in settlement. Then, by setting a reasonable preset acquisition period, the sliding window method is used to identify time series intervals where the settlement change rate continuously exceeds a first preset threshold, eliminating interference from single-period data fluctuations and ensuring that what is captured is a continuous anomaly in the settlement change rate, rather than accidental fluctuations. Finally, through a spatial proximity analysis algorithm, the monitoring grids with continuously exceeding time series intervals are spatially aggregated, and continuously distributed monitoring grid areas are selected to obtain anomaly segments in the settlement change rate. This achieves dual screening from continuous temporal anomalies to continuous spatial anomalies, ensuring the accuracy and completeness of anomaly segment identification and providing a precise anomaly area range for subsequent identification of differential settlement zones and cause analysis.

[0061] As can be seen from the above, this embodiment calculates the settlement change rate using the differential method, ensuring the accuracy of the settlement change rate calculation and truly reflecting the dynamic settlement change trend of each monitoring grid. The generation of the settlement change rate time series provides a clear data carrier for the subsequent identification of continuous anomalies. The setting of the preset acquisition period and the application of the sliding window method effectively avoid misjudgment caused by single-period data fluctuations and improve the accuracy of identifying continuous over-limit time series intervals. The spatial aggregation process, through spatial proximity analysis, ensures the spatial continuity of the settlement change rate anomaly segment, avoiding misjudging scattered over-limit monitoring grids as anomaly segments, further improving the accuracy and reliability of settlement change rate anomaly segment identification, and laying a precise regional foundation for subsequent in-depth analysis of settlement anomalies.

[0062] In one embodiment of this application, based on the settlement difference between each monitoring grid and its adjacent monitoring grids, cluster analysis is performed on all monitoring grids to identify continuous areas where the settlement difference exceeds a second preset threshold, as differential settlement zones, including:

[0063] Based on the construction settlement dataset, the settlement difference between each monitoring grid and all its adjacent monitoring grids is calculated, resulting in a set of settlement difference values ​​between each monitoring grid and all its adjacent monitoring grids;

[0064] The statistical characteristics of the set of settlement difference values ​​corresponding to each monitoring grid are used as the clustering feature vector to characterize the degree of uneven settlement of the monitoring grid.

[0065] Cluster analysis is performed on the feature vectors of each cluster to divide the multiple monitoring grids into multiple clusters;

[0066] Clusters with a mean cluster feature vector exceeding a second preset threshold and whose monitoring grids form a continuous surface in space are selected from multiple clusters. The area corresponding to this cluster is designated as the differential settlement zone.

[0067] In this embodiment, adjacent monitoring grids refer only to directly adjacent monitoring grids, that is, monitoring grids directly above, below, to the left, or to the right of a given monitoring grid. Diagonally adjacent monitoring grids are excluded from the adjacent monitoring grid range because the spatial distance exceeds 10m (the side length of the monitoring grid), ensuring the spatial correlation of adjacent monitoring grids and avoiding the ineffectiveness of settlement difference values ​​due to excessive spatial distance. Based on the cumulative settlement value of each monitoring grid in the construction settlement data, the settlement difference between that monitoring grid and all its directly adjacent monitoring grids is calculated. The settlement difference is calculated using absolute values, and the specific formula is as follows: ,in This represents the cumulative settlement value of the current monitoring grid. The cumulative settlement value of adjacent monitoring grids is used, with the settlement difference value in mm, to ensure the non-negativity and comparability of the calculation results. The settlement difference values ​​of each monitoring grid and all adjacent monitoring grids are summarized to form the settlement difference value set of that monitoring grid. If a monitoring grid is an edge grid (such as a monitoring grid at the boundary of the construction area) and the number of adjacent monitoring grids is less than 4, the settlement difference value is calculated according to the actual number of adjacent monitoring grids to form the settlement difference value set.

[0068] In this embodiment, the statistical feature of the settlement difference set can be the mean. The arithmetic mean of all settlement differences in the settlement difference set of each monitoring grid is calculated. This mean accurately characterizes the degree of uneven settlement in the local area surrounding the monitoring grid. The larger the mean, the more significant the settlement difference around the monitoring grid. Using the mean settlement difference of each monitoring grid as a unique feature, a clustering feature vector characterizing the degree of uneven settlement in the local area of ​​the monitoring grid is constructed. The dimension of the clustering feature vector is 1, which facilitates subsequent processing and calculation by the clustering algorithm.

[0069] In this embodiment, the density clustering algorithm (DBSCAN) can be used to perform cluster analysis on the clustering feature vectors of all monitored grids. This algorithm does not require pre-setting the number of clusters and can automatically identify continuous areas with high density, making it suitable for clustering areas with abnormal settlement differences. In practical applications, other clustering algorithms can also be selected, but this embodiment does not limit them. The core parameter of the algorithm is set as: neighborhood radius. =5m, this parameter matches the side length of the monitoring grid, ensuring that the clustering relationship between adjacent monitoring grids can be captured; Minimum number of cluster points MinPts=8, this parameter avoids misclassifying a small number of scattered monitoring grids into a single cluster, ensuring the reliability of the clustering results. During the clustering analysis, the algorithm groups monitoring grids with similar clustering feature vectors and close spatial distances into one cluster, ultimately dividing all monitoring grids into multiple clusters, each cluster corresponding to a local region. First, the arithmetic mean of the cluster feature vectors (mean of settlement difference) of all monitoring grids within each cluster is calculated. Clusters with this mean exceeding a second preset threshold (0.3 mm) are selected, and the areas corresponding to these clusters are considered areas with abnormal settlement differences. Then, the spatial continuity of the selected clusters is assessed using a spatial connectivity analysis algorithm to determine whether all monitoring grids within a cluster can form a continuous planar region. The criteria for a continuous planar region are: any two monitoring grids within a cluster can be connected through adjacent monitoring grids (spatial distance ≤ 10 m), and the area covered by the monitoring grids within the cluster is ≥ 300 mm. (i.e., ≥3 monitoring grids of 10m×10m); finally, the area corresponding to the cluster that meets the above two conditions is determined as the differential settlement zone, and the spatial range, maximum settlement difference and average settlement difference of the differential settlement zone are recorded.

[0070] In this embodiment, the scope of adjacent monitoring grids is first defined to ensure the rationality and correlation of the settlement difference calculation. By calculating the settlement difference between each monitoring grid and its adjacent grids, a settlement difference set is formed to capture the settlement differences around each monitoring grid. Then, by calculating the mean of the settlement difference set, a clustering feature vector is constructed to quantify the local uneven settlement of each monitoring grid into a single feature value, which facilitates subsequent clustering analysis. Next, a density clustering algorithm is used to divide the monitoring grids into multiple clusters based on the clustering feature vector and spatial distance, realizing the aggregation of areas with similar settlement differences. Finally, through dual screening (mean of clustering feature vector exceeding the limit, spatially continuous areal), differential settlement zones are screened from multiple clusters, realizing the quantitative identification of differential settlement zones, eliminating the reliance on manual experience, and ensuring that the identified differential settlement zones not only meet the requirements of settlement difference anomalies but also have spatial continuity, and can truly reflect the uneven settlement distribution of the construction area.

[0071] As can be seen from the above, the reasonable definition of adjacent monitoring grids in this embodiment ensures the pertinence and relevance of settlement difference calculation and avoids interference from invalid settlement differences. The construction of the settlement difference set and cluster feature vector transforms the scattered settlement difference data into quantitative feature indicators, providing a scientific data foundation for cluster analysis. The application of the density clustering algorithm does not require a preset number of clusters and can automatically identify continuous areas with abnormal settlement differences, improving the flexibility and accuracy of cluster analysis. The setting of dual screening conditions effectively eliminates invalid areas where the mean settlement difference exceeds the limit but the space is not continuous, ensuring the accuracy and completeness of differential settlement zone identification. It can accurately reflect the uneven settlement distribution in the construction area and provide a precise range of abnormal areas for subsequent time-series correlation analysis and cause diagnosis, further improving the practicality of the entire construction settlement monitoring method.

[0072] In one embodiment of this application, based on settlement time sequence characteristics and construction procedure time sequence characteristics, the correlation coefficient between each construction behavior and settlement anomaly is obtained, including:

[0073] The settlement time sequence characteristics and the corresponding construction procedure time sequence characteristics are aligned and combined according to time steps to obtain a combined time sequence characteristic sequence;

[0074] Based on the combined temporal feature sequence, and through the encoding process of the combined temporal feature sequence using a temporal correlation model, the deep correlation features between construction behavior and settlement anomalies in the time dimension are obtained; the temporal correlation model is trained based on the correspondence between historical settlement temporal features and historical construction procedure temporal features.

[0075] A linear transformation is applied to the deep association features to map them to one-dimensional original association values;

[0076] The one-dimensional original correlation value is transformed nonlinearly using the first formula to map the one-dimensional original correlation value into the correlation coefficient between each construction behavior and settlement anomaly.

[0077] The first formula is:

[0078] ;

[0079] in, Let Y represent the correlation coefficient between the i-th construction action and the settlement anomaly, and Y represent the one-dimensional original correlation value.

[0080] In this embodiment, the time steps of settlement time sequence characteristics and construction procedure time sequence characteristics are aligned. The acquisition cycle of the ground grid settlement monitoring instrument with the lowest acquisition frequency (1 time / 2 hours) is used as the unified time step. The construction procedure time sequence characteristics (such as shield tunneling speed, 1 time / second) are downsampled. The downsampling adopts the mean sampling method. The arithmetic mean of multiple construction procedure time sequence characteristic data within each unified time step is taken as the construction procedure time sequence characteristic value of that time step, ensuring that the time steps of settlement time sequence characteristics and construction procedure time sequence characteristics are completely consistent. After alignment, the settlement time series features and construction procedure time series features at the same time step are spliced ​​together to form a combined time series feature sequence. The length of the combined time series feature sequence is N (the number of effective data collections in the 72 hours before the anomaly occurs, N=864). The dimension of the combined feature at each time step is the sum of the dimension of the settlement time series feature and the dimension of the construction procedure time series feature. The dimension of the settlement time series feature is 3 (cumulative settlement value, settlement change rate, and settlement difference between adjacent grids), and the dimension of the construction procedure time series feature for a single construction behavior is 1. Therefore, the dimension of the combined time series feature sequence corresponding to a single construction behavior is 4.

[0081] In this embodiment, the temporal correlation model can be a bidirectional long short-term memory network model, which consists of an input layer, a forward long short-term memory network layer, a backward long short-term memory network layer, and a fusion layer. The number of neurons in the input layer is consistent with the dimension of the combined temporal feature sequence (4). The number of neurons in both the forward and backward long short-term memory network layers is set to 64. This number can accurately capture the long-term dependencies of temporal features and avoid model overfitting. In practical applications, other values ​​can be selected for the number of neurons, and this embodiment does not limit this. The fusion layer uses a concatenation method to concatenate the output features of the forward and backward long short-term memory network layers to obtain the initial vector of deep correlation features. The training samples for the model consist of the correspondence data between historical settlement time-series features and historical construction procedure time-series features, with a sample size of no less than 1000 sets to ensure sufficient training of the model. During training, the mean squared error is used as the loss function, the learning rate is set to 0.001, and the number of iterations is set to 100 rounds. The weights and bias parameters of the model are continuously optimized through the backpropagation algorithm until the prediction accuracy of the model validation set reaches more than 95%, at which point training stops, and the trained time-series correlation model is obtained.

[0082] The combined temporal feature sequence is input into the trained temporal correlation model. The input layer passes the combined temporal feature sequence to the forward and backward long short-term memory network layers. The forward long short-term memory network layer extracts the forward temporal correlation features of "construction procedure change → settlement change" from time step 1 to N. The backward long short-term memory network layer extracts the reverse temporal correlation features of "settlement change ← construction procedure change" from time step N to 1. The fusion layer concatenates the correlation features extracted from the forward and backward layers to obtain the deep correlation features between construction behavior and settlement anomaly in the time dimension. The deep correlation features are high-dimensional vectors with a dimension of 128 (64-dimensional forward features + 64-dimensional backward features). This vector deeply encodes the temporal correlation information between construction behavior and settlement anomaly.

[0083] In this embodiment, a fully connected layer is used to perform a linear transformation on the deep association features. The fully connected layer has 128 input neurons (consistent with the dimension of the deep association features) and 1 output neuron. The linear transformation is performed using the formula... The high-dimensional deep association features are mapped to a one-dimensional original association value Y, where W is the weight matrix of the fully connected layer (dimension 1×128), H is the deep association feature vector (dimension 128×1), and b is the bias term of the fully connected layer (dimension 1×1). W and b are the optimal parameters learned during the training of the temporal association model, ensuring that the linear transformation can accurately map the association information in the deep association features. The one-dimensional original association value Y has no fixed range and can be any real number, which is used for subsequent nonlinear normalization processing.

[0084] The original one-dimensional correlation value Y is nonlinearly transformed using the first formula, mapping it to the [0,1] interval, thus obtaining the correlation coefficient between the i-th construction behavior and settlement anomaly. In the first formula, e is the natural constant, approximately equal to 2.71828; the larger the one-dimensional original correlation value Y, the stronger the temporal correlation between construction behavior and settlement anomalies. The closer Y is to 1, the weaker the temporal correlation. The closer the coefficient is to 0, the better it matches the quantitative requirements of the correlation coefficient. For each construction activity, the correlation coefficient is calculated independently according to the above steps, ensuring that the coefficients do not interfere with each other and guaranteeing their independence and accuracy.

[0085] In this embodiment, firstly, by aligning and combining time steps, the settlement time-series features and construction procedure time-series features are fused into a combined time-series feature sequence, ensuring a one-to-one correspondence between the two types of time-series features in the time dimension, laying the foundation for subsequent time-series correlation analysis. Then, the combined time-series feature sequence is input into the trained time-series correlation model. The model uses forward and backward encoding of a bidirectional long short-term memory network to deeply extract the forward and backward correlation features between construction behavior and settlement anomalies in the time dimension. Through a fusion layer, deep correlation features are obtained, achieving in-depth mining of time-series correlation information. Next, through linear transformation of a fully connected layer, the high-dimensional deep correlation features are mapped to one-dimensional original correlation values, simplifying the representation of correlation information. Finally, through nonlinear normalization using the first formula, the one-dimensional original correlation values ​​are mapped to the [0,1] interval to obtain quantified correlation coefficients, achieving accurate quantification of the correlation between construction behavior and settlement anomalies, providing accurate input features for subsequent calculation of influence weights.

[0086] As can be seen from the above, the time step alignment and combination processing in this embodiment ensures the time synchronization of settlement time sequence characteristics and construction procedure time sequence characteristics, avoiding correlation analysis errors caused by time deviations; the bidirectional coding design of the time sequence correlation model can simultaneously capture forward and reverse time sequence correlation characteristics, deeply explore the long-term dependency relationship between construction behavior and settlement anomalies, and improve the comprehensiveness and accuracy of correlation feature extraction; the linear transformation converts high-dimensional correlation features into one-dimensional original correlation values, simplifying the processing difficulty of correlation information and facilitating subsequent normalization operations; the application of the first formula realizes the standardization and normalization of the original correlation values, fixing the value range of the correlation coefficient to [0,1], which facilitates the intuitive interpretation of the correlation strength between each construction behavior and settlement anomalies, while ensuring the comparability of correlation coefficients, providing accurate quantitative basis for the subsequent calculation of influence weights and the determination of dominant construction behaviors, completely getting rid of the drawbacks of manual experience-based qualitative judgment, and improving the scientificity and accuracy of the entire monitoring method.

[0087] In one embodiment of this application, the influence weight of each construction action on differential settlement is obtained based on the correlation coefficient, including:

[0088] An importance analysis was performed on the correlation coefficients corresponding to each construction activity to obtain the initial impact weight of each construction activity.

[0089] Obtain the temporal variation characteristics of construction activity data corresponding to each construction activity within the historical period associated with the differential settlement zone;

[0090] Based on the time-series variation characteristics, the initial influence weights of each construction activity are modified to obtain the influence weight of each construction activity on differential settlement.

[0091] In this embodiment, the Random Forest Feature Importance Analysis Algorithm can be used to analyze the importance of the correlation coefficients corresponding to each construction behavior. The algorithm uses the correlation coefficients of each construction behavior as the only input feature. Through the ensemble learning of multiple independent decision trees, the node impurity reduction value of each construction behavior is calculated. Then, through the proportion normalization process, the initial influence weight of each construction behavior is obtained. The sum of the initial influence weights of all construction behaviors is 1, and the value range of the initial influence weight is [0,1].

[0092] In this embodiment, the historical time period associated with the differential settlement zone is set to 72 hours prior to the time when the differential settlement zone was identified. This time period can fully cover the changes in construction behavior before the occurrence of differential settlement and can be fine-tuned according to the monitoring accuracy requirements, with a fine-tuning range of 48-96 hours. The temporal change characteristics are the time series change patterns of construction activity data corresponding to each construction behavior within this historical time period, specifically including the initial value, termination value, change amplitude, change rate, and whether abrupt changes occur of construction parameters. These are acquired through the multi-type sensor data mentioned above, with the acquisition frequency consistent with the construction activity data, ensuring that the temporal change characteristics can truly reflect the dynamic changes in construction behavior.

[0093] In this embodiment, the initial influence weight can be corrected using a correction coefficient method. The correction coefficient is determined based on the temporal variation characteristics of the construction behavior, and its value range can be set to 0.8-1.2. The corrected influence weight = initial influence weight × correction coefficient. The sum of the influence weights of all construction behaviors after correction remains 1 (after secondary normalization). The specific correction logic is as follows: if the temporal variation characteristics of the construction behavior show that its construction parameters have obvious abnormal changes in the historical period (such as the change amplitude exceeding the normal construction threshold), then the correction coefficient is set to 1.0-1.2; if no abnormal changes or the changes are gradual, then the correction coefficient is set to 0.8-1.0, ensuring that the corrected influence weight can match the actual contribution of the construction behavior to differential settlement.

[0094] In this embodiment, the correlation coefficients corresponding to each construction activity are first used as input. A random forest feature importance analysis algorithm is used to calculate the initial influence weight of each construction activity, thus initially quantifying the contribution of each activity to differential settlement. Then, the historical time period associated with the differential settlement zone is determined, and construction activity data corresponding to each activity within this time period are collected. The temporal variation characteristics of construction parameters are extracted to capture the dynamic change patterns of construction activities before differential settlement occurs. Finally, a corresponding correction coefficient is set based on the temporal variation characteristics to correct the initial influence weight and perform secondary normalization. This ensures that the corrected influence weight not only reflects the correlation between construction activities and settlement anomalies but also reflects the actual impact of abnormal changes in the construction activities themselves on differential settlement, improving the calculation accuracy of the influence weight and providing a more reliable quantitative basis for the accurate determination of subsequent settlement anomaly causes.

[0095] As can be seen from the above, this embodiment calculates the initial influence weights using the random forest feature importance analysis algorithm, ensuring the quantitative accuracy of the initial weights and maintaining consistency with the calculation logic of the correlation coefficients mentioned above, thus improving the coherence of the technical solution. The reasonable setting of historical time periods can fully capture the changes in construction behavior before differential settlement occurs, avoiding the omission of key change features. The application of the correction coefficient method enables accurate correction of the initial influence weights, making up for the deficiency of the initial weights not considering the temporal changes of construction behavior, so that the influence weights can more realistically reflect the actual contribution of construction behavior to differential settlement. The calculation accuracy of the corrected influence weights is improved compared with the initial weights, further improving the accuracy and reliability of subsequent cause determination.

[0096] In one embodiment of this application, the initial influence weights of each construction activity are corrected based on the temporal variation characteristics to obtain the influence weight of each construction activity on differential settlement, including:

[0097] Based on the characteristics of temporal changes, determine whether the construction parameters of each construction activity have undergone abnormal changes in the historical period, and obtain the start time and magnitude of the abnormal changes.

[0098] Based on the chronological relationship between the start time and the time when the differential settlement zone was identified, as well as the degree of deviation between the change magnitude and the corresponding normal construction threshold range, the initial influence weight of each construction behavior is dynamically adjusted to obtain the influence weight of each construction behavior on differential settlement.

[0099] The dynamic adjustment includes: increasing the initial impact weight for construction activities that occur earlier than the settlement anomaly identification time and whose change exceeds the normal construction threshold range by a preset multiple; and decreasing the initial impact weight for construction activities that occur later than the settlement anomaly identification time or whose change does not exceed the normal construction threshold range.

[0100] In this embodiment, based on the extracted temporal change characteristics, the values ​​of construction parameters corresponding to the construction behavior in historical time periods are compared with preset normal construction threshold ranges. If the construction parameter at a certain moment exceeds the normal construction threshold range and the duration is ≥1 collection cycle, it is determined that the construction parameter of the construction behavior has undergone abnormal changes. The normal construction threshold range is consistent with the previous description, specifically: shield tunneling speed 30-50mm / min, segment assembly axial pressure 2-3MPa, segment assembly circumferential gap 2-5mm, daily water level variation in dewatering wells ≤0.5m, and retaining structure support axial force ≥80% of the design value. The start time of the abnormal change is the time when the construction parameter first exceeds the normal threshold range, and the change amplitude is the absolute value of the difference between the maximum value of the construction parameter during the abnormal period and the upper limit (or lower limit, for lower thresholds such as support axial force) of the normal threshold range, with the unit consistent with the corresponding construction parameter.

[0101] In this embodiment, the preset multiplier is set to 1.5 times. This multiplier can accurately distinguish between slight fluctuations and obvious anomalies. It can be fine-tuned according to engineering geological conditions, with a fine-tuning range of 1.2-1.8 times. The adjustment range of the initial influence weight is set to a fixed gradient, specifically: the adjustment range is 10%-20% of the initial influence weight, and the adjustment range is 10%-30% of the initial influence weight. The adjustment range can be flexibly adjusted according to the degree of deviation of the change range. The greater the deviation, the greater the adjustment range.

[0102] In this embodiment, the initial influence weights are dynamically adjusted:

[0103] Increased weighting: If the start time of abnormal changes in construction parameters is earlier than the time when the differential settlement zone is identified, and the magnitude of the change exceeds the normal construction threshold range by a preset multiple (1.5 times), then the initial impact weight of this construction behavior will be increased. For example, if the normal threshold for shield tunneling speed is 30-50 mm / min, and the abnormal change is 40 mm / min (exceeding the upper limit by 10 mm / min, reaching 1.8 times the upper limit), and the abnormal start time is 6 hours earlier than the settlement anomaly identification time, then its initial impact weight (e.g., 0.85) will be increased by 20%, and the corrected weight will be 1.02. After subsequent secondary normalization, the sum will be ensured to be 1.

[0104] Adjustment: If the start time of abnormal changes in construction parameters is later than the time when the differential settlement zone is identified, or if the change amplitude does not exceed the normal construction threshold range (or the exceedance amplitude does not reach a preset multiple), then the initial impact weight of this construction behavior will be reduced. For example, if the water level change amplitude of the dewatering well is 0.6m (exceeding the threshold by 0.1m, but not reaching 1.5 times), then its initial impact weight (e.g., 0.03) will be reduced by 10%, and the corrected weight will be 0.027; if the start time of abnormal segment assembly pressure is later than the time of settlement anomaly identification, then its initial impact weight will be reduced by 30%.

[0105] After all construction activities have been dynamically adjusted, the corrected weights are processed a second time using a proportion normalization formula to ensure that the sum of the influence weights of all construction activities is strictly 1. The normalization formula is consistent with the previous one, that is, the corrected weight = the corrected weight of a certain construction activity / the sum of the corrected weights of all construction activities.

[0106] In this embodiment, firstly, based on the temporal variation characteristics of construction parameters and combined with a preset normal construction threshold range, it is determined whether abnormal changes have occurred in construction behavior, and the start time and magnitude of abnormal changes are accurately extracted to provide a quantitative basis for dynamic adjustment. Then, a clear preset multiple and adjustment range are set, using "the order of abnormal time" and "the degree of deviation" as dual adjustment standards to dynamically adjust the initial influence weights. Construction behaviors that occur earlier than settlement anomalies and deviate significantly from the threshold contribute more to the settlement anomalies, thus increasing their weights; construction behaviors that occur later than settlement anomalies or deviate slightly contribute less, thus decreasing their weights. Finally, through secondary normalization, it is ensured that the corrected influence weights meet the value requirements, enabling the weights to accurately match the actual contribution of construction behaviors to differential settlement, further improving the scientific nature and accuracy of the weight calculation.

[0107] As can be seen from the above, this embodiment clarifies the criteria for judging abnormal changes in construction parameters, ensuring the accuracy and consistency of abnormal information extraction and avoiding adjustment deviations caused by fuzzy judgments. The quantitative setting of preset multiples and adjustment ranges provides a clear basis for the dynamic adjustment of initial influence weights, solving the problem of lack of quantitative standards for correction in existing technologies. The application of dual adjustment standards enables targeted adjustment of initial influence weights, making the corrected weights more consistent with the actual impact of construction behavior and significantly improving the calculation accuracy of influence weights. The secondary normalization process ensures the standardization of weights, meets the needs of subsequent cause judgment, provides more reliable quantitative support for the accurate locking of settlement anomaly causes, and further reduces the probability of misjudgment of causes.

[0108] In one embodiment of this application, determining the construction behavior that leads to abnormal settlement based on influence weights includes:

[0109] All construction behaviors are ranked according to their influence weights, and a predetermined number of the top-ranked construction behaviors are selected as candidate construction behaviors.

[0110] For each candidate construction behavior, the temporal variation characteristics of its corresponding construction activity data within a preset historical period before the time point when the differential settlement zone was identified are obtained, and the abnormal start time and abnormal variation magnitude of the candidate construction behavior are determined based on the temporal variation characteristics.

[0111] If the anomaly starts earlier than the time when the differential settlement zone is identified, the first time condition is met; if the magnitude of the anomaly exceeds the preset anomaly magnitude threshold, the second magnitude condition is met.

[0112] Candidate construction behaviors that simultaneously meet the first time condition and the second magnitude condition are identified as construction behaviors that lead to abnormal settlement.

[0113] In this embodiment, all construction behaviors are first sorted in descending order of their influence weight, with a preset number of 3. This number ensures coverage of all possible dominant construction behaviors while avoiding inefficiency due to an excessive number of candidates. The number can be fine-tuned based on the total number of construction behaviors, ranging from 2 to 5. The top 3 ranked construction behaviors are selected as candidate construction behaviors, and their corresponding influence weight values ​​are retained for priority determination after subsequent verification.

[0114] In this embodiment, the preset historical time period is set to 72 hours before the differential settlement zone is identified. This time period can completely cover the abnormal changes in construction behavior before the settlement anomaly occurs, ensuring that abnormal information of candidate construction behaviors can be captured. It can be fine-tuned according to the monitoring accuracy requirements, with a fine-tuning range of 48-96 hours. Through multiple types of sensors, the temporal change characteristics of construction activity data corresponding to each candidate construction behavior are collected within the preset historical time period. The collection frequency is consistent with the construction activity data (shield tunneling parameters once / second, dewatering well and retaining structure parameters once / 30 minutes), ensuring the completeness and accuracy of the temporal change characteristics.

[0115] In this embodiment, based on the acquired temporal change characteristics and combined with the preset normal construction threshold range, the abnormal start time and abnormal change amplitude of each candidate construction behavior are determined. The abnormal start time is the time point when the construction parameter corresponding to the candidate construction behavior first exceeds the normal construction threshold range and the duration is ≥ 1 collection cycle. The abnormal change amplitude is the absolute value of the difference between the maximum value of the construction parameter during the abnormal period and the upper limit (or lower limit, for the lower limit threshold such as the axial force of the retaining structure support) of the normal construction threshold range.

[0116] In this embodiment, the setting and verification of time and amplitude conditions are as follows:

[0117] First time condition: The anomaly start time is earlier than the time when the differential settlement zone was identified, and the time difference is ≥1 acquisition cycle (at least 2 hours), ensuring that the abnormal change in the candidate construction behavior occurred before the settlement anomaly, and has a temporal basis for causation; if the anomaly start time is later than or equal to the time when the differential settlement zone was identified, the first time condition is not met.

[0118] The second amplitude condition: The preset abnormal amplitude threshold is set to 1.5 times the upper (or lower) limit of the normal construction threshold range, consistent with the preset multiple of the previously mentioned correction weight, to ensure the consistency of the technical solution. It can be fine-tuned according to the engineering geological conditions, with a fine-tuning range of 1.2-1.8 times. If the abnormal change amplitude of the candidate construction behavior exceeds this abnormal amplitude threshold, the second amplitude condition is met; if it does not exceed it, the second amplitude condition is not met.

[0119] Determination of the dominant cause: For each candidate construction behavior, a dual-condition verification is performed. The candidate construction behavior that simultaneously satisfies the first time condition and the second amplitude condition is determined as the construction behavior that leads to abnormal settlement. If there are multiple candidate construction behaviors that simultaneously satisfy the dual conditions, the one with the largest influence weight is selected as the dominant construction behavior. If none of the candidate construction behaviors simultaneously satisfy the dual conditions, the construction behavior ranked 4th or 5th is reselected as the candidate construction behavior, and the above verification process is repeated until at least one construction behavior that leads to abnormal settlement is determined.

[0120] For example, a differential settlement zone was identified in a subway tunnel construction section at 12:00 on [Date] (denoted as T0). The construction behavior that caused the settlement anomaly was determined according to the aforementioned method: the construction behavior was first sorted in descending order of influence weight, and the top 3 were selected as candidates, namely, the tunnel advance speed (weight 0.85), the circumferential gap of the segment assembly (weight 0.12), and the water level of the dewatering well (weight 0.03). The preset historical period was 72 hours before T0, the anomaly amplitude threshold was 1.5 times the normal construction threshold, and the first time condition required that the anomaly start time was earlier than T0 and the time difference was ≥2 hours.

[0121] The temporal variation characteristics of the candidate construction behaviors were collected and verified: the normal threshold for shield tunneling speed is 30-50 mm / min, and its abnormal start time is 18:00, 72 hours before T0, which is earlier than T0 and the time difference meets the requirements. The abnormal maximum value is 75 mm / min, and the variation range is 25 mm / min, which reaches 1.5 times the upper limit of the normal threshold, thus meeting the dual conditions. The normal threshold for the circumferential gap of the segment assembly is 2-5 mm, and the abnormal maximum value is 6 mm. The variation range does not reach 1.5 times the threshold, thus not meeting the second amplitude condition. The abnormal start time of the dewatering well water level is 1 hour before T0, and the time difference is less than 2 hours, thus not meeting the first time condition.

[0122] Verification revealed that only the tunnel boring machine's advance speed met both conditions simultaneously, thus it was determined to be the construction activity that caused this settlement anomaly and was therefore the core target for handling this settlement anomaly.

[0123] In this embodiment, construction behaviors are first ranked according to their influence weights to filter out candidate construction behaviors with higher weights, thus narrowing the scope of cause determination and improving verification efficiency. Then, a preset historical period prior to the occurrence of settlement anomalies is determined, and the temporal change characteristics of candidate construction behaviors within this period are collected to accurately extract the anomaly start time and the magnitude of the anomaly change, providing quantitative evidence for cause verification. Next, clear time and magnitude conditions are set to perform dual verification on candidate construction behaviors. The time condition ensures that the anomaly of the candidate construction behavior occurs before the settlement anomaly, establishing a causal relationship; the magnitude condition ensures that the anomaly of the candidate construction behavior reaches a level sufficient to trigger the settlement anomaly. Finally, candidate construction behaviors that simultaneously meet both conditions are identified as construction behaviors causing settlement anomalies. If multiple conditions exist, the one with the highest weight is selected as the dominant cause, achieving accurate cause determination and avoiding misjudgment and omission.

[0124] As can be seen from the above, the screening of candidate construction behaviors narrows the scope of cause verification, improves the efficiency of cause determination, and ensures that no possible dominant causes are overlooked; the reasonable setting of preset historical time periods ensures that abnormal change information of candidate construction behaviors can be fully captured, providing sufficient data support for verification; the verification of both time and magnitude conditions effectively eliminates candidate construction behaviors that have no direct causal relationship with settlement anomalies, solves the problem of cause misjudgment in existing technologies, and significantly improves the accuracy of cause determination; the priority determination of multiple candidate construction behaviors ensures that the dominant construction behavior can be accurately identified, providing a clear and reliable basis for targeted handling of settlement anomalies, further improving the practicality of construction settlement monitoring methods, and helping to reduce the probability of construction settlement accidents.

[0125] Based on the same principle as the construction settlement monitoring method provided in the embodiments of this application, the embodiments of this application also provide a construction settlement monitoring device, such as... Figure 2 As shown, the construction settlement monitoring device 20 may specifically include: a data acquisition module 21, a dataset generation module 22, a settlement analysis module 23, a correlation calculation module 24, and a settlement monitoring module 25.

[0126] Data acquisition module 21 is used to acquire surface settlement data and construction activity data in the construction tunnel area. Both surface settlement data and construction activity data are time-stamped. Surface settlement data is data that characterizes the overall settlement distribution in the construction tunnel area and the surrounding preset area. Construction activity data is data that reflects the construction status and behavioral parameters during the subway construction process.

[0127] The dataset generation module 22 is used to divide the construction tunnel area into multiple monitoring grids and label the corresponding spatial coordinates for each monitoring grid; it matches the construction activity data and areal settlement data with the spatial coordinates of the corresponding monitoring grids to generate a construction settlement dataset.

[0128] Settlement analysis module 23 is used to identify, based on the construction settlement dataset, the continuous area corresponding to the monitoring grid where the settlement change rate exceeds the first preset threshold, as the settlement change rate abnormal segment; and to perform cluster analysis on all monitoring grids based on the settlement difference between each monitoring grid and adjacent monitoring grids, to identify the continuous area where the settlement difference exceeds the second preset threshold, as the differential settlement zone.

[0129] The correlation calculation module 24 is used to extract the settlement time sequence characteristics and construction procedure time sequence characteristics associated with the settlement change rate anomaly segment and differential settlement zone; and based on the settlement time sequence characteristics and construction procedure time sequence characteristics, obtain the correlation coefficient between each construction behavior and settlement anomaly;

[0130] Settlement monitoring module 25 is used to obtain the influence weight of each construction behavior on differential settlement based on the correlation coefficient, and to determine the construction behavior that leads to settlement anomalies based on the influence weight.

[0131] In one embodiment of this application, the settlement analysis module 23 is specifically used for:

[0132] Based on the construction settlement dataset, the settlement change rate of each monitoring grid is calculated over time, and a time series of the settlement change rate of each monitoring grid is generated.

[0133] In the time series of settlement change rate, identify the time interval in which the settlement change rate continuously exceeds the first preset threshold within the preset collection period;

[0134] Spatially aggregate the monitoring grids corresponding to the time series intervals to obtain the abnormal settlement change rate segments.

[0135] In one embodiment of this application, the settlement analysis module 23 is further used for:

[0136] Based on the construction settlement dataset, the settlement difference between each monitoring grid and all its adjacent monitoring grids is calculated, resulting in a set of settlement difference values ​​between each monitoring grid and all its adjacent monitoring grids;

[0137] The statistical characteristics of the set of settlement difference values ​​corresponding to each monitoring grid are used as the clustering feature vector to characterize the degree of uneven settlement of the monitoring grid.

[0138] Cluster analysis is performed on the feature vectors of each cluster to divide the multiple monitoring grids into multiple clusters;

[0139] Clusters with a mean cluster feature vector exceeding a second preset threshold and whose monitoring grids form a continuous surface in space are selected from multiple clusters. The area corresponding to this cluster is designated as the differential settlement zone.

[0140] In one embodiment of this application, the association calculation module 24 is specifically used for:

[0141] The settlement time sequence characteristics and the corresponding construction procedure time sequence characteristics are aligned and combined according to time steps to obtain a combined time sequence characteristic sequence;

[0142] Based on the combined temporal feature sequence, and through the encoding process of the combined temporal feature sequence using a temporal correlation model, the deep correlation features between construction behavior and settlement anomalies in the time dimension are obtained; the temporal correlation model is trained based on the correspondence between historical settlement temporal features and historical construction procedure temporal features.

[0143] A linear transformation is applied to the deep association features to map them to one-dimensional original association values;

[0144] The one-dimensional original correlation value is transformed nonlinearly using the first formula to map the one-dimensional original correlation value into the correlation coefficient between each construction behavior and settlement anomaly.

[0145] The first formula is:

[0146] ;

[0147] in, Let Y represent the correlation coefficient between the i-th construction action and the settlement anomaly, and Y represent the one-dimensional original correlation value.

[0148] In one embodiment of this application, the settlement monitoring module 25 is specifically used for:

[0149] An importance analysis was performed on the correlation coefficients corresponding to each construction activity to obtain the initial influence weight of each construction activity.

[0150] Obtain the temporal variation characteristics of construction activity data corresponding to each construction behavior within the historical period associated with the differential settlement zone;

[0151] Based on the time-series variation characteristics, the initial influence weights of each construction activity are corrected to obtain the influence weight of each construction activity on differential settlement.

[0152] In one embodiment of this application, the settlement monitoring module 25 is further configured to:

[0153] Based on the characteristics of temporal changes, determine whether the construction parameters of each construction activity have undergone abnormal changes in the historical period, and obtain the start time and magnitude of the abnormal changes.

[0154] Based on the chronological relationship between the start time and the time when the differential settlement zone was identified, as well as the degree of deviation between the change magnitude and the corresponding normal construction threshold range, the initial influence weight of each construction behavior is dynamically adjusted to obtain the influence weight of each construction behavior on differential settlement.

[0155] The dynamic adjustment includes: increasing the initial impact weight for construction activities that occur earlier than the settlement anomaly identification time and whose change exceeds the normal construction threshold range by a preset multiple; and decreasing the initial impact weight for construction activities that occur later than the settlement anomaly identification time or whose change does not exceed the normal construction threshold range.

[0156] In one embodiment of this application, the settlement monitoring module 25 is further configured to:

[0157] All construction behaviors are ranked according to their influence weights, and a predetermined number of the top-ranked construction behaviors are selected as candidate construction behaviors.

[0158] For each candidate construction behavior, the temporal variation characteristics of its corresponding construction activity data within a preset historical period before the time point when the differential settlement zone was identified are obtained, and the abnormal start time and abnormal variation magnitude of the candidate construction behavior are determined based on the temporal variation characteristics.

[0159] If the anomaly starts earlier than the time when the differential settlement zone is identified, the first time condition is met; if the magnitude of the anomaly exceeds the preset anomaly magnitude threshold, the second magnitude condition is met.

[0160] Candidate construction behaviors that simultaneously meet the first time condition and the second magnitude condition are identified as construction behaviors that lead to abnormal settlement.

[0161] The apparatus in this application embodiment can execute the method provided in this application embodiment, and the implementation principle is similar. The actions performed by each module in the apparatus of each embodiment of this application correspond to the steps in the method of each embodiment of this application. For detailed functional descriptions of each module of the apparatus, please refer to the descriptions in the corresponding methods shown above, which will not be repeated here.

[0162] Figure 3 A schematic diagram of the structure of an electronic device to which this application embodiment applies is shown, such as... Figure 3 As shown, the electronic device can be used to implement the methods provided in any embodiment of this application.

[0163] like Figure 3 As shown, the electronic device 300 may primarily include at least one processor 301. Figure 3 The diagram shows components such as a memory 302, a communication module 303, and an input / output interface 304. Optionally, these components can be connected and communicate with each other via a bus 305. It should be noted that... Figure 3 The structure of the electronic device 300 shown is merely illustrative and does not constitute a limitation on the electronic devices to which the methods provided in the embodiments of this application are applicable.

[0164] The memory 302 can be used to store operating systems and applications, etc. The applications can include computer programs that implement the methods shown in the embodiments of this application when invoked by the processor 301, and can also include programs for implementing other functions or services. The memory 302 can be ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices that can store information and computer programs, or it can be EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited thereto.

[0165] Processor 301 is connected to memory 302 via bus 305 and implements corresponding functions by calling the application programs stored in memory 302. Processor 301 can be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 301 can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0166] Electronic device 300 can connect to a network via communication module 303 (which may include, but is not limited to, components such as a network interface) to communicate with other devices (such as user terminals or servers) through the network and achieve data interaction, such as sending data to or receiving data from other devices. Communication module 303 may include wired network interfaces and / or wireless network interfaces, meaning the communication module may include at least one of wired or wireless communication modules.

[0167] The electronic device 300 can connect to necessary input / output devices, such as a keyboard and display device, via the input / output interface 304. The electronic device 300 itself may have a display device, and other display devices can also be connected externally via the interface 304. Optionally, a storage device, such as a hard drive, can also be connected via the interface 304 to store data from the electronic device 300, read data from the storage device, or store data from the storage device in the memory 302. It is understood that the input / output interface 304 can be a wired interface or a wireless interface. Depending on the actual application scenario, the device connected to the input / output interface 304 can be a component of the electronic device 300 or an external device connected to the electronic device 300 when needed.

[0168] The bus 305 used to connect the components may include a path for transmitting information between the components. The bus 305 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Depending on its function, the bus 305 may be divided into an address bus, a data bus, a control bus, etc.

[0169] Optionally, for the solution provided in the embodiments of this application, the memory 302 can be used to store a computer program that executes the solution of this application, and the processor 301 runs the computer program. When the processor 301 runs the computer program, it implements the operation of the method or apparatus provided in the embodiments of this application.

[0170] Based on the same principle as the method provided in the embodiments of this application, the embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, can implement the corresponding content of the aforementioned method embodiments.

[0171] This application also provides a computer program product, which includes a computer program that, when executed by a processor, can implement the corresponding content of the aforementioned method embodiments.

[0172] It should be noted that the terms "first," "second," "third," "fourth," "1," "2," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in a sequence other than that shown in the figures or text.

[0173] In the embodiments of this application, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0174] It should be understood that although arrows indicate various operation steps in the flowcharts of this application's embodiments, the order in which these steps are implemented is not limited to the order indicated by the arrows. Unless explicitly stated herein, in some implementation scenarios of this application's embodiments, the implementation steps in each flowchart can be executed in other orders as required. Furthermore, some or all steps in each flowchart, based on the actual implementation scenario, may include multiple sub-steps or multiple stages. Some or all of these sub-steps or stages can be executed at the same time, and each sub-step or stage can also be executed at different times. In scenarios where execution times differ, the execution order of these sub-steps or stages can be flexibly configured according to requirements, and this application's embodiments do not limit this.

[0175] The above description is only an optional implementation method for some implementation scenarios of this application. It should be noted that for those skilled in the art, other similar implementation methods based on the technical concept of this application without departing from the technical concept of this application also fall within the protection scope of the embodiments of this application.

Claims

1. A method for monitoring construction settlement, characterized in that, include: Acquire surface settlement data and construction activity data of the construction tunnel area. Both surface settlement data and construction activity data carry timestamps. The surface settlement data is data that characterizes the overall settlement distribution of the construction tunnel area and the surrounding preset area. The construction activity data is data that reflects the construction status and behavioral parameters during the subway construction process. The construction tunnel area is divided into multiple monitoring grids, and the corresponding spatial coordinates are marked for each monitoring grid. The construction activity data and the areal settlement data are matched with the spatial coordinates of the corresponding monitoring grids to generate a construction settlement dataset. Based on the construction settlement dataset, continuous areas corresponding to monitoring grids whose settlement change rate exceeds a first preset threshold are identified as abnormal settlement change rate segments. Based on the settlement difference between each monitoring grid and its adjacent monitoring grids, cluster analysis is performed on all monitoring grids to identify continuous areas where the settlement difference exceeds a second preset threshold, which are then used as differential settlement zones. Extract settlement time sequence features and construction procedure time sequence features associated with the settlement change rate anomaly segment and the differential settlement zone; and based on the settlement time sequence features and the construction procedure time sequence features, obtain the correlation coefficient between each construction behavior and the settlement anomaly; Based on the correlation coefficient, the influence weight of each construction action on differential settlement is obtained, and the construction action that leads to abnormal settlement is determined based on the influence weight.

2. The construction settlement monitoring method as described in claim 1, characterized in that, The step of identifying continuous regions corresponding to monitoring grids whose settlement change rate exceeds a first preset threshold based on the construction settlement dataset, as abnormal settlement change rate segments, includes: Based on the construction settlement dataset, the settlement change rate of each monitoring grid is calculated over time, and a time series of the settlement change rate of each monitoring grid is generated. In the time series of the settlement change rate, identify the time interval in which the settlement change rate continuously exceeds the first preset threshold within the preset collection period; Spatially aggregate the monitoring grids corresponding to the time intervals to obtain the abnormal settlement change rate segments.

3. The construction settlement monitoring method as described in claim 1, characterized in that, The method involves clustering all monitoring grids based on the settlement difference between each monitoring grid and its adjacent grids to identify continuous areas where the settlement difference exceeds a second preset threshold, as differential settlement zones. Based on the construction settlement dataset, the settlement difference between each monitoring grid and all its adjacent monitoring grids is calculated to obtain the set of settlement difference values ​​between each monitoring grid and all its adjacent monitoring grids; The statistical characteristics of the set of settlement difference values ​​corresponding to each monitoring grid are used as the clustering feature vector to characterize the degree of uneven settlement of the monitoring grid. Cluster analysis is performed on each of the clustering feature vectors to divide the multiple monitoring grids into multiple clusters; From the plurality of clusters, select clusters whose average cluster feature vector exceeds the second preset threshold and whose monitoring grids form a continuous surface in space, and designate the region corresponding to the cluster as the differential settlement zone.

4. The construction settlement monitoring method as described in claim 1, characterized in that, Based on the settlement time sequence characteristics and the construction procedure time sequence characteristics, the correlation coefficients between each construction behavior and settlement anomalies are obtained, including: The settlement time sequence features and the corresponding construction procedure time sequence features are aligned and combined according to time steps to obtain a combined time sequence feature sequence. Based on the combined time-series feature sequence, and by encoding the combined time-series feature sequence through a time-series correlation model, a deep correlation feature between construction behavior and settlement anomaly in the time dimension is obtained; the time-series correlation model is trained based on the correspondence between historical settlement time-series features and historical construction procedure time-series features. A linear transformation is performed on the deep association features to map them into one-dimensional original association values; The one-dimensional original correlation value is nonlinearly transformed by the first formula to map the one-dimensional original correlation value into the correlation coefficient between each construction behavior and settlement anomaly. The first formula is: ； in, Let Y represent the correlation coefficient between the i-th construction action and the settlement anomaly, and Y represent the one-dimensional original correlation value.

5. The construction settlement monitoring method as described in claim 1, characterized in that, Based on the correlation coefficient, the weight of each construction action on differential settlement is obtained, including: An importance analysis was performed on the correlation coefficients corresponding to each construction activity to obtain the initial influence weight of each construction activity. Obtain the temporal variation characteristics of construction activity data corresponding to each construction activity within the historical period associated with the differential settlement zone; Based on the aforementioned temporal variation characteristics, the initial influence weights of each construction activity are corrected to obtain the influence weight of each construction activity on differential settlement.

6. The construction settlement monitoring method as described in claim 5, characterized in that, The initial influence weights of each construction activity are corrected based on the time-series change characteristics to obtain the influence weight of each construction activity on differential settlement, including: Based on the time-series change characteristics, determine whether the construction parameters of each construction activity have undergone abnormal changes during the historical period, and obtain the start time and magnitude of the abnormal changes. Based on the chronological relationship between the start time and the time point when the differential settlement zone was identified, and the degree of deviation between the change amplitude and the corresponding normal construction threshold range, the initial influence weight of each construction behavior is dynamically adjusted to obtain the influence weight of each construction behavior on differential settlement. The dynamic adjustment includes: increasing the initial impact weight for construction activities that occur earlier than the settlement anomaly identification time and whose change exceeds the normal construction threshold range by a preset multiple; and decreasing the initial impact weight for construction activities that occur later than the settlement anomaly identification time or whose change does not exceed the normal construction threshold range.

7. The construction settlement monitoring method as described in claim 1, characterized in that, The determination of construction behaviors leading to abnormal settlement based on the influence weights includes: All construction behaviors are sorted according to the influence weights, and a predetermined number of construction behaviors with the highest ranking are selected as candidate construction behaviors. For each candidate construction behavior, the temporal variation characteristics of its corresponding construction activity data within a preset historical period before the time point when the differential settlement zone is identified are obtained, and the abnormal start time and abnormal variation magnitude of the candidate construction behavior are determined based on the temporal variation characteristics. If the anomaly starts earlier than the time when the differential settlement zone is identified, the first time condition is satisfied; if the magnitude of the anomaly exceeds a preset anomaly magnitude threshold, the second magnitude condition is satisfied. Candidate construction behaviors that simultaneously satisfy the first time condition and the second amplitude condition are identified as construction behaviors that lead to the settlement anomaly.

8. A construction settlement monitoring device, characterized in that, include: The data acquisition module is used to acquire surface settlement data and construction activity data in the construction tunnel area. Both the surface settlement data and the construction activity data carry timestamps. The surface settlement data is data that characterizes the overall settlement distribution in the construction tunnel area and the surrounding preset area, and the construction activity data is data that reflects the construction status and behavioral parameters during the subway construction process. The dataset generation module is used to divide the construction tunnel area into multiple monitoring grids and label the corresponding spatial coordinates of each monitoring grid; the construction activity data and the areal settlement data are matched with the spatial coordinates of the corresponding monitoring grids to generate a construction settlement dataset; The settlement analysis module is used to identify, based on the construction settlement dataset, the continuous area corresponding to the monitoring grid where the settlement change rate exceeds a first preset threshold, as an abnormal settlement change rate segment. Based on the settlement difference between each monitoring grid and its adjacent monitoring grids, cluster analysis is performed on all monitoring grids to identify continuous areas where the settlement difference exceeds a second preset threshold, which are then used as differential settlement zones. The correlation calculation module is used to extract the settlement time sequence features and construction procedure time sequence features associated with the settlement change rate anomaly segment and the differential settlement zone; and based on the settlement time sequence features and the construction procedure time sequence features, to obtain the correlation coefficient between each construction behavior and the settlement anomaly. The settlement monitoring module is used to obtain the influence weight of each construction behavior on differential settlement based on the correlation coefficient, and to determine the construction behavior that causes settlement abnormalities based on the influence weight.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the construction settlement monitoring method according to any one of claims 1 to 7 when running the computer program.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the construction settlement monitoring method according to any one of claims 1 to 7.

Images