A method for safety evaluation of open-cut tunnel foundation pit support structure considering time-space effect

Abstract

The application discloses a kind of open cut tunnel foundation pit support structure safety evaluation method considering space-time effect, it is related to tunnel engineering and foundation pit engineering safety monitoring technical field;Method includes: obtaining open cut tunnel foundation pit geological parameter, design parameter, construction parameter and construction progress node sequence;Similar engineering is filtered based on geological parameter;Similarity is calculated based on design, construction parameter and construction sequence matching coefficient, and effective sample is filtered;Similarity is used as weight to construct prior distribution;Bayesian update is carried out in combination with the measured data of node that has been constructed, and the expected response interval of each partition under each node is generated;According to the measured data of current node, the path deviation degree relative to the median value in expected interval is calculated;Linear regression is carried out to the deviation degree of multiple nodes, and the trajectory form is identified according to regression slope and goodness of fit, and risk early warning is determined when it is divergent trajectory;The application is helpful to identify potential risk in the early stage of foundation pit support structure response deviating from normal evolution trajectory.

Description

Technical Field

[0001] This invention relates to the field of safety monitoring technology for tunnel engineering and foundation pit engineering. More specifically, this invention relates to a safety assessment method for the support structure of open-cut tunnel foundation pit that takes into account spatiotemporal effects. Background Technology

[0002] In cut-and-cover tunnel construction, the safety assessment of the foundation pit support structure is a crucial step in ensuring project safety. Currently, the project primarily uses a method of comparing real-time monitoring data with preset thresholds for safety assessment; an alarm is triggered when the monitoring data exceeds the threshold.

[0003] However, this method is a reactive alarm; when the data triggers the threshold, the support structure may already be in a dangerous state, making it difficult to allow sufficient time for construction adjustments. Furthermore, existing assessment methods typically use uniform static thresholds, failing to adequately consider the stress differences at different spatial locations (such as corners and the center) and the impact of construction progress (such as excavation and support procedures) on the structural response, resulting in a lack of specificity in the assessment benchmark.

[0004] Existing methods for assessing the safety status of retaining structures calculate bending moments by smoothing and optimizing inclinometer data and then comparing them with the bending moment limit values. While this method improves the reliability of inclinometer data, its assessment is still based on a comparison between current measured data and design limit values, constituting a post-hoc judgment, and it does not consider the cumulative impact of construction progress on the internal forces of the structure. Other methods for assessing the performance of foundation pit support extract signal features through waveform analysis and combine them with dynamic evaluation models for performance evaluation.

[0005] Although this method introduces a dynamic adjustment mechanism, its evaluation benchmark is mainly based on the changing trend of the real-time data itself, and it does not make full use of the experience information of similar historical projects, making it difficult to identify risks in the early stages when the structural response deviates from the normal trajectory. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a safety assessment method for open-cut tunnel foundation pit support structures that considers spatiotemporal effects. This method solves the problems of existing safety assessment methods for open-cut tunnel foundation pit support structures, such as reliance on post-event threshold triggering, insufficient consideration of spatiotemporal effects leading to a lack of specificity in the assessment benchmark, and difficulty in identifying risks in the early stages when the structural response deviates from the normal evolution trajectory.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a safety assessment method for open-cut tunnel foundation pit support structures considering spatiotemporal effects, comprising the following steps: S1: Establishing a spatiotemporal parameter system Geological, design, and construction parameters for each spatial zone of the current open-cut tunnel foundation pit project are obtained to establish basic data reflecting the spatial differences within the foundation pit. The entire foundation pit excavation process is discretized into a sequence of construction progress nodes according to key procedures, with a total number of nodes. This is used to align the structural response at different construction stages. The spatial partition is divided into multiple main segments along the longitudinal direction, and each main segment is divided into multiple sub-regions along the transverse direction, forming a two-dimensional grid, thereby assessing the safety status of different spatial locations separately.

[0008] S2: Screening of Geologically Similar Projects Based on the aforementioned geological parameters, the spatial zoning geological parameters of all projects in the historical engineering database are clustered to form multiple geological pattern categories, which quantify the similarity of geological conditions. The Euclidean distance between the geological parameters of each spatial zoning of the current project and each cluster center is calculated, and the projects are assigned to the category with the smallest distance, thereby identifying the group of historical projects most similar to the current project in terms of geological conditions. Historical projects belonging to the same category and having the same longitudinal segment and lateral position are selected to form a geologically similar project set to ensure spatial alignment.

[0009] S3: Calculation of similarity between design and construction parameters and selection of effective samples For each spatial partition, based on design parameters, construction parameters, and construction sequence matching coefficients, the similarity between the current project and each historical project in the geologically similar project set is calculated. Historical projects with similarity greater than a preset threshold are selected as the effective similar sample set, thereby further filtering out projects from the geologically similar projects that have highly consistent design, construction, and construction sequence. The construction sequence matching coefficient takes a first preset value or a second preset value depending on whether the longitudinal segmented excavation sequence is consistent, to reflect the impact of differences in construction timing on the structural response.

[0010] S4: Prior Distribution Construction Structural response monitoring data of historical projects in the same spatial partition and at the same construction progress node are extracted from the effective similar sample set. The weighted mean and weighted variance are calculated with similarity as the weight, and a Gaussian prior distribution is constructed to quantify the typical response patterns and dispersion of similar projects under the same spatiotemporal conditions.

[0011] S5: Bayesian Update and Spatiotemporal-Response Benchmark Map Generation The measured structural response data of the currently constructed nodes of the project are obtained, and Bayesian updates are performed in combination with the prior distribution to obtain the posterior mean and posterior variance of the constructed nodes. This integrates historical experience with current measured information and reduces uncertainty.

[0012] For nodes not yet constructed, a linear fit is performed on the difference between the measured values ​​and the prior mean of constructed nodes. The fitted value is used as a time extrapolation term to capture the evolution trend of response deviation with construction progress. The posterior deviations of adjacent sub-regions at the same node are weighted and averaged using the inverse of distance as the weight. This weighted average is used as a spatial extrapolation term to correct predictions using spatial correlation. The time and spatial extrapolation terms are then weighted and fused according to the proportion of constructed nodes to the total number of nodes to obtain the posterior mean of nodes not yet constructed. This approach relies more on spatial information in the early stages of construction and more on temporal trends in the later stages.

[0013] The posterior variance of unconstructed nodes is calculated as the maximum posterior variance of constructed nodes multiplied by (2 - time goodness of fit). A higher time goodness of fit results in less variance amplification, reflecting the reliability of the extrapolated prediction. The quantiles corresponding to the pre-set confidence level are used. The posterior standard deviation is used as the interval half-width to obtain the expected response intervals for each construction progress node and each spatial partition, forming a spatiotemporal-response benchmark map, which provides a dynamic comparison benchmark for subsequent real-time monitoring.

[0014] S6: Real-time Path Deviation Calculation The current construction progress node is determined based on the real-time construction progress. Measured data for the current node and its corresponding spatial partition are obtained. The expected median is read from the baseline map. The path deviation is calculated as: (Measured data - Expected median) / (|Expected median| + ... ,in A preset positive number is used to make the degree of deviation dimensionless and avoid the denominator being too small, thereby obtaining a deviation index that can be compared across partitions and nodes.

[0015] S7: Deviation Pattern Recognition and Risk Warning For the same spatial partition continuous Linear regression was performed on the path deviation of each node to obtain the slope and goodness of fit. A preset positive integer is used to quantify the changing trend and linearity of the deviation sequence. When this... When all nodes have the same sign of path deviation, and the absolute value of the slope is greater than a preset slope threshold, and the goodness of fit is greater than a preset goodness of fit threshold, the trajectory is identified as divergent and an early warning is issued, thus identifying potential risks when the deviation trend is significant and stable. If at least two adjacent spatial partitions are on the same contiguous... If all nodes within a window are identified as divergent trajectories with the same path deviation sign, the warning level is raised to reflect that spatial inconsistencies may originate from overall construction issues.

[0016] Furthermore, the longitudinal main section includes the end well section, the standard section, and the intersection section with the existing structure; the transverse sub-regions include the left wall region, the middle section of the foundation, the right wall region, and the corner region, respectively corresponding to the stress characteristics of different locations in the tunnel foundation pit. The construction progress node sequence is as follows: completion of the first layer of earthwork excavation for each longitudinal segment, erection of the first support and application of prestress, completion of the second layer of earthwork excavation, erection of the second support, and completion of the tunnel floor slab pouring, to cover the key procedures of the entire excavation and support process.

[0017] Furthermore, the clustering in step S2 employs the K-means algorithm, with the number of clusters preset to a fixed positive integer, to group historical projects with similar geological characteristics into the same category. Before calculating the Euclidean distance, each component of the geological parameters is subjected to min-max normalization to eliminate the influence of dimensions and ensure comparability of different parameters in the distance calculation.

[0018] Furthermore, the design parameters in step S3 include: pile diameter or wall thickness, pile length or wall depth, reinforcement ratio, concrete strength grade, support stiffness, support spacing, and support prestress, to comprehensively characterize the bearing capacity and stiffness characteristics of the support structure. The construction parameters include: layer thickness, excavation duration for each layer, time nodes for each support erection, and longitudinal segmented excavation sequence, to reflect the temporal impact of the construction process on the structural response. Each parameter is min-max normalized before calculating similarity to unify the dimensions. The similarity is calculated based on the product of weighted Euclidean distance and the construction sequence matching coefficient. The weights of the weighted Euclidean distance are determined by the analytic hierarchy process (AHP) or the entropy weight method to reflect the differences in the degree of influence of each parameter on the structural response.

[0019] Furthermore, the geological parameters include soil layer thickness, unit weight, cohesion, internal friction angle, and compression modulus for each spatial zone, in order to fully describe the mechanical and deformation characteristics of the soil and provide a basis for judging geological similarity.

[0020] Furthermore, the prior distribution is a Gaussian distribution to simplify Bayesian calculations by utilizing its mathematical closure. In the Bayesian update, the observation error variance is a preset positive number to characterize the measurement uncertainty of the monitoring instrument. The posterior mean of the constructed nodes is a weighted average of the prior mean and the measured value, with the weights determined by the prior variance and the observation error variance, thereby achieving adaptive fusion of prior information and measured data.

[0021] Furthermore, in the path deviation calculation formula... Let be a preset positive number, and It is much smaller than the typical order of magnitude of the expected median, so as to prevent numerical anomalies caused by an excessively small denominator when the expected median is close to zero, without affecting the relative comparison of the deviation.

[0022] Furthermore, the preset positive integer The preset slope threshold and preset goodness-of-fit threshold are set according to the engineering risk level to adjust the sensitivity and robustness of trend identification. The condition for raising the warning level is that at least two adjacent spatial partitions are in the same contiguous... Within a window comprised of individual nodes, all trajectories are identified as divergent and share the same path deviation sign, thus identifying spatial consistency anomalies that may be caused by overall construction factors. The warning information includes spatial partition identifiers, current node numbers, path deviation sequences, regression slopes, spatiotemporal consistency verification results, and suggested construction procedures for investigation, guiding targeted measures on-site.

[0023] Furthermore, this also includes: after the completion of the current project, adding the parameters of each spatial partition and the monitoring data of all construction progress nodes to the historical project database after min-max normalization, in order to expand the historical sample library and obtain richer empirical basis for the similarity retrieval and prior distribution update of subsequent projects.

[0024] Furthermore, the preset positive integer The preset slope threshold, preset goodness-of-fit threshold, the first preset value, the second preset value, and the preset similarity threshold in step S3, and the quantile corresponding to the preset confidence level in step S5 are all fixed values ​​preset according to the engineering risk level, used to ensure that the rigor of the evaluation method matches the engineering safety requirements.

[0025] This invention provides a safety assessment method for the support structure of open-cut tunnel foundation pits that considers spatiotemporal effects. Compared with existing technologies, it has the following advantages: This invention obtains detailed geological parameters, retaining structure design parameters, and construction parameters for each spatial zone of the foundation pit, and discretizes the entire excavation process into a sequence of construction progress nodes according to key procedures. This process concretizes abstract spatiotemporal effects into quantifiable spatial zones and construction progress nodes, laying the foundation for subsequently constructing an evaluation benchmark that matches spatiotemporal conditions. By establishing zones for different locations of the foundation pit (such as corners and the middle) and defining progress nodes according to construction procedures (such as excavation completion and support erection), the evaluation model can establish personalized response expectations for each spatial location at each construction stage. This helps to improve the problem of the lack of specificity in the evaluation benchmark caused by the use of a uniform static threshold in traditional methods.

[0026] This invention uses geological parameter clustering to screen projects from a historical engineering database that have similar geological conditions to current engineering projects. Then, based on weighted similarity calculations using design and construction parameters, it further filters out historical projects that are highly similar in geology, design, and construction as valid samples. This multi-level similarity screening mechanism helps ensure that historical experience data has high reference value. Using similarity as weight, a prior distribution (Gaussian distribution) of the structural response of each zone at each construction progress node is constructed, quantifying the typical response patterns and dispersion of similar projects under the same conditions. This prior distribution provides a reasonable initial estimate for subsequent Bayesian updates, ensuring that the evaluation benchmark is supported by historical experience.

[0027] This invention combines measured data from completed nodes of the current project to perform a Bayesian update on the prior distribution, generating a posterior distribution and calculating the expected response interval. For nodes not yet constructed, the posterior mean is extrapolated by linearly fitting the observed deviation (the difference between the measured value and the prior mean), allowing the baseline spectrum to be dynamically adjusted as construction progresses. This update mechanism integrates historical experience with measured information from the current project, ensuring that the expected response interval inherits the common patterns of similar projects while reflecting the unique characteristics of the current project. By incorporating prior uncertainty and observation errors into the calculation using Bayesian formulas, the resulting expected response interval has a clear probabilistic meaning, providing a quantitative basis for subsequent deviation assessment.

[0028] This invention transforms the structural response state into a dimensionless relative deviation index by calculating the path deviation of measured data relative to the median of the expected response interval. This index helps eliminate comparison barriers caused by differences in dimensions and numerical ranges between different zones and nodes, providing a unified measurement standard for the degree of deviation. Furthermore, a sliding window linear regression is performed on the deviation of multiple consecutive construction progress nodes within the same zone to calculate the regression slope and goodness of fit. When multiple consecutive nodes have the same sign of deviation and the absolute value of the regression slope exceeds a preset threshold, while the goodness of fit is also higher than a preset threshold, it is identified as a divergent trajectory, and a risk warning is issued. This trend identification mechanism helps to capture risks in the early stages when the structural response has not yet exceeded the traditional threshold but has already continuously deviated from the expected trajectory, allowing engineers time to take countermeasures. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] First Embodiment Please see Figure 1 This application provides a safety assessment method for open-cut tunnel foundation pit support structures that considers spatiotemporal effects, comprising the following steps: S1: Obtain the geological parameters, design parameters, and construction parameters for each spatial zone of the current open-cut tunnel foundation pit project; discretize the entire foundation pit excavation process into a sequence of construction progress nodes according to key procedures, with a total number of nodes. The spatial partition is divided into multiple main segments along the longitudinal direction, and each main segment is divided into multiple sub-regions along the transverse direction, forming a two-dimensional grid. S2: Based on the geological parameters, cluster the spatial partition geological parameters of all projects in the historical engineering database to form multiple geological model categories; calculate the Euclidean distance between the geological parameters of each spatial partition of the current project and each cluster center, and classify them into the category with the smallest distance; select historical projects that belong to the same category and have the same vertical segment and horizontal position to form a set of geologically similar projects; S3: For each spatial partition, based on design parameters, construction parameters, and construction sequence matching coefficient, calculate the similarity between the current project and each historical project in the geologically similar project set, and select historical projects with similarity greater than a preset threshold as the effective similar sample set; the construction sequence matching coefficient takes a first preset value or a second preset value depending on whether the longitudinal segment excavation sequence is consistent. S4: Extract structural response monitoring data of each historical project in the same spatial partition and the same construction progress node from the effective similar sample set, calculate the weighted mean and weighted variance with similarity as the weight, and construct a Gaussian prior distribution; S5: Obtain the measured structural response data of the currently constructed nodes of the project, combine it with the prior distribution to perform Bayesian update, and obtain the posterior mean and posterior variance of the constructed nodes. For nodes that have not been constructed, the difference between the measured value and the prior mean of the constructed nodes is linearly fitted, and the fitted value is used as the time extrapolation term. The posterior bias of adjacent partitions at the same node is weighted by the inverse of the distance, and the weighted average is used as the spatial extrapolation term. The time extrapolation term and the spatial extrapolation term are weighted and merged according to the proportion of the number of constructed nodes to the total number of nodes to obtain the posterior mean of the unconstructed nodes. The maximum posterior variance of the constructed nodes multiplied by (2 - time goodness of fit) is taken as the posterior variance of the unconstructed nodes. Based on the quantiles corresponding to the preset confidence level The posterior standard deviation is used as the interval half-width to obtain the expected response intervals for each construction progress node and each spatial partition, thus forming a spatiotemporal-response benchmark map. S6: Determine the current construction progress node based on the real-time construction progress, obtain the measured data of the current node and the corresponding spatial partition, read the expected median from the benchmark map, and calculate the path deviation = (measured data - expected median) / (|expected median| + ... ),in It is a preset positive number; S7: Continuous within the same spatial partition Linear regression was performed on the path deviation of each node to obtain the slope and goodness of fit. It is a preset positive integer; when this When the path deviations of all nodes have the same sign, and the absolute value of the slope is greater than the preset slope threshold, and the goodness of fit is greater than the preset goodness of fit threshold, it is judged as a divergent trajectory and an early warning is issued; if at least two adjacent spatial partitions are in the same continuous... If all nodes within a window are identified as divergent trajectories and have the same path deviation sign, the warning level will be raised.

[0032] Second Embodiment In a specific implementation process, this embodiment is a further implementation of embodiment one, and specifically includes: I. Data Acquisition and Establishment of Spatiotemporal Parameter System First, based on the geometry of the narrow and elongated pit of the open-cut tunnel, the longitudinal variability of geological conditions, and the stiffness distribution of the support structure, the pit is spatially divided into multiple zones.

[0033] The division method is as follows: along the tunnel mileage direction, based on the change points of the geological survey profile, the change points of the support structure design, and the intersection with the existing underground structure, the foundation pit is divided into multiple longitudinal main sections.

[0034] Within each longitudinal main segment, the area along the width of the excavation pit is divided into the left wall zone, the middle section of the foundation, the right wall zone, and the corner zone. The left and right wall zones are located near the left and right retaining structures, respectively; the middle section of the foundation is the remaining central area; and the corner zone is the area where the longitudinal main segment intersects with the transverse boundary.

[0035] This creates a two-dimensional spatial partitioned grid, with each partition denoted as . It has independent geological parameters, design parameters, and construction parameters.

[0036] For each spatial partition Extract layered geological parameters from the engineering geological survey report, including the thickness of each soil layer. (Unit: m) For soil layer number; natural unit weight of each soil layer (Unit: kN / m) 3 ); Cohesion of each soil layer (Unit: kPa); Internal friction angle of each soil layer (Unit: °); Compression modulus of each soil layer (Unit: MPa). Arrange these parameters into a vector form in order of depth.

[0037] Extracting design parameter vectors from support structure design drawings This includes: geometric parameters of the retaining structure (pile diameter) Or wall thickness Unit: m; pile length or wall depth (Unit: m); Material parameters (reinforcement ratio) Dimensionless; axial compressive strength corresponding to concrete strength grade (Unit: MPa); Support system parameters (stiffness of steel or concrete supports) Unit: MN / m; Horizontal spacing of supports Unit: m; Vertical spacing of supports Unit: m; Support prestress value (Unit: kN)

[0038] Extracting construction parameter vectors from the construction organization design scheme This includes: layered excavation parameters (layer thickness) Unit: m; Duration of excavation for each layer of earthwork (Unit: days); Support erection time parameters (the delay time of erection of each support relative to the completion of excavation of the corresponding layer, unit: days); Longitudinal segmented excavation sequence (e.g., excavating sequentially from one end to the other, excavating from both ends to the middle, or skipping segments).

[0039] The entire process of excavating the foundation pit, from the start of excavation to the completion of the bottom slab pouring, is discretized into a sequence of construction progress nodes according to key procedures, with a total number of nodes. .

[0040] The node sequence is constructed as follows: a node is set when the first layer of earthwork excavation of each longitudinal segment is completed; a node is set when the first support of each longitudinal segment is erected and prestressed; a node is set when the second layer of earthwork excavation of each longitudinal segment is completed; a node is set when the second support of each longitudinal segment is erected and prestressed; and so on, until all layers and supports are completed; finally, the last node is set when the tunnel floor concrete is poured.

[0041] The node sequence records both the layering order in the depth direction and the segmentation order in the vertical direction.

[0042] II. Screening of Geologically Similar Projects Establish or access a historical engineering database in advance, which stores data on multiple completed open-cut tunnel foundation projects.

[0043] The data for each project includes: project number and basic attributes; geological parameters, design parameters, and construction parameters for each spatial zone; and structural response monitoring data for each spatial zone at each construction progress node (e.g., deep horizontal displacement, support axial force, and steel reinforcement stress).

[0044] For each spatial partition of each project in the historical engineering database, extract its geological parameters to form a geological feature vector. .

[0045] The extraction method is as follows: Select several layers (usually 3 to 5 layers) with relatively large thickness within the excavation depth range of the foundation pit, and calculate the weighted average unit weight. Weighted average cohesion Weighted average internal friction angle Weighted average compression modulus ,but .

[0046] right Perform min-max normalization on each component: ,in and These are the global minimum and maximum values ​​of this component in the historical database, respectively.

[0047] The K-means algorithm is used to cluster all normalized feature vectors in the historical database. The preset number of clusters is used. It is a fixed positive integer, which is preset based on engineering experience before the system starts.

[0048] The K-means algorithm takes a normalized set of feature vectors as input and outputs the class label of each feature vector and the cluster center of each class. ( ).

[0049] For each spatial partition of the current project The geological feature vectors were extracted and normalized using the same method. .calculate With each cluster center Euclidean distance ,Pick The smallest category is taken as the geological model category to which the partition belongs.

[0050] Then, all historical engineering projects belonging to the same geological model category and having the same vertical segment and the same horizontal position are selected from the historical database to form a set of geologically similar projects for this partition, denoted as […]. .

[0051] III. Calculation of similarity between design and construction parameters and selection of effective samples For geologically similar engineering sets Each historical project Extract its design parameter vector and construction parameter vector The design parameter vector for the current project is denoted as... The construction parameter vector is denoted as .

[0052] right and Each component is subjected to min-max normalization, and the minimum and maximum values ​​of the normalization are taken from the historical database. and The global extrema of each component.

[0053] The weighting coefficients of each design parameter and each construction parameter were determined using the Analytic Hierarchy Process (AHP).

[0054] For the design parameters, a judgment matrix is ​​constructed. Experts compare the importance of each pair of parameters pairwise (using a 1-9 scale). The eigenvector corresponding to the largest eigenvalue of the judgment matrix is ​​calculated, and the weight vector is obtained after normalization. ( ),in The number of design parameters is specified. A consistency check is performed; if the consistency ratio is less than 0.1, the weight allocation is considered reasonable.

[0055] Similarly, a weight vector is obtained from the construction parameters. ( ), This refers to the number of construction parameters.

[0056] For each historical project Calculate its weighted Euclidean distance from the current project. :

[0057] in , These are the current project and the historical project, respectively. Normalized values ​​of each design parameter , The first The normalized values ​​of each construction parameter.

[0058] Define construction sequence matching coefficient Compare the current longitudinal segmented excavation sequence with the excavation sequence of historical projects: if the sequences are completely identical, then... Take the first preset value If the order is different (e.g., reversed or skipped), then Take the second preset value . and All are positive numbers less than or equal to 1, and .

[0059] The formula for calculating similarity is: ,in It is a natural constant. Set a preset similarity threshold. Filter out The historical projects constitute a valid set of similar samples, denoted as... .

[0060] like If empty, then gradually decrease. Re-filter until at least one sample is obtained; if the sample is still empty after dropping to the preset lower limit, use the default prior distribution: take the equally weighted mean of all samples from the same spatial partition in the historical database as the prior distribution. Equal-weighted variance as .

[0061] IV. Construction of Prior Distribution For valid similar sample sets Each historical project Extract its partitions in the same space and each construction progress node ( Structural response monitoring data under ( ) is denoted as Structural response monitoring data can include deep horizontal displacement, support axial force, or steel reinforcement stress.

[0062] If the monitoring data collection time of historical projects is not completely consistent with the completion time of nodes, linear interpolation or the average value within a specified time window before and after that time is used as the representative value.

[0063] For each construction progress node Based on similarity Calculate the weighted mean using the weights. and weighted variance :

[0064] Gaussian distribution As a spatial partition At the construction progress milestones The prior distribution is given below.

[0065] V. Bayesian Update and Spatiotemporal-Response Benchmark Map Generation As construction progresses, the measured structural response data of the current construction progress nodes are obtained and denoted as follows: ,in , This represents the number of completed nodes. Measured data comes from on-site automated monitoring systems (such as inclinometers and axial force gauges). The sampling frequency is set according to project requirements, and the most recent valid reading after the node completion time is taken.

[0066] Let the observation error variance It is a preset positive number, the value of which is determined according to the accuracy level of the monitoring instrument.

[0067] like (If there are no completed construction nodes yet), then the posterior distribution of all construction progress nodes is directly taken as the prior distribution, i.e. , Posterior standard deviation Then, it jumps to the expected response range for calculation.

[0068] like Then, a Bayesian update is performed on the nodes that have already been constructed. For each constructed node... , prior distribution With likelihood function Combined, the posterior distribution is calculated using Bayes' theorem. The posterior distribution is still a Gaussian distribution, with a mean of [missing value]. and variance for:

[0069] Posterior standard deviation .

[0070] For progress milestones that have not yet been constructed We use a weighted fusion method of temporal extrapolation and spatial extrapolation to predict posterior parameters.

[0071] The time extrapolation term is calculated as follows: First, calculate the difference between the constructed nodes. .like Point to point Perform least-squares linear fitting to obtain the fitted line. ,in For node sequence number, The intercept is... The slope is given. The fitting parameters are obtained by solving the normal equation.

[0072] For unconstructed node numbers Time extrapolation item This is the fitted value. The goodness of fit of this fit is denoted as . .like If this happens, linear fitting is not possible, and the time extrapolation term degenerates into... (Using a unique difference as a constant extrapolation), and Take 0.

[0073] The spatial extrapolation term is calculated as follows: for nodes that have not been constructed... Consider the current partition Adjacent partition set ,in Defined as a partition The set of all spatially shared boundaries (including vertically adjacent front and rear segments, and horizontally adjacent left and right sub-regions).

[0074] For each adjacent partition If the partition is on node The posterior mean has already been calculated. and prior mean Then calculate .

[0075] Distance between the center of the current partition and the center of the adjacent partition The reciprocal of the product is used as the weight to calculate the weighted average. The summation is performed only on adjacent partitions containing existing data. If no adjacent partitions contain data, then... .

[0076] Define fusion coefficient , The total deviation is .

[0077] Unconstructed nodes posterior mean Posterior variance ,in This is the maximum value of the post-hoc variance of the constructed nodes, i.e. Posterior standard deviation .

[0078] For each construction progress node (completed node) and unconstructed nodes ), based on the standard normal distribution quantiles corresponding to the preset confidence level. Multiply by the corresponding posterior standard deviation to obtain the interval half-width, and you get the expected response interval.

[0079] The expected response range for completed nodes is... The expected response range for unconstructed nodes is .

[0080] Partition each space Under each construction progress node It is stored as a two-dimensional table, which is called the spatiotemporal-response benchmark map.

[0081] VI. Real-time monitoring and path deviation calculation During construction, the completed work processes are reported in real time through an automated monitoring system or a manual reporting system. The system matches the reported work process descriptions with a preset sequence of construction progress nodes to determine the current construction progress node and its number. .

[0082] Obtain the current node from the field monitoring system. and corresponding spatial partitions The measured structural response data is denoted as The monitoring system typically includes inclinometers, axial force gauges, and stress gauges, with the acquisition frequency set according to engineering requirements. The most recent valid reading after the node completion time is taken as... If no valid data is available within the specified time window after that moment, the system will wait for or use the last valid data.

[0083] Read the same partition from the spatiotemporal-response benchmark map At the same node The expected median (i.e., posterior mean). Path deviation. The calculation formula is: ,in Let be a pre-defined positive number, and Much smaller than the typical order of magnitude of |d|.

[0084] The system maintains a path deviation sequence for each spatial partition. Stored in order of node number.

[0085] VII. Deviation Pattern Recognition and Risk Warning For each spatial partition Take the most recent consecutive Path deviation of each construction progress node ,in It is a preset positive integer.

[0086] Let the node number within the window be... (Relative index), calculate the mean , .

[0087] Regression slope The calculation formula is:

[0088] Goodness of fit The calculation formula is:

[0089] The current partition is determined to have a divergent trajectory when all three of the following conditions are met: All positive or all negative; |k| is greater than the preset slope threshold ; Greater than the preset goodness-of-fit threshold .

[0090] When a divergent trajectory is identified, the system immediately issues a risk warning, with the default warning level being Level 1.

[0091] If at least two adjacent spatial partitions (partitions that share a spatial boundary) are in the same contiguous... If all nodes within a window are identified as divergent trajectories, and the path deviation signs of these partitions are the same, then the system determines it as "spatiotemporal consistency anomaly" and raises the warning level by one level.

[0092] Warning information is sent to relevant personnel via system pop-ups, SMS, emails, etc. The information includes: spatial partition identifier of the risk occurrence; current construction progress node number and corresponding process description; path deviation sequence that caused the warning; regression slope; spatiotemporal consistency verification results; and construction processes to be investigated (e.g., checking support axial force, verifying earthwork excavation speed, checking dewatering operation status, etc.).

[0093] Third Embodiment This embodiment is an optional embodiment of the above embodiments, and specifically includes: 8. Database closed-loop update Optionally, after the current foundation pit project is completed, the parameters of each spatial partition and the monitoring data of all construction progress nodes are added to the historical project database after min-max normalization.

[0094] During normalization, the global minimum and maximum values ​​of each component in the database are recalculated, and the clustering model is updated (K-means clustering can be re-executed, or an incremental clustering algorithm can be used).

[0095] IX. Implementation Method of Preset Parameters All preset parameters are pre-set according to the project risk level and can be defined in the monitoring system's configuration file. Specific details are as follows: Preset positive integer (Sliding window size): The value is determined based on the total density of construction progress nodes and the requirements for trend sensitivity.

[0096] Preset slope threshold The value is set according to the project risk level; the higher the risk, the smaller the value.

[0097] Preset goodness threshold : Set according to the requirement of linear trend.

[0098] Construction sequence matching coefficient first preset value Second preset value : All of them are positive numbers less than or equal to 1.

[0099] Preset similarity threshold : Set according to the requirements for the quality of historical samples.

[0100] Preset quantiles corresponding to the confidence level Determined based on the required safety margin.

[0101] Observation error variance : Based on the accuracy preset of the monitoring instrument.

[0102] Path deviation denominator protection item Take a positive number that is much smaller than |d|.

[0103] The above preset parameters are set via configuration file before system startup, and can also be dynamically adjusted by the system during construction based on fluctuations in actual monitoring data.

[0104] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0105] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for safety assessment of open-cut tunnel foundation pit support structures considering spatiotemporal effects, characterized in that, Includes the following steps: S1: Obtain the geological parameters, design parameters, and construction parameters for each spatial zone of the current open-cut tunnel foundation pit project; discretize the entire foundation pit excavation process into a sequence of construction progress nodes according to key procedures, with a total number of nodes. The spatial partition is divided into multiple main segments along the longitudinal direction, and each main segment is divided into multiple sub-regions along the transverse direction, forming a two-dimensional grid. S2: Cluster the spatial partition geological parameters of all projects in the historical engineering database to form multiple geological model categories; calculate the Euclidean distance between the geological parameters of each spatial partition of the current project and each cluster center, and assign them to the category with the smallest distance; Historical projects belonging to the same category and having the same longitudinal segment and transverse position are selected to form a set of geologically similar projects; S3: Based on design parameters, construction parameters, and construction sequence matching coefficient, calculate the similarity between the current project and each historical project in the geologically similar project set, and select historical projects with similarity greater than a preset threshold as the effective similar sample set; the construction sequence matching coefficient takes a first preset value or a second preset value depending on whether the longitudinal segmented excavation sequence is consistent. S4: Extract structural response monitoring data of each historical project in the same spatial partition and the same construction progress node from the effective similar sample set, calculate the weighted mean and weighted variance with similarity as the weight, and construct a Gaussian prior distribution; S5: Obtain the measured structural response data of the currently constructed nodes of the project, combine it with the prior distribution to perform Bayesian update, and obtain the posterior mean and posterior variance of the constructed nodes. Linear fitting is performed on the difference between the measured values ​​and the prior mean of the constructed nodes, and the fitted value is used as the time extrapolation term. The posterior bias of adjacent partitions at the same node is weighted by the inverse of the distance, and the weighted average is used as the spatial extrapolation term. The time extrapolation term and the spatial extrapolation term are weighted and merged according to the proportion of the number of constructed nodes to the total number of nodes to obtain the posterior mean of the unconstructed nodes. The maximum posterior variance of the constructed nodes multiplied by (2 - time goodness of fit) is taken as the posterior variance of the unconstructed nodes. Based on the quantiles corresponding to the preset confidence level The posterior standard deviation is used as the interval half-width to obtain the expected response intervals for each construction progress node and each spatial partition, thus forming a spatiotemporal-response benchmark map. S6: Determine the current construction progress node based on the real-time construction progress, obtain the measured data of the current node and the corresponding spatial partition, read the expected median from the benchmark map, and calculate the path deviation = (measured data - expected median) / (|expected median| + ... ),in It is a preset positive number; S7: Continuous within the same spatial partition Linear regression was performed on the path deviation of each node to obtain the slope and goodness of fit. It is a preset positive integer; when this When the path deviation of each node has the same sign, the absolute value of the slope is greater than the preset slope threshold, and the goodness of fit is greater than the preset goodness of fit threshold, it is determined to be a divergent trajectory and an early warning is issued. If at least two adjacent spatial partitions are in the same contiguous zone If all nodes within a window are identified as divergent trajectories and have the same path deviation sign, the warning level will be raised.

2. The method for safety assessment of open-cut tunnel foundation pit support structure considering spatiotemporal effects according to claim 1, characterized in that, The longitudinal main section includes the end section, the standard section, and the existing structural intersection section; The horizontal sub-region includes the left wall region, the middle section of the base, the right wall region, and the corner region; The construction progress milestones are as follows: completion of the first layer of earthwork excavation in each longitudinal segment, erection of the first support and application of prestress, completion of the second layer of earthwork excavation, erection of the second support, and completion of the tunnel bottom slab pouring.

3. The method for safety assessment of open-cut tunnel foundation pit support structure considering spatiotemporal effects according to claim 1, characterized in that, The clustering in step S2 uses the K-means algorithm, with the number of clusters preset to a fixed positive integer; before calculating the Euclidean distance, the components of the geological parameters are subjected to minimum-maximum normalization.

4. The method for safety assessment of open-cut tunnel foundation pit support structure considering spatiotemporal effects according to claim 1, characterized in that, The design parameters mentioned in step S3 include: pile diameter or wall thickness, pile length or wall depth, reinforcement ratio, concrete strength grade, support stiffness, support spacing, and support prestress. The construction parameters include: layer thickness, duration of excavation for each layer, time nodes for erection of each support, and longitudinal segmented excavation sequence. Each parameter is min-max normalized before similarity calculation; The similarity is calculated based on the product of weighted Euclidean distance and construction sequence matching coefficient, and the weights of the weighted Euclidean distance are determined by the analytic hierarchy process or the entropy weight method.

5. The method for safety assessment of open-cut tunnel foundation pit support structure considering spatiotemporal effects according to claim 1, characterized in that, The geological parameters include soil layer thickness, unit weight, cohesion, internal friction angle, and compression modulus for each spatial zone.

6. The method for safety assessment of open-cut tunnel foundation pit support structure considering spatiotemporal effects according to claim 1, characterized in that, The prior distribution is a Gaussian distribution; In Bayesian updates, the variance of the observation error is a preset positive number; The posterior mean of the constructed nodes is a weighted average of the prior mean and the measured value, with the weights determined by the prior variance and the observation error variance.

7. The method for safety assessment of open-cut tunnel foundation pit support structure considering spatiotemporal effects according to claim 1, characterized in that, The path deviation calculation formula Let be a preset positive number, and It is much smaller than the typical order of magnitude of the expected median value.

8. A method for safety assessment of open-cut tunnel foundation pit support structure considering spatiotemporal effects according to claim 1, characterized in that, The preset positive integer The preset slope threshold and preset goodness-of-fit threshold are set according to the engineering risk level; The condition for raising the warning level is that at least two adjacent spatial partitions are in the same contiguous zone. All nodes within the window are classified as divergent trajectories and have the same sign for path deviation. The early warning information includes spatial partition identifier, current node number, path deviation sequence, regression slope, spatiotemporal consistency verification results, and suggested construction procedures for investigation.

9. A method for safety assessment of open-cut tunnel foundation pit support structure considering spatiotemporal effects according to claim 1, characterized in that, Also includes: After the current project is completed, the parameters of each spatial zone and the monitoring data of all construction progress nodes will be added to the historical project database after min-max normalization.

10. A method for safety assessment of open-cut tunnel foundation pit support structure considering spatiotemporal effects according to claim 1, characterized in that, The preset positive integer The preset slope threshold, preset goodness-of-fit threshold, the first preset value, the second preset value, and the preset similarity threshold in step S3, and the quantile corresponding to the preset confidence level in step S5 are all fixed values ​​preset according to the engineering risk level.

Images