A foundation pit monitoring risk state multi-parameter dynamic determination method and system

Abstract

This invention relates to the technical field of foundation pit engineering monitoring, and provides a method and system for dynamic determination of multi-parameter risk status in foundation pit monitoring. The method includes: analyzing on-site monitoring data of the foundation pit to identify correlation trend information among multiple monitoring parameters caused by changes in deep soil; when the identified correlation trend information meets preset triggering conditions, initiating a risk scenario simulation; during the risk scenario simulation, based on the current state of the digital model, simulating the foundation pit response under preset external factors, simulating the chain reaction caused by changes in deep soil, and calculating the probability and expected time of the chain reaction reaching a preset safety limit in the future; generating early warning information of foundation pit risk based on the probability and expected time, and displaying the risk evolution trend graphically. This invention improves the safety and reliability of foundation pit engineering.

Description

Technical Field

[0001] This invention relates to the technical field of foundation pit engineering monitoring, specifically to a multi-parameter dynamic determination method and system for foundation pit monitoring risk status. Background Technology

[0002] In deep foundation pit engineering, especially in environments with complex and variable geological conditions, ensuring the stability of the pit and the safety of surrounding facilities is crucial. Traditional monitoring methods often rely on setting fixed safety limits for each monitoring point, issuing an alarm only when a single data point exceeds this limit. However, the actual situation is far more complex. Multiple factors, such as soil deformation, groundwater changes, and settlement of surrounding buildings, often interact with each other. A small change may not immediately trigger an alarm, but when multiple small changes occur simultaneously and are interconnected, they may indicate potential risks.

[0003] Existing foundation pit engineering monitoring systems rely on independent threshold values ​​for each monitoring parameter for judgment. They treat minor increases in pore water pressure, slow movement of diaphragm walls, slight stress changes in steel supports, and minor settlement in subway tunnels as independent, unrelated events. The system fails to recognize the inherent causal relationships and spatial interactions among these seemingly insignificant data changes that do not reach their respective warning thresholds. It lacks a mechanism to understand that subtle changes in pore water pressure in deep, weak interlayers are the initial driving factor behind a chain reaction leading to slow deformation of the overlying clay, wall deformation, redistribution of stress on supports, and ground settlement. Furthermore, the system cannot perform comprehensive risk analysis based on these multi-source, below-critical data, nor can it adjust the importance of different monitoring parameters in the overall risk assessment according to actual conditions.

[0004] This reliance on a single indicator leads to a significant delay in risk warnings. For example, when extremely small, localized changes in pore water pressure occur in deep soil or weak interlayers due to excavation disturbance, these changes trigger a series of slow, continuous, and interconnected subcritical changes through the interaction between the soil and the support structure. These include slow deformation of the overlying soft soil layer, slow local movement of the support structure, redistribution of stress in the internal support structure, and slight settlement of surrounding sensitive structures. Existing monitoring systems typically treat these multiple data changes, distributed in different locations, accumulating over time, and none of which have reached their respective warning thresholds, as independent and unrelated events. They cannot identify in advance a systemic risk arising from deep geological responses from these combinations of parameters below their respective warning thresholds. This results in risk warnings often being triggered only after local instability has actually occurred, and a certain monitoring value has changed drastically and exceeded a single threshold, thus missing the optimal opportunity to take preventative measures and making subsequent emergency response extremely passive and difficult.

[0005] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0006] This application discloses a method and system for dynamic determination of multi-parameter risk status in foundation pit monitoring, which aims to solve the problem that existing foundation pit engineering monitoring systems cannot effectively identify the correlation trend information between multiple parameters under complex geological conditions, resulting in delayed risk warning and missed opportunities for preventive measures.

[0007] The technical solution of this application is as follows: Firstly, this application discloses a multi-parameter dynamic determination method for the risk status of foundation pit monitoring, specifically including: Obtain a digital model of the foundation pit project; the digital model includes the geometric structure of the foundation pit, the parameters of the support structure, and the physical and mechanical parameters of each soil layer; the physical and mechanical parameters of the soil layers include the model parameters of the weak soil layer area. Receive on-site monitoring data of the foundation pit; based on the on-site monitoring data of the foundation pit, calibrate the model parameters of the digital model so that the digital model reflects the current state of the foundation pit; the calibration process includes updating the model parameters of the weak soil layer area according to the trend of changes in the deep soil. Analyze the on-site monitoring data of the foundation pit to identify the correlation trend information between multiple monitoring parameters caused by changes in deep soil. When the correlation trend information meets the preset trigger conditions, initiate risk scenario simulation. The correlation trend information is a set of information consisting of the time synchronization, spatial proximity and trend consistency of multiple monitoring parameters. The information set includes at least synchronization indicators, amplitude deviation indicators, external load interference suppression results and deformation correlation results, which are used as the triggering basis for initiating risk scenario simulation. During the risk scenario simulation, based on the current state of the digital model, the response of the foundation pit is simulated under the influence of preset external factors. The chain reaction caused by the change of deep soil is simulated, and the probability and expected time of the chain reaction reaching the preset safety limit in the future are calculated. Based on probability and expected time, early warning information on foundation pit risks is generated, and the risk evolution trend is displayed graphically.

[0008] Through this technical solution, this application can effectively integrate multi-source monitoring data, and through digital model calibration and correlation trend information identification, realize dynamic judgment and early warning of the risk status of the foundation pit, thereby overcoming the limitations of traditional single indicator judgment and significantly improving the accuracy of risk identification and the timeliness of early warning.

[0009] Secondly, this application also discloses a multi-parameter dynamic determination system for foundation pit monitoring risk status, used to perform multi-parameter dynamic determination of foundation pit monitoring risk status, specifically including: The digital model acquisition module is used to acquire the digital model of the foundation pit project. The digital model includes the geometric structure of the foundation pit, the parameters of the support structure, and the physical and mechanical parameters of each soil layer. The physical and mechanical parameters of the soil layer include the model parameters of the weak soil layer area. The monitoring data receiving module is used to receive on-site monitoring data of the foundation pit; based on the on-site monitoring data of the foundation pit, it calibrates the model parameters of the digital model so that the digital model reflects the current state of the foundation pit; the calibration process includes updating the model parameters of the weak soil layer area according to the trend of changes in the deep soil. The correlation trend information module is used to analyze on-site monitoring data of the foundation pit and identify correlation trend information between multiple monitoring parameters caused by changes in deep soil. When the correlation trend information is found to meet the preset trigger conditions, a risk scenario simulation is initiated. The risk scenario simulation module is used to simulate the foundation pit response under the influence of preset external factors based on the current state of the digital model during risk scenario simulation, to deduce the chain reaction caused by changes in deep soil, and to calculate the probability and estimated time of the chain reaction reaching the preset safety limit in the future. The early warning information generation module is used to generate early warning information for foundation pit risks based on probability and expected time, and to display the risk evolution trend in a graphical manner.

[0010] Through this technical solution, this application can provide a system that integrates digital modeling, monitoring data reception, correlation trend information identification, risk scenario simulation, and early warning information generation, thereby enabling efficient and accurate multi-parameter dynamic determination of the risk status of foundation pit monitoring, and providing comprehensive technical support for the safety management of foundation pit engineering.

[0011] Beneficial Effects: The multi-parameter dynamic determination method for the risk status of foundation pit monitoring disclosed in this application acquires a digital model of the foundation pit project and performs real-time calibration, enabling it to accurately reflect the current state of the foundation pit. Particularly regarding the trend of deep soil changes, it can promptly update the model parameters in weak soil layers, thereby improving the model's fit to the actual situation. This method further analyzes on-site monitoring data of the foundation pit to identify the correlation trend information between multiple monitoring parameters caused by changes in deep soil, and uses this as the trigger for initiating risk scenario simulations. This identification of multi-parameter correlation trend information overcomes the limitations of traditional single-indicator judgments, enabling the early detection of potential systemic risks from multiple seemingly independent, minor changes. In the risk scenario simulation stage, based on the current state of the digital model, this application simulates the foundation pit response under the influence of preset external factors, simulates the chain reaction caused by changes in deep soil, and calculates the probability and expected time of the chain reaction reaching a preset safety limit in the future, thus achieving quantitative prediction of the risk evolution trend. Finally, based on the calculated probability and estimated time, early warning information for foundation pit risks is generated and the risk evolution trend is displayed intuitively in a graphical format. This provides project managers with timely and comprehensive risk information, enabling them to take preventative measures in advance and avoiding the problems of delayed risk warnings and missed optimal response opportunities inherent in traditional monitoring systems. In summary, this application, through digital model calibration, multi-parameter correlation trend identification, risk scenario simulation, and early warning, effectively solves the technical problem in existing technologies that cannot identify the inherent correlation between multi-source, below-critical-state data, leading to delayed risk warnings, and significantly improves the safety and reliability of foundation pit engineering. Attached Figure Description

[0012] Figure 1 This is a flowchart of a method for dynamic determination of multi-parameter risk status of foundation pit monitoring in one embodiment of the present invention; Figure 2 This is a flowchart of a method for dynamic determination of multi-parameter risk status in foundation pit monitoring according to another embodiment of the present invention; Figure 3 This is a system block diagram of a multi-parameter dynamic determination system for monitoring the risk status of a foundation pit, according to another embodiment of the present invention. Explanation of reference numerals in the attached figures: 1. Multi-parameter dynamic judgment system for foundation pit monitoring risk status; 11. Digital model acquisition module; 12. Monitoring data receiving module; 13. Correlation trend information module; 14. Risk scenario simulation module; 15. Early warning information generation module. Detailed Implementation

[0013] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0014] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0015] This application proposes a multi-parameter dynamic determination method for the risk status of foundation pit monitoring, combined with... Figure 1 As shown, it specifically includes: S1, Obtain the digital model of the foundation pit project; the digital model includes the geometric structure of the foundation pit, the parameters of the support structure, and the physical and mechanical parameters of each soil layer; the physical and mechanical parameters of the soil layer include the model parameters of the weak soil layer area. S2, Receive on-site monitoring data of the foundation pit; Based on the on-site monitoring data of the foundation pit, calibrate the model parameters of the digital model so that the digital model reflects the current state of the foundation pit; The calibration process includes updating the model parameters of the weak soil layer area according to the trend of changes in the deep soil. S3, analyze the on-site monitoring data of the foundation pit and identify the correlation trend information between multiple monitoring parameters caused by changes in deep soil; when the correlation trend information meets the preset trigger conditions, initiate risk scenario simulation; wherein, the correlation trend information is a set of information composed of the time synchronization, spatial proximity and trend consistency between multiple monitoring parameters, and the information set includes at least synchronization index, amplitude deviation index, external load interference suppression results and deformation correlation results, which are used as the trigger basis for initiating risk scenario simulation; S4. During the risk scenario simulation, based on the current state of the digital model, the foundation pit response is simulated under the influence of preset external factors. The chain reaction caused by the change of deep soil is simulated, and the probability and expected time of the chain reaction reaching the preset safety limit in the future are calculated. S5 generates early warning information on foundation pit risks based on probability and expected time, and displays the risk evolution trend graphically.

[0016] This application relates to a multi-parameter dynamic determination method for the risk status of foundation pit monitoring. This method acquires a digital model of the foundation pit project and dynamically calibrates the model parameters using on-site monitoring data, enabling the digital model to represent the actual state of the foundation pit at the current moment. Based on this, correlation analysis is performed on the monitoring data to identify the correlation trends between multiple monitoring parameters caused by changes in deep soil. When the correlation trends meet preset trigger conditions, a risk scenario simulation is initiated. During the risk scenario simulation, based on the current state of the digital model, the foundation pit response is simulated under preset external factors, and the chain reactions that may be caused by changes in deep soil are deduced. The probability and estimated time of the chain reactions reaching preset safety limits in the future are further calculated, and early warning information for foundation pit risk is generated accordingly. The risk evolution trend is also displayed graphically. This method overcomes the lag and one-sidedness of traditional foundation pit monitoring relying on isolated judgments of single indicators, achieving early, dynamic, and multi-parameter comprehensive determination of the risk status of foundation pits.

[0017] To facilitate understanding of the technical solution of this application, the key terms used in this document will be explained in a unified manner. "Digital model" refers to a digital representation model of a foundation pit project, used to characterize the geometric structure, support structure state, and soil mechanical state of the foundation pit. The digital model can be a three-dimensional finite element model, a three-dimensional discrete element model, or other computational models capable of reflecting the coupling relationship between the foundation pit structure and the soil. The digital model includes at least the excavation depth, excavation boundary, support structure layout, support structure material parameters, physical and mechanical parameters of each soil layer, and model parameters for the weak soil layer area. The model parameters for the weak soil layer area are used to characterize the state of local areas in the deep soil that are more sensitive to deformation propagation, seepage transmission, and stability evolution.

[0018] "Foundation pit on-site monitoring data" refers to the data set collected in real time or near real time by monitoring equipment deployed on the foundation pit itself, the support structure, and the surrounding environment. Foundation pit on-site monitoring data can include data such as horizontal displacement of the retaining structure, column settlement, deep horizontal displacement of the soil, pore water pressure, earth pressure, anchor cable internal force, deformation of surrounding buildings and structures, and response of surrounding rail transit structures, which are used to characterize the actual stress and deformation state during the construction and operation of the foundation pit.

[0019] "Correlation trend information" refers to a set of information comprised of the coupled changes of multiple monitoring parameters in the time, spatial, and directional dimensions. Correlation trend information includes at least synchronicity indicators, amplitude deviation indicators, external load disturbance suppression results, and deformation correlation results. Specifically, synchronicity indicators characterize the consistency of different monitoring parameters over time; amplitude deviation indicators characterize the degree of deviation of the current monitoring parameter from its normal fluctuation range or baseline trend; external load disturbance suppression results mitigate the impact of short-term external disturbances such as rainfall, temporary surcharges, and construction machinery disturbances on the monitoring parameters; and deformation correlation results characterize the transmission relationship between deep soil changes and the deformation of the support structure, surface settlement, and the response of surrounding structures. As the basis for initiating risk scenario simulations, correlation trend information aims to identify and highlight multi-parameter linked anomalies that, while not yet reaching individual alarm thresholds, already possess potential risk evolution significance.

[0020] "Risk scenario simulation" refers to the process of predictively simulating the subsequent response of a foundation pit by considering the effects of preset external factors, based on the current state calibration of the digital model. Risk scenario simulation is not only used to analyze the stress and deformation trends of the foundation pit structure and soil under the current state, but also to simulate the chain reaction path caused by changes in deep soil, including the transmission and extension process of deep soil deformation to the upper soil, support structure and surrounding environment.

[0021] "Preset safety limits" refer to the safety control boundaries used to determine whether the risk state of the foundation pit has entered the danger zone. Preset safety limits may include deformation limits of the retaining structure, ground settlement limits, internal force limits of the support structure, deformation limits of surrounding structures, and other engineering control thresholds related to the safe operation of the foundation pit.

[0022] "Early warning information" refers to risk alerts generated based on risk scenario simulations. Early warning information includes at least the risk level, risk type, estimated time to reach preset safety limits, probability of risk development, and recommended preventative measures. It is used to provide decision support to management before the risk evolves into a manifest instability event.

[0023] Based on the above terminology definitions, the multi-parameter dynamic determination method for foundation pit monitoring risk status provided in this application revolves around digital model acquisition, on-site monitoring data reception, model parameter calibration, identification of related trend information, risk scenario simulation, and early warning output.

[0024] During implementation, the first step is to obtain a digital model of the foundation pit project. This digital model can be built based on engineering design drawings, geological survey data, support construction parameters, and on-site survey results. The digital model includes information on the foundation pit's geometric structure, support structure parameters, physical and mechanical parameters of each soil layer, and model parameters for weak soil layers. Support structure parameters may include the thickness of the diaphragm wall, pile dimensions, support stiffness, anchor length, and prestress level. Physical and mechanical parameters of each soil layer may include density, elastic modulus, Poisson's ratio, internal friction angle, cohesion, permeability coefficient, and consolidation parameters. Model parameters for weak soil layers primarily reflect the mechanical properties and seepage characteristics of deep, weak interlayers, saturated sensitive layers, or locally disturbed areas. By establishing a digital model containing the above information, a computational foundation can be provided for subsequent dynamic calibration and risk simulation based on the current state.

[0025] After acquiring the digital model, on-site monitoring data of the foundation pit is received, and the model parameters of the digital model are calibrated based on this data to ensure that the digital model reflects the current state of the foundation pit. The calibration process is not a one-time static correction of the digital model, but rather a dynamic update of parameters related to the current state in the digital model based on the changes in deformation, stress, and seepage reflected in the monitoring data. Specifically, when the trend of deep soil changes is significant in the weak soil layer area, the model parameters for the weak soil layer area can be updated first, such as adjusting the permeability coefficient, compression modulus, consolidation coefficient, strength reduction parameter, or local contact conditions, so that the digital model's representation of the deep soil state is closer to the actual site conditions. The calibration process can be implemented using data assimilation, ensuring that the monitoring data and the calculation model form a more consistent state representation at the current moment.

[0026] After the digital model completes its current state calibration, the on-site monitoring data of the foundation pit is analyzed to identify the correlation trends among multiple monitoring parameters caused by changes in deep soil. The identification process uses changes in deep soil as the primary clue, analyzing whether there are synchronous enhancement, spatial proximity transmission, and consistent trend relationships between these changes and the responses of the support structure, surface deformation, and the surrounding environment. Specifically, when a certain type of monitoring parameter in the deep soil shows continuous small changes, the judgment is not made directly based on whether a single parameter exceeds its limit. Instead, further analysis is conducted to see whether other parameters that are synchronous in time, adjacent in space, and coupled in trend change simultaneously. For example, if the deep pore water pressure continues to rise while the lateral displacement of the retaining structure, surface settlement, or the response of the surrounding rail transit structure shows synchronous accelerated changes, it indicates that the current state may have expanded from local deep soil disturbance to a larger-scale structural response. The resulting synchronicity indicators, amplitude deviation indicators, external load interference suppression results, and deformation correlation results together constitute the correlation trend information.

[0027] After identifying the associated trend information, it is determined whether the associated trend information meets the preset trigger conditions. When the associated trend information meets the preset trigger conditions, a risk scenario simulation is initiated. The preset trigger conditions are used to distinguish between general short-term fluctuations and linked changes with risk evolution significance. Their setting can be completed by combining historical engineering cases, risk assessment models, or expert experience. The preset trigger conditions do not require a single indicator to have reached the alarm threshold, but rather emphasize the degree to which multiple parameters meet the combined requirements in terms of time synchronization, amplitude deviation, stability after interference removal, and deformation transmission logic. Thus, the system can intervene in the risk chain that may be caused by changes in deep soil before the traditional single-indicator alarm is triggered.

[0028] After the risk scenario simulation is initiated, based on the current state of the digital model, the response of the foundation pit is simulated under the influence of preset external factors. The simulation extrapolates the chain reactions triggered by changes in deep soil, and calculates the probability and estimated time for these chain reactions to reach the preset safety limit in the future. Preset external factors may include rainfall, groundwater level changes, seismic input, additional loads from surrounding construction, traffic load fluctuations, or other external conditions that disturb the foundation pit's state. During the simulation, the digital model starts from the calibrated current state and analyzes whether changes in deep soil under the influence of external factors will further cause chain reactions such as additional settlement of the overlying soil, redistribution of internal forces in the retaining structure, amplification of deformation of the support system, expansion of surface settlement, and additional responses from surrounding buildings and structures. Furthermore, by performing probability analysis and time estimation on the simulation results, the probability and estimated time for the chain reactions to reach the preset safety limit in the future are obtained. The probability characterizes the likelihood of the risk developing into a significant safety event, and the estimated time characterizes the time scale required for the risk to evolve to the control boundary.

[0029] After obtaining the probability and estimated time of reaching the preset safety limit, early warning information for foundation pit risks is generated based on the probability and estimated time, and the risk evolution trend is displayed graphically. Early warning information may include risk level, risk type, main risk sources, estimated time interval for reaching the safety limit, and recommended mitigation measures. Graphical methods may include time-risk curves, spatial heat map, risk transmission path diagram, three-dimensional deformation evolution diagram, or other visualization forms that can intuitively reflect the risk development trend. Through graphical display, managers can more intuitively understand the current stage of the risk, its evolution speed, and the potential scope of impact, thereby enabling them to organize risk mitigation and adjust construction measures in advance.

[0030] Optional, combined Figure 2 As shown, the analysis of on-site monitoring data of the foundation pit identifies the correlation trend information among multiple monitoring parameters caused by changes in deep soil. When the identified correlation trend information meets the preset trigger conditions, the steps to initiate risk scenario simulation include: A1. Time series decomposition was performed on the data from multiple pore water pressure sensors, lateral displacement sensors of the foundation pit retaining structure, and vertical displacement observation points of the adjacent rail transit structure in the field monitoring data of the foundation pit, and the trend components and high-frequency fluctuation components corresponding to each type of data were separated. A2. Spatial difference analysis was performed on the trend components of the pore water pressure sensor data to obtain a synchronicity index to characterize the degree of synchronous change at multiple points, and an amplitude deviation index to characterize the degree of spatial difference. A3. Frequency characteristic analysis is performed on the high-frequency fluctuation components of the pore water pressure sensor data to identify periodic fluctuations that match the characteristics of external loads. The periodic fluctuations are then suppressed or eliminated to obtain the external load interference suppression results. A4. Based on the results of external load interference suppression, the trend component of the pore water pressure sensor data is corrected. In the corrected trend component, the synchronous change characteristics of multiple pore water pressure sensor data showing a continuous small increase are identified according to the synchronicity index. At the same time, the spatial differential distribution characteristics of the continuous small increase are identified according to the amplitude deviation index. The corresponding pore water pressure mode is determined based on the synchronous change characteristics and differential distribution characteristics. A5. The correlation between the lateral displacement sensor data of the foundation pit retaining structure and the vertical displacement observation point data of the adjacent rail transit structure and the pore water pressure mode in terms of spatial proximity and temporal synchronization is analyzed to obtain the deformation correlation results. A6 uses amplitude deviation index, synchronicity index, external load interference suppression results and deformation correlation results as correlation trend information; A7. When the amplitude deviation index corresponding to the pore water pressure mode exceeds the preset threshold, and the synchronization index reaches the preset synchronization threshold, and the external load interference suppression result indicates that the external periodic load is insufficient to explain the pore water pressure mode, and the deformation correlation result indicates that the lateral displacement of the retaining structure and the vertical displacement of the rail transit structure show a deformation acceleration trend that is spatially adjacent to and time-synchronous with the pore water pressure mode, the correlation trend information is determined to meet the preset conditions, and the risk scenario simulation is initiated.

[0031] Specifically, when processing the on-site monitoring data of the foundation pit, the data from pore water pressure sensors in the deep, weak interlayer area, lateral displacement sensors of the foundation pit retaining structure, and vertical displacement observation points of the adjacent rail transit structure are first decomposed into time series data. This decomposition process aims to separate the raw monitoring data into long-term trend components and short-term high-frequency fluctuation components, facilitating subsequent analysis of data with different properties. The trend component reflects the long-term evolution of the monitoring parameters, while the high-frequency fluctuation component contains instantaneous disturbances and noise information.

[0032] Furthermore, spatial difference analysis was performed on the trend components of the pore water pressure sensor data. This analysis calculated synchronicity and amplitude deviation indices by comparing the pore water pressure trends at different locations. The synchronicity index quantifies the degree of synchronization of pore water pressure changes across multiple monitoring points; for example, a higher synchronicity index indicates that multiple sensor data simultaneously show an upward or downward trend. The amplitude deviation index characterizes the spatial differences among these synchronous changes; for example, if the pore water pressure increase in some areas is significantly greater than in other areas, the amplitude deviation index reflects this spatial inhomogeneity.

[0033] Simultaneously, frequency characteristic analysis was performed on the high-frequency fluctuation components of the pore water pressure sensor data. This analysis aimed to identify periodic fluctuations consistent with external load characteristics, such as periodic pressure changes caused by subway operation, vibrations from surrounding construction, or tidal effects. Once these periodic fluctuations were identified, suppression or removal processing was performed to eliminate interference from external loads on the deep soil variation signals, thereby obtaining purer external load interference suppression results.

[0034] Based on this, the trend components of the pore water pressure sensor data were corrected using the results of external load interference suppression. The corrected trend components can more accurately reflect the true changes in deep soil. In the corrected trend components, combined with the previously obtained synchronicity and amplitude deviation indices, multiple synchronous changes exhibiting continuous, small increases in pore water pressure sensor data were identified, along with the spatially differentiated distribution characteristics of these continuous, small increases. By integrating these synchronous and differentiated distribution characteristics, the corresponding pore water pressure pattern can be determined, which is an important indicator of potential anomalies in deep soil.

[0035] Furthermore, it is necessary to analyze the correlation between the lateral displacement sensor data of the foundation pit retaining structure and the vertical displacement observation data of the adjacent rail transit structure and the aforementioned pore water pressure pattern. This correlation analysis focuses on spatial proximity and temporal synchronization, that is, determining whether the deformation of the retaining structure and the rail transit structure is spatially close to the pore water pressure pattern region and temporally synchronized with the changes in the pore water pressure pattern. Through this analysis, deformation correlation results can be obtained, further verifying the impact of deep soil changes on the surrounding structures.

[0036] Finally, the amplitude deviation index, synchronicity index, external load disturbance suppression results, and deformation correlation results are integrated as correlation trend information. When the amplitude deviation index corresponding to the pore water pressure model exceeds the preset threshold, the synchronicity index reaches the preset synchronization threshold, the external load disturbance suppression results indicate that the external periodic load is insufficient to explain the pore water pressure model, and the deformation correlation results indicate that the lateral displacement of the retaining structure and the vertical displacement of the rail transit structure show an accelerating deformation trend that is spatially adjacent to and temporally synchronized with the pore water pressure model, it can be determined that the correlation trend information meets the preset conditions, and risk scenario simulation is initiated.

[0037] In some preferred embodiments, it is assumed that multiple pore water pressure sensors are deployed in the deep, weak interlayer area of ​​a foundation pit project, while lateral displacement sensors are installed on the foundation pit retaining structure, and vertical displacement observation points are set up next to the adjacent rail transit structure. When the monitoring system receives these on-site monitoring data, it first performs time series decomposition on all data, separating the pore water pressure, lateral displacement, and vertical displacement data into trend components and high-frequency fluctuation components, respectively.

[0038] Specifically, the system performs spatial difference analysis on the trend components of the pore water pressure sensor data. For example, if it is found that multiple pore water pressure sensor data in a region all show a continuous slight upward trend, and their synchronicity index reaches the preset 0.8 (indicating 80% synchronicity), but at the same time the amplitude deviation index shows that the upward amplitude in the northeast corner of the foundation pit is significantly higher than that in other areas, then a pore water pressure pattern is initially identified.

[0039] Meanwhile, frequency characteristic analysis was performed on the high-frequency fluctuation components of the pore water pressure sensor data. If the analysis results showed periodic fluctuations consistent with the operating frequency of the surrounding subway, the system would perform suppression processing on these fluctuations to eliminate the instantaneous impact of subway vibration on pore water pressure, thus obtaining the external load interference suppression results. The results indicate that even after excluding the subway load, the upward trend in pore water pressure still exists and cannot be fully explained by external periodic loads.

[0040] Subsequently, the system combines the corrected pore water pressure trend components with previously identified synchronicity and amplitude deviation indicators to further confirm the characteristics of the pore water pressure pattern. For example, it confirms that the pattern manifests as a continuous, slight increase in pore water pressure in deep soil, with a significant differential distribution in the northeastern region.

[0041] Next, the system analyzes the correlation between the lateral displacement sensor data of the foundation pit retaining structure and the vertical displacement observation data of the adjacent rail transit structure and the pore water pressure pattern. If it is found that the lateral displacement of the retaining structure at the northeast corner of the foundation pit and the vertical displacement of the adjacent rail transit structure both show an accelerated deformation trend that is spatially adjacent and temporally synchronized with the pore water pressure pattern—for example, if the lateral displacement rate of the retaining structure and the vertical displacement rate of the rail transit structure both increase significantly within about 24 hours after the pore water pressure begins to rise—then a clear deformation correlation result is obtained.

[0042] Finally, when the amplitude deviation index (e.g., the increase in the northeast corner area exceeds 20% of the average of other areas), the synchronicity index (0.8), the external load disturbance suppression result (indicating that the external load is insufficient to explain the pore water pressure pattern), and the deformation correlation result (indicating that the retaining structure and the rail transit structure exhibit synchronous accelerated deformation) all meet the preset trigger conditions, the system will determine that the correlation trend information meets the preset conditions and immediately initiate risk scenario simulation to assess the potential foundation pit instability risk and the expected occurrence time.

[0043] Optionally, the steps of analyzing on-site monitoring data of the foundation pit and identifying the correlation trend information among multiple monitoring parameters caused by changes in deep soil include: Read the monitoring parameters from deep soil, foundation pit retaining structure and surrounding sensitive structures from the on-site monitoring data of the foundation pit, and form a set of monitoring parameters; Time series analysis was performed on the set of monitoring parameters to extract the changing trends of each monitoring parameter; Establish a correlation strength assessment model for each monitoring parameter pair; a monitoring parameter pair is a combination formed by selecting any two monitoring parameters from each monitoring parameter, used to characterize the correlation relationship between the two monitoring parameters and participate in the calculation of the correlation strength index; Based on the current stage of foundation pit excavation, the trend of groundwater level change, and the soil consolidation creep rate, update the correlation weight coefficients of the correlation strength assessment model. Based on the updated correlation weight coefficient and correlation strength assessment model, the correlation strength index between each pair of monitoring parameters in the monitoring parameter set is calculated. Based on the correlation strength index, the monitoring parameter pair that most accurately reflects the potential risks at the current stage is determined as the parameter correlation combination, and the parameter correlation combination and the corresponding correlation strength index are used as correlation trend information.

[0044] Specifically, when reading monitoring parameters, the construction of the monitoring parameter set aims to comprehensively cover key areas and structures that may be affected by changes in deep soil. Monitoring parameters from deep soil can include deep soil displacement, pore water pressure, and earth pressure, directly reflecting the deformation and stress state of the deep soil. Monitoring parameters from the foundation pit retaining structure can include lateral displacement, tilt, and stress, characterizing the retaining structure's response to soil changes. Monitoring parameters from surrounding sensitive structures can include settlement, tilt, and cracks of surrounding buildings, assessing the impact of the foundation pit project on the surrounding environment. By collecting this multi-source, heterogeneous monitoring data, a comprehensive data foundation can be provided for subsequent correlation analysis.

[0045] Furthermore, time series analysis of the monitoring parameter set aims to extract the long-term trends of each monitoring parameter from the raw data, which may contain noise and short-term fluctuations. This can be achieved through various time series analysis methods, such as moving average, exponential smoothing, wavelet analysis, or trend decomposition, to clearly reveal the inherent laws governing the evolution of parameters over time.

[0046] This involves establishing correlation strength assessment models for each pair of monitoring parameters. A monitoring parameter pair refers to a combination of any two monitoring parameters selected from the set of monitoring parameters. For example, deep soil displacement and lateral displacement of the retaining structure can constitute a monitoring parameter pair, as can pore water pressure and settlement of surrounding buildings. The correlation strength assessment model aims to quantify the degree and nature of the mutual influence between these two parameters. It can be constructed using statistical or machine learning models, such as correlation coefficients, mutual information, and Granger causality.

[0047] In practical applications, updating the correlation weight coefficients of the correlation strength assessment model is crucial to this scheme. The excavation stage of the foundation pit, the trend of groundwater level changes, and the soil consolidation creep rate are important factors affecting the stability of foundation pit projects, and they dynamically change the correlation strength between different monitoring parameters. For example, in the stage where deep soil consolidation creep is significant, the correlation between pore water pressure and deep soil displacement may increase, while the correlation with the displacement of the retaining structure may decrease. By dynamically adjusting the correlation weight coefficients based on this real-time working condition information, the assessment model can more accurately reflect the true correlation between parameters under the current foundation pit condition.

[0048] Therefore, based on the updated correlation weight coefficients and correlation strength assessment model, the correlation strength index between each pair of monitoring parameters in the monitoring parameter set can be calculated. The correlation strength index is a quantitative value used to characterize the tightness and direction of the correlation between two monitoring parameters under a specific operating condition.

[0049] Finally, based on the correlation strength index, the monitoring parameter pair that most accurately reflects the potential risks at the current stage can be determined as the parameter correlation combination. For example, if the correlation strength index of a certain monitoring parameter pair (such as deep soil pore water pressure and lateral displacement of the retaining structure) increases significantly, and its trend is highly consistent with the known risk pattern, then this parameter pair and its correlation strength index will be determined as the parameter correlation combination for the current stage and used as part of the correlation trend information to initiate risk scenario simulation.

[0050] Optionally, the steps of determining the monitoring parameter pair that most accurately reflects the potential risk at the current stage as a parameter association combination based on the correlation strength index, and using the parameter association combination and the corresponding correlation strength index as correlation trend information, include: Establish a risk cascading path library; the risk cascading path library stores risk cascading paths; the risk cascading path defines the initial parameter pair, initial condition, intermediate transmission mechanism, trigger condition, and final impact parameter; The correlation strength index between each monitoring parameter pair is continuously read, and the trend of the correlation strength index is extracted to obtain the correlation strength change trend; Based on the trend of association strength change and the starting conditions of each risk cascade path in the risk cascade path library, the trend matching degree between the trend of association strength change and each risk cascade path is calculated. Cascade path candidate set is generated by filtering risk cascade paths from the risk cascade path library whose trend matching degree is greater than or equal to a preset threshold. Based on the current digital model status of the foundation pit, for each risk cascading path in the cascading path candidate set, the evolution process of each risk cascading path in the cascading path candidate set is dynamically deduced, and the risk triggering potential assessment results corresponding to the risk cascading path are output; the risk triggering potential assessment results include at least the probability of triggering subsequent chain reactions and the expected time. Based on the risk triggering potential assessment results, each risk cascading path in the cascading path candidate set is sorted, and the risk cascading path with the highest risk triggering potential and exceeding the preset risk potential threshold is determined as the target cascading path. The initial parameter pairs corresponding to the target cascade path are determined as the parameter association combinations that most accurately reflect the potential risks at the current stage, and the parameter association combinations and corresponding association strength indicators are used as association trend information.

[0051] The risk cascading path library is a pre-built knowledge base that stores various possible risk cascading paths for foundation pits. Each risk cascading path is defined in detail, including its initial parameter pair, the initial conditions that trigger the risk, the intermediate mechanisms by which the risk propagates between different monitoring parameters, the triggering conditions for risk escalation, and the final potential impact parameters. For example, one path might describe how an abnormal drop in pore water pressure in deep soil causes accelerated lateral displacement of the retaining structure through soil consolidation, ultimately leading to settlement of adjacent buildings.

[0052] In practical applications, the system continuously reads the correlation strength indices calculated between various monitoring parameter pairs. To better capture the dynamic evolution of risk, these correlation strength indices are further subjected to trend extraction to obtain the correlation strength change trend. This trend reflects the changing pattern of correlation strength over time, such as whether it is continuously strengthening, weakening, or fluctuating.

[0053] Subsequently, the extracted correlation strength change trends are compared with the starting conditions of each preset risk cascade path in the risk cascade path library, and the trend matching degree is calculated. The trend matching measure quantifies the similarity between the correlation trend reflected by the current monitoring data and the starting conditions of a specific risk cascade path. For example, if the starting condition of a risk cascade path is "the correlation strength between pore water pressure and lateral displacement of the retaining structure continues to rise," and the currently monitored correlation strength change trend also shows a similar upward trend, then the trend matching degree will be high.

[0054] When the trend matching degree reaches or exceeds the preset threshold, it indicates that the currently monitored associated trend is highly consistent with a potential risk cascading path. At this time, the corresponding risk cascading path will be screened out and added to the cascading path candidate set.

[0055] For each risk cascading path in the candidate set, the system dynamically simulates the current state of the foundation pit's digital model. This simulation examines how the foundation pit responds to pre-set external factors (such as continuous rainfall or nearby construction) and how the chain reaction triggered by changes in deep soil will evolve. Through this simulation, the system can output a risk triggering potential assessment result for each risk cascading path. This result includes at least the probability and estimated time of triggering subsequent chain reactions, providing a quantitative basis for risk warning.

[0056] Finally, based on these risk triggering potential assessment results, all risk cascading paths in the candidate cascading path set are ranked. The risk cascading path with the highest risk triggering potential that exceeds the preset risk potential threshold is selected and determined as the target cascading path. The initial parameter pair corresponding to this target cascading path is determined as the parameter association combination that most accurately reflects the potential risk at the current stage. At the same time, this parameter association combination and the corresponding association strength index will be used as association trend information for subsequent risk scenario extrapolation.

[0057] Optionally, the steps for calculating the trend matching degree of risk cascading paths based on the trend of association strength change and the starting conditions of each risk cascading path in the risk cascading path library include: The correlation strength index between each monitoring parameter pair is smoothed over a time window to filter out short-term fluctuations and noise, resulting in a smoothed correlation strength index. Based on the smoothed association strength index, the trend of association strength change is extracted, and the trend of association strength change is feature extracted to obtain the trend feature vector. Real-time acquisition of the working condition information of the foundation pit, and use of the working condition information to contextualize the starting conditions of the risk cascading path; For each risk cascading path in the risk cascading path library, the trend matching degree of the risk cascading path is calculated based on the trend feature vector and the initial conditions after contextual constraints.

[0058] Specifically, smoothing the correlation strength index between monitoring parameter pairs over a time window involves using moving averages, exponential smoothing, or other filtering techniques to process continuous correlation strength index data within a preset time window. This eliminates random noise and short-term fluctuations in the data, thereby better revealing its inherent long-term or medium-term trends. The smoothed correlation strength index can more stably reflect the true evolution of the correlation strength between parameter pairs.

[0059] Based on the smoothed association strength index, the trend of association strength change is extracted, and features of the association strength change trend are extracted to obtain a trend feature vector. This can be understood as identifying the rising, falling, or stable trend of the association strength index using mathematical methods (such as linear regression, polynomial fitting, or difference operations) on the basis of smoothed data. The trend feature vector is a set of numerical values ​​that quantify these trends, such as the slope, acceleration, duration, fluctuation amplitude, and whether there is an inflection point. These features together constitute the "fingerprint" of the trend, which is used for subsequent matching analysis.

[0060] In practical applications, real-time acquisition of the excavation pit's operational status information is used to contextualize the initial conditions of risk cascading paths. Specifically, this involves the system continuously monitoring and collecting real-time operational status data of the excavation pit, such as the current excavation depth, groundwater level changes, and surrounding environmental loads (e.g., traffic loads, construction loads). This operational status information is used to dynamically adjust or weight the preset initial conditions of risk cascading paths. For example, if the initial condition of a certain risk path is more sensitive under the "deep excavation stage," and the current operational condition is indeed deep excavation, then this initial condition will be given a higher weight or a stricter judgment threshold to better reflect the current actual situation.

[0061] Optionally, for each risk cascading path in the risk cascading path library, the steps for calculating the trend matching degree of the risk cascading path based on the trend feature vector and the initial conditions after contextual constraints include: Retrieve the starting condition template of the preset target risk cascade path from the risk cascade path library; the starting condition template shall at least include the starting parameter pair, the starting trend pattern characteristics, the applicable time window scale, and the spatial proximity constraint. The trend feature vector is normalized to match the time window scale of the starting condition template. Based on the normalized trend feature vector and the initial trend morphology features in the initial condition template, the trend similarity index is calculated. For the initial conditions after contextual constraints are applied, a contextual consistency index is generated; the contextual consistency index characterizes the degree of conformity between the initial conditions after contextual constraints and the spatial proximity constraints of the initial condition template. When the context consistency index falls below a preset threshold, a penalty factor is generated, and the trend similarity index is penalized and corrected based on the penalty factor. Based on the trend similarity index and the contextual consistency index after penalty correction, the trend matching degree of the risk cascade path is calculated by combining them according to preset weights.

[0062] Specifically, when calculating the trend matching degree of a risk cascading path, the initial condition template of the target risk cascading path needs to be retrieved from the risk cascading path library. This initial condition template is a set of predefined standard features for a specific risk cascading path, which includes at least initial parameter pairs, initial trend morphology characteristics, applicable time window scales, and spatial proximity constraints. The initial parameter pairs refer to the combination of two or more key monitoring parameters that trigger the risk cascading path; the initial trend morphology characteristics refer to the specific pattern that these parameters should exhibit in terms of trend changes, such as continuous rise or accelerated decline; the applicable time window scale defines the length of time required to observe these trend characteristics; and the spatial proximity constraints specify the geographical location or structural area where these parameter changes occur. This template serves as a standardized reference for evaluating the degree of conformity between the trend characteristics reflected in real-time monitoring data and the predefined risk pattern.

[0063] Furthermore, to ensure the comparability of the trend feature vector with the initial condition template, time-scale normalization of the trend feature vector is necessary. This normalization aims to ensure that the trend feature vector maintains consistency with the time window scale corresponding to the initial condition template. For example, if the template defines a 7-day time window, but the real-time extracted trend feature vector is based on 5 days of data, then interpolation, resampling, or other time series processing methods are needed to normalize the 5-day data to a 7-day time scale to eliminate comparison bias caused by inconsistent time windows.

[0064] Based on this, a trend similarity index is calculated using the normalized trend feature vector and the initial trend morphology features in the initial condition template. The trend similarity index quantifies the degree of morphological similarity between the real-time observed trend features and the preset risk model. It can be calculated using methods such as correlation coefficient, Euclidean distance, and dynamic time warping (DTW). A higher index value indicates that the trend morphology is closer to the risk model defined in the template.

[0065] Simultaneously, a contextual consistency index is generated for the initial conditions after contextual constraints are applied. The contextual consistency index characterizes the degree of conformity between the initial conditions after contextual constraints and the spatial proximity constraints of the initial condition template. For example, if the template requires the risk to occur in a specific area of ​​the foundation pit, while the initial conditions after contextual constraints indicate that the current trend occurs in an adjacent but non-critical area, the contextual consistency index will be low. The generation of this index helps assess whether the current risk scenario is consistent with the context of the pre-set risk pattern.

[0066] In a preferred implementation, when the contextual consistency index falls below a preset threshold, a penalty factor is generated, and the trend similarity index is penalized and corrected based on this penalty factor. The preset threshold is the standard for judging whether contextual consistency is acceptable; once it falls below this threshold, it indicates that the current context significantly deviates from the template requirements. In this case, the generated penalty factor reduces the weight of the trend similarity index or directly reduces it to reflect the negative impact of contextual mismatch on risk assessment, avoiding misjudgment of risk due to trend pattern similarity when the context is inappropriate.

[0067] Finally, the trend matching degree of the risk cascading path is calculated by combining the penalized trend similarity index and the contextual consistency index according to preset weights. The preset weights can be set based on experience or expert knowledge to balance the contributions of trend pattern similarity and contextual consistency to the final matching degree. For example, a higher weight can be assigned to the trend similarity index, while the contextual consistency index serves as a correction factor.

[0068] Optionally, the steps of extracting the trend of association strength change based on the smoothed association strength index and extracting features from the association strength change trend to obtain the trend feature vector can be further refined into the following operations: Within a preset sliding time window, the trend of the smoothed correlation strength index is extracted to obtain the correlation strength change trend sequence. Based on the correlation strength change trend sequence, the trend slope, trend acceleration, fluctuation amplitude, and persistence index are calculated; the persistence index is used to characterize the degree to which the trend changes in the same direction within a continuous window. Identify inflection points and corresponding inflection point strengths in a sequence of trends in correlation strength changes. The trend slope, trend acceleration, fluctuation amplitude, persistence indicators, and inflection point strength are normalized. The normalized trend slope, trend acceleration, fluctuation amplitude, persistence index, and inflection point strength are combined in a preset order to obtain the trend feature vector.

[0069] The preset sliding time window setting aims to dynamically capture the changing characteristics of correlation strength indicators at different time scales. The size of this window can be flexibly configured according to the monitoring frequency of the foundation pit project, the response time of soil deformation, and the expected speed of risk evolution. For example, it can be set to several hours, several days, or several weeks to adapt to monitoring needs under different working conditions. Trend extraction can employ various time series analysis methods, such as moving average, exponential smoothing, linear regression, or Kalman filtering, to separate the long-term trend component from the original data, filter out short-term fluctuations and noise, and thus obtain a more stable and realistic correlation strength change trend sequence.

[0070] Furthermore, based on the obtained correlation strength change trend sequence, several key trend features are calculated. The trend slope characterizes the rate and direction of correlation strength change over time; for example, a positive slope indicates increasing correlation strength, while a negative slope indicates decreasing correlation strength. Trend acceleration reflects the rate of change of the trend slope, i.e., the acceleration or deceleration of the correlation strength change trend. Fluctuation amplitude quantifies the range of fluctuation of correlation strength around the trend line, reflecting its stability or activity level. The persistence index measures the degree to which the trend maintains the same direction of change within a continuous time window; for example, the number of consecutive upward or downward windows can be calculated. Its purpose is to distinguish between short-term fluctuations and persistent trends, thereby improving the reliability of trend determination. In addition, by analyzing the correlation strength change trend sequence, inflection points can be identified—points where the trend direction or rate changes significantly—and the corresponding inflection point strength can be calculated. This strength quantifies the drastic nature of the trend change at the inflection point, and these inflection points often foreshadow critical moments when the risk status of the foundation pit may undergo a qualitative change.

[0071] To enable effective comparison and combination of trend features with different dimensions and numerical ranges, it is necessary to normalize the calculated trend slope, trend acceleration, fluctuation amplitude, persistence index, and inflection point strength. Normalization methods can include min-max normalization and Z-score standardization, mapping each feature value to a unified numerical range and eliminating the influence of dimensional differences. Finally, the normalized trend features are combined in a predetermined order to form a multi-dimensional trend feature vector, which comprehensively reflects the dynamic changes in the correlation strength index.

[0072] Optionally, the steps of acquiring real-time working condition information of the foundation pit and using this information to contextualize constraints on the initial conditions of the risk cascading path include: Real-time acquisition of working condition information of the foundation pit; working condition information includes at least the excavation stage, groundwater level change trend and changes in surrounding environmental load; For the initial conditions of each risk cascade path in the risk cascade path library, extract the constraint terms corresponding to the working condition information; the constraint terms should include at least applicable excavation stage constraints, applicable water level change constraints, and applicable load change constraints. The consistency score of the scenario is calculated by comparing the working condition information with the constraints. When the contextual consistency score is lower than the preset score, a contextual weight adjustment factor is generated and bound to the starting condition of the corresponding risk cascade path to obtain the starting condition after contextual constraints. The contextual weight adjustment factor is a penalty coefficient used to characterize the degree of deviation of the contextual consistency score. It is used to adjust the cost of deviation of the starting condition in the trend matching degree calculation to reduce the trend matching degree of the risk cascade path with inconsistent contextualities.

[0073] Specifically, real-time acquisition of foundation pit operational information refers to the system continuously acquiring current operational status data of the foundation pit from on-site monitoring equipment, construction management systems, or other data sources. This operational information is dynamically changing and includes at least the current excavation stage of the foundation pit (e.g., earthwork excavation, main structure construction, etc.), the trend of groundwater level changes (e.g., water level rising, falling, or stabilizing), and changes in surrounding environmental loads (e.g., adjacent construction, traffic vibrations, etc.). This information is a key input for assessing the risk scenario of the foundation pit.

[0074] Specifically, for each risk cascade path in the risk cascade path library, extracting constraints corresponding to the working conditions means that for each predefined risk cascade path in the library, its initial conditions typically include requirements for the external environment or working conditions. This application identifies and extracts specific constraints corresponding to the currently acquired working condition information from these initial conditions. For example, a risk path may only be applicable during the "deep foundation pit excavation stage" or may only be triggered when the groundwater level continues to decline. These constraints include at least applicable excavation stage constraints, applicable water level change constraints, and applicable load change constraints, which define the applicability of the risk path under specific working conditions.

[0075] In practical applications, comparing the consistency of working condition information with constraints to calculate a situational consistency score involves comparing the real-time acquired foundation pit working condition information with the constraints extracted from the initial conditions of the risk cascade path. For example, if the applicable excavation stage constraint for a certain risk path is "earthwork excavation," while the current working condition information shows that the foundation pit is in the "main structure construction stage," then the two are inconsistent. The situational consistency score is a quantitative assessment of the degree of this consistency or inconsistency; a higher score indicates a closer match between the current working condition and the initial conditions of the risk path.

[0076] Furthermore, when the situational consistency score is lower than the preset score, a situational weight adjustment factor is generated and bound to the starting conditions of the corresponding risk cascade path, resulting in the starting conditions after contextual constraints. This means that if the current working condition does not closely match the starting conditions of a certain risk cascade path (i.e., the situational consistency score is lower than the preset threshold), the system will generate a situational weight adjustment factor. This adjustment factor is essentially a penalty coefficient used to characterize the degree to which the situational consistency score deviates from the preset score. This factor is bound to the starting conditions of the risk cascade path, thereby reducing the trend matching degree of the situationally inconsistent risk cascade path in subsequent trend matching degree calculations by adjusting the cost of the deviation of the starting conditions. For example, if the situational consistency score is very low, the adjustment factor will significantly reduce the trend matching degree of the risk path, or even make it not considered, thereby avoiding triggering unnecessary risk warnings under unsuitable working conditions.

[0077] Optionally, for each risk cascading path in the risk cascading path library, the steps for calculating the trend matching degree of the risk cascading path based on the trend feature vector and the initial conditions after contextual constraints include: Obtain the trend feature vector and the initial conditions after applying contextual constraints; Extract the initial condition constraint set from the initial conditions after contextualizing the constraints; the initial condition constraint set shall include at least the allowable range constraints and directional constraints on the trend slope, trend acceleration, fluctuation amplitude, persistence index and inflection point strength. For each trend feature vector, calculate the out-of-bounds amount of each trend feature relative to the allowable range constraint, and calculate the inconsistency amount of each trend feature relative to the directional constraint, to obtain the deviation amount of each dimension of the trend feature vector. The deviations of each dimension are weighted and summed according to preset weights to obtain the initial condition deviation value. When the initial conditions after contextual constraints carry a contextual weight adjustment factor, the contextual weight adjustment factor is applied to the initial condition deviation value to obtain the corrected deviation value. Based on the correction deviation cost, the trend matching degree of the risk cascade path is generated.

[0078] This series of steps aims to provide the necessary data input for subsequent matching degree calculations. The trend feature vector is a quantitative representation of the changing trend of the correlation strength between monitored parameter pairs, while the initial conditions following the contextual constraints contain the specific triggering requirements of a particular risk cascading path under the current operating conditions. Specifically, the set of initial condition constraints is a detailed definition of the triggering conditions for risk cascading paths, specifying the specific requirements that the trend feature vector should meet in each dimension. These constraints include allowable range constraints and directional constraints on trend slope, trend acceleration, fluctuation amplitude, persistence indicators, and inflection point strength. The allowable range constraints limit the numerical range of these features, while the directional constraints specify their direction of change, such as continuous rise or fall.

[0079] Out-of-bounds quantity refers to the degree to which a certain feature value of the trend feature vector exceeds the allowable range defined in the set of initial condition constraints. Inconsistency quantity refers to the degree to which the direction of change of the trend feature vector does not conform to the expected direction defined in the set of initial condition constraints. By calculating these quantities, the degree of deviation between the current monitoring data trend and the preset risk triggering conditions can be quantified. Since different trend features may have different importance in risk assessment, weighting and summing the deviation quantities of each dimension through preset weights can more accurately reflect the overall degree of deviation. The higher the initial condition deviation cost value, the more mismatched the current trend is with the initial conditions of the risk cascade path. The scenario weight adjustment factor is used to characterize the degree of deviation between the consistency score of the current working condition and the applicable scenario of the risk cascade path. When the scenario consistency score is low, this adjustment factor is generated and applied to the initial condition deviation cost value to further penalize those risk cascade paths that are not applicable under the current working condition, thereby reducing their trend matching degree. Finally, the trend matching degree is the inverse measure of the correction deviation cost value, that is, the smaller the deviation cost value, the higher the matching degree. This can be obtained by normalizing the corrected bias cost and subtracting the normalized bias cost from the maximum matching degree, or by using other inverse mapping functions.

[0080] This application proposes a multi-parameter dynamic determination system for foundation pit monitoring risk status, used to perform multi-parameter dynamic determination of foundation pit monitoring risk status, combined with... Figure 3 As shown, the multi-parameter dynamic judgment system 1 for monitoring the risk status of foundation pits includes: The digital model acquisition module 11 is used to acquire the digital model of the foundation pit project. The digital model includes the geometric structure of the foundation pit, the parameters of the support structure, and the physical and mechanical parameters of each soil layer. The physical and mechanical parameters of the soil layer include the model parameters of the weak soil layer area. The monitoring data receiving module 12 is used to receive on-site monitoring data of the foundation pit; based on the on-site monitoring data of the foundation pit, it calibrates the model parameters of the digital model so that the digital model reflects the current state of the foundation pit; the calibration process includes updating the model parameters of the weak soil layer area according to the trend of deep soil changes. The correlation trend information module 13 is used to analyze the on-site monitoring data of the foundation pit and identify the correlation trend information between multiple monitoring parameters caused by changes in deep soil. When the correlation trend information is found to meet the preset trigger conditions, the risk scenario simulation is initiated. The risk scenario simulation module 14 is used to simulate the foundation pit response under the influence of preset external factors based on the current state of the digital model during the risk scenario simulation, to simulate the chain reaction caused by the change of deep soil, and to calculate the probability and expected time of the chain reaction reaching the preset safety limit in the future. The early warning information generation module 15 is used to generate early warning information of foundation pit risk based on probability and expected time, and to display the risk evolution trend in a graphical manner.

[0081] The specific steps and principles of the multi-parameter dynamic determination method for foundation pit monitoring risk status have been described in the above embodiments, and will not be repeated here. It should be emphasized that this application further provides a multi-parameter dynamic determination system for foundation pit monitoring risk status, which is implemented by encapsulating the above method steps into different functional modules.

[0082] Specifically, the digital model acquisition module can be configured to acquire digital models of foundation pit projects through various methods. For example, this module can be implemented as a data interface for importing 3D model data of the foundation pit from external CAD / BIM systems, or extracting topographic, geological stratification, and surrounding environmental data from a geographic database through an integrated Geographic Information System (GIS) interface. Alternatively, the digital model acquisition module can also include a modeling tool that allows users to manually input the geometric structure of the foundation pit, support structure parameters, and physical and mechanical parameters of each soil layer, including model parameters for weak soil zones.

[0083] The monitoring data receiving module can be configured to receive on-site monitoring data from the foundation pit and calibrate the model parameters of the digital model based on this data. For example, this module can be implemented as a data acquisition and processing unit, connecting to the on-site sensor network via wired or wireless communication interfaces (such as Modbus, MQTT, LoRaWAN, etc.) to receive monitoring data in real time from devices such as displacement sensors, stress sensors, and pore water pressure sensors. The received data can be stored in a local database. The calibration function can be implemented as a data processing algorithm, such as using statistical methods like least squares or Bayesian estimation, to adjust relevant parameters in the digital model based on the received monitoring data, so that the digital model can reflect the current state of the foundation pit.

[0084] The correlation trend information module can be configured to analyze on-site monitoring data of the foundation pit and identify correlation trend information between multiple monitoring parameters caused by changes in deep soil. This module can be implemented as a data analysis engine containing various algorithms, such as time series analysis algorithms (e.g., ARIMA, LSTM), correlation analysis algorithms (e.g., Pearson correlation coefficient, Granger causality), and cluster analysis algorithms (e.g., K-means, DBSCAN). These algorithms are used to process the monitoring data to identify the temporal synchronicity, spatial proximity, and trend consistency between parameters. When the identified correlation trend information meets preset trigger conditions, the module will issue an instruction to initiate a risk scenario simulation.

[0085] The risk scenario simulation module can be configured to simulate the foundation pit response under preset external factors based on the current state of the digital model during risk scenario simulation. For example, this module can be implemented as a numerical simulator that integrates finite element analysis (FEA) or discrete element analysis (DEM) software interfaces, allowing it to take the digital model as input and perform simulation calculations based on preset external factors (such as rainfall intensity, seismic waveforms, and surrounding construction loads). Through simulation, this module can deduce the chain reaction triggered by changes in deep soil and calculate the probability and estimated time of the chain reaction reaching a preset safety limit in the future. The calculation of probability and time can be based on Monte Carlo simulation or reliability analysis methods.

[0086] The early warning information generation module can be configured to generate early warning information for foundation pit risks based on probability and expected time, and to graphically display the risk evolution trend. For example, this module can be implemented as a visualization engine, capable of transforming risk projection results into an intuitive graphical interface. Early warning information may include risk level, possible risk types, expected occurrence time, and recommended measures. Graphical display methods may include time-risk curves, risk heatmaps, 3D visualization models, and risk propagation path animations, enabling users to intuitively understand the risk development trend.

[0087] Traditional foundation pit monitoring systems rely on setting fixed safety limits for each monitoring point, issuing an alarm only when a single data point exceeds this limit. This approach treats minor increases in pore water pressure, slow movement of diaphragm walls, slight stress changes in steel supports, and minor settlements in subway tunnels as independent and unrelated events. It fails to recognize the inherent causal relationships and spatial interactions among these seemingly insignificant data changes that do not reach their respective warning thresholds. This single-indicator approach leads to severely delayed risk warnings, often triggered only after local instability has already occurred and a monitoring value has drastically changed and exceeded a single threshold, thus missing the optimal opportunity to take preventative measures.

[0088] To address this issue, this application proposes a multi-parameter dynamic judgment system for foundation pit monitoring risk status. By integrating functional modules such as digital model acquisition, monitoring data reception, correlation trend information identification, risk scenario simulation, and early warning information generation, it achieves early, dynamic, and comprehensive multi-parameter judgment of foundation pit risks. Compared to existing technologies, this system can identify potential systemic risks from seemingly independent, minute changes. For example, by identifying the synchronicity, proximity, and trend consistency among multiple monitoring parameters caused by changes in deep soil through the correlation trend information module, dynamic prediction is performed in the risk scenario simulation module, and finally, a forward-looking early warning is provided through the early warning information generation module. This modular design enables the system to process multi-source heterogeneous data more efficiently and accurately, and to perform complex risk simulations, significantly improving the intelligence level of foundation pit engineering safety management and the timeliness and accuracy of risk early warning, transforming foundation pit engineering safety management from passive response to proactive prevention.

[0089] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A multi-parameter dynamic determination method for the risk status of foundation pit monitoring, characterized in that, include: Obtain a digital model of the foundation pit project; the digital model includes the geometric structure of the foundation pit, the parameters of the support structure, and the physical and mechanical parameters of each soil layer. The physical and mechanical parameters of the soil layer include the model parameters of the weak soil layer zone; Receive on-site monitoring data of the foundation pit; Based on the on-site monitoring data of the foundation pit, the model parameters of the digital model are calibrated so that the digital model reflects the current state of the foundation pit; wherein, the calibration process includes updating the model parameters of the weak soil layer area according to the trend of changes in the deep soil mass; Analyze the on-site monitoring data of the foundation pit to identify the correlation trend information between multiple monitoring parameters caused by changes in deep soil; when the correlation trend information is found to meet the preset trigger conditions, a risk scenario simulation is initiated; wherein, the correlation trend information is a set of information composed of the time synchronization, spatial proximity and trend consistency between multiple monitoring parameters, and the information set includes at least synchronization index, amplitude deviation index, external load interference suppression result and deformation correlation result, which are used as the trigger basis for initiating the risk scenario simulation; During the risk scenario simulation, based on the current state of the digital model, the foundation pit response is simulated under the influence of preset external factors, the chain reaction caused by changes in deep soil is simulated, and the probability and expected time of the chain reaction reaching the preset safety limit in the future are calculated. Based on the probability and expected time, early warning information on foundation pit risk is generated, and the risk evolution trend is displayed graphically.

2. The multi-parameter dynamic determination method for the risk status of foundation pit monitoring according to claim 1, characterized in that, The analysis of the on-site monitoring data of the foundation pit identifies the correlation trend information between multiple monitoring parameters caused by changes in deep soil. When the identified correlation trend information meets the preset triggering conditions, the steps for initiating a risk scenario simulation include: Time series decomposition was performed on the data from multiple pore water pressure sensors in the deep weak interlayer area, the lateral displacement sensor data of the foundation pit retaining structure, and the vertical displacement observation point data of the adjacent rail transit structure in the on-site monitoring data of the foundation pit, and the trend component and high-frequency fluctuation component corresponding to each type of data were separated. Spatial difference analysis was performed on the trend components of the pore water pressure sensor data to obtain a synchronicity index to characterize the degree of synchronous change at multiple points, and an amplitude deviation index to characterize the degree of spatial difference. Frequency characteristic analysis is performed on the high-frequency fluctuation components of the pore water pressure sensor data to identify periodic fluctuations that match the characteristics of external loads, and suppression or elimination processing is performed on the periodic fluctuations to obtain the external load interference suppression results. Based on the external load interference suppression results, the trend component of the pore water pressure sensor data is corrected. In the corrected trend component, the synchronous change characteristics of multiple pore water pressure sensor data showing a continuous small increase are identified according to the synchronicity index. At the same time, the spatial differential distribution characteristics of the continuous small increase are identified according to the amplitude deviation index. The corresponding pore water pressure mode is determined based on the synchronous change characteristics and the differential distribution characteristics. The correlation between the lateral displacement sensor data of the foundation pit retaining structure and the vertical displacement observation point data of the adjacent rail transit structure and the pore water pressure mode in terms of spatial proximity and temporal synchronization was analyzed to obtain the deformation correlation results. The amplitude deviation index, the synchronicity index, the external load disturbance suppression result, and the deformation correlation result are used as correlation trend information; When the amplitude deviation index corresponding to the pore water pressure mode exceeds a preset threshold, and the synchronization index reaches a preset synchronization threshold, and the external load interference suppression result indicates that the external periodic load is insufficient to explain the pore water pressure mode, and the deformation correlation result indicates that the lateral displacement of the retaining structure and the vertical displacement of the rail transit structure show a deformation acceleration trend that is spatially adjacent to and temporally synchronized with the pore water pressure mode, it is determined that the correlation trend information meets the preset conditions, and risk scenario simulation is initiated.

3. The multi-parameter dynamic determination method for the risk status of foundation pit monitoring according to claim 1, characterized in that, The steps of analyzing the on-site monitoring data of the foundation pit and identifying the correlation trend information among multiple monitoring parameters caused by changes in deep soil include: Read the monitoring parameters from deep soil, foundation pit retaining structure and surrounding sensitive structures from the on-site monitoring data of the foundation pit to form a set of monitoring parameters; Time series analysis was performed on the set of monitoring parameters to extract the changing trends of each monitoring parameter; A correlation strength assessment model is established for each monitoring parameter pair; the monitoring parameter pair is a combination formed by selecting any two monitoring parameters from each monitoring parameter, which is used to characterize the correlation relationship between the two monitoring parameters and participate in the calculation of the correlation strength index; Based on the current excavation stage of the foundation pit, the trend of groundwater level changes, and the soil consolidation creep rate, update the correlation weight coefficients of the correlation strength assessment model. Based on the updated correlation weight coefficients and the correlation strength assessment model, the correlation strength index between each pair of monitoring parameters in the monitoring parameter set is calculated. Based on the correlation strength index, the monitoring parameter pair that most accurately reflects the potential risk at the current stage is determined as the parameter correlation combination, and the parameter correlation combination and the corresponding correlation strength index are used as the correlation trend information.

4. The multi-parameter dynamic determination method for the risk status of foundation pit monitoring according to claim 3, characterized in that, The step of determining the monitoring parameter pair that most accurately reflects the potential risk at the current stage as a parameter association combination based on the correlation strength index, and using the parameter association combination and the corresponding correlation strength index as the correlation trend information, includes: Establish a risk cascading path library; the risk cascading path library stores risk cascading paths; the risk cascading path defines the starting parameter pair, starting condition, intermediate transmission mechanism, triggering condition, and final impact parameter; The correlation strength index between each monitoring parameter pair is continuously read, and the trend of the correlation strength index is extracted to obtain the correlation strength change trend. Based on the trend of the correlation strength change and the starting conditions of each risk cascade path in the risk cascade path library, the trend matching degree between the trend of the correlation strength change and each risk cascade path is calculated. Risk cascade paths with a trend matching degree greater than or equal to a preset threshold are selected from the risk cascade path library to generate a cascade path candidate set; Based on the current digital model status of the foundation pit, for each risk cascading path in the cascading path candidate set, the evolution process of each risk cascading path in the cascading path candidate set is dynamically deduced, and the risk triggering potential assessment result corresponding to the risk cascading path is output; the risk triggering potential assessment result includes at least the probability of triggering subsequent chain reactions and the expected time. Based on the risk triggering potential assessment results, each risk cascading path in the cascading path candidate set is sorted, and the risk cascading path with the highest risk triggering potential and exceeding the preset risk potential threshold is determined as the target cascading path. The initial parameter pair corresponding to the target cascade path is determined as the parameter association combination that most accurately reflects the potential risks at the current stage, and the parameter association combination and the corresponding association strength index are used as the association trend information.

5. The multi-parameter dynamic determination method for the risk status of foundation pit monitoring according to claim 4, characterized in that, The step of calculating the trend matching degree between the trend of association strength change and each risk cascade path in the risk cascade path library, based on the trend of association strength change and the starting conditions of each risk cascade path, includes: The correlation strength index between each monitoring parameter pair is smoothed over a time window to filter out short-term fluctuations and noise, resulting in a smoothed correlation strength index. Based on the smoothed association strength index, the trend of association strength change is extracted, and feature extraction is performed on the trend of association strength change to obtain a trend feature vector. Real-time acquisition of the working condition information of the foundation pit, and use of the working condition information to contextualize the starting conditions of the risk cascading path; For each risk cascading path in the risk cascading path library, the trend matching degree of the risk cascading path is calculated based on the trend feature vector and the starting conditions after contextual constraints.

6. The multi-parameter dynamic determination method for the risk status of foundation pit monitoring according to claim 5, characterized in that, The step of calculating the trend matching degree of each risk cascading path in the risk cascading path library, based on the trend feature vector and the initial conditions after contextual constraints, includes: Retrieve the preset starting condition template of the target risk cascading path from the risk cascading path library; the starting condition template includes at least the starting parameter pair, the starting trend morphology characteristics, the applicable time window scale, and the spatial proximity constraint. The trend feature vector is normalized to match the time window scale of the starting condition template. Based on the normalized trend feature vector and the initial trend morphology features in the initial condition template, a trend similarity index is calculated. For the initial conditions after contextual constraints are applied, a contextual consistency index is generated; the contextual consistency index characterizes the degree of conformity between the initial conditions after contextual constraints and the spatial proximity constraints of the initial condition template. When the context consistency index is lower than a preset threshold, a penalty factor is generated, and the trend similarity index is penalized and corrected based on the penalty factor. The trend matching degree of the risk cascading path is obtained by combining the trend similarity index after penalty correction with the context consistency index according to preset weights.

7. The multi-parameter dynamic determination method for the risk status of foundation pit monitoring according to claim 5, characterized in that, The steps of extracting the correlation strength change trend based on the smoothed correlation strength index and extracting features from the correlation strength change trend to obtain a trend feature vector include: Within a preset sliding time window, the trend of the smoothed correlation strength index is extracted to obtain the correlation strength change trend sequence. Based on the correlation strength change trend sequence, the trend slope, trend acceleration, fluctuation amplitude, and persistence index are calculated; the persistence index is used to characterize the degree to which the trend maintains the same direction of change within a continuous window; Identify the inflection points and corresponding inflection point inflection point in the correlation strength change trend sequence; The trend slope, trend acceleration, fluctuation amplitude, persistence indicators, and inflection point strength are normalized. The normalized trend slope, trend acceleration, fluctuation amplitude, persistence index, and inflection point strength are combined in a preset order to obtain the trend feature vector.

8. The multi-parameter dynamic determination method for the risk status of foundation pit monitoring according to claim 5, characterized in that, The step of acquiring the working condition information of the foundation pit in real time and using the working condition information to contextualize the starting conditions of the risk cascading path includes: Real-time acquisition of working condition information of the foundation pit; the working condition information includes at least the excavation stage, the trend of groundwater level change, and the change of surrounding environmental load; For the starting conditions of each risk cascade path in the risk cascade path library, extract the constraint terms corresponding to the working condition information; the constraint terms include at least applicable excavation stage constraints, applicable water level change constraints, and applicable load change constraints; The consistency score is calculated by comparing the working condition information with the constraint terms. When the context consistency score is lower than the preset score, a context weight adjustment factor is generated, and the context weight adjustment factor is bound to the starting condition of the corresponding risk cascade path to obtain the starting condition after contextual constraints; wherein, the context weight adjustment factor is a penalty coefficient used to characterize the degree of deviation of the context consistency score, and is used to adjust the deviation cost of the starting condition in the trend matching degree calculation, so as to reduce the trend matching degree of the risk cascade path of context inconsistency.

9. The multi-parameter dynamic determination method for the risk status of foundation pit monitoring according to claim 8, characterized in that, The step of calculating the trend matching degree of each risk cascading path in the risk cascading path library, based on the trend feature vector and the initial conditions after contextual constraints, includes: Obtain the trend feature vector and the initial conditions after applying contextual constraints; Extract the initial condition constraint set from the initial conditions after contextualizing the constraints; the initial condition constraint set includes at least the allowable range constraints and directional constraints on the trend slope, trend acceleration, fluctuation amplitude, persistence index, and inflection point strength. For the trend feature vector, calculate the out-of-bounds amount of each trend feature relative to the allowable range constraint, and calculate the inconsistency amount of each trend feature relative to the directional constraint to obtain the deviation amount of each dimension of the trend feature vector. The deviations of each dimension are weighted and summed according to preset weights to obtain the initial condition deviation value. When the initial condition after contextual constraints carries the context weight adjustment factor, the context weight adjustment factor is applied to the initial condition deviation value to obtain the corrected deviation value. Based on the corrected deviation cost, a trend matching degree for the risk cascade path is generated.

10. A multi-parameter dynamic determination system for foundation pit monitoring risk status, used to perform multi-parameter dynamic determination of foundation pit monitoring risk status, characterized in that, include: The digital model acquisition module is used to acquire the digital model of the foundation pit project. The digital model includes the geometric structure of the foundation pit, the parameters of the support structure, and the physical and mechanical parameters of each soil layer; the physical and mechanical parameters of the soil layers include the model parameters of the weak soil layer area. The monitoring data receiving module is used to receive on-site monitoring data of the foundation pit; Based on the on-site monitoring data of the foundation pit, the model parameters of the digital model are calibrated so that the digital model reflects the current state of the foundation pit; wherein, the calibration process includes updating the model parameters of the weak soil layer area according to the trend of changes in the deep soil mass; The correlation trend information module is used to analyze the on-site monitoring data of the foundation pit and identify the correlation trend information between multiple monitoring parameters caused by changes in deep soil. When the correlation trend information is found to meet the preset triggering conditions, a risk scenario simulation is initiated. The risk scenario simulation module is used to simulate the foundation pit response under the influence of preset external factors based on the current state of the digital model during the risk scenario simulation, to deduce the chain reaction caused by the change of deep soil, and to calculate the probability and expected time of the chain reaction reaching the preset safety limit in the future. The early warning information generation module is used to generate early warning information for foundation pit risks based on the probability and expected time, and to display the risk evolution trend in a graphical manner.

Images