BIM (Building Information Modeling)-based pile support foundation pit whole-process safety management system

Abstract

The invention discloses a BIM (Building Information Modeling)-based pile support foundation pit whole-process safety management system, relates to the technical field of foundation pit safety management, and aims to solve the technical problem of frequent false alarm caused by insufficient construction interference identification capability of a fixed threshold alarm mechanism. And a decision intervention module. According to the method, the multi-source information atlas is constructed, and construction activities, equipment states, environmental factors and monitoring data are dynamically associated. When the data is abnormal, the root cause correlation analysis unit is started, and the intensity and duration characteristics of the abnormal signal are calculated by using a time sequence characteristic extraction algorithm and are compared with the historical pattern library. The mechanism can intelligently judge whether data abnormity is derived from construction interference such as substantive soil body / structure instability or heavy equipment operation, the alarm accuracy is remarkably improved, and the problem of frequent false alarm caused by insufficient construction interference identification capability of a fixed threshold value alarm mechanism is solved.

Description

Technical Field

[0001] This invention relates to the field of foundation pit safety management technology, and more specifically, to a BIM-based safety management system for the entire process of pile-supported foundation pits. Background Technology

[0002] In the field of foundation pit engineering safety management, with the continuous development of urban underground space, the interaction between foundation pit construction and the surrounding pipelines, soil, and support structure forms an increasingly complex "support-soil-pipeline" system. In particular, in complex working conditions involving pile-supported structures and adjacent important underground pipelines (such as high-pressure gas pipes, water supply pipes, etc.), the precise prevention and control of systemic instability risks has become a key problem that the industry urgently needs to solve.

[0003] Traditional methods heavily rely on alarm mechanisms based on fixed thresholds. Their core logic involves triggering alarms based on preset values ​​(e.g., pipeline settlement exceeding 30mm). However, the complex environment of foundation pit construction sites, the movement of heavy equipment (such as cranes and excavators), and the instantaneous vibrations and load changes generated by support installation operations can all cause temporary fluctuations in monitoring data. These fluctuations are not caused by substantial structural instability or soil damage. Traditional systems cannot distinguish between these benign disturbances caused by normal construction activities and genuine structural instability risks. Once the data fluctuation reaches the fixed threshold, a high-level alarm (such as a red alert) is triggered, leading to frequent false alarms. This not only wastes management resources but also causes "alarm fatigue," making managers insensitive to genuine danger signals and creating significant safety hazards. Therefore, we propose a BIM-based safety management system for the entire process of pile-supported foundation pit construction. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the existing technology, adapt to the needs of reality, and provide a BIM-based safety management system for the entire process of pile-supported foundation pits, so as to solve the technical problem of frequent false alarms caused by the insufficient recognition ability of the current fixed threshold alarm mechanism due to construction interference.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a BIM-based safety management system for the entire process of pile-supported foundation pits, the system including a multi-source synchronous sensing module, a digital twin intelligent module, and a decision intervention module; The digital twin intelligent module is connected to the multi-source synchronous sensing module and is used to build and update a dynamic digital twin based on real-time data, and to perform risk simulation and diagnosis. The digital twin intelligent module includes a dynamic risk simulation module and a multi-source causal diagnosis module. The dynamic risk simulation module includes a real-time digital twin unit, a mechanism-data fusion unit, and a risk prediction unit; The real-time digital twin unit is used to achieve high-fidelity synchronous mapping between physical entities and virtual models; The mechanism-data fusion unit is used to fuse mechanical mechanism models and data-driven algorithms for state simulation and prediction. The risk prediction unit is used to predict risk trends based on the output of the fusion model and identify the deviation between the predicted value and the actual value to generate a trend warning. The multi-source causal diagnosis module includes a multi-source information graph unit, a root cause correlation analysis unit, and a responsibility tracing unit; The multi-source information graph unit is used to construct and manage the relationships between construction, monitoring, environmental and behavioral data; The root cause correlation analysis unit is used to perform cross-domain causal analysis based on the multi-source information graph when monitoring data is abnormal, and to locate the root cause of the risk. The responsibility tracing unit is used to associate the diagnosed root causes of risks with specific construction activities, equipment, or responsible persons.

[0006] Preferably, the risk prediction unit is configured to trigger a trend warning when the actual value of the monitoring data deviates significantly from the calculated prediction value of the mechanism-data fusion unit. The triggering priority of this warning is higher than that of a threshold alarm based on a fixed threshold.

[0007] Preferably, the root cause correlation analysis unit is configured to distinguish between real risk alarms caused by substantial soil or structural instability and instantaneous data anomalies caused by construction interference such as heavy equipment operation and temporary loading, thereby reducing false alarms.

[0008] Preferably, the multi-source synchronous sensing module is used to collect multi-source data in real time during the foundation pit construction process, including an Internet of Things (IoT) monitoring unit, a video behavior recognition unit, and an environmental and social data analysis unit; The IoT monitoring unit is used to acquire data on the stress, displacement and deformation of the support structure and the surrounding environment through a built-in sensor network, and to dynamically link the designed BIM model with the actual construction progress. The video behavior recognition unit uses computer vision algorithms to automatically identify personnel violations, equipment status, and typical construction procedures in the construction area. The environmental and social data analysis unit establishes and processes the correlation model between environmental and social data, analyzes the coupling influence relationship between multi-source data, and converts the processed effective data and anomaly analysis results into structured data that conforms to the unified system standard, and transmits them to the real-time digital twin unit and the mechanism-data fusion unit.

[0009] Preferably, the decision intervention module is connected to the digital twin intelligent module and is used to provide risk warnings and decision suggestions based on the central output, including a risk warning unit, a decision support unit, and an instruction execution unit; The risk warning unit includes trend warning and threshold alarm. The trend warning analyzes the risk evolution trend, rate and coupling effect to predict potential risks in advance and finally generate graded warning information. Threshold alarms trigger an immediate alarm when the data exceeds the threshold range by comparing risk parameters with preset safety thresholds in real time. The decision support unit is used to simulate the effects of different measures in a digital twin; The instruction execution unit is used to send early warning information to relevant responsible persons according to the risk level, track the execution results of the intervention instructions recorded by the digital twin unit, and feed them back to the digital twin intelligent module to optimize the model.

[0010] Preferably, the real-time digital twin unit uses a physical-virtual model synchronization error control algorithm to achieve high-fidelity mapping. The algorithm is as follows: ; like This triggers a real-time model update; in, for Virtual-physical synchronization error at any moment for The position coordinates of the support structure in the BIM virtual model at any time. For The stress values ​​of the support structure in the BIM virtual model at any given time. for The measured coordinates of the support structure within the physical entity of the foundation pit at any given time. for Measured stress values ​​of the support structure in the physical entity of the foundation pit at any given time. This represents the maximum permissible synchronization error.

[0011] Preferably, the mechanism-data fusion unit adopts an adaptive fusion prediction model, and the algorithm is as follows: ; in, for The combined predicted value of the support structure state based on the stress and displacement at any given time. for Real-time adaptive fusion weights for The calculated value of the mechanical mechanism model at any moment. for Time-based data drives model predictions. for The prediction error variance of the time-machining mechanism model. for Time-based data drives the model's prediction error variance.

[0012] Preferably, the risk prediction unit uses a deviation significance test algorithm to trigger trend warning. The algorithm is as follows: ; like and If so, a trend warning will be triggered; in, for The relative deviation rate between the monitored values ​​and the fused predicted values ​​at any given time. for Measured values ​​of the support structure condition at all times. for The combined predicted value of stress and displacement of the support structure at any given time. To set a preset deviation threshold, The length of the sliding window. For the first The relative deviation rate at any given time.

[0013] Preferably, the root cause association analysis unit uses a time series feature extraction algorithm to distinguish between real risks and transient anomalies, as shown in the formula: ; like If it is true, it is determined to be a real risk; otherwise, it is a transient anomaly. in, for abnormal signal strength at any time for Monitor the changes in data constantly. The time interval for data changes. for The duration of the time-based abnormal data, As the baseline duration, This is the signal strength threshold.

[0014] Preferably, the decision support unit uses a multi-objective optimization evaluation algorithm to screen disposal schemes, as shown in the following formula: ; Where E represents the overall effectiveness index of the disposal measures. The risk level before treatment, For the first The risk level after the implementation of these measures Based on the benchmark disposal cost, For the first The implementation cost of such disposal measures For the first The implementation time of these measures Based on the implementation time, , =0.2, and =0.2.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention constructs a multi-source information map to dynamically correlate construction activities, equipment status, environmental factors, and monitoring data. When data anomalies occur, the root cause analysis unit is activated, applying a time-series feature extraction algorithm to calculate the intensity and duration characteristics of the abnormal signal and comparing it with a historical pattern database. This mechanism can intelligently determine whether data anomalies originate from substantial soil / structure instability or construction interference such as heavy equipment operation, thereby effectively filtering out the vast majority of false alarms caused by temporary construction activities, significantly improving alarm accuracy, and solving the problem of frequent false alarms caused by insufficient construction interference identification capabilities in fixed threshold alarm mechanisms.

[0016] 2. This invention also deploys an adaptive fusion prediction model that integrates mechanical mechanisms with real-time monitoring data to dynamically predict the state of support structures and pipelines. Furthermore, by utilizing a deviation significance test algorithm, it analyzes the relative deviation rate between actual and predicted values ​​of monitoring data and their cumulative effect, enabling the identification of abnormal trends before physical quantities exceed limits. This allows the system to issue trend warnings, with higher priority than traditional threshold alarms, thereby achieving early detection and early warning of hidden risks such as the formation of cavities under pipelines. This advances the risk response window by several hours, completely resolving the problem of missed reports caused by the failure to identify risk evolution trends.

[0017] 3. This invention also employs a physical-virtual model synchronization error control algorithm to ensure high-fidelity synchronization between the BIM virtual model and the physical entity of the foundation pit throughout the entire process. This not only provides accurate initial and boundary conditions for the aforementioned dynamic predictions but also enables the decision support unit to perform multi-objective optimization evaluation algorithms for different emergency response plans within the digital twin, quantitatively assessing the comprehensive effects of each plan in terms of risk control, cost, and schedule. This assists managers in making scientific and optimal decisions, changing the traditional extensive management model that relies on experience. Attached Figure Description

[0018] Figure 1 This is a system block diagram of the present invention; Figure 2 This is a detailed system block diagram of the multi-source synchronous sensing module of the present invention; Figure 3 This is a specific system block diagram of the decision intervention module of the present invention. Detailed Implementation

[0019] Example: Figures 1 to 3 As shown, the present invention relates to a BIM-based safety management system for the entire process of pile-supported foundation pits. The system includes a multi-source synchronous sensing module, a digital twin intelligent module, and a decision intervention module. The digital twin intelligent module is connected to the multi-source synchronous sensing module and is used to build and update a dynamic digital twin based on real-time data, and to perform risk simulation and diagnosis. The digital twin intelligent module includes a dynamic risk simulation module and a multi-source causal diagnosis module. The dynamic risk simulation module includes a real-time digital twin unit, a mechanism-data fusion unit, and a risk prediction unit; The real-time digital twin unit is used to achieve high-fidelity synchronous mapping between physical entities and virtual models. To ensure mapping accuracy, a physical-virtual model synchronization error control algorithm is adopted, as follows: ; like This triggers real-time model updates.

[0020] in, for Virtual-physical synchronization error at any moment for The position coordinates of the support structure in the BIM virtual model at any time. For The stress values ​​of the support structure in the BIM virtual model at any given time. for The measured coordinates of the support structure within the physical entity of the foundation pit at any given time. for Measured stress values ​​of the support structure in the physical entity of the foundation pit at any given time. This represents the maximum permissible synchronization error.

[0021] This algorithm ensures that the deviation between the virtual model and the physical entity is always kept within the error range, providing a high-fidelity virtual scenario for subsequent risk simulation and response rehearsal.

[0022] The mechanism-data fusion unit is used to fuse mechanical mechanism models and data-driven algorithms for state simulation and prediction. An adaptive fusion prediction model is adopted, and the algorithm is as follows: ; in, for The combined predicted value of the support structure state based on the stress and displacement at any given time. for The time-adaptive fusion weights are used to balance the contribution ratios of the two types of models, and their values ​​range from [0,1]. for The calculated value of the mechanical mechanism model at any moment. for The model's predictions are driven by real-time data and are generated using an LSTM neural network trained on historical monitoring and environmental data. for The variance of the prediction error of the time-matter mechanical mechanism model, corresponding to The physical dimensions of . for Time-based data drives the model prediction error variance, corresponding to The physical dimensions of .

[0023] in, Based on the theory of elastic foundation beams: ; in Earth pressure, For the span of the support piles, The bending stiffness of the pile; Among them, the variance of the prediction error of the real-time computer model is... Variance of prediction error in data-driven models It supports adaptive fusion of weights and trend early warning determination, and adopts a simplified sliding window method. The specific implementation is as follows: a. Parameter settings: Window length This corresponds to 5 hours of monitoring data, with a sampling frequency of 30 minutes per sampling. The data selection logic balances the reliability of error variance calculation with system computational efficiency, capturing changes in model prediction accuracy without requiring excessively long periods, and balancing computational efficiency with data representativeness. The sliding step size... The window updates synchronously with each new monitoring data point added. b: Initialize the window dataset: ; in , , For the mechanism model Prediction error at any time For the data-driven model The prediction error is determined by taking the initial data from the first 10 sampling periods after system startup. c: Add After obtaining the time data, calculate the new error. : , Update the dataset: Remove oldest data ,join in , Similarly; d: Variance calculation: ; in, For the first The prediction error of the time-matter mechanical mechanism model For the first The prediction error of the model is driven by time-matter data. For the first Measured values ​​of the support structure condition at all times. This represents the length of the sliding window.

[0024] The model is through The advantages of the two types of models are dynamically balanced: when there is sufficient data, the data-driven model is emphasized, and when the construction stage changes, the mechanism model is emphasized, so as to reduce the average error of the fused prediction value and balance the prediction accuracy and anti-interference ability.

[0025] The multi-source causal diagnosis module includes a multi-source information graph unit, a root cause correlation analysis unit, and a responsibility tracing unit; The multi-source information graph unit is used to construct and manage the correlation relationships between construction, monitoring, environmental, and behavioral data. A multi-source data correlation strength calculation algorithm is used to quantify the correlation relationships, as detailed below: ; in, for Time data type and The correlation strength, with values ​​ranging from -1 to 1, shows that the larger the absolute value, the stronger the correlation. for Time data type and covariance, for Time data type variance for Time data type variance for Time of the first The values ​​that can be taken for data types, such as pile displacement and earth pressure, for Time of the first The values ​​that can be taken from data types, such as traffic flow and weather data, Domain weights are set based on engineering experience, such as the relationship between pile displacement and earth pressure. With traffic flow This algorithm can quickly identify risk-related factors, solving the problem of disorganized and difficult-to-correlate multi-source data.

[0026] Preferably, the risk prediction unit is configured to trigger a trend warning when the actual value of the monitoring data deviates significantly from the calculated prediction value of the mechanism-data fusion unit. The triggering priority of this warning is higher than that of a threshold alarm based on a fixed threshold.

[0027] The risk prediction unit is used to predict risk trends based on the output of the fusion model and identify the deviation between the predicted value and the actual value to generate a trend warning. This unit is configured to trigger trend alerts using a deviation significance test algorithm, as follows: ; like and This will trigger a trend warning.

[0028] in, for The relative deviation rate between the monitored values ​​and the fused predicted values ​​at any given time. for Measured values ​​of the support structure condition at all times. for The combined predicted value of stress and displacement of the support structure at any given time. The preset deviation threshold is set according to the safety level of the foundation pit; for a Class I foundation pit, it is set at 5% to 8%. The value represents the sliding window length, corresponding to a 24-hour monitoring cycle, with a sampling frequency of 30 minutes per instance. The logic behind this value is to cover the short-term trend cycle of risk evolution. By accumulating the deviation over 24 hours, false alarms due to instantaneous fluctuations are avoided, ensuring the accuracy of trend warnings. The determination of whether a trend warning is triggered is based on the accumulated relative deviation rate over multiple time points. For the first The relative deviation rate at any given time. This warning has a higher triggering priority than threshold alarms based on fixed thresholds. By determining the dual conditions of instantaneous deviation and cumulative deviation, it can provide warnings 6-12 hours earlier than traditional single threshold alarms, enabling early detection and early warning of risks.

[0029] By deploying an adaptive fusion prediction model that integrates mechanical mechanisms with real-time monitoring data, the system dynamically predicts the condition of support structures and pipelines. Furthermore, by utilizing a deviation significance test algorithm, and analyzing the relative deviation rate between actual and predicted values ​​of monitoring data and its cumulative effect, abnormal trends can be identified before physical quantities exceed limits. This enables the system to issue trend warnings, with higher priority than traditional threshold alarms, thereby achieving early detection and early warning of hidden risks such as the formation of cavities under pipelines. This advances the risk intervention window by several hours, completely resolving the problem of missed reports caused by the failure to identify risk evolution trends.

[0030] Preferably, the root cause correlation analysis unit is configured to distinguish between real risk alarms caused by substantial soil or structural instability and instantaneous data anomalies caused by construction interference such as heavy equipment operation and temporary loading, thereby reducing false alarms.

[0031] The root cause correlation analysis unit is used to perform cross-domain causal analysis based on the multi-source information map when monitoring data anomalies, to locate the root cause of the risk. This unit is configured to use a time series feature extraction algorithm to distinguish between real risks and transient anomalies, and to analyze signal strength. The specific formula is as follows: ; like If it is true, it is considered a real risk; otherwise, it is considered a transient anomaly.

[0032] in, for The intensity of abnormal signals at any given time is a normalized indicator. for Monitor the changes in data constantly. The time interval for data changes. for The duration of the time-based abnormal data, As the baseline duration, This is the signal strength threshold.

[0033] Signal strength threshold The derivation logic is as follows: a. Collect two types of historical monitoring data for similar foundation pit projects: actual risks (soil instability, structural cracking, etc.) and transient disturbances (equipment operation, temporary loading, etc.); b. Based on signal strength ,when Calculate the scores of all samples Discover real risk samples All ≥0.62, transient interference samples All ≤0.58; c. Take 0.6 as the threshold between the two types of samples to ensure that no real risks are missed and to effectively eliminate instantaneous interference, thus meeting the accuracy requirements for practical engineering applications.

[0034] When heavy equipment is in operation Big but ,at this time This is determined to be a transient anomaly; when the soil becomes unstable, large and ,at this time This was determined to be a real risk.

[0035] By constructing a multi-source information map, construction activities, equipment status, environmental factors, and monitoring data are dynamically correlated. When data anomalies occur, the root cause analysis unit is activated, applying a time-series feature extraction algorithm to calculate the intensity and duration characteristics of the abnormal signal and comparing it with a historical pattern database. This mechanism can intelligently determine whether data anomalies originate from substantial soil / structural instability or construction interference such as heavy equipment operation, thereby effectively filtering out the vast majority of false alarms caused by temporary construction activities, significantly improving alarm accuracy, and solving the problem of frequent false alarms caused by insufficient construction interference identification capabilities of fixed threshold alarm mechanisms.

[0036] The responsibility tracing unit is used to associate the diagnosed risk sources with specific construction activities, equipment, or responsible persons. Combined with the construction logs, equipment operation data, and personnel attendance information recorded by the multi-source information graph unit, it enables accurate tracing of risk responsibility.

[0037] Preferably, the multi-source synchronous sensing module is used to collect multi-source data in real time during the foundation pit construction process, including an Internet of Things (IoT) monitoring unit, a video behavior recognition unit, and an environmental and social data analysis unit; The IoT monitoring unit is used to acquire data on the stress, displacement and deformation of the support structure and the surrounding environment through the built-in sensor network, and dynamically associates the designed BIM model with the actual construction progress to provide basic data support for subsequent fusion prediction and synchronous mapping. The video behavior recognition unit uses computer vision algorithms to automatically identify personnel violations in the construction area, such as not wearing safety protective equipment or illegally entering dangerous areas; equipment status such as equipment failure or improper operation; and typical construction procedures such as foundation pit excavation and support installation. The recognition results are synchronized to the multi-source information map unit. The environmental and social data analysis unit establishes and processes the correlation model between environmental data (meteorological and geological data) and social data (traffic and public opinion data), analyzes the coupling influence relationship between multi-source data, and converts the processed effective data and anomaly analysis results into structured data that conforms to the unified system standard, and transmits them to the real-time digital twin unit and the mechanism-data fusion unit.

[0038] Preferably, the decision intervention module is connected to the digital twin intelligent module and is used to provide risk warnings and decision suggestions based on the central output, including a risk warning unit, a decision support unit, and an instruction execution unit; The risk warning unit includes trend warning and threshold alarm. The trend warning analyzes the risk evolution trend, rate and coupling effect through the deviation significance test algorithm of the above-mentioned risk prediction unit, predicts potential risks in advance, and finally generates graded warning information of four levels: blue, yellow, orange and red. Threshold alarms are triggered by comparing risk parameters with preset safety thresholds in real time. When the data exceeds the threshold range, an immediate alarm is triggered. Trend warnings have a higher triggering priority than threshold alarms. The decision support unit is used to simulate the effects of different treatment measures; A multi-objective optimization evaluation algorithm is used to select the optimal solution, as follows: ; Where E is the comprehensive effectiveness index of the treatment measures, with a value range of [0,1]. The larger the value, the better the effect. The risk level before intervention is categorized into levels 1 to 4, corresponding to a score of 4 to 1. For the first The risk level after the implementation of these measures is classified into four levels, from 1 to 4, corresponding to a score of 4 to 1. Based on the benchmark disposal cost, For the first The implementation cost of such disposal measures For the first The implementation time of these measures Based on the implementation time, , ,and The weighting coefficients are set to 0.6, 0.2, and 0.2 respectively, prioritizing the risk reduction effect. This algorithm can quickly screen out the optimal solution with high risk reduction rate, low cost, and short construction period, solving the problem of traditional decision-making relying on experience and lacking quantitative basis.

[0039] The instruction execution unit is used to send early warning information to relevant responsible persons according to the risk level, track and record the execution results of intervention instructions, and provide early warnings through multiple channels such as sound and light alarms, platform push, and SMS notifications. At the same time, it tracks and records the execution results of intervention instructions and feeds them back to the digital twin intelligent module to iteratively optimize the core algorithm parameters such as the fusion prediction model and the synchronous error control algorithm, thereby improving the system's subsequent risk prediction and handling capabilities.

[0040] Through a physical-virtual model synchronization error control algorithm, the system ensures high-fidelity synchronization between the BIM virtual model and the physical entity of the foundation pit throughout the entire process. This not only provides accurate initial and boundary conditions for the aforementioned dynamic predictions, but also enables the decision support unit to conduct multi-objective optimization evaluation algorithm simulations of different emergency response plans within the digital twin. It quantitatively evaluates the comprehensive effectiveness of each plan in terms of risk control, cost, and schedule, thereby assisting managers in making scientific and optimal decisions and changing the traditional extensive management model that relies on experience.

[0041] The embodiments disclosed in this invention are preferred embodiments, but are not limited thereto. Those skilled in the art can easily understand the spirit of this invention based on the above embodiments and make different extensions and variations, but as long as they do not depart from the spirit of this invention, they are all within the protection scope of this invention.

Claims

1. A BIM-based safety management system for the entire process of pile-supported foundation pit construction, characterized in that, The system includes a multi-source synchronous sensing module, a digital twin intelligence module, and a decision intervention module; The digital twin intelligent module is connected to the multi-source synchronous sensing module and is used to build and update a dynamic digital twin based on real-time data, and to perform risk simulation and diagnosis. The digital twin intelligent module includes a dynamic risk simulation module and a multi-source causal diagnosis module. The dynamic risk simulation module includes a real-time digital twin unit, a mechanism-data fusion unit, and a risk prediction unit; The real-time digital twin unit is used to achieve high-fidelity synchronous mapping between physical entities and virtual models; The mechanism-data fusion unit is used to fuse mechanical mechanism models and data-driven algorithms for state simulation and prediction. The risk prediction unit is used to predict risk trends based on the output of the fusion model and identify the deviation between the predicted value and the actual value to generate a trend warning. The multi-source causal diagnosis module includes a multi-source information graph unit, a root cause correlation analysis unit, and a responsibility tracing unit; The multi-source information graph unit is used to construct and manage the relationships between construction, monitoring, environmental and behavioral data; The root cause correlation analysis unit is used to perform cross-domain causal analysis based on the multi-source information graph when monitoring data is abnormal, and to locate the root cause of the risk. The responsibility tracing unit is used to associate the diagnosed root causes of risks with specific construction activities, equipment, or responsible persons.

2. The BIM-based safety management system for the entire process of pile-supported foundation pits according to claim 1, characterized in that, The risk prediction unit is configured to trigger a trend warning when the actual value of the monitoring data deviates significantly from the calculated prediction value of the mechanism-data fusion unit. The triggering priority of this warning is higher than that of a threshold alarm based on a fixed threshold.

3. The BIM-based safety management system for the entire process of pile-supported foundation pits according to claim 2, characterized in that, The root cause correlation analysis unit is configured to distinguish between real risk alarms caused by substantial soil or structural instability and instantaneous data anomalies caused by construction interference such as heavy equipment operation and temporary loading, thereby reducing false alarms.

4. A BIM-based safety management system for the entire process of pile-supported foundation pits according to claim 3, characterized in that, The multi-source synchronous sensing module is used to collect multi-source data in real time during the foundation pit construction process, including an Internet of Things (IoT) monitoring unit, a video behavior recognition unit, and an environmental and social data analysis unit. The IoT monitoring unit is used to acquire data on the stress, displacement and deformation of the support structure and the surrounding environment through a built-in sensor network, and to dynamically link the designed BIM model with the actual construction progress. The video behavior recognition unit uses computer vision algorithms to automatically identify personnel violations, equipment status, and typical construction procedures in the construction area. The environmental and social data analysis unit establishes and processes the correlation model between environmental and social data, analyzes the coupling influence relationship between multi-source data, and converts the processed effective data and anomaly analysis results into structured data that conforms to the unified system standard, and transmits them to the real-time digital twin unit and the mechanism-data fusion unit.

5. A BIM-based safety management system for the entire process of pile-supported foundation pits according to claim 4, characterized in that, The decision intervention module is connected to the digital twin intelligent module and is used to provide risk warnings and decision suggestions based on the output of the central system. It includes a risk warning unit, a decision support unit, and an instruction execution unit. The risk warning unit includes trend warning and threshold alarm. The trend warning analyzes the risk evolution trend, rate and coupling effect to predict potential risks in advance and finally generate graded warning information. Threshold alarms trigger an immediate alarm when the data exceeds the threshold range by comparing risk parameters with preset safety thresholds in real time. The decision support unit is used to simulate the effects of different measures in a digital twin; The instruction execution unit is used to send early warning information to relevant responsible persons according to the risk level, track the execution results of the intervention instructions recorded by the digital twin unit, and feed them back to the digital twin intelligent module to optimize the model.

6. A BIM-based safety management system for the entire process of pile-supported foundation pits according to claim 5, characterized in that, The real-time digital twin unit employs a physical-virtual model synchronization error control algorithm to achieve high-fidelity mapping. The algorithm is as follows: ； like This triggers a real-time model update; in, for Virtual-physical synchronization error at any moment for The position coordinates of the support structure in the BIM virtual model at any time. For The stress values ​​of the support structure in the BIM virtual model at any given time. for The measured coordinates of the support structure within the physical entity of the foundation pit at any given time. for Measured stress values ​​of the support structure in the physical entity of the foundation pit at any given time. This represents the maximum permissible synchronization error.

7. A BIM-based safety management system for the entire process of pile-supported foundation pits according to claim 2, characterized in that, The mechanism-data fusion unit employs an adaptive fusion prediction model, and the algorithm is as follows: ； in, for The combined predicted value of the support structure state based on the stress and displacement at any given time. for Real-time adaptive fusion weights for The calculated value of the mechanical mechanism model at any moment. for Time-based data drives model predictions. for The prediction error variance of the time-machining mechanism model. for Time-based data drives the model's prediction error variance.

8. A BIM-based safety management system for the entire process of pile-supported foundation pits according to claim 2, characterized in that, The risk prediction unit uses a deviation significance test algorithm to trigger trend warnings. The algorithm is as follows: ； like and If so, a trend warning will be triggered; in, for The relative deviation rate between the monitored values ​​and the fused predicted values ​​at any given time. for Measured values ​​of the support structure condition at all times. for The combined predicted value of stress and displacement of the support structure at any given time. To set a preset deviation threshold, The length of the sliding window. For the first The relative deviation rate at any given time.

9. A BIM-based safety management system for the entire process of pile-supported foundation pits according to claim 3, characterized in that, The root cause association analysis unit uses a time series feature extraction algorithm to distinguish between real risks and transient anomalies, as shown in the formula: ； like If it is true, it is determined to be a real risk; otherwise, it is a transient anomaly. in, for abnormal signal strength at any time for Monitor the changes in data constantly. The time interval for data changes. for The duration of the time-based abnormal data, As the baseline duration, This is the signal strength threshold.

10. A BIM-based safety management system for the entire process of pile-supported foundation pits according to claim 5, characterized in that, The decision support unit uses a multi-objective optimization evaluation algorithm to select treatment options, as shown in the formula: ； Where E represents the overall effectiveness index of the disposal measures. The risk level before treatment, For the first The risk level after the implementation of these measures Based on the benchmark disposal cost, For the first The implementation cost of such disposal measures For the first The implementation time of these measures Based on the implementation time, , =0.2, and =0.2.

Images