Foundation pit digital twinning monitoring method and system and application thereof

Abstract

The invention discloses a foundation pit digital twinning monitoring method and system and application thereof. The system comprises a foundation pit data acquisition module, a foundation pit digital model simulation module, a foundation pit multi-source data processing and analysis module, a foundation pit abnormal state intelligent prediction module, a foundation pit monitoring and early warning decision module and a foundation pit data storage and backup module. The invention further discloses a foundation pit digital twinning monitoring method, deep learning modeling is carried out in combination with multi-source historical data, and the accuracy and the intelligent level of foundation pit abnormal state monitoring are further improved.

Description

Technical Field

[0001] The present invention relates to computer science and digital twin technology, and is applied to foundation pit monitoring. Background Art

[0002] Digital twin technology enables real-time updating and monitoring of the state of physical entities by creating an accurate virtual copy of the physical entity in a computer system. In the field of engineering, due to its large depth and area, as well as the complexity of the surrounding terrain, manual monitoring is difficult to achieve in real time, and the response is slow in special circumstances. By applying digital twin technology, we can transform the foundation pit into a virtual model and integrate advanced sensor equipment to generate a real-time state diagram, thereby achieving continuous and real-time monitoring of the foundation pit. This method not only reduces labor costs, but also significantly improves the efficiency of monitoring. Despite this, the current foundation pit monitoring technology based on digital twins still has the problem of low accuracy in identifying abnormal states, and the level of intelligence needs to be improved. In addition, the existing technology has not been able to fully utilize multi-source data, which limits its monitoring capabilities. In order to further improve the accuracy and intelligence level of monitoring, future research and development needs to focus on improving the accuracy of abnormal state monitoring and exploring how to more effectively integrate and analyze data from different sources to achieve more comprehensive and intelligent foundation pit monitoring. Summary of the invention

[0003] In view of the shortcomings of the existing detection methods, the present invention proposes a foundation pit digital twin monitoring system, which includes: a foundation pit data acquisition module, a foundation pit digital model simulation module, a foundation pit multi-source data processing and analysis module, a foundation pit abnormal state intelligent prediction module, a foundation pit monitoring and early warning decision module, and a foundation pit data storage and backup module.

[0004] The foundation pit data acquisition module is used to collect physical parameter data such as displacement, stress, settlement, cracks, etc. from sensors of the foundation pit and its surrounding environment in real time, and transmit the data to the processing center through a communication interface.

[0005] The foundation pit digital model simulation module is used to construct a three-dimensional digital model of the foundation pit and synchronize it with the real-time collected data to simulate the actual state and changes of the foundation pit in a virtual environment.

[0006] The foundation pit multi-source data processing and analysis module is used to integrate and fuse information from different sensors and data sources, extract key features, and analyze the changing trend of the foundation pit state to provide data support for subsequent abnormal state prediction.

[0007] The intelligent prediction module for abnormal conditions of foundation pits is used to learn historical data using deep learning algorithms, establish prediction models, identify abnormal patterns, and assess the risks that may be caused by abnormal conditions, providing a scientific basis for early warning decisions.

[0008] The foundation pit monitoring and early warning decision-making module is responsible for determining whether an early warning needs to be triggered based on the results of the intelligent prediction module, and issuing early warning information to relevant personnel through various communication channels, while providing emergency response plans and guiding on-site personnel to take corresponding measures.

[0009] The foundation pit data storage and backup module is responsible for storing all collected data, analysis results and warning records in the database, and regularly backing up data to ensure the security, integrity and traceability of information.

[0010] The present invention also relates to a foundation pit digital twin monitoring method, which is implemented by the above-mentioned foundation pit digital twin monitoring system and comprises the following steps: Step S1, foundation pit data collection: Use a variety of equipment to obtain key data of the foundation pit, including various sensors and high-definition cameras. The types of sensors include displacement sensors, stress sensors, pressure sensors, tilt sensors, water level sensors and temperature sensors. The sensors are strategically deployed at various key locations of the foundation pit. Their main responsibility is to create accurate three-dimensional models for digital twins and continuously monitor the state changes of the foundation pit. The high-definition cameras are installed in key areas of the foundation pit to capture visual evidence such as wall crack development and structural deformation to enhance the comprehensiveness and accuracy of the monitoring system.

[0011] Step S2, simulation modeling of the digital twin model of the foundation pit: using BIM (Building Information Modeling) technology and GIS (Geographic Information System) data, based on the geographic information and engineering drawings of the foundation pit, and the basic information collected by the sensor, a three-dimensional digital twin model of the foundation pit is established. The collected sensor data is associated with the digital twin model of the foundation pit to achieve dynamic update of the foundation pit indicators to reflect the actual status of the foundation pit.

[0012] Step S3, pit data processing: a series of standardization processes are performed on the sensor data and camera images collected in step S1 to improve the availability and consistency of the data. The processing flow includes data normalization, data mapping, and filling.

[0013] Step S4, extraction of multi-source historical features of foundation pits: implement a continuous data collection strategy, integrate and record time series data from sensors, cameras, and three-dimensional models, ensure the continuity and integrity of the data in the time dimension, and use it to monitor the changing trend of the foundation pit status.

[0014] Step S5, modeling of foundation pit state change trend: involves using a deep learning algorithm in combination with historical data and multi-source data to model the foundation pit state trend and predict the future state of the foundation pit for abnormal state assessment.

[0015] Step S6, risk assessment of abnormal state of foundation pit: Use classification algorithm to classify and evaluate the predicted future state of foundation pit. Once the prediction result shows that the future state of foundation pit may exceed the safety threshold, the system will automatically trigger the early warning mechanism. The assessment result will be used to update the risk management strategy so as to take appropriate preventive measures to reduce potential engineering risks.

[0016] Step S7, foundation pit monitoring and early warning: When the foundation pit abnormal state risk assessment step detects an abnormal situation, that is, the state or predicted state of the foundation pit exceeds the preset safety threshold, the foundation pit monitoring and early warning system will be activated immediately. The early warning information will be promptly conveyed to project managers and on-site engineers through multiple channels, including but not limited to mobile application (APP) push notifications, text messages, emails and phone calls. Ensure that when the foundation pit is in an abnormal state, relevant personnel can quickly take measures, such as suspending construction, evacuating personnel or conducting on-site inspections, to prevent potential safety accidents and project losses.

[0017] The present invention also discloses a non-volatile storage medium, characterized in that the non-volatile storage medium includes a stored program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute the above method.

[0018] The present invention also discloses an electronic device, characterized in that it comprises a processor and a memory; the memory stores computer-readable instructions, and the processor is used to execute the computer-readable instructions, wherein the computer-readable instructions execute the above method when they are executed. Beneficial Effects

[0019] The present invention discloses a foundation pit digital twin monitoring method, system and application thereof. The system comprises: a foundation pit data acquisition module, a foundation pit digital model simulation module, a foundation pit multi-source data processing and analysis module, a foundation pit abnormal state intelligent prediction module, a foundation pit monitoring early warning decision module, and a foundation pit data storage and backup module.

[0020] The present invention also discloses a digital twin monitoring method for foundation pits, which combines multi-source historical data for deep learning modeling to further improve the accuracy and intelligence level of abnormal state monitoring of foundation pits. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is an architecture diagram of a foundation pit digital twin monitoring system of the present invention. DETAILED DESCRIPTION Example 1

[0022] The present invention is further described below in conjunction with the accompanying drawings. Embodiment 1 of the present invention relates to a foundation pit digital twin monitoring system, the system comprising: a foundation pit data acquisition module, a foundation pit digital model simulation module, a foundation pit multi-source data processing and analysis module, a foundation pit abnormal state intelligent prediction module, a foundation pit monitoring early warning decision module, and a foundation pit data storage and backup module.

[0023] The foundation pit data acquisition module is used to collect physical parameter data such as displacement, stress, settlement, cracks, etc. from sensors of the foundation pit and its surrounding environment in real time, and transmit the data to the processing center through a communication interface.

[0024] The foundation pit digital model simulation module is used to construct a three-dimensional digital model of the foundation pit and synchronize it with the real-time collected data to simulate the actual state and changes of the foundation pit in a virtual environment.

[0025] The foundation pit multi-source data processing and analysis module is used to integrate and fuse information from different sensors and data sources, extract key features, and analyze the changing trend of the foundation pit state to provide data support for subsequent abnormal state prediction.

[0026] The intelligent prediction module for abnormal conditions of foundation pits is used to learn historical data using deep learning algorithms, establish prediction models, identify abnormal patterns, and assess the risks that may be caused by abnormal conditions, providing a scientific basis for early warning decisions.

[0027] The foundation pit monitoring and early warning decision-making module is responsible for determining whether an early warning needs to be triggered based on the results of the intelligent prediction module, and issuing early warning information to relevant personnel through various communication channels, while providing emergency response plans and guiding on-site personnel to take corresponding measures.

[0028] The foundation pit data storage and backup module is responsible for storing all collected data, analysis results and warning records in the database, and regularly backing up data to ensure the security, integrity and traceability of information.

[0029] The foundation pit data acquisition module includes a sensor integration submodule, a data acquisition device submodule and a communication interface submodule.

[0030] The sensor integration submodule is responsible for integrating and deploying various sensors, such as displacement sensors, stress sensors, and vibration sensors, to collect data from the foundation pit and its surrounding environment in real time. The data acquisition submodule is responsible for collecting data from sensors and performing preliminary processing. The communication interface submodule is responsible for transmitting the collected data to the foundation pit digital model simulation module and the foundation pit multi-source data processing and analysis module via wireless or wired networks.

[0031] The foundation pit digital model simulation module includes a virtual model construction submodule and a model synchronization submodule. The virtual model construction submodule is responsible for building a three-dimensional digital model of the foundation pit, including structural and geological information. The model synchronization submodule is used to synchronize the real-time collected data with the digital model to reflect the current status and changes of the foundation pit.

[0032] The foundation pit multi-source data processing and analysis module includes a data fusion submodule, a feature extraction submodule and a trend analysis submodule.

[0033] The data fusion submodule integrates information from different sensors and data sources to perform multi-source data fusion. The feature extraction submodule is responsible for extracting key features from the fused data for subsequent analysis and prediction. The trend analysis submodule is used to analyze the changing trend of the foundation pit status and provide a basis for prediction.

[0034] The intelligent prediction module for abnormal state of foundation pit includes an intelligent prediction submodule and a risk assessment submodule.

[0035] The intelligent prediction submodule uses a deep learning algorithm to learn historical data and establish a prediction model. The risk assessment submodule assesses the risks that may be caused by abnormal conditions and provides a basis for decision-making.

[0036] The foundation pit monitoring and early warning decision module includes an early warning trigger submodule, an information release submodule and an emergency response submodule. The early warning trigger submodule determines whether to trigger an early warning based on the prediction and evaluation results. The information release submodule is responsible for releasing the early warning information to relevant personnel through various channels, such as text messages and APP push. The emergency response submodule is responsible for providing an emergency response plan and guiding on-site personnel to take corresponding measures.

[0037] The foundation pit data storage and backup module includes a data storage submodule and a data backup submodule.

[0038] The data storage submodule is responsible for storing the collected data, analysis results, warning records and other information in the database. The data backup submodule is used to regularly back up the database to ensure the security and integrity of the data. Example 2

[0039] The present invention discloses a foundation pit digital twin monitoring method, which is implemented by the system of Example 1 and specifically includes the following steps: Step S1, foundation pit data collection: Use a variety of equipment to obtain key data of the foundation pit, including various sensors and high-definition cameras. The types of sensors include displacement sensors, stress sensors, pressure sensors, tilt sensors, water level sensors and temperature sensors. The sensors are strategically deployed at various key locations of the foundation pit. Their main responsibility is to create accurate three-dimensional models for digital twins and continuously monitor the state changes of the foundation pit. The high-definition cameras are installed in key areas of the foundation pit to capture visual evidence such as wall crack development and structural deformation to enhance the comprehensiveness and accuracy of the monitoring system.

[0040] Step S2, simulation modeling of the digital twin model of the foundation pit: using BIM (Building Information Modeling) technology and GIS (Geographic Information System) data, based on the geographic information and engineering drawings of the foundation pit, and the basic information collected by the sensor, a three-dimensional digital twin model of the foundation pit is established. The collected sensor data is associated with the digital twin model of the foundation pit to achieve dynamic update of the foundation pit indicators to reflect the actual status of the foundation pit.

[0041] Step S3, pit data processing: A series of standardization processes are performed on the sensor data and camera images collected in step S1 to improve the availability and consistency of the data. The processing flow includes data normalization, data mapping, and filling. The following is a detailed description of the data processing flow using a key area of ​​the pit as an example, and the remaining areas will follow the same steps.

[0042] Step S31, image data processing: For the image I captured by the camera, first convert it to RGB color mode and crop it as needed. The data dimension after cropping is [W,H,3], where the first dimension size is W, representing the length of the image, the second dimension size is H, representing the height of the image, and the third dimension size is 3, representing the number of channels. The number of channels of image data in RGB format is 3.

[0043] Step S32, sensor data processing: Due to the limited number of sensors, for areas not covered by sensors, the averaging method is used to fill the blank points with data. The processed sensor is denoted as S, and the data dimension is [W, H, 1], where the first dimension size is W, the second dimension size is H, and the third dimension size is 1, which is used to keep consistent with the format of the image data for subsequent analysis and model training.

[0044] Step S33, processing of the three-dimensional digital twin model of the foundation pit: mapping the three-dimensional model data D into two-dimensional data to facilitate subsequent training. The steps and formulas are as follows: Step S331: Calculate the covariance matrix C of the three-dimensional data D:

[0045] Where n represents the number of three-dimensional data, D' represents the average value of D, and T represents the transposition operation.

[0046] Step S332: Calculate eigenvalues ​​and eigenvectors:

[0047] Among them, C is the covariance matrix, v is the eigenvector, and α is the eigenvalue.

[0048] Step S333: Mapping the three-dimensional data into two-dimensional data:

[0049] Q represents the digital twin model data of the foundation pit after being mapped into two-dimensional data. It means taking the first k columns of the feature vector to achieve dimensionality reduction.

[0050] Step S34: standardize all data and scale the data to a specific range. The standardization formula used in the present invention is as follows:

[0051] Among them, x refers to the original data, μ is the mean value of the data, σ is the standard deviation of the data, and x' is the standardized data. The sensor data S, image data I, and two-dimensional foundation pit digital twin model data Q are standardized respectively.

[0052] Step S4, extraction of multi-source historical features of foundation pits: implement a continuous data collection strategy, integrate and record time series data from sensors, cameras, and three-dimensional models, ensure the continuity and integrity of the data in the time dimension, and use it to monitor the changing trend of the foundation pit status.

[0053] Step S41: For sensor data S in the past period of time, record it as S*=[S1, S2, ..., St-1], where the subscripts 1, 2, ..., t-1 represent a certain time node in the past.

[0054] Step S42: For the image data I within a period of time in the past, record it as I*=[I1, I2, ..., It-1], where the subscripts 1, 2, ..., t-1 represent a certain time node in the past.

[0055] Step S43: For the two-dimensional foundation pit digital twin model data Q within the past period of time, record it as Q*=[Q1,Q2,...,Qt-1], where the subscripts 1,2,...,t-1 represent a certain time node in the past.

[0056] Step S44: The multi-source data at time t are integrated into Xt, Xt=[St,It,Qt], where St represents the sensor data at time t, It represents the image data at time t, and Qt represents the two-dimensional foundation pit digital twin model data at time t.

[0057] Step S5, modeling of foundation pit state change trend: involves using a deep learning algorithm in combination with historical data and multi-source data to model the foundation pit state trend and predict the future state of the foundation pit for abnormal state assessment.

[0058] Step S51, input and output processing: predict the future state Ht of the foundation pit based on multi-source historical data X*=[X1, X2, ..., Xt-1], where X* represents multi-source historical data, subscripts 1, 2, ..., t represent a certain time node in the past, and Xt-1 represents multi-source data at time t-1.

[0059] Step S52, extracting features from multi-source data, includes the following steps: Feature extraction is performed on multi-source data at each time. For multi-source data Xi at time i, the formula is as follows:

[0060] in is the state at the previous time i-1, the initial state is a zero matrix, To combine the multi-source data Xi and the state of the previous time The result after connection.

[0061]

[0062] Where We represents the convolution kernel, the size is e, and the above formula is processed After convolution processing, it is recorded as .

[0063]

[0064] Among them, sigmoid() is the activation function, is the parameter matrix, is the bias vector obtained by modeling learning, Ri represents the recursive state at time i, Gi represents the input state at time i, Represents the weight at time i, which is calculated by the recursive state Ri, and function() represents the calculation function.

[0065]

[0066] in is the state at time i, is the state at the previous time i-1, represents the multiplication of elements. Other variables are calculated by the above formula and will not be repeated here.

[0067] Step S53, historical feature extraction: For the future state Ht of the foundation pit, the feature extraction of each time point data in the multi-source historical data X*=[X1,X2,...,Xt-1] is performed one by one in step S52, so as to obtain the future state Ht of the foundation pit at the final time t.

[0068] Multi-source historical data feature extraction can analyze historical data and current monitoring data, identify potential risk patterns for in-depth analysis, and the evaluation model can predict changes in the state of the foundation pit in the future.

[0069] Step S6, risk assessment of abnormal state of foundation pit: Use classification algorithm to classify and evaluate the predicted future state of foundation pit. Once the prediction result shows that the future state of foundation pit may exceed the safety threshold, the system will automatically trigger the early warning mechanism. The assessment result will be used to update the risk management strategy so as to take appropriate preventive measures to reduce potential engineering risks.

[0070] Step S7, foundation pit monitoring and early warning: When the foundation pit abnormal state risk assessment step detects an abnormal situation, that is, the state or predicted state of the foundation pit exceeds the preset safety threshold, the foundation pit monitoring and early warning system will be activated immediately. The early warning information will be promptly conveyed to project managers and on-site engineers through multiple channels, including but not limited to mobile application (APP) push notifications, text messages, emails and phone calls. Ensure that when the foundation pit is in an abnormal state, relevant personnel can quickly take measures, such as suspending construction, evacuating personnel or conducting on-site inspections, to prevent potential safety accidents and project losses.

[0071] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. A foundation pit digital twin monitoring system, characterized by: The system includes: a foundation pit data acquisition module, a foundation pit digital model simulation module, a foundation pit multi-source data processing and analysis module, a foundation pit abnormal state intelligent prediction module, a foundation pit monitoring and early warning decision module, and a foundation pit data storage and backup module; The foundation pit data acquisition module is used to collect displacement, stress, settlement, crack physical parameter data from sensors of the foundation pit and its surrounding environment in real time, and transmit the data to the processing center through a communication interface; The foundation pit digital model simulation module is used to construct a three-dimensional digital model of the foundation pit and synchronize it with the real-time collected data to simulate the actual state and changes of the foundation pit in a virtual environment; The foundation pit multi-source data processing and analysis module is used to integrate and fuse information from different sensors and data sources, extract key features, and analyze the changing trend of the foundation pit state to provide data support for subsequent abnormal state prediction; The intelligent prediction module for abnormal state of foundation pit is used to learn historical data by using deep learning algorithm, establish prediction model, identify abnormal patterns, and evaluate the risks that may be caused by abnormal state, so as to provide scientific basis for early warning decision-making; it includes intelligent prediction submodule and risk assessment submodule; the intelligent prediction submodule uses deep learning algorithm to learn historical data and establish prediction model; the risk assessment submodule evaluates the risks that may be caused by abnormal state and provides basis for decision-making; The foundation pit monitoring and early warning decision module is responsible for determining whether an early warning needs to be triggered based on the results of the intelligent prediction module, and issuing early warning information to relevant personnel through various communication channels, while providing emergency response plans and guiding on-site personnel to take corresponding measures; The foundation pit data storage and backup module is responsible for storing all collected data, analysis results and warning records in the database, and regularly backing up data to ensure the security, integrity and traceability of information.

2. A foundation pit digital twin monitoring system according to claim 1, characterized in that: The foundation pit data acquisition module includes a sensor integration submodule, a data acquisition submodule and a communication interface submodule; The sensor integration submodule is responsible for integrating and deploying various sensors, such as displacement sensors, stress sensors, and vibration sensors, for collecting data on the foundation pit and its surrounding environment in real time; The data collector submodule is responsible for collecting data from sensors and performing preliminary processing; The communication interface submodule is responsible for transmitting the collected data to the foundation pit digital model simulation module and the foundation pit multi-source data processing and analysis module through a wireless or wired network.

3. According to claim 1, a foundation pit digital twin monitoring system is characterized in that: The foundation pit digital model simulation module includes a virtual model building submodule and a model synchronization submodule; The virtual model building submodule is responsible for building a three-dimensional digital model of the foundation pit, including structural and geological information; The model synchronization submodule is used to synchronize the real-time collected data with the digital model to reflect the current status and changes of the foundation pit.

4. According to claim 1, a foundation pit digital twin monitoring system is characterized in that: The foundation pit multi-source data processing and analysis module includes a data fusion submodule, a feature extraction submodule and a trend analysis submodule; The data fusion submodule integrates information from different sensors and data sources to perform multi-source data fusion; The feature extraction submodule is responsible for extracting key features from the fused data for subsequent analysis and prediction; The trend analysis submodule is used to analyze the changing trend of the foundation pit status and provide a basis for prediction.

5. The foundation pit digital twin monitoring system according to claim 1, characterized in that: The foundation pit monitoring early warning decision module includes an early warning trigger submodule, an information release submodule and an emergency response submodule; The warning triggering submodule determines whether to trigger a warning based on the prediction and evaluation results; The information release submodule is responsible for releasing the warning information to relevant personnel through various channels, such as SMS and APP push; The emergency response submodule is responsible for providing emergency response plans and guiding on-site personnel to take corresponding measures.

6. A foundation pit digital twin monitoring system according to claim 1, characterized in that: The foundation pit data storage and backup module includes a data storage submodule and a data backup submodule; The data storage submodule is responsible for storing the collected data, analysis results, and warning record information in the database; The data backup submodule is used to regularly back up the database to ensure the security and integrity of the data.

7. A digital twin monitoring method for a foundation pit, the method being implemented by the digital twin monitoring system for a foundation pit according to any one of claims 1 to 6, characterized in that: The following steps are involved: Step S1, foundation pit data collection: Use a variety of equipment to obtain key data of the foundation pit, the equipment includes sensors and high-definition cameras. The types of sensors include displacement sensors, stress sensors, pressure sensors, tilt sensors, water level sensors and temperature sensors. The sensors are strategically deployed at various key locations of the foundation pit. Their main responsibility is to create an accurate three-dimensional model for the digital twin and continuously monitor the state changes of the foundation pit; the high-definition cameras are installed in key areas of the foundation pit to capture visual evidence of wall crack development and structural deformation to enhance the comprehensiveness and accuracy of the monitoring system; Step S2, simulation modeling of the digital twin model of the foundation pit: using BIM technology and GIS data, according to the geographic information and engineering drawings of the foundation pit, and the basic information collected by the sensor, a three-dimensional digital twin model of the foundation pit is established; the collected sensor data is associated with the digital twin model of the foundation pit to realize the dynamic update of the foundation pit indicators to reflect the actual status of the foundation pit; Step S3, foundation pit data processing: a series of standardized processing is performed on the sensor data and camera images collected in step S1, aiming to improve the availability and consistency of the data; the processing flow includes data normalization, data mapping and filling; Step S4, extracting multi-source historical features of foundation pits: implementing a continuous data collection strategy, integrating and recording time series data from sensors, cameras, and three-dimensional models, ensuring the continuity and integrity of the data in the time dimension, and used to monitor the changing trend of foundation pit status; Step S5, modeling the trend of foundation pit state changes: involving deep learning algorithms combined with historical data and multi-source data to model the foundation pit state trend and predict the future state of the foundation pit for abnormal state assessment; Step S6, risk assessment of abnormal state of foundation pit: use classification algorithm to classify and assess the predicted future state of foundation pit. Once the prediction result shows that the future state of foundation pit is likely to exceed the safety threshold, the system will automatically trigger the early warning mechanism. The assessment result will be used to update the risk management strategy so as to take appropriate preventive measures to reduce potential engineering risks. Step S7, foundation pit monitoring and early warning: When the foundation pit abnormal state risk assessment step detects an abnormal situation, that is, the state or predicted state of the foundation pit exceeds the preset safety threshold, the foundation pit monitoring and early warning system will be activated immediately; the early warning information will be promptly conveyed to project managers and on-site engineers through multiple channels to ensure that when the foundation pit is in an abnormal state, relevant personnel can take prompt measures to prevent potential safety accidents and engineering losses.

8. A foundation pit digital twin monitoring method according to claim 7, characterized in that: The step S3 specifically includes: Step S31, image data processing: for the image I captured by the camera, first convert it into RGB color mode and crop it as needed. The data dimension after cropping is [W,H,3], where the first dimension size is W, representing the length of the image, the second dimension size is H, representing the height of the image, and the third dimension size is 3, representing the number of channels. The number of channels of the image data in RGB format is 3; Step S32, sensor data processing: Due to the limited number of sensors, for areas not covered by sensors, the averaging method is used to fill the blank points with data; the processed sensor is recorded as S, and the data dimension is [W, H, 1], where the first dimension size is W, the second dimension size is H, and the third dimension size is 1, which is used to keep consistent with the format of the image data, so as to facilitate subsequent analysis and model training; Step S33, processing of the three-dimensional digital twin model of the foundation pit: mapping the three-dimensional model data D into two-dimensional data to facilitate subsequent training. The steps and formulas are as follows: Step S331: Calculate the covariance matrix C of the three-dimensional data D: ； Where n represents the number of three-dimensional data, D' represents the average value of D, and T represents the transposition operation; Step S332: Calculate eigenvalues ​​and eigenvectors: ； Where C is the covariance matrix, v is the eigenvector, and α is the eigenvalue; Step S333: Mapping the three-dimensional data into two-dimensional data: ； Q represents the digital twin model data of the foundation pit after being mapped into two-dimensional data. It represents taking the first k columns of the feature vector to achieve dimensionality reduction; Step S34: Standardize all data and scale the data to a specific range using the following standardization formula: ； Among them, x refers to the original data, μ is the mean value of the data, σ is the standard deviation of the data, and x' is the standardized data; the sensor data S, image data I, and two-dimensional foundation pit digital twin model data Q are standardized respectively; The step S4 specifically includes: Step S41: for sensor data S in the past period of time, record it as S*=[S1, S2, ..., St-1], where the subscripts 1, 2, ..., t-1 represent a certain time node in the past; Step S42: for image data I within a period of time in the past, record it as I*=[I1,I2,...,It-1], where the subscripts 1,2,...,t-1 represent a certain time point in the past; Step S43: For the two-dimensional foundation pit digital twin model data Q within a past period of time, record it as Q*=[Q1,Q2,...,Qt-1], where the subscripts 1,2,...,t-1 represent a certain time node in the past; Step S44: The multi-source data at time t are integrated into Xt, Xt=[St,It,Qt], where St represents the sensor data at time t, It represents the image data at time t, and Qt represents the two-dimensional foundation pit digital twin model data at time t.

9. A foundation pit digital twin monitoring method according to claim 8, characterized in that: The step S5 specifically includes: Step S51, input and output processing: predicting the future state Ht of the foundation pit based on multi-source historical data X*=[X1, X2, ..., Xt-1], where X* represents multi-source historical data, subscripts 1, 2, ..., t represent a certain time node in the past, and Xt-1 represents multi-source data at time t-1; Step S52, extracting features from multi-source data, includes the following steps: Feature extraction is performed on multi-source data at each time. For multi-source data Xi at time i, the formula is as follows: ； in is the state at the previous time i-1, the initial state is a zero matrix, To combine the multi-source data Xi and the state of the previous time The result after connection; ； Where We represents the convolution kernel, the size is e, and the above formula is processed After convolution processing, it is recorded as ; ； Among them, sigmoid() is the activation function, is the parameter matrix, is the bias vector obtained by modeling learning, Ri represents the recursive state at time i, Gi represents the input state at time i, Represents the weight at time i, which is calculated by the recursive state Ri, and function() represents the calculation function; ； in is the state at time i, is the state at the previous time i-1, represents the multiplication of elements, and other variables are calculated by the above formula and will not be repeated here; Step S53, historical feature extraction: For the future state Ht of the foundation pit, the feature extraction of each time point data in the multi-source historical data X*=[X1,X2,...,Xt-1] is performed one by one in step S52, so as to obtain the future state Ht of the foundation pit at the final time t.

10. An electronic device, characterized in that: It comprises a processor and a memory; the memory stores computer-readable instructions, and the processor is used to run the computer-readable instructions, wherein the digital twin monitoring method for foundation pits according to claim 7 is executed when the computer-readable instructions are run.

Images