Deep foundation pit excavation deformation real-time monitoring system based on relative measurement

Abstract

The invention discloses a deep foundation pit excavation deformation real-time monitoring system based on relative measurement. The deep foundation pit excavation deformation real-time monitoring system comprises a relative measurement sensor group, a data acquisition and preprocessing module, a wireless data transmission module, a cloud data processing and analysis module, an intelligent risk assessment module, a visual alarm module and a solar power supply and energy storage management module. The relative measurement sensor group collects deformation data of the deep foundation pit, the data acquisition and preprocessing module performs primary processing on the data, the wireless data transmission module stably transmits the data to the cloud, the cloud data processing and analysis module performs deformation prediction, and the intelligent risk assessment module performs risk grading. The visual alarm module gives an alarm in real time and carries out emergency notification according to different risk levels, and the solar power supply and energy storage management module guarantees continuous operation of the system. The system solves the problems of measurement precision, data transmission, risk assessment, power supply stability and the like of an existing monitoring system, provides powerful support for safe construction of the deep foundation pit, and has remarkable benefits.

Description

Technical Field

[0001] The invention relates to the technical field of geotechnical engineering investigation for power tunnels, and in particular to a real-time monitoring system for deep foundation pit excavation deformation based on relative measurement. Background Art

[0002] In the construction of power tunnel projects, deep foundation pit excavation is a critical and challenging task. With the increase in the depth of the foundation pit and the increasing complexity of the surrounding environment, deformation monitoring during the foundation pit excavation process is crucial to ensure construction safety and project quality. Traditional deformation monitoring methods mainly rely on manual regular measurements, which have many disadvantages: data acquisition is not timely, and it is difficult to reflect the dynamics of foundation pit deformation in real time; measurement accuracy is easily affected by human factors and has large errors; it is impossible to effectively predict the deformation trend and it is difficult to take preventive measures in advance.

[0003] Although the existing real-time monitoring system for deep foundation pit excavation deformation based on relative measurement has developed to a certain extent, there are still some problems that need to be solved. In terms of measurement accuracy, it is difficult for the existing system to effectively reduce the cumulative error caused by absolute measurement, resulting in inaccurate measurement results. During data transmission, stability is difficult to guarantee and it is easily affected by external environmental factors, resulting in data loss or transmission interruption. For the risk assessment of deep foundation pit deformation, the existing systems are mostly based on a single evaluation indicator or a simple evaluation model, which cannot comprehensively and accurately consider the combined effects of multiple factors, resulting in unreliable risk assessment results.

[0004] In addition, the existing system also has deficiencies in power supply stability, and cannot ensure continuous and stable power support for monitoring equipment during long-term monitoring, especially under severe weather conditions. These problems limit the effective application of existing monitoring systems in the field of geotechnical engineering surveys for power tunnels. A more accurate, stable, comprehensive and self-sufficient deep foundation pit excavation deformation real-time monitoring system is urgently needed to meet engineering needs. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings existing in the prior art and to propose a real-time monitoring system for deep foundation pit excavation deformation based on relative measurement.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions: The real-time monitoring system for deep foundation pit excavation deformation based on relative measurement includes: relative measurement sensor group, data acquisition and preprocessing module, wireless data transmission module, cloud data processing and analysis module, intelligent risk assessment module, visual alarm module and solar power supply and energy storage management module; The relative measurement sensor group consists of multiple distributed optical fiber sensors and high-precision inclinometers. The distributed optical fiber sensors are arranged in a grid pattern along the retaining structure of the deep foundation pit and the soil at the bottom of the pit, and are used to obtain real-time information on the relative displacement changes inside the retaining structure and the soil. The high-precision inclinometers are installed at key parts of the retaining structure and on the ground around the foundation pit, and are used to monitor the changes in the inclination angles of the structure and the ground. The measurement accuracy is improved and the absolute measurement error is reduced through the relative measurement method; The data acquisition and preprocessing module is connected to the relative measurement sensor group, and is used to synchronously collect sensor data and perform preprocessing such as filtering, noise reduction, and linear correction on the collected raw data to initially improve the data quality; The wireless data transmission module is connected to the data acquisition and preprocessing module using multi-hop relay communication technology, and stably transmits the preprocessed data to the cloud data processing and analysis module. When the signal is interfered, it automatically switches the communication frequency band to ensure the continuity and integrity of data transmission; The cloud data processing and analysis module receives the data sent by the wireless data transmission module, and uses a deformation feature extraction algorithm based on deep learning and a finite element analysis model to deeply analyze the deformation data during the excavation of the deep foundation pit, establish a multi-physical field coupling model including the stress and deformation of the retaining structure, the flow change of the soil, and the influence of the groundwater level, realize high-precision real-time prediction of the deformation trend of the foundation pit, and compare it with the preset dynamic safety threshold; Based on the prediction results of the cloud data processing and analysis module, the intelligent risk assessment module combines the surrounding environment information of the foundation pit (including surrounding buildings, underground pipelines, traffic flow, etc.) and historical deformation data, and uses the fuzzy comprehensive evaluation method to dynamically classify and evaluate the deformation risk of the deep foundation pit. The evaluation results are divided into four levels: low risk, medium risk, high risk, and extremely high risk; The visualization alarm module is connected to the intelligent risk assessment module, and performs real-time alarms in different visualization forms on the local monitoring screen and the remote mobile terminal according to the risk level (low risk is displayed in green and stable, medium risk is displayed in yellow and flashing, high risk is displayed in orange and flashing quickly, extremely high risk is displayed in red and flashing explosively), and automatically sends an emergency text message notification with accurate location information to the preset emergency rescue personnel in the case of extremely high risk; The solar power supply and energy storage management module provides power support for the entire monitoring system, including high-efficiency solar panels, intelligent energy storage battery packs, and an energy management system. The solar panels convert solar energy into electrical energy and store it in the energy storage battery packs. The energy management system automatically optimizes the power distribution according to the power consumption requirements and real-time power conditions of each module, ensuring that the system can continue to operate stably for no less than X hours under non-illuminated conditions such as rainy days and nights. At the same time, it has overcharge and over-discharge protection functions and real-time power monitoring functions, and transmits the power information to the cloud data processing and analysis module through wireless signals for recording and management. When the power is lower than the preset threshold, the system automatically sends a charging reminder message to the maintenance personnel.

[0007] As a further technical solution of the present invention, the distributed optical fiber sensor in the relative measurement sensor group obtains the relative displacement change information inside the enclosure structure and the soil body in real time, specifically including: Based on the optical fiber Brillouin scattering principle, the distributed optical fiber sensor uses the relationship between Brillouin frequency shift and optical fiber strain to obtain the relative displacement change information inside the enclosure structure and the soil body. The specific relational formula is:

[0008] In the formula: is the Brillouin frequency shift at temperature T and strain S; is the initial temperature , and the initial strain of the Brillouin frequency shift; Through the change of the Brillouin frequency shift , the optical fiber strain can be calculated:

[0009] In the formula: is the change amount of the Brillouin frequency shift; is the change amount of the Brillouin frequency shift caused by the temperature change; is the temperature coefficient; is the initial Brillouin frequency shift.

[0010] As a further technical solution of the present invention, the high-precision inclinometer in the relative measurement sensor group improves the measurement accuracy of the inclination angle change between the structure and the ground through relative measurement and reduces the absolute measurement error, specifically including: Before the foundation pit excavation, the high-precision inclinometer is calibrated for reference, and the initial inclination angle value is recorded. During the subsequent monitoring process, the current inclination angle value is collected in real time, and the inclination angle change amount is calculated through the following formula :

[0011] To reduce the absolute measurement error, the following algorithm is adopted to correct the zero drift and temperature drift error of the sensor:

[0012] In the formula: is the corrected tilt angle; is the tilt angle of the original measurement; is the correction value of the zero drift and temperature drift error related to temperature, and its calculation formula is:

[0013] In the formula: is the temperature coefficient; is the constant term; is the current temperature.

[0014] As a further technical solution of the present invention, the specific steps of the cloud data processing and analysis module using a deformation feature extraction algorithm based on deep learning and a finite element analysis model to deeply analyze the deformation data during the deep foundation pit excavation process include: S1: Deformation feature extraction algorithm based on deep learning: Use a convolutional neural network (CNN) to extract spatial features from the monitoring data of multi-dimensional time series. The multi-dimensional time series Xt,n composed of the standardized historical data of multiple monitoring points is used as the input data. Xt,n is a two-dimensional matrix, where one row represents the monitoring values of n monitoring points at a certain moment, and one column represents the monitoring values of a certain monitoring point within the time step t. After inputting this sequence into the CNN part, convolutional operations are performed to extract spatial features, and multiple one-dimensional sequences composed of spatial features are output; then use a long short-term memory network (LSTM) to perform time series learning and regression prediction on the spatial feature sequence extracted by the CNN, output multiple predicted values of spatial features, and finally integrate the predicted values of spatial features through a fully connected layer to output the predicted values of the monitoring data; S2: Finite element analysis model: Establish a multi-physical field coupling model including the stress and deformation of the retaining structure, the flow change of the soil mass, and the influence of the groundwater level; S3: Real-time prediction: Combine the deformation feature data extracted by the deep learning algorithm with the finite element analysis model, and through data fusion and numerical simulation methods, achieve high-precision real-time prediction of the foundation pit deformation trend; specifically, use the deformation feature values predicted by the deep learning algorithm as the boundary conditions or initial conditions of the finite element analysis model, and obtain the predicted deformation values of the foundation pit at the next moment through numerical solution; S4: Compare with the dynamic safety threshold: The preset dynamic safety threshold is determined according to the design parameters of the foundation pit, the surrounding environmental conditions, as well as relevant engineering experience and technical specifications; during the real-time prediction process, the predicted foundation pit deformation value is compared with the preset dynamic safety threshold . When the predicted deformation value exceeds the safety threshold, the alarm mechanism is triggered to timely remind the construction personnel to take corresponding measures.

[0015] As a further technical solution of the present invention, in the above S1, a convolutional neural network (CNN) is used to extract spatial features from the multi-dimensional time series monitoring data, and its convolution operation formula is:

[0016] In the formula: is the output feature map after convolution operation; is the input multi-dimensional time series monitoring data matrix; is the convolution kernel weight matrix, with the size of , where is the convolution kernel size; is the convolution kernel bias vector; is the activation function, using the ReLU function, ; The long short-term memory network (LSTM) is used to perform time series learning and regression prediction on the spatial feature sequence extracted by the CNN. The unit state update formula of the LSTM is:

[0017] In the formula: is the unit state at the current time ; is the output of the forget gate, which controls how much information of the previous unit state is forgotten. The calculation formula is: .

[0018] is the unit state of the previous moment; is the output of the input gate, which controls how much new information is stored in the unit state at the current time. The calculation formula is: .

[0019] is the candidate unit state, and the calculation formula is ; are the weight matrices of the forget gate, input gate and candidate unit state respectively; Bias vectors for the forget gate, input gate, and candidate cell state, respectively; Denote the hidden state at the previous moment at time t And the input at the current moment Perform a concatenation operation; Is the sigmoid activation function, ; Tanh is the hyperbolic tangent function, ; Is an element-wise multiplication operation; Finally, the spatial feature prediction value is integrated through a fully connected layer to output the prediction value of the monitoring data. The calculation formula of the fully connected layer is:

[0020] In the formula: Is the prediction value of the monitoring data; Is the weight matrix of the fully connected layer; Is the hidden state vector output by the LSTM; Is the bias vector of the fully connected layer.

[0021] As a further technical solution of the present invention, in the S2, the multi-physical field coupling model includes the finite element equation of the stress and deformation of the enclosure structure, the finite element equation of the soil flow change, and the finite element equation of the influence of the groundwater level; The finite element equation of the stress and deformation of the enclosure structure is: In the formula: Is the stiffness matrix, Is the nodal displacement vector, Is the nodal force vector; The finite element equation of the soil flow change is: In the formula: Is the seepage velocity vector, Is the permeability coefficient tensor of the soil, Is the dynamic viscosity coefficient of the fluid, Is the pressure of the fluid; The finite element equation of the influence of the groundwater level is: In the formula: Is the head height, Is the permeability coefficient tensor.

[0022] As a further technical solution of the present invention, in the S3, the prediction formula is:

[0023] In the formula: is the node displacement vector at the next moment; is the node displacement vector at the current moment; is the time step; is the residual vector based on the current node displacement and the prediction value of deep learning, and is obtained by iterative solution of the finite element equation.

[0024] As a further technical solution of the present invention, in S4, the mathematical expression of the preset dynamic safety threshold is:

[0025] In the formula: is the maximum allowable displacement of the retaining structure at the position ; is the allowable displacement at the design stage of the retaining structure; is the displacement safety factor based on the surrounding environment and engineering experience; is the maximum allowable stress of the soil mass at the position ; is the design allowable stress of the soil mass; is the soil stress safety factor based on the surrounding environment and engineering experience; The predicted foundation pit deformation value and the preset dynamic safety threshold The comparison formula is: Alarm = In the formula: Alarm is the alarm signal, 1 indicates triggering an alarm, and 0 indicates normal; is the predicted displacement of the retaining structure at the position ; is the predicted stress of the soil mass at the position ;

[0026] As a further technical solution of the present invention, the intelligent risk assessment module uses the fuzzy comprehensive evaluation method to dynamically classify and evaluate the deformation risk of the deep foundation pit, specifically including: Sa: Construct an evaluation index system: The factors related to the deformation risk of the deep foundation pit are used as evaluation indexes, including the horizontal displacement and vertical settlement of the retaining structure, the horizontal displacement and vertical settlement of the soil mass, the change of the groundwater level, and the surrounding environment factors (such as the distance of surrounding buildings, the distribution of underground pipelines, etc.). Let the evaluation index set be , where represents the th evaluation index; Sb: Determine the fuzzy evaluation level: Set the fuzzy evaluation level set as , which respectively represent four levels of low risk, medium risk, high risk and extremely high risk; Sc: Construct a fuzzy membership function: For each evaluation index Construct a fuzzy membership function based on its actual measured value or predicted value to determine its membership degree for each evaluation level; Sd: Establish a fuzzy evaluation matrix: Based on the membership degrees of each evaluation index for each evaluation level, establish a fuzzy evaluation matrix , which is composed of single-factor evaluation sets as rows of the matrix. Each single-factor evaluation set contains the membership degrees of this evaluation index for each evaluation level, that is , where represents the membership degree of the th evaluation index for the th evaluation level; Se: Determine the weight vector: Determine the weight vector of each evaluation index through the analytic hierarchy process , where represents the weight of the th evaluation index and satisfies ; Sf: Conduct fuzzy comprehensive evaluation: Use fuzzy operations to perform weighted summation on the evaluation indexes to obtain the fuzzy comprehensive evaluation result, and adopt the weighted average model:

[0027] In the formula: is the fuzzy comprehensive evaluation result vector, representing the membership degrees of the deformation risk of the deep foundation pit at each evaluation level; Sg: Defuzzification: In order to obtain a specific evaluation result, defuzzify the fuzzy comprehensive evaluation result, and adopt the maximum membership degree method: Risk Level That is, select the evaluation level with the largest membership degree as the final deformation risk level of the deep foundation pit.

[0028] As a further technical solution of the present invention, in the Sc, for the horizontal displacement of the retaining structure , use a triangular membership function to determine its membership degrees for low risk , medium risk , high risk and extremely high risk :

[0029] In the formula: is a measure of belonging to the evaluation level , and the value range is between ; For example, , it means that when the horizontal displacement of the retaining structure is 5 mm, there is The possibility belongs to the low - risk level; is a parameter of the triangular membership function, which is determined according to engineering experience and relevant specifications. For example, for low - risk , it can be taken as (unit: mm), which means that when the horizontal displacement is less than 0 mm, the membership degree to low - risk is 0; when the horizontal displacement is between 0 mm and 5 mm, the membership degree increases linearly; when the horizontal displacement is between 5 mm and 10 mm, the membership degree decreases linearly; when the horizontal displacement is greater than 10 mm, the membership degree to low - risk is 0; is the evaluation index The actual measured value or predicted value of, for example, the horizontal displacement of the retaining structure is 5 mm.

[0030] The beneficial effects of the present invention are as follows: 1. Improvement in measurement accuracy: Through the innovative design of the relative measurement sensor group, including the combination of distributed optical fiber sensors and high - precision inclinometers, the absolute measurement error is effectively reduced, greatly improving the accuracy of deep foundation pit deformation monitoring and providing a more accurate data basis for subsequent risk assessment.

[0031] 2. Enhancement of the stability of data transmission and processing: The wireless data transmission module is connected to the data acquisition and pre - processing module using multi - hop relay communication technology, and can automatically switch the communication frequency band when the signal is interfered, ensuring the continuity and integrity of data transmission; the cloud data processing and analysis module realizes high - precision real - time prediction of the deformation trend of the foundation pit, solves the problems of easy interruption of data transmission and incomplete risk assessment in the existing monitoring system, and improves the stability and reliability of the monitoring system.

[0032] 3. Improvement in the scientificity and reliability of risk assessment: The intelligent risk assessment module uses the fuzzy comprehensive evaluation method to dynamically classify and evaluate the deformation risk of the deep foundation pit. Through a series of scientific evaluation steps, the risk assessment results are more comprehensive and accurate, can timely and accurately reflect the deformation risk status of the deep foundation pit, and provide a reliable decision - making basis for construction personnel.

[0033] 4. Guarantee of power supply stability: The solar power supply and energy storage management module provides power support for the entire monitoring system, solves the problem of poor power supply stability in the existing monitoring system, ensures the long - term stable operation of the monitoring system, and reduces the risk of missing monitoring data caused by power supply interruption. Description of the Drawings

[0034] Figure 1 is the system diagram of the real - time monitoring system for deep foundation pit excavation deformation based on relative measurement proposed by the present invention. Specific Embodiments

[0035] To make the technical means, creative features, achieved objectives and effects of the present invention easy to understand, the present invention will be further described below in conjunction with specific embodiments.

[0036] Please refer to the attached Figure 1 , a real-time monitoring system for the deformation of deep foundation pit excavation based on relative measurement, including: a relative measurement sensor group, a data acquisition and preprocessing module, a wireless data transmission module, a cloud data processing and analysis module, an intelligent risk assessment module, a visual alarm module, and a solar power supply and energy storage management module;

[0037] The relative measurement sensor group consists of multiple distributed optical fiber sensors and high-precision inclination sensors. The distributed optical fiber sensors are arranged in a grid pattern along the retaining structure and the soil at the bottom of the deep foundation pit to obtain real-time information on the relative displacement changes inside the retaining structure and the soil. The high-precision inclination sensors are installed at key parts of the retaining structure and the ground around the foundation pit to monitor the inclination angle changes of the structure and the ground. The measurement accuracy is improved and the absolute measurement error is reduced through relative measurement. The distributed optical fiber sensor is based on the fiber Bragg grating (FBG) sensing principle, and the fiber grating inside is sensitive to the tiny strain inside the retaining structure and the soil. When the relative displacement of the retaining structure or the soil changes, the period and effective refractive index of the fiber grating change accordingly, resulting in a shift in the wavelength of the reflected light. By monitoring the change in the wavelength of the reflected light in real time with a high-precision optical wavelength demodulator and combining the strain-wavelength conversion coefficient of the sensor, the relative displacement change amount at the corresponding position is accurately calculated to achieve real-time acquisition of the relative displacement change information inside the retaining structure and the soil.

[0038] The high-precision inclination sensor adopts a dual-axis capacitive tilt measurement technology, with two mutually perpendicular pendulum mass blocks and capacitor plates built in. Under the action of gravity, the swing angle of the mass block has a sine relationship with the tilt angle of the sensor. By measuring the capacitance change generated between the capacitor plates due to the swing of the mass block, it is converted into an electrical signal proportional to the tilt angle through the signal processing circuit inside the sensor and output.

[0039] To improve the measurement accuracy and reduce the absolute measurement error, the system adopts a relative measurement method, that is, the high-precision inclination sensor is calibrated before the foundation pit excavation, and the initial tilt angle value is recorded. During the subsequent monitoring process, the current tilt angle value is collected in real time and the relative difference is calculated with the initial value. The zero drift and temperature drift errors of the sensor are corrected through an algorithm, so as to obtain a more accurate change amount of the tilt angle of the structure and the ground, effectively reducing the cumulative error and environmental interference error caused by absolute measurement.

[0040] The data acquisition and preprocessing module is connected to the relative measurement sensor group, which is used to synchronously collect sensor data and perform preprocessing such as filtering, denoising, and linear correction on the collected raw data to preliminarily improve the data quality; The wireless data transmission module is connected to the data acquisition and preprocessing module using multi-hop relay communication technology, stably transmitting the preprocessed data to the cloud data processing and analysis module, and automatically switching the communication frequency band when the signal is interfered to ensure the continuity and integrity of data transmission; The cloud data processing and analysis module receives the data sent by the wireless data transmission module, and uses a deformation feature extraction algorithm based on deep learning and a finite element analysis model to deeply analyze the deformation data during the deep foundation pit excavation process, establish a multi-physics field coupling model including the stress and deformation of the retaining structure, the flow change of the soil mass, and the influence of the groundwater level, realize high-precision real-time prediction of the foundation pit deformation trend, and compare it with the preset dynamic safety threshold; Based on the prediction results of the cloud data processing and analysis module, the intelligent risk assessment module combines the information of the foundation pit surrounding environment (including surrounding buildings, underground pipelines, and traffic flow, etc.) and historical deformation data, and uses the fuzzy comprehensive evaluation method to dynamically classify and evaluate the deformation risk of the deep foundation pit. The evaluation results are divided into four levels: low risk, medium risk, high risk, and extremely high risk; The visualization alarm module is connected to the intelligent risk assessment module, and performs real-time alarms in different visualization forms (low risk is displayed in green and stable, medium risk is displayed in yellow and flashing, high risk is displayed in orange and flashing quickly, extremely high risk is displayed in red and exploding and flashing) on the local monitoring screen and remote mobile terminals according to the risk level, and automatically sends an emergency text message with accurate location information to the preset emergency rescue personnel in the case of extremely high risk; The solar power supply and energy storage management module provides power support for the entire monitoring system, including high-efficiency solar panels, intelligent energy storage battery packs, and an energy management system. The solar panels convert solar energy into electrical energy and store it in the energy storage battery packs. The energy management system automatically optimizes the power distribution according to the power consumption requirements and real-time power conditions of each module to ensure that the system can continue to operate stably for no less than X hours under non-light conditions such as rainy days and nights. At the same time, it has overcharge and over-discharge protection functions and real-time power monitoring functions, and transmits the power information to the cloud data processing and analysis module through wireless signals for recording and management. When the power is lower than the preset threshold, the system automatically sends a charging reminder text message to the maintenance personnel.

[0041] In a preferred embodiment, the distributed optical fiber sensors in the relative measurement sensor group to obtain the relative displacement change information inside the retaining structure and the soil mass specifically include: Distributed fiber optic sensors are based on the principle of fiber optic Brillouin scattering. The relationship between Brillouin frequency shift and fiber strain is used to obtain information on the relative displacement changes inside the retaining structure and soil mass. The specific relationship is as follows:

[0042] In the formula: is the Brillouin frequency shift at temperature T and strain S; is the initial temperature , and the initial strain is the Brillouin frequency shift at; Through the change in Brillouin frequency shift , the fiber strain can be calculated as:

[0043] In the formula: is the change in Brillouin frequency shift; is the change in Brillouin frequency shift caused by temperature change; is the temperature coefficient; is the initial Brillouin frequency shift.

[0044] In a preferred embodiment, the high-precision inclinometer in the relative measurement sensor group improves the measurement accuracy of the inclination angle changes of the structure and the ground through relative measurement and reduces the absolute measurement error. Specifically, it includes: Before the foundation pit excavation, the high-precision inclinometer is calibrated for reference, and the initial inclination angle value is recorded , and during the subsequent monitoring process, the current inclination angle value is collected in real time , and the inclination angle change is calculated through the following formula :

[0045] To reduce the absolute measurement error, the following algorithm is used to correct the zero drift and temperature drift errors of the sensor:

[0046] In the formula: is the corrected inclination angle; is the originally measured inclination angle; is the zero drift and temperature drift error correction value related to temperature, and its calculation formula is:

[0047] In the formula: is the temperature coefficient; is the constant term; is the current temperature.

[0048] In a preferred embodiment, the cloud data processing and analysis module uses a deformation feature extraction algorithm based on deep learning and a finite element analysis model to deeply analyze the deformation data during the deep foundation pit excavation process, specifically including: S1: Deformation feature extraction algorithm based on deep learning: Use a convolutional neural network (CNN) to extract spatial features from the monitoring data of multi-dimensional time series. The multi-dimensional time series Xt,n composed of the standardized historical data of multiple monitoring points is used as the input data. Xt,n is a two-dimensional matrix, where one row represents the monitoring values of n monitoring points at a certain moment, and one column represents the monitoring values of a certain monitoring point within the time step t. After inputting this sequence into the CNN part, perform convolution operations to extract spatial features and output multiple one-dimensional sequences composed of spatial features; then use a long short-term memory network (LSTM) to perform time series learning and regression prediction on the spatial feature sequence extracted by the CNN, output the predicted values of multiple spatial features, and finally integrate the predicted values of spatial features through a fully connected layer to output the predicted values of the monitoring data; S2: Finite element analysis model: Establish a multi-physics field coupling model including the stress and deformation of the retaining structure, the flow change of the soil mass, and the influence of the groundwater level; S3: Real-time prediction: Combine the deformation feature data extracted by the deep learning algorithm with the finite element analysis model, and achieve high-precision real-time prediction of the foundation pit deformation trend through data fusion and numerical simulation; specifically, use the deformation feature values predicted by the deep learning algorithm as the boundary conditions or initial conditions of the finite element analysis model, and obtain the predicted deformation values of the foundation pit at the next moment through numerical solution; S4: Comparison with the dynamic safety threshold: The preset dynamic safety threshold is determined according to the design parameters of the foundation pit, the surrounding environmental conditions, and relevant engineering experience and technical specifications; during the real-time prediction process, compare the predicted foundation pit deformation value with the preset dynamic safety threshold When the predicted deformation value exceeds the safety threshold, trigger the alarm mechanism to timely remind the construction personnel to take corresponding measures.

[0049] In a preferred embodiment, in S1, a convolutional neural network (CNN) is used to extract spatial features from the monitoring data of multi-dimensional time series, and its convolution operation formula is:

[0050] Where: is the output feature map after the convolution operation; is the input multi-dimensional time series monitoring data matrix; is the convolution kernel weight matrix, with a size of , where is the convolution kernel size; is the convolution kernel bias vector; is the activation function, using the ReLU function, ; Use the long short-term memory network (LSTM) to perform time series learning and regression prediction on the spatial feature sequence extracted by the CNN. The unit state update formula of the LSTM is:

[0051] In the formula: is the current time of the unit state; is the output of the forget gate, controlling how much information of the previous unit state is forgotten. The calculation formula is: .

[0052] is the unit state of the previous moment; is the output of the input gate, controlling how much new information is stored in the unit state at the current moment. The calculation formula is: .

[0053] is the candidate unit state, and the calculation formula is ; are the weight matrices of the forget gate, input gate, and candidate unit state respectively; are the bias vectors of the forget gate, input gate, and candidate unit state respectively; represents concatenating the hidden state of the previous moment and the input at the current moment; is the sigmoid activation function, ; tanh is the hyperbolic tangent function, ; is the element-wise multiplication operation; Finally, integrate the spatial feature prediction value through the fully connected layer to output the prediction value of the monitoring data. The calculation formula of the fully connected layer is:

[0054] In the formula: is the prediction value of the monitoring data; is the weight matrix of the fully connected layer; is the hidden state vector output by the LSTM; Is the bias vector of the fully connected layer.

[0055] In a preferred embodiment, in S2, the multi-physical field coupling model includes the finite element equation of the stress and deformation of the retaining structure, the finite element equation of the soil flow change, and the finite element equation of the influence of the groundwater level. The finite element equation of the stress and deformation of the retaining structure is: In the formula: Is the stiffness matrix, Is the nodal displacement vector, Is the nodal force vector; The finite element equation of the soil flow change is: In the formula: Is the seepage velocity vector, Is the permeability coefficient tensor of the soil, Is the dynamic viscosity coefficient of the fluid, Is the pressure of the fluid; The finite element equation of the influence of the groundwater level is: In the formula: Is the water head height, Is the permeability coefficient tensor.

[0056] In a preferred embodiment, in S3, the prediction formula is:

[0057] In the formula: Is the nodal displacement vector at the next moment; Is the nodal displacement vector at the current moment; Is the time step; Is the residual vector based on the current nodal displacement and the deep learning prediction value, and is obtained by iterative solution of the finite element equation.

[0058] In a preferred embodiment, in S4, the mathematical expression of the preset dynamic safety threshold is:

[0059] In the formula: Is the maximum allowable displacement of the retaining structure at the position ; Is the allowable displacement at the design stage of the retaining structure; Is the displacement safety factor based on the surrounding environment and engineering experience; Is the maximum allowable stress of the soil at the position ; Is the design allowable stress of the soil; is the soil stress safety factor based on the surrounding environment and engineering experience; The predicted foundation pit deformation value is compared with the preset dynamic safety threshold The comparison formula is: Alarm = In the formula: Alarm is the alarm signal, 1 indicates triggering the alarm, and 0 indicates normal; is the predicted displacement of the retaining structure at the position ; is the predicted stress of the soil at the position ;

[0060] In a preferred embodiment, the intelligent risk assessment module uses the fuzzy comprehensive evaluation method to dynamically classify and evaluate the deformation risk of the deep foundation pit, which specifically includes: Sa: Construct an evaluation index system: The factors related to the deformation risk of the deep foundation pit are used as evaluation indexes, including the horizontal displacement and vertical settlement of the retaining structure, the horizontal displacement and vertical settlement of the soil, the change of the groundwater level, and the surrounding environment factors (such as the distance of surrounding buildings, the distribution of underground pipelines, etc.). Let the evaluation index set be where represents the th evaluation index; Sb: Determine the fuzzy evaluation level: Set the fuzzy evaluation level set as , which respectively represent four levels: low risk, medium risk, high risk, and extremely high risk; Sc: Construct a fuzzy membership function: For each evaluation index , according to its actual measured value or predicted value, construct a fuzzy membership function to determine its membership degree to each evaluation level; Sd: Establish a fuzzy evaluation matrix: According to the membership degrees of each evaluation index to each evaluation level, establish a fuzzy evaluation matrix , which is composed of single-factor evaluation sets as rows. Each single-factor evaluation set contains the membership degrees of this evaluation index to each evaluation level, that is , where represents the membership degree of the th evaluation index to the th evaluation level; Se: Determine the weight vector: Determine the weight vector of each evaluation index through the analytic hierarchy process , where represents the weight of the th evaluation index, and satisfies ; Sf: Conduct fuzzy comprehensive evaluation: Use fuzzy operations to perform weighted summation on evaluation indicators to obtain the fuzzy comprehensive evaluation result, and adopt the weighted average model:

[0061] Where: is the fuzzy comprehensive evaluation result vector, representing the membership degree of the deformation risk of the deep foundation pit at each evaluation level; Sg: Defuzzification: In order to obtain a specific evaluation result, defuzzify the fuzzy comprehensive evaluation result, and adopt the maximum membership degree method: Risk Level That is, select the evaluation level with the largest membership degree as the final deformation risk level of the deep foundation pit.

[0062] In a preferred embodiment, in Sc, for the horizontal displacement of the retaining structure , use the triangular membership function to determine its membership degrees to low risk , medium risk , high risk and extremely high risk :

[0063] Where: is a measure of belonging to the evaluation level , and the value range is between ; for example, , it means that when the horizontal displacement of the retaining structure is 5 mm, there is possibility of belonging to the low risk level; are the parameters of the triangular membership function, which are determined according to engineering experience and relevant specifications. For example, for low risk , it can be taken as (unit: mm), which means that when the horizontal displacement is less than 0 mm, the membership degree to low risk is 0; when the horizontal displacement is between 0 mm and 5 mm, the membership degree increases linearly; when the horizontal displacement is between 5 mm and 10 mm, the membership degree decreases linearly; when the horizontal displacement is greater than 10 mm, the membership degree to low risk is 0; is the actual measured value or predicted value of the evaluation indicator , for example, the horizontal displacement of the retaining structure is 5 mm.

[0064] From the above description, it can be seen that the above embodiments of the present invention achieve the following technical effects: 1. Improvement in measurement accuracy: Through the innovative design of the relative measurement sensor group, including the combination of distributed optical fiber sensors and high-precision inclination sensors. The distributed optical fiber sensors accurately obtain the relative displacement change information inside the retaining structure and soil mass based on the fiber Bragg grating (FBG) sensing principle. The high-precision inclination sensors adopt dual-axis capacitive tilt measurement technology and improve the measurement accuracy of the tilt angle changes of the structure and the ground through relative measurement, effectively reducing the absolute measurement error, greatly improving the accuracy of deep foundation pit deformation monitoring, and providing a more accurate data basis for subsequent risk assessment.

[0065] 2. Enhancement of the stability of data transmission and processing: The wireless data transmission module is connected to the data acquisition and preprocessing module using multi-hop relay communication technology, and can automatically switch the communication frequency band when the signal is interfered, ensuring the continuity and integrity of data transmission; The cloud data processing and analysis module uses a deformation feature extraction algorithm based on deep learning and a finite element analysis model to deeply analyze the deformation data during the excavation of the deep foundation pit, establish a multi-physical field coupling model including the stress and deformation of the retaining structure, soil mass flow changes, and the influence of the groundwater level, and realize high-precision real-time prediction of the foundation pit deformation trend, solving the problems of easy interruption of data transmission and insufficient comprehensive risk assessment in the existing monitoring system, and improving the stability and reliability of the monitoring system.

[0066] 3. Improvement in the scientificity and reliability of risk assessment: The intelligent risk assessment module uses the fuzzy comprehensive evaluation method to dynamically classify and evaluate the deformation risk of the deep foundation pit, comprehensively considering multiple evaluation indicators such as the horizontal displacement and vertical settlement of the retaining structure, the horizontal displacement and vertical settlement of the soil mass, the change of the groundwater level, and the surrounding environmental factors. Through a series of scientific assessment steps such as constructing the evaluation index system, determining the fuzzy evaluation level, constructing the fuzzy membership function, establishing the fuzzy evaluation matrix, determining the weight vector, and performing fuzzy comprehensive evaluation and defuzzification, the risk assessment result is more comprehensive and accurate, can timely and accurately reflect the deformation risk status of the deep foundation pit, and provide a reliable decision-making basis for construction personnel.

[0067] 4. Guarantee of power supply stability: The solar power supply and energy storage management module provides power support for the entire monitoring system, including high-efficiency solar panels, intelligent energy storage battery packs, and an energy management system; The solar panels convert solar energy into electrical energy and store it in the energy storage battery packs. The energy management system automatically optimizes the power distribution according to the power consumption requirements and real-time power conditions of each module, ensuring that the system can continue to operate stably for at least X hours under conditions of no sunlight such as rainy days and nights, and at the same time having overcharge and over-discharge protection functions and real-time power monitoring functions, solving the problem of poor power supply stability in the existing monitoring system, guaranteeing the long-term stable operation of the monitoring system, and reducing the risk of missing monitoring data caused by power supply interruption.

[0068] Those of ordinary skill in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of the present invention (including the claims) is limited to these examples; within the concept of the present invention, the technical features in the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations in different aspects of the present invention as described above, which are not provided in detail for the sake of brevity.

[0069] The present invention is intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A real-time monitoring system for the deformation of deep foundation pit excavation based on relative measurement, characterized in that Including: A relative measurement sensor group, a data acquisition and preprocessing module, a wireless data transmission module, a cloud data processing and analysis module, an intelligent risk assessment module, a visualization alarm module, and a solar power supply and energy storage management module; The relative measurement sensor group consists of multiple distributed optical fiber sensors and high-precision inclination sensors; The data acquisition and preprocessing module is connected to the relative measurement sensor group and is used to synchronously acquire sensor data and perform preprocessing; The wireless data transmission module is connected to the data acquisition and preprocessing module using multi-hop relay communication technology and stably transmits the preprocessed data to the cloud data processing and analysis module; The cloud data processing and analysis module receives the data sent by the wireless data transmission module and deeply analyzes the deformation data during the deep foundation pit excavation process; The intelligent risk assessment module dynamically classifies and assesses the deformation risk of the deep foundation pit based on the prediction results of the cloud data processing and analysis module; The visualization alarm module is connected to the intelligent risk assessment module and gives real-time alarms according to the risk level; The solar power supply and energy storage management module provides power support for the entire monitoring system.

2. The real-time monitoring system for the deformation of deep foundation pit excavation based on relative measurement according to claim 1, wherein, The distributed optical fiber sensors in the relative measurement sensor group obtain the relative displacement change information inside the retaining structure and the soil mass in real time, specifically including: Distributed optical fiber sensors are based on the principle of fiber Brillouin scattering. The relationship between Brillouin frequency shift and fiber strain is used to obtain the relative displacement change information inside the enclosure structure and soil mass. The specific relationship is as follows: In the formula: is the Brillouin frequency shift at temperature T and strain S; is the initial temperature , the initial strain and the Brillouin frequency shift under By the change of Brillouin frequency shift , the fiber strain can be calculated : In the formula: is the change in Brillouin frequency shift; is the change amount of Brillouin frequency shift caused by temperature change; is the temperature coefficient; is the initial Brillouin frequency shift.

3. The real-time monitoring system for the deformation of deep foundation pit excavation based on relative measurement according to claim 1, characterized in that, The high-precision inclination sensors in the relative measurement sensor group improve the measurement accuracy of the inclination angle change of the structure and the ground and reduce the absolute measurement error through relative measurement, specifically including: Before the foundation pit excavation, calibrate the high-precision inclination sensor to record the initial inclination angle value , and during the subsequent monitoring process, collect the current inclination angle value in real time , and calculate the inclination angle change through the following formula : To reduce the absolute measurement error, use the following algorithm to correct the zero drift and temperature drift errors of the sensor: In the formula: is the corrected inclination angle; is the tilt angle measured originally; is the zero drift and temperature drift error correction value related to temperature, and its calculation formula is: Where: is the temperature coefficient; is the constant term; is the current temperature.

4. The real-time monitoring system for deep foundation pit excavation deformation based on relative measurement according to claim 1, characterized in that The cloud data processing and analysis module uses a deformation feature extraction algorithm based on deep learning and a finite element analysis model to deeply analyze the deformation data during the deep foundation pit excavation process, specifically including: S1: Deformation feature extraction algorithm based on deep learning: Use a convolutional neural network to extract spatial features from the monitoring data of multi-dimensional time series. Take the multi-dimensional time series Xt,n composed of the standardized historical data of multiple monitoring points as the input data. After inputting this sequence into the CNN part, perform convolution operations to extract spatial features and output multiple one-dimensional sequences composed of spatial features; then use a long short-term memory network to perform time series learning and regression prediction on the spatial feature sequence extracted by the CNN, output the predicted values of multiple spatial features, and finally integrate the predicted values of spatial features through a fully connected layer to output the predicted values of the monitoring data; S2: Finite element analysis model: Establish a multi-physics field coupling model including the stress and deformation of the retaining structure, the flow change of the soil mass, and the influence of the groundwater level; S3: Real-time prediction: Combine the deformation feature data extracted by the deep learning algorithm with the finite element analysis model, and through data fusion and numerical simulation methods, achieve high-precision real-time prediction of the foundation pit deformation trend; specifically, use the deformation feature values predicted by the deep learning algorithm as the boundary conditions or initial conditions of the finite element analysis model, and obtain the deformation prediction value of the foundation pit at the next moment through numerical solution; S4: Compare with the dynamic safety threshold: The preset dynamic safety threshold is determined based on the design parameters of the foundation pit, the surrounding environmental conditions, as well as relevant engineering experience and technical specifications; during the real-time prediction process, compare the predicted foundation pit deformation value with the preset dynamic safety threshold When the predicted deformation value exceeds the safety threshold, trigger the alarm mechanism to promptly remind the construction personnel to take corresponding measures.

5. The real-time monitoring system for the deformation of deep foundation pit excavation based on relative measurement according to claim 4, characterized in that In the above S1, when using a convolutional neural network to extract spatial features from the monitoring data of multi-dimensional time series, its convolution operation formula is: In the formula: is the output feature map after convolution operation; is the input multi-dimensional time series monitoring data matrix; is the convolution kernel weight matrix, with the size of , where is the convolution kernel size; is the convolution kernel bias vector; is the activation function, using the ReLU function, ; Use a long short-term memory network to perform time series learning and regression prediction on the spatial feature sequence extracted by CNN. The unit state update formula of LSTM is as follows: In the formula: is the unit state at the current moment ; is the output of the forget gate, which controls how much information of the cell state at the previous moment is forgotten. The calculation formula is: . is the cell state at the previous moment; is the input gate output, which controls how much new information is stored in the cell state at the current moment. The calculation formula is: . is the candidate cell state, and the calculation formula is ; They are the weight matrices of the forget gate, input gate, and candidate cell state respectively; They are the bias vectors of the forget gate, input gate, and candidate cell state respectively; It represents concatenating the hidden state at the previous moment and the input at the current moment at time t; is the sigmoid activation function, ; The tanh is the hyperbolic tangent function, ; is an element-level multiplication operation; Finally, the spatial feature prediction values are integrated through the fully connected layer to output the prediction values of the monitoring data. The calculation formula of the fully connected layer is as follows: In the formula: is the predicted value of the monitoring data; is the weight matrix of the fully connected layer; is the hidden state vector output by the LSTM; is the bias vector of the fully connected layer.

6. The real-time monitoring system for the deformation of deep foundation pit excavation based on relative measurement according to claim 4, characterized in that In S2, the multi-physical field coupling model includes the finite element equation for the stress and deformation of the enclosure structure, the finite element equation for the flow change of the soil mass, and the finite element equation for the influence of the groundwater level. The finite element equation for the stress and deformation of the enclosure structure is as follows: In the formula: is the stiffness matrix, is the nodal displacement vector, is the nodal force vector; The finite element equation for the flow change of soil mass is as follows: In the formula: is the seepage velocity vector, is the permeability coefficient tensor of the soil mass, is the dynamic viscosity coefficient of the fluid, is the pressure of the fluid; The finite element equation affected by the groundwater level is as follows: In the formula: is the water head height, is the permeability coefficient tensor.

7. The real-time monitoring system for the deformation of deep foundation pit excavation based on relative measurement according to claim 4, wherein In S3, the prediction formula is as follows: In the formula: is the nodal displacement vector at the next moment; is the nodal displacement vector at the current moment; is the time step; is the residual vector based on the current nodal displacement and the prediction value of deep learning, which is obtained by iterative solution of the finite element equation.

8. The real-time monitoring system for deformation during deep foundation pit excavation based on relative measurement according to claim 4, wherein, In S4, the mathematical expression of the preset dynamic safety threshold is as follows: In the formula: is the maximum allowable displacement of the retaining structure at the position ; is the allowable displacement at the design stage of the retaining structure; is the displacement safety factor based on the surrounding environment and engineering experience; is the maximum allowable stress of the soil mass at the position ; is the design allowable stress of the soil mass; is the soil stress safety factor based on the surrounding environment and engineering experience; The predicted foundation pit deformation value and the preset dynamic safety threshold The comparison formula is: Alarm = Where: Alarm is the alarm signal, 1 indicates triggering an alarm, and 0 indicates normal; is the predicted displacement of the retaining structure at the position ; is the predicted stress of the soil mass at the position .

9. The real-time monitoring system for deep foundation pit excavation deformation based on relative measurement according to claim 1, wherein The intelligent risk assessment module uses the fuzzy comprehensive evaluation method to dynamically classify and evaluate the deformation risk of the deep foundation pit, which specifically includes: Sa: Construct an evaluation index system: Consider the factors related to the deformation risk of deep foundation pits as evaluation indicators, including the horizontal displacement, vertical settlement of the retaining structure, horizontal displacement, vertical settlement of the soil body, groundwater level change, and surrounding environmental factors. Let the evaluation index set be , where represents the th evaluation indicator; Sb: Determine the fuzzy evaluation level: Set the fuzzy evaluation level set as , which respectively represent four levels of low risk, medium risk, high risk and extremely high risk; Sc: Construct the fuzzy membership function: For each evaluation index , construct the fuzzy membership function according to its actual measured value or predicted value, and determine its membership degree to each evaluation level; Sd: Establish a fuzzy evaluation matrix: According to the membership degrees of each evaluation index to each evaluation level, establish a fuzzy evaluation matrix , which is composed of single-factor evaluation sets as rows to form a matrix. Each single-factor evaluation set contains the membership degrees of this evaluation index to each evaluation level, that is , where represents the membership degree of the th evaluation index to the th evaluation level; Se: Determine the weight vector: Determine the weight vector of each evaluation index through the Analytic Hierarchy Process , where represents the weight of the -th evaluation index, and satisfies ; Sf: Conduct a fuzzy comprehensive evaluation: Use fuzzy operations to perform weighted summation on the evaluation indicators to obtain the fuzzy comprehensive evaluation result, and adopt the weighted average model: In the formula: is the fuzzy comprehensive evaluation result vector, representing the membership degrees of the deformation risk of the deep foundation pit at each evaluation level; Sg: Defuzzification: To obtain a specific evaluation result, defuzzification is performed on the fuzzy comprehensive evaluation result using the maximum membership degree method: Risk Level That is, select the evaluation level with the largest membership degree as the final deformation risk level of the deep foundation pit.

10. The real-time monitoring system for deep foundation pit excavation deformation based on relative measurement according to claim 9, characterized in that, In the Sc, for the horizontal displacement of the enclosure structure , a triangular membership function is used to determine its membership degrees for low risk , medium risk , high risk and extremely high risk : In the formula: is a measure of belonging to the evaluation level The degree, and the value range is within ; is the parameter of the triangular membership function, which is determined according to engineering experience and relevant specifications; is the evaluation index The actual measured value or predicted value of

Images