Bridge safety early warning method and system based on construction monitoring

Abstract

The invention relates to a bridge safety early warning method and system based on construction monitoring, and belongs to the technical field of bridge engineering.The early warning method comprises the steps that multi-dimensional monitoring data collected by sensor networks deployed on a bridge girder and a bridge key section in real time are obtained; performing edge calculation processing on the multi-dimensional monitoring data to obtain a cleaned monitoring data set; based on the cleaned monitoring data set, combining a finite element model to construct a three-dimensional dynamic stress field, and generating a dynamic stress nephogram matrix; dynamically correcting a pre-configured stress allowable threshold value according to the environment monitoring data; predicting a stress prediction matrix in a future time window based on historical stress data and the dynamic stress nephogram matrix; the real-time stress value and the stress prediction matrix are compared with a preset stress allowable threshold value, and an early warning score is calculated; and determining a corresponding early warning level according to the early warning score and triggering a maintenance control strategy. According to the invention, the accuracy of bridge structure risk early warning can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of bridge engineering, and in particular to a bridge safety early warning method and system based on construction monitoring. Background Art

[0002] As one of the important infrastructures, the construction quality of bridges is directly related to traffic safety and sustainable social and economic development. With the acceleration of urbanization, the scale of bridge construction has continued to expand, and various types of bridges have been built and put into use. During the construction and use of bridges, it is crucial to ensure the structural safety and long-term stability of bridges. However, during the construction process, bridges are often affected by many factors, such as construction loads, material strength, environmental changes, etc. These factors bring potential risks to the stability of bridge structures. How to effectively monitor the health of bridges and promptly discover and solve structural problems has become a key issue in modern bridge engineering.

[0003] In recent years, with the development of intelligent technology and sensor technology, more and more modern monitoring methods have been applied to bridge safety management. By deploying various types of sensors (such as strain gauges, accelerometers, environmental sensors, etc.) at key parts of bridges, multi-dimensional data such as stress, vibration, and environmental factors of bridges can be collected in real time, which provides technical support for real-time monitoring of bridges. These monitoring data can reflect the stress condition and overall health status of the bridge to a certain extent, providing a reference for decision makers.

[0004] However, since the stress changes in bridge structures during the construction phase are affected by multiple factors such as the environment and construction loads, common construction monitoring methods often cannot fully take into account the changes in these factors. In particular, the structural state of the bridge often changes dramatically during the construction process, but due to the lack of precise early warning monitoring and environmental adaptability, it is difficult to accurately warn of structural risks during the construction process. Summary of the invention

[0005] In order to improve the accuracy of bridge structure risk warning, the present application provides a bridge safety warning method and system based on construction monitoring.

[0006] In the first aspect, the present application provides a bridge safety early warning method based on construction monitoring, which adopts the following technical solution: A bridge safety early warning method based on construction monitoring, the early warning method comprising: Acquire multi-dimensional monitoring data collected in real time by a sensor network deployed on the bridge main beam and key sections of the bridge; the multi-dimensional monitoring data includes real-time stress values, environmental monitoring data and construction loads; Performing edge computing processing on the multidimensional monitoring data to obtain a cleaned monitoring data set; Based on the cleaned monitoring data set, a three-dimensional dynamic stress field is constructed in combination with the finite element model to generate a dynamic stress cloud map matrix; Dynamically modifying a pre-configured stress allowable threshold value according to the environmental monitoring data; Predicting a stress prediction matrix within a future time window based on historical stress data and the dynamic stress cloud map matrix; Comparing the real-time stress value and the stress prediction matrix with the preset stress allowable threshold value respectively, and calculating the early warning score; The corresponding warning level is determined according to the warning score and a maintenance control strategy is triggered.

[0007] By adopting the above technical solutions, a highly intelligent bridge safety monitoring and early warning system has been built based on multi-dimensional perception technology, dynamic modeling, data prediction and intelligent control. By collecting multi-dimensional monitoring data in real time and combining future trend prediction, the system can effectively identify and predict safety risks in bridge construction, greatly improving the safety of bridge construction. This technical solution can not only provide accurate real-time monitoring, but also dynamically adjust the early warning threshold according to environmental changes, reduce the occurrence of false alarms and missed alarms, and ultimately improve the safety and efficiency of bridge construction.

[0008] Optionally, based on the cleaned monitoring data set, the steps of constructing a three-dimensional dynamic stress field in combination with a finite element model and generating a dynamic stress cloud map matrix include: Obtain bridge design parameters and build an initial finite element model based on the cleaned monitoring data set; Divide the finite element mesh nodes based on the initial finite element model and load the construction phase boundary conditions to simulate and generate a theoretical stress reference field matrix; Obtaining the spatial distribution coordinates of the sensor network, calculating the weight of each sensor corresponding node according to the finite element grid nodes, and constructing a mapping matrix; Dynamically correct the initial finite element model parameters according to the real-time stress value and the environmental monitoring data to obtain a corrected finite element model; Based on the mapping matrix, the real-time stress value is fused with the theoretical stress reference field matrix by weighted interpolation to generate a dynamic stress cloud map matrix.

[0009] By adopting the above technical solutions, dynamic high-precision reconstruction and risk control of the stress field during bridge construction are achieved. By combining precise modeling with dynamic correction, the stress distribution of the bridge structure can be accurately predicted, and the model can adapt to environmental changes in real time, thereby facilitating real-time monitoring of safety risks during the construction process, early warning of potential dangers, and avoiding the occurrence of major safety accidents.

[0010] Optionally, the step of dynamically correcting the pre-configured stress allowable threshold according to the environmental monitoring data includes: Extracting real-time temperature data and wind speed data according to the environmental monitoring data; Based on the pre-configured reference stress allowable threshold, dynamic correction is performed according to the temperature data and wind speed data, and the correction formula is: ; Among them, σ 允许 is the allowable threshold value of modified stress, σ 基准 is the allowable threshold of the baseline stress, ΔT is the temperature change, V is the wind speed, β is the temperature compensation coefficient, and γ is the wind load reduction coefficient.

[0011] By adopting the above technical solution, combined with the correction mechanism of environmental compensation coefficient and wind load reduction coefficient, dynamic correction of stress allowable threshold is realized. This technical solution not only takes into account the static stress threshold during design, but also can timely adapt to environmental changes during construction and later use. The corrected stress threshold can more accurately reflect the real safety status of the bridge under different environmental conditions, avoiding misjudgment or omission due to environmental changes.

[0012] Optionally, the step of predicting the stress prediction matrix in the future time window based on the historical stress data and the dynamic stress cloud map matrix includes: Extracting stress gradient characteristics of each finite element mesh node according to the dynamic stress cloud map matrix, and constructing a node stress gradient matrix; Based on the historical stress data, a spatiotemporal prediction model is constructed and trained according to the modified finite element model topological relationship to obtain a trained spatiotemporal prediction model; Acquire construction machinery positioning data on the bridge, adjust node weights of local finite element mesh nodes, and weight the node stress gradient matrix; The weighted node stress gradient matrix is ​​input into the spatiotemporal prediction model to obtain a stress prediction matrix in a future time window.

[0013] By adopting the above technical solution, combining the dynamic stress cloud map matrix with historical data, and using spatiotemporal feature fusion and dynamic modeling technology, it is possible to achieve high-precision prediction of stress trends during bridge construction. By extracting node stress gradient characteristics, building a spatiotemporal prediction model, dynamically weighting the impact of construction machinery, and correcting thresholds in combination with actual environmental parameters, this technical solution effectively improves the risk assessment accuracy and early warning response capabilities during bridge construction.

[0014] Optionally, based on the historical stress data, the spatiotemporal prediction model is constructed and trained according to the modified finite element model topological relationship to obtain the trained spatiotemporal prediction model, the steps include: Based on the modified finite element model topological relationship, preprocessing the historical stress data to obtain a sample characteristic tensor; Dividing the sample feature tensor into a training set and a test set; Construct a hybrid prediction model consisting of a graph convolutional network and a bidirectional long short-term memory network; Inputting the training set into the hybrid prediction model for dynamic training, optimizing model parameters, and obtaining a spatiotemporal prediction model; The spatiotemporal prediction model is tested and model parameters are corrected based on the test set until the loss function of the model meets a preset condition or the number of model iterations reaches a preset number, thereby obtaining the trained spatiotemporal prediction model.

[0015] By adopting the above technical solutions, based on the combination of graph convolutional networks and bidirectional long short-term memory networks, an efficient spatiotemporal prediction model is constructed, which can achieve high-precision stress trend prediction in bridge construction safety monitoring. The hybrid model architecture can not only integrate spatial and temporal features, but also improve the model's prediction ability through dynamic training optimization, provide accurate prediction support in complex environments, and help improve the safety and efficiency of bridge construction and operation.

[0016] Optionally, the step of comparing the real-time stress value and the stress prediction matrix with a preset stress allowable threshold value respectively, and calculating the early warning score includes: The real-time stress value is used as the first column of data of the stress prediction matrix to form a complete time window data matrix; Based on the complete time window data matrix, the preset stress allowable threshold is expanded into a stress threshold matrix with consistent dimension; Performing point-by-point comparison between the complete time window data matrix and the stress threshold matrix to generate an over-limit marking matrix; A weighted calculation is performed on the over-limit marking matrix, and a warning score is obtained in combination with a time decay coefficient.

[0017] By adopting the above technical solution, combined with the weighted scoring mechanism of real-time stress value and predicted stress trend, the current safety status of the bridge and future potential risks can be accurately reflected. By gradually comparing the real-time stress value, predicted stress value and the corrected stress threshold, and combining multi-level over-limit marking, weighted calculation and time attenuation coefficient, the system can provide more detailed and accurate early warning scores. This technical method not only improves the accuracy of stress prediction, but also can identify potential risks in advance, improve the real-time and accuracy of bridge safety monitoring, and provide sufficient time and decision-making basis for emergency response.

[0018] In the second aspect, the present application provides a bridge safety early warning system based on construction monitoring, which adopts the following technical solutions: A bridge safety early warning system based on construction monitoring, the early warning system includes: An acquisition module is used to acquire multi-dimensional monitoring data collected in real time by a sensor network deployed on the bridge main beam and key sections of the bridge; the multi-dimensional monitoring data includes real-time stress values, environmental monitoring data and construction loads; A data processing module, used to perform edge computing processing on the multi-dimensional monitoring data to obtain a cleaned monitoring data set; The cloud map matrix generation module is used to construct a three-dimensional dynamic stress field based on the cleaned monitoring data set and the finite element model to generate a dynamic stress cloud map matrix; A dynamic correction module, used for dynamically correcting the pre-configured stress allowable threshold according to the environmental monitoring data; A stress prediction module, used for predicting a stress prediction matrix in a future time window based on historical stress data and the dynamic stress cloud map matrix; A scoring module, used to compare the real-time stress value and the stress prediction matrix with the preset stress allowable threshold value, and calculate the early warning score; The early warning module is used to determine the corresponding early warning level according to the early warning score and trigger the maintenance control strategy.

[0019] As an implementation of the cloud map matrix generation module, the cloud map matrix generation module includes: A model building unit, used to obtain bridge design parameters and build an initial finite element model based on the cleaned monitoring data set; A simulation unit, used for dividing finite element mesh nodes based on the initial finite element model and loading construction phase boundary conditions, so as to simulate and generate a theoretical stress reference field matrix; A mapping matrix construction unit, used to obtain the spatial distribution coordinates of the sensor network, calculate the weight of each sensor corresponding node according to the finite element grid node, and construct a mapping matrix; A dynamic correction unit, used to dynamically correct the initial finite element model parameters according to the real-time stress value and environmental monitoring data to obtain a corrected finite element model; The interpolation fusion unit is used to perform weighted interpolation fusion on the real-time stress value and the theoretical stress reference field matrix based on the mapping matrix to generate a dynamic stress cloud map matrix.

[0020] In a third aspect, the present application provides a computer device, which adopts the following technical solution: A computer device comprises a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method as described in the first aspect.

[0021] In a fourth aspect, the present application provides a computer-readable storage medium, which adopts the following technical solution: A computer-readable storage medium stores a computer program that can be loaded by a processor and execute any one of the methods in the first aspect.

[0022] In summary, the present application includes at least one of the following beneficial technical effects: through real-time acquisition of multi-dimensional monitoring data (such as stress values, environmental data and construction loads) and edge computing processing, combined with the finite element model to generate a dynamic stress cloud map matrix and perform future stress prediction. By dynamically correcting the stress allowable threshold, comparing the real-time data with the predicted data, and calculating the early warning score, the bridge status can be monitored in real time and the early warning and maintenance control strategy can be triggered. This technical solution can accurately evaluate and warn the bridge during the construction process, greatly improve the safety of bridge construction, reduce the occurrence of potential risks, and ensure the smooth progress of the construction process. It has significant practical significance and application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a first flow chart of a bridge safety early warning method according to one embodiment of the present application.

[0024] Figure 2 It is a second flow chart of the bridge safety early warning method according to one embodiment of the present application.

[0025] Figure 3 It is a third flow chart of a bridge safety early warning method according to one embodiment of the present application.

[0026] Figure 4 It is a fourth flow chart of a bridge safety early warning method according to one embodiment of the present application.

[0027] Figure 5 It is a fifth flow chart of a bridge safety early warning method according to one embodiment of the present application.

[0028] Figure 6 It is a sixth flow chart of a bridge safety early warning method according to one embodiment of the present application. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solutions and advantages of this application more clear, the following Figure 1-6 It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0030] The embodiment of the present application discloses a bridge safety early warning method based on construction monitoring.

[0031] Reference Figure 1 , a bridge safety early warning method based on construction monitoring, the early warning method includes: Step S101, obtaining multi-dimensional monitoring data collected in real time by a sensor network deployed on the bridge main beam and key sections of the bridge; Among them, multi-dimensional monitoring data include real-time stress values, environmental monitoring data and construction loads; Specifically, during the construction of a bridge, the primary task to ensure structural safety is to obtain high-precision and high-frequency monitoring data. This embodiment uses multi-dimensional monitoring technology to deploy different types of sensors at key locations on the bridge to collect multiple data dimensions including stress, vibration, environment, and construction load. Specifically, fiber grating sensors are used to monitor stress changes on the bridge, and MEMS accelerometers are used to monitor vibration conditions in real time, especially dynamic responses caused by changes in the external environment (such as wind speed); environmental sensors are used to monitor meteorological factors such as temperature and wind speed, which have a very important impact on the bridge structure.

[0032] It should be noted that during the data collection process, multiple sensors need to be calibrated in space and time according to their positions and monitoring targets. Space and time calibration refers to synchronizing the spatial position and time of data collected by different sensors through algorithms to eliminate errors caused by layout or time differences. For example, fiber grating sensors are arranged on the main beam of the bridge, and MEMS accelerometers are arranged at the root of the pier. Therefore, the collected data of these sensors need to be aligned in space and time to ensure that accurate and comparable data are obtained at the same time point, thereby forming a monitoring network covering the entire bridge, reflecting the health status of the structure in real time, and ensuring that the accuracy of each monitoring data reaches the millimeter level.

[0033] Step S102, performing edge computing processing on the multi-dimensional monitoring data to obtain a cleaned monitoring data set; Specifically, in the bridge construction monitoring system, the quality of data is crucial, because the monitoring data is often disturbed by external noise, especially during the construction process, the vibration and noise generated by construction machinery (such as tower cranes, etc.) may pollute the monitoring data. Therefore, edge computing can be used to process and clean the data in real time. Edge computing is a method of processing data on computing devices near the data source (such as sensors), which can reduce data transmission while improving data processing efficiency.

[0034] In one of the embodiments of the present application, a Kalman filter algorithm can be used to remove interference signals caused by construction machinery (such as tower crane operation, concrete pouring, etc.). Kalman filtering is a recursive algorithm based on a state-space model. It can efficiently remove noise from data through a prediction and update process to ensure that the remaining data is more accurate. For abnormal data detection, an isolation forest algorithm can be used to identify abnormal points that deviate from the normal data distribution by constructing multiple random trees. These abnormal data will be automatically eliminated or compensated to avoid affecting the normal operation of the system.

[0035] Step S103, based on the cleaned monitoring data set, a three-dimensional dynamic stress field is constructed in combination with a finite element model to generate a dynamic stress cloud map matrix; In order to better understand the stress distribution of the bridge structure during the construction process, the embodiment of the present application constructs a theoretical stress field of the bridge through a finite element model (FEM). The finite element model is a numerical calculation method that simulates physical phenomena by decomposing a continuous medium into discrete units. In bridge construction, the finite element model can simulate the stress, deformation and other behaviors of the structure and provide theoretical stress distribution.

[0036] Specifically, the stress field data obtained through finite element analysis is combined with the stress data measured by the actual sensor, and the inverse distance weighted interpolation method (IDW) is used for data fusion. The inverse distance weighted interpolation method is an algorithm that performs weighted averaging based on the distance between data points. The closer the distance, the greater the weight of the data point. This method effectively fuses the theoretical stress field and the measured data, thereby dynamically correcting the parameters in the finite element model (such as elastic modulus E and Poisson's ratio ν) to generate dynamic stress cloud maps. These cloud maps can intuitively show the stress distribution of various parts of the bridge, especially during the construction process, where concentrated areas of local stress may become potential risk points.

[0037] Step S104, dynamically modifying the pre-configured stress allowable threshold according to the environmental monitoring data; Among them, the stress tolerance threshold of the bridge will be affected by external environmental factors, such as temperature and wind speed. For example, under extreme temperature conditions, the elastic modulus of the bridge material may change, resulting in fluctuations in the stress tolerance threshold. In order to make the stress threshold more in line with the actual situation, the embodiment of the present application adopts a dynamic correction method to adjust the stress tolerance threshold in real time according to environmental parameters.

[0038] Specifically, the system will dynamically adjust the preset stress threshold using a formula based on real-time monitored temperature and wind speed data. Changes in temperature and wind speed will affect the upper and lower limits of the stress threshold respectively. The higher the temperature, the lower the elastic modulus of the material, and the threshold will be adjusted down accordingly. Increased wind speed may cause greater dynamic loads, and the threshold needs to be adjusted down to cope with sudden wind loads. This dynamic correction formula can adjust the stress threshold in a timely manner according to different environmental factors, so that the system can adapt to different construction environments.

[0039] Step S105, predicting a stress prediction matrix in a future time window based on historical stress data and a dynamic stress cloud map matrix; Among them, the structural health management of the bridge should not only focus on the current state, but also predict the future stress change trend so as to intervene in advance. The embodiment of the present application uses LSTM (Long Short-Term Memory) network for stress prediction. LSTM is a neural network model that is particularly suitable for time series prediction and can process and predict data with long-term dependencies.

[0040] Specifically, in the present application, the input of the LSTM network includes historical stress data, stress gradient (from the dynamic stress cloud map matrix) and environmental changes (such as temperature change rate), and its output is the evolution trend of bridge stress in the future (such as 15 minutes). By training the LSTM model, the system can learn the law of stress change and predict future stress values ​​based on the input data. If the system predicts that the stress of a certain part will exceed the preset allowable threshold in the future, it will issue an early warning. This early warning mechanism can provide sufficient response time for bridge maintenance personnel and avoid the occurrence of emergencies.

[0041] Step S106, comparing the real-time stress value and the stress prediction matrix with the preset stress allowable threshold value, and calculating the early warning score; Specifically, in the early warning system, simple threshold judgment may have the risk of misjudgment and omission. By introducing the early warning scoring system, the early warning scoring quantifies the current safety status of the bridge by comprehensively considering the relationship between the real-time stress value and the predicted stress trend. The scoring model compares the real-time stress with the dynamically corrected stress threshold, and also considers the relationship between the predicted value and the threshold.

[0042] For example, when the real-time stress value or the predicted stress trend approaches or exceeds the allowable threshold, the score will increase accordingly. For example, when the score reaches a preset threshold (such as above 0.9), the system will trigger a red warning.

[0043] Step S107, determining the corresponding warning level according to the warning score and triggering the maintenance control strategy.

[0044] Among them, based on the pre-configured warning level mapping table, the corresponding score interval is found according to the warning score, the corresponding warning level is determined, and then the corresponding maintenance response strategy is triggered according to the warning level.

[0045] In some embodiments, for yellow warnings, the system will call on drones for inspections, equipped with 4K high-definition cameras and using the YOLOv5 algorithm for crack detection. These high-precision inspection equipment can identify cracks as small as 0.1mm; for red warnings, the system will link the hydraulic support system to automatically adjust the support force according to real-time load changes, while avoiding construction machinery to prevent collision accidents.

[0046] In the above implementation, a highly intelligent bridge safety monitoring and early warning system is constructed based on multi-dimensional sensing technology, dynamic modeling, data prediction and intelligent control. By collecting multi-dimensional monitoring data in real time and combining future trend prediction, the system can effectively identify and predict safety risks in bridge construction, greatly improving the safety of bridge construction. This technical solution can not only provide accurate real-time monitoring, but also dynamically adjust the early warning threshold according to environmental changes, reduce the occurrence of false alarms and missed alarms, and ultimately improve the safety and efficiency of bridge construction.

[0047] Reference Figure 2 As an implementation method of step S103, based on the cleaned monitoring data set, a three-dimensional dynamic stress field is constructed in combination with a finite element model, and the steps of generating a dynamic stress cloud map matrix include: Step S201, obtaining bridge design parameters and constructing an initial finite element model based on the cleaned monitoring data set; Among them, based on the basic parameters of bridge design (such as elastic modulus, geometric dimensions and cross-sectional properties of materials, etc.), combined with the cleaned monitoring data (including the layout position of sensors and historical stress data), the initial finite element model is constructed using parametric modeling tools (such as ANSYS, ABAQUS, etc.). In this process, the characteristics of the construction stage, such as the location of temporary supports and the age strength curve of concrete, need to be embedded in the model to ensure that the boundary conditions of the model are consistent with the actual construction sequence.

[0048] Step S202, dividing the finite element mesh nodes based on the initial finite element model and loading the construction phase boundary conditions, and simulating and generating a theoretical stress reference field matrix; The theoretical stress reference field matrix contains the theoretical stress components of each finite element mesh node.

[0049] Specifically, meshing is performed based on the initial finite element model, and an adaptive meshing algorithm is used to ensure that the mesh is refined to a smaller size (such as 0.5m) in areas with large stress gradients (such as the root of the bridge pier, anchorage area, etc.), while the mesh size can be relaxed to a larger value (such as 2m) in areas with gentle stress changes. Subsequently, loads during the construction phase, such as concrete pumping impact force, wind load spectrum, etc., are loaded, and static simulation is performed to calculate the theoretical stress component of each mesh node, thereby generating a theoretical stress reference field matrix.

[0050] Step S203, obtaining the spatial distribution coordinates of the sensor network, calculating the weight of each sensor corresponding node according to the finite element grid nodes, and constructing a mapping matrix; The spatial distribution coordinates of the sensor network are obtained, and the shape function interpolation principle is used to calculate the weight of the finite element grid node corresponding to each sensor. In this way, a sparse mapping matrix can be constructed, which describes the relationship between the stress data of the sensor and the stress data of the grid node. The weight coefficient can be used to realize the conversion between the discrete data collected by the sensor and the continuous field in the finite element model.

[0051] For example, assuming that a sensor is located at a certain position on the bridge, the data collected by the sensor is related to the stress of three adjacent nodes around its location, with weights of 0.6, 0.3, and 0.1 respectively. By calculating these weights, the measured stress data of the sensor can be converted into stress data of the grid nodes.

[0052] Step S204, dynamically correcting the initial finite element model parameters according to the real-time stress value and the environmental monitoring data to obtain a corrected finite element model; Among them, the accuracy of the model is improved by dynamically correcting the parameters of the finite element model, and the elastic modulus is adjusted using the temperature correction formula according to environmental monitoring data (such as temperature sensor data). For example, temperature changes will affect the elastic modulus of the material, so the elastic modulus is adjusted in real time according to temperature changes. At the same time, when the deviation between the measured stress and the theoretical value of certain finite element units continues to be large, the model parameters can be optimized online through data-driven optimization methods to minimize stress errors.

[0053] Step S205, based on the mapping matrix, weighted interpolation fusion is performed on the real-time stress value and the theoretical stress reference field matrix to generate a dynamic stress cloud map matrix.

[0054] Among them, the real-time stress data and the theoretical stress reference field matrix are weighted interpolated and fused using a mapping matrix. In some embodiments, the inverse distance weighted method (IDW) can be used to calculate the stress increment, where the weight coefficient changes with the construction stage. This method performs weighted adjustment based on the difference between the stress measured by the sensor and the theoretical stress field, thereby generating a dynamic stress cloud map matrix. These cloud map matrices can accurately display the stress distribution of various parts of the bridge and provide a basis for subsequent risk warnings.

[0055] In the above implementation, dynamic high-precision reconstruction and risk control of the stress field during the bridge construction period are achieved. By combining precise modeling with dynamic correction, the stress distribution of the bridge structure can be accurately predicted, and the model can adapt to environmental changes in real time, thereby facilitating real-time monitoring of safety risks during the construction process, early warning of potential dangers, and avoiding the occurrence of major safety accidents.

[0056] Reference Figure 3 As an implementation of step S104, the step of dynamically correcting the pre-configured stress allowable threshold according to the environmental monitoring data includes: Step S301, extracting real-time temperature data and wind speed data according to environmental monitoring data; Among them, the environmental monitoring system collects and extracts temperature and wind speed data from bridge monitoring points in real time. These data are usually collected through environmental sensors (such as temperature sensors, wind speed sensors, etc.). Real-time temperature data reflects the temperature changes in the environment around the bridge, while wind speed data reflects the impact of wind on the bridge structure. Temperature changes and wind speed are important environmental factors that affect the stress of bridge structures. Therefore, these two data are the basis for dynamically adjusting the allowable threshold of bridge stress.

[0057] Step S302, based on the pre-configured reference stress allowable threshold, dynamic correction is performed according to temperature data and wind speed data, and the correction formula is: ; Among them, σ 允许 is the allowable threshold value of modified stress, σ 基准 is the allowable threshold of the baseline stress, ΔT is the temperature change, V is the wind speed, β is the temperature compensation coefficient, and γ is the wind load reduction coefficient.

[0058] In the above implementation, the correction mechanism of the environmental compensation coefficient and the wind load reduction coefficient is combined to realize the dynamic correction of the stress allowable threshold. This technical solution not only takes into account the static stress threshold during design, but also can timely adapt to environmental changes during construction and later use. The corrected stress threshold can more accurately reflect the real safety status of the bridge under different environmental conditions, avoiding misjudgment or missed judgment due to environmental changes.

[0059] Reference Figure 4 As an implementation of step S105, the step of predicting the stress prediction matrix in the future time window based on the historical stress data and the dynamic stress cloud map matrix includes: Step S401, extracting stress gradient characteristics of each finite element mesh node according to the dynamic stress cloud map matrix, and constructing a node stress gradient matrix; Among them, the stress gradient reflects the rate and direction of stress change in different parts of the bridge, and is a key feature for understanding the health status of the bridge structure and predicting future stress changes. By calculating the stress gradient of each node in the stress field (i.e., the first-order partial derivative in three-dimensional space), the stress change trend of the bridge in different areas can be revealed. The node stress gradient matrix contains the stress gradient characteristics of each node, which is one of the inputs of the spatiotemporal prediction model.

[0060] Step S402, based on the historical stress data, construct a spatiotemporal prediction model according to the modified finite element model topological relationship and perform training to obtain a trained spatiotemporal prediction model; Among them, a spatiotemporal prediction model is constructed based on the historical stress data and the topological relationship of the revised finite element model. The revised finite element model includes possible changes (such as load, support position, etc.) during the bridge construction process.

[0061] In one embodiment of the present application, the core of the spatiotemporal prediction model is a hybrid architecture that combines a graph convolutional network (GCN) and a long short-term memory network (LSTM). GCN is used to capture the topological relationship of the bridge structure and the spatial association between nodes, while LSTM can process the time series features in the historical stress data and learn the time-varying laws of stress data. The trained spatiotemporal prediction model will be used to predict future stress trends.

[0062] For example, assuming that the historical stress data includes the stress time series data of the past month, the modified finite element model includes the boundary conditions and construction progress of the bridge structure. The connection relationship between the nodes of the bridge is modeled through GCN, and the time series data is learned through LSTM, and finally a spatiotemporal prediction model that can combine spatial and temporal information is obtained.

[0063] Step S403, obtaining the construction machinery positioning data on the bridge, adjusting the node weights of the local finite element mesh nodes, and weighting the node stress gradient matrix; Among them, construction machinery (such as tower cranes, concrete pump trucks, etc.) has a great impact on the stress of certain parts of the bridge. By obtaining the position of the construction machinery and weighting the finite element nodes within the influence range based on the position of the machinery, the weights of the nodes affected by the machinery are made higher. The weighted node stress gradient matrix will provide more accurate input data for the spatiotemporal prediction model.

[0064] Step S404: input the weighted node stress gradient matrix into the spatiotemporal prediction model to obtain the stress prediction matrix in the future time window.

[0065] Among them, the prediction matrix will contain the stress prediction value of each node in the future. By combining the local weighted matrix with the spatiotemporal prediction model, the actual stress state of the construction environment and structure can be fully considered to ensure the high accuracy of the prediction results.

[0066] In the above implementation, the dynamic stress cloud map matrix and historical data are combined, and the spatiotemporal feature fusion and dynamic modeling technology are used to achieve high-precision prediction of stress trends during bridge construction. By extracting node stress gradient characteristics, building a spatiotemporal prediction model, dynamically weighting the impact of construction machinery, and correcting thresholds in combination with actual environmental parameters, this technical solution effectively improves the risk assessment accuracy and early warning response capabilities during bridge construction.

[0067] Reference Figure 5 As an implementation method of step S402, based on the historical stress data, the spatiotemporal prediction model is constructed and trained according to the modified finite element model topological relationship, and the steps of obtaining the trained spatiotemporal prediction model include: Step S501, preprocessing the historical stress data based on the modified finite element model topological relationship to obtain a sample feature tensor; Among them, the modified finite element model topology is used to guide the data preprocessing process. Through the modified topological structure, the mapping relationship between the sensor and the bridge node can be established, and then the historical stress data can be processed accordingly. This processing includes resampling the data (for example, obtaining data once a minute) and interpolation filling (for example, using linear interpolation of adjacent nodes to fill missing data). Through these preprocessing operations, sample feature tensors are generated as input data for model training.

[0068] For example, suppose that there are multiple stress sensors on the bridge, and the data of these sensors are inconsistent in time and space. By using the modified finite element model topology, it is ensured that the historical stress data of each sensor can be matched with its corresponding node. Then, through resampling and interpolation processing, the missing data is filled, and finally a complete and standardized sample feature tensor is obtained.

[0069] Step S502, dividing the sample feature tensor into a training set and a test set; The processed sample feature tensor is divided into a training set and a test set. The training set is used to train the model, and the test set is used to evaluate the performance of the model. Usually, the ratio of the training set and the test set is set according to the specific application scenario, for example, 70% for training and 30% for testing. Through reasonable division, it can be ensured that the model is fully learned during the training process and its generalization ability can be evaluated on the test set.

[0070] Step S503, constructing a hybrid prediction model including a graph convolutional network and a bidirectional long short-term memory network; In one embodiment of the present application, the hybrid prediction model combines a graph convolutional network (GCN) and a bidirectional long short-term memory network (Bi-LSTM). GCN can use the topological information of the bridge structure to learn the relationship between nodes, while Bi-LSTM can process the time series characteristics in the historical stress data. Through the combination of the two, the model can consider both spatial and temporal factors, thereby achieving accurate prediction of the health status of the bridge structure.

[0071] Specifically, the graph convolutional network can help capture the spatial relationship between the nodes of the bridge structure and learn the connection pattern between the nodes, while the Bi-LSTM will perform time series analysis on the stress data of each node and learn the changing trend of stress. After combining the two, the model can be optimized in both space and time.

[0072] Step S504, inputting the training set into the hybrid prediction model for dynamic training, optimizing the model parameters, and obtaining a spatiotemporal prediction model; The training set is input into the hybrid prediction model, and the model parameters are optimized through dynamic training. The training process uses a standard optimization algorithm (such as the Adam optimizer) to minimize the loss function (such as the mean square error), thereby continuously adjusting the model parameters until the model can better fit the training set data.

[0073] Step S505, testing the spatiotemporal prediction model and calibrating model parameters based on the test set until the loss function of the model meets a preset condition or the number of model iterations reaches a preset number, thereby obtaining a trained spatiotemporal prediction model.

[0074] The spatiotemporal prediction model is evaluated through the test set, and the loss function (such as mean square error) on the test set is calculated. If the loss function of the model meets the preset conditions (for example, below a certain threshold), the training ends; if the loss function does not meet the conditions, the training continues by adjusting the model parameters until the stopping condition is met. For example, assuming that the initial value of the model's loss function is 0.1, the preset stopping condition is that the loss function is lower than 0.05. After multiple rounds of training, until the loss function drops below 0.05, the training process stops and the model is completed.

[0075] In the above implementation, based on the combination of graph convolutional network and bidirectional long short-term memory network, an efficient spatiotemporal prediction model is constructed, which can achieve high-precision stress trend prediction in bridge construction safety monitoring. The hybrid model architecture can not only integrate spatial and temporal features, but also improve the prediction ability of the model through dynamic training optimization, provide accurate prediction support in complex environments, and help improve the safety and efficiency of bridge construction and operation.

[0076] Reference Figure 6 As an implementation method of step S106, the real-time stress value and the stress prediction matrix are respectively compared with the preset stress allowable threshold value, and the step of calculating the early warning score includes: Step S601, using the real-time stress value as the first column of data of the stress prediction matrix to form a complete time window data matrix; The real-time stress value is used as the first column of the prediction matrix and is combined with the stress prediction matrix in the future time window to form a complete time window data matrix. The real-time stress value reflects the current state of the bridge structure, while the stress prediction matrix contains the predicted data of stress in the future. By combining the two into a complete matrix, it can be ensured that the model takes into account the relationship between the current stress state and future trends when calculating the score.

[0077] Step S602, based on the complete time window data matrix, expanding the preset stress allowable threshold into a stress threshold matrix with consistent dimensions; The preset stress allowable threshold is expanded into a stress threshold matrix that is consistent with the dimension of the complete time window data matrix. The preset stress allowable threshold is usually the safe stress value defined during bridge design, which represents the boundary of structural safety. By expanding the stress threshold matrix, it is ensured that when comparing the stress value at each moment, the threshold corresponds to the corresponding stress value.

[0078] Step S603, performing point-by-point comparison based on the complete time window data matrix and the stress threshold matrix to generate an over-limit marking matrix; Among them, the system compares each element in the complete time window data matrix (i.e., real-time stress value and predicted stress value) with the corresponding value in the stress threshold matrix one by one. When the stress value exceeds the preset threshold, it is marked as "overlimit". Multi-level marking rules can be used here: for example, real-time stress value overlimit is marked as "2" (emergency overlimit), predicted value overlimit is marked as "1" (warning overlimit), otherwise it is marked as "0" (normal state). In this way, we get an overlimit marking matrix that reflects the safety status of the bridge at each time point.

[0079] Step S604: weighted calculation is performed on the over-limit marking matrix, and a warning score is obtained in combination with the time decay coefficient.

[0080] Among them, the overlimit marking matrix includes real-time overlimit marking and predicted value overlimit marking. The generated overlimit marking matrix is ​​weighted. For the overlimit marking of the real-time stress value, a higher weight (such as 3 times) is given, because the real-time stress value represents the actual state of the current bridge and has a more direct impact on safety. The predicted stress value is given a lower weight (such as 1 times). In addition, the time decay coefficient is introduced, and the impact of future predicted values ​​gradually weakens over time. Finally, all weighted results are combined to calculate a comprehensive early warning score.

[0081] In the above implementation, the weighted scoring mechanism combining real-time stress values ​​and predicted stress trends can accurately reflect the current safety status of the bridge and potential future risks. By gradually comparing the real-time stress value, predicted stress value and the corrected stress threshold, and combining multi-level over-limit marking, weighted calculation and time attenuation coefficient, the system can provide more detailed and accurate early warning scores. This technical method not only improves the accuracy of stress prediction, but also can identify potential risks in advance, improve the real-time and accuracy of bridge safety monitoring, and provide sufficient time and decision-making basis for emergency response.

[0082] The embodiment of the present application also discloses a bridge safety early warning system based on construction monitoring.

[0083] A bridge safety early warning system based on construction monitoring, the early warning system includes: An acquisition module is used to acquire multi-dimensional monitoring data collected in real time by a sensor network deployed on the bridge main beam and key sections of the bridge; the multi-dimensional monitoring data includes real-time stress values, environmental monitoring data and construction loads; The data processing module is used to perform edge computing on the multi-dimensional monitoring data to obtain a cleaned monitoring data set; The cloud map matrix generation module is used to construct a three-dimensional dynamic stress field based on the cleaned monitoring data set and the finite element model to generate a dynamic stress cloud map matrix; A dynamic correction module is used to dynamically correct the pre-configured stress allowable threshold according to environmental monitoring data; Stress prediction module, used to predict the stress prediction matrix in the future time window based on historical stress data and dynamic stress cloud map matrix; The scoring module is used to compare the real-time stress value and the stress prediction matrix with the preset stress allowable threshold value and calculate the early warning score; The early warning module is used to determine the corresponding early warning level according to the early warning score and trigger the maintenance control strategy.

[0084] As an implementation of the cloud map matrix generation module, it includes: A model building unit, used to obtain bridge design parameters and build an initial finite element model based on the cleaned monitoring data set; The simulation unit is used to divide the finite element mesh nodes based on the initial finite element model and load the boundary conditions of the construction stage, and simulate and generate the theoretical stress reference field matrix; A mapping matrix construction unit is used to obtain the spatial distribution coordinates of the sensor network, calculate the weight of each sensor corresponding node according to the finite element grid node, and construct a mapping matrix; A dynamic correction unit is used to dynamically correct the initial finite element model parameters according to the real-time stress value and environmental monitoring data to obtain a corrected finite element model; The interpolation fusion unit is used to perform weighted interpolation fusion of the real-time stress value and the theoretical stress reference field matrix based on the mapping matrix to generate a dynamic stress cloud map matrix.

[0085] A bridge safety warning system based on construction monitoring in an embodiment of the present application can implement any of the above-mentioned bridge safety warning methods, and the specific working process of each module in the bridge safety warning system can refer to the corresponding process in the above-mentioned method embodiment.

[0086] In the several embodiments provided in this application, it should be understood that the provided methods and systems can be implemented in other ways. For example, the system embodiments described above are only illustrative; for example, the division of a certain module is only a logical function division, and there may be other division methods in actual implementation, such as multiple modules can be combined or integrated into another system, or some features can be ignored or not executed.

[0087] The embodiment of the present application also discloses a computer device.

[0088] The computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, a bridge safety early warning method based on construction monitoring as described above is implemented.

[0089] The embodiment of the present application also discloses a computer-readable storage medium.

[0090] A computer-readable storage medium stores a computer program that can be loaded by a processor and execute any one of the above-mentioned bridge safety early warning methods based on construction monitoring.

[0091] Among them, computer-readable storage media can be any tangible medium that contains or stores a program, which can be used by or in conjunction with an instruction execution system, apparatus or device; the program code contained on the computer-readable medium can be transmitted using any appropriate medium, including but not limited to wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0092] It should be noted that in the above embodiments, the description of each embodiment has different emphases, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0093] The above are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Any feature disclosed in this specification (including the abstract and drawings), unless otherwise stated, can be replaced by other equivalent or alternative features with similar purposes. That is, unless otherwise stated, each feature is only an example of a series of equivalent or similar features.

Claims

1. A bridge safety early warning method based on construction monitoring, characterized in that: The early warning method comprises: Acquire multi-dimensional monitoring data collected in real time by a sensor network deployed on the bridge main beam and key sections of the bridge; the multi-dimensional monitoring data includes real-time stress values, environmental monitoring data and construction loads; Performing edge computing processing on the multidimensional monitoring data to obtain a cleaned monitoring data set; Based on the cleaned monitoring data set, a three-dimensional dynamic stress field is constructed in combination with the finite element model to generate a dynamic stress cloud map matrix; Dynamically modifying a pre-configured stress allowable threshold value according to the environmental monitoring data; Predicting a stress prediction matrix within a future time window based on historical stress data and the dynamic stress cloud map matrix; Comparing the real-time stress value and the stress prediction matrix with the preset stress allowable threshold value respectively, and calculating the early warning score; The corresponding warning level is determined according to the warning score and a maintenance control strategy is triggered.

2. A bridge safety early warning method based on construction monitoring according to claim 1, characterized in that: Based on the cleaned monitoring data set, the three-dimensional dynamic stress field is constructed in combination with the finite element model, and the steps of generating the dynamic stress cloud map matrix include: Obtain bridge design parameters and build an initial finite element model based on the cleaned monitoring data set; Divide the finite element mesh nodes based on the initial finite element model and load the construction phase boundary conditions to simulate and generate a theoretical stress reference field matrix; Obtaining the spatial distribution coordinates of the sensor network, calculating the weight of each sensor corresponding node according to the finite element grid nodes, and constructing a mapping matrix; Dynamically correct the initial finite element model parameters according to the real-time stress value and the environmental monitoring data to obtain a corrected finite element model; Based on the mapping matrix, the real-time stress value is fused with the theoretical stress reference field matrix by weighted interpolation to generate a dynamic stress cloud map matrix.

3. A bridge safety early warning method based on construction monitoring according to claim 1, characterized in that: The step of dynamically correcting the pre-configured stress allowable threshold according to the environmental monitoring data includes: Extracting real-time temperature data and wind speed data according to the environmental monitoring data; Based on the pre-configured reference stress allowable threshold, dynamic correction is performed according to the temperature data and wind speed data, and the correction formula is: ； Among them, σ 允许 is the allowable threshold value of modified stress, σ 基准 is the allowable threshold of the baseline stress, ΔT is the temperature change, V is the wind speed, β is the temperature compensation coefficient, and γ is the wind load reduction coefficient.

4. A bridge safety early warning method based on construction monitoring according to claim 2, characterized in that: The step of predicting the stress prediction matrix in the future time window based on the historical stress data and the dynamic stress cloud map matrix includes: Extracting stress gradient characteristics of each finite element mesh node according to the dynamic stress cloud map matrix, and constructing a node stress gradient matrix; Based on the historical stress data, a spatiotemporal prediction model is constructed and trained according to the modified finite element model topological relationship to obtain a trained spatiotemporal prediction model; Acquire construction machinery positioning data on the bridge, adjust node weights of local finite element mesh nodes, and weight the node stress gradient matrix; The weighted node stress gradient matrix is ​​input into the spatiotemporal prediction model to obtain a stress prediction matrix in a future time window.

5. A bridge safety early warning method based on construction monitoring according to claim 4, characterized in that: Based on the historical stress data, a spatiotemporal prediction model is constructed and trained according to the modified finite element model topological relationship, and the steps of obtaining the trained spatiotemporal prediction model include: Based on the modified finite element model topological relationship, preprocessing the historical stress data to obtain a sample characteristic tensor; Dividing the sample feature tensor into a training set and a test set; Construct a hybrid prediction model consisting of a graph convolutional network and a bidirectional long short-term memory network; Inputting the training set into the hybrid prediction model for dynamic training, optimizing model parameters, and obtaining a spatiotemporal prediction model; The spatiotemporal prediction model is tested and model parameters are corrected based on the test set until the loss function of the model meets a preset condition or the number of model iterations reaches a preset number, thereby obtaining the trained spatiotemporal prediction model.

6. A bridge safety early warning method based on construction monitoring according to any one of claims 1 to 5, characterized in that: The steps of comparing the real-time stress value and the stress prediction matrix with the preset stress allowable thresholds and calculating the early warning score include: The real-time stress value is used as the first column of data of the stress prediction matrix to form a complete time window data matrix; Based on the complete time window data matrix, the preset stress allowable threshold is expanded into a stress threshold matrix with consistent dimension; Performing point-by-point comparison between the complete time window data matrix and the stress threshold matrix to generate an over-limit marking matrix; A weighted calculation is performed on the over-limit marking matrix, and a warning score is obtained in combination with a time decay coefficient.

7. A bridge safety early warning system based on construction monitoring, characterized in that: The early warning system includes: An acquisition module is used to acquire multi-dimensional monitoring data collected in real time by a sensor network deployed on the bridge main beam and key sections of the bridge; the multi-dimensional monitoring data includes real-time stress values, environmental monitoring data and construction loads; A data processing module, used to perform edge computing processing on the multi-dimensional monitoring data to obtain a cleaned monitoring data set; The cloud map matrix generation module is used to construct a three-dimensional dynamic stress field based on the cleaned monitoring data set and the finite element model to generate a dynamic stress cloud map matrix; A dynamic correction module, used for dynamically correcting the pre-configured stress allowable threshold according to the environmental monitoring data; A stress prediction module, used for predicting a stress prediction matrix in a future time window based on historical stress data and the dynamic stress cloud map matrix; A scoring module, used to compare the real-time stress value and the stress prediction matrix with the preset stress allowable threshold value, and calculate the early warning score; The early warning module is used to determine the corresponding early warning level according to the early warning score and trigger the maintenance control strategy.

8. A bridge safety early warning system based on construction monitoring according to claim 7, characterized in that: The cloud map matrix generation module includes: A model building unit, used to obtain bridge design parameters and build an initial finite element model based on the cleaned monitoring data set; A simulation unit, used for dividing finite element mesh nodes based on the initial finite element model and loading construction phase boundary conditions, so as to simulate and generate a theoretical stress reference field matrix; A mapping matrix construction unit, used to obtain the spatial distribution coordinates of the sensor network, calculate the weight of each sensor corresponding node according to the finite element grid node, and construct a mapping matrix; A dynamic correction unit, used to dynamically correct the initial finite element model parameters according to the real-time stress value and environmental monitoring data to obtain a corrected finite element model; The interpolation fusion unit is used to perform weighted interpolation fusion on the real-time stress value and the theoretical stress reference field matrix based on the mapping matrix to generate a dynamic stress cloud map matrix.

9. A computer device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 6 is implemented.

10. A computer-readable storage medium, characterized in that: A computer program is stored which can be loaded by a processor and execute the method according to any one of claims 1 to 6.

Images