Structure monitoring system and method for civil engineering

Abstract

The invention belongs to the technical field of structure monitoring systems, and discloses a structure monitoring system and method for civil engineering, and the method comprises the steps that a data collection module collects structure data; the cruise monitoring mechanism collects a high-definition image, an infrared image and laser point cloud data; the environment data acquisition module acquires environment data; the image analysis module identifies abnormal conditions; the infrared data analysis module detects internal defects; the laser data analysis module accurately measures the deformation of the structure; the comprehensive analysis module performs comprehensive evaluation; the risk prediction module fuses multi-source data and predicts potential problems and risks; when an abnormal condition or a potential problem risk is monitored, the alarm module gives an alarm in time. According to the invention, through combination of sensor monitoring and unmanned aerial vehicle cruise monitoring, the structure of civil engineering is monitored in an omnibearing manner; the comprehensive analysis module is used for integrating the multi-source monitoring data to carry out comprehensive evaluation and abnormity judgment; and the risk prediction module fuses monitoring results of multiple modules and predicts potential problems and risks.

Description

Technical Field

[0001] This application relates to the technical field of structural monitoring systems, and more specifically, to a structural monitoring system and method for civil engineering. Background Art

[0002] The monitoring of civil engineering structures involves knowledge and technologies in multiple disciplinary fields. By monitoring civil engineering structures in real time, key data such as stress, strain, displacement, and vibration of the structures can be obtained in a timely manner. For example, during the construction and use of large bridges, the monitoring system can capture the deformation of the bridge under the action of vehicle loads, wind loads, etc. in real time. Once problems such as abnormal deformation or stress concentration are detected, early warnings can be issued in a timely manner so that corresponding measures can be taken to avoid serious damage or even collapse accidents of the structure. Long-term monitoring data can help engineers accurately evaluate the durability and remaining service life of the structure. Taking high-rise buildings as an example, by monitoring the carbonation degree of concrete, the corrosion situation of steel bars, etc., the performance degradation law of structural materials can be understood, and maintenance and reinforcement plans can be formulated in advance to ensure the safety and reliability of the structure within the designed service life.

[0003] The prior art document with the publication number CN213365020U provides a real-time monitoring device for civil engineering buildings of building structures, belonging to the technical field of building monitoring. It includes a mounting frame, a receiving antenna, a transmitting antenna, a receiver, a transmitter, a fixing frame, a main control unit, and an acoustic-optic alarm module. The bottom of the mounting frame is inserted into the top of the fixing frame, and the receiving antenna is arranged on the upper surface of the mounting frame. This utility model is installed on one side of a load-bearing column or load-bearing beam through the mounting frame, with simple installation and no damage to the building's main body structure. The transmitter will emit radar waves into the building structure through the transmitting antenna, and the radar waves reflected back from inside the building are received by the receiving antenna and transmitted to the receiver. The receiver then feeds back the processed radar wave data to the main control unit to monitor the inside of the building in real time. When abnormal situations such as concrete cracking and steel bar fracture are detected, the detailed positions of the abnormal points will be recorded by a marker, and the acoustic-optic alarm module will issue an alarm.

[0004] Although the existing technical solutions mentioned above can achieve relevant beneficial effects through the structures of the existing technologies, they still have the following defects: 1. Most of the existing technologies rely on ground manual inspections and limited fixed sensor layouts. When facing complex terrains such as mountain bridges and deep foundation pits, or civil engineering projects with complex structures such as large venues, it is difficult for humans to reach some areas, and fixed sensors cannot cover comprehensively, resulting in blind spots in monitoring. 2. Existing monitoring technologies often can only obtain data from a single perspective. For example, only relying on ground cameras to monitor the surface of the structure, the assessment of the overall health status of the structure is not comprehensive enough. Most existing technologies only analyze single-type data. For example, only judging the structural health status based on structural deformation data, resulting in low accuracy of abnormal judgment. 3. It is impossible to comprehensively process multi-source data in a timely manner to predict potential problems and risks.

[0005] In view of this, we propose a structural monitoring system and method for civil engineering. Summary of the Invention

[0006] 1. Technical problems to be solved

[0007] The purpose of this application is to provide a structural monitoring system and method for civil engineering, which solves the technical problems raised in the above background technology, realizes the combination of sensor monitoring and drone cruise monitoring, and achieves all-round monitoring of the structures of civil engineering; the comprehensive analysis module integrates multi-source monitoring data to comprehensively evaluate the structural conditions and make abnormal judgments; the risk prediction module fuses the monitoring results of multiple modules and historical data to predict potential problems and risks of the structure.

[0008] 2. Technical solutions

[0009] The technical solution of this application provides a structural monitoring system for civil engineering, including:

[0010] Data collection module: Collect a large amount of data of civil engineering structures (including geological data, drawings, construction data, maintenance data, etc.);

[0011] Structural data acquisition module: Reasonably arrange several sensors on the civil engineering structure, including strain sensors, displacement sensors, vibration sensors, temperature sensors, etc., and evenly deploy them at key parts of the civil engineering structure to collect data such as stress and strain, displacement, vibration, and temperature of the structure in real time.

[0012] Route planning module: Plan the optimal flight route of the drone according to the civil engineering structure data and the parameter data of the drone;

[0013] Cruise monitoring mechanism: including a drone, which is equipped with a high-definition camera, a laser scanner, and an infrared thermal imager on the drone to collect high-definition images, infrared images, and laser point cloud data of the civil engineering structure;

[0014] Environmental data acquisition module: It acquires environmental data in real time, including temperature, humidity, wind direction and speed, earthquake, rainfall, etc.;

[0015] Image analysis module: It preprocesses, extracts features and conducts recognition and analysis on the acquired images, and promptly identifies abnormal situations; it analyzes diseases such as cracks and spalling on the surface of the structure through image recognition technology;

[0016] Infrared data analysis module: It analyzes the acquired infrared data and promptly identifies abnormal situations; an infrared thermal imager obtains the temperature distribution of the structure to detect internal defects;

[0017] Laser data analysis module: It acquires the laser point cloud data of civil engineering structures in real time; a laser scanner generates a three-dimensional point cloud model of the structure to accurately measure the deformation of the structure.

[0018] Comprehensive analysis module: It combines the monitoring results of the structure data acquisition module, environmental data acquisition module, laser data analysis module, image analysis module and infrared data analysis module to comprehensively evaluate the situation of civil engineering structures and comprehensively judge abnormal situations;

[0019] Risk prediction module: It integrates the monitoring results of the structure data acquisition module, environmental data acquisition module, laser data analysis module, image analysis module and infrared data analysis module, and combines historical data to predict potential problems and risks of civil engineering structures.

[0020] Alarm module: It includes an alarm. When abnormal situations or potential problem risks are detected, it promptly issues an alarm.

[0021] Central control unit: It is network-connected to the data collection module, structure data acquisition module, route planning module, cruise monitoring mechanism, laser data analysis module, environmental data acquisition module, image analysis module, comprehensive analysis module, risk prediction module and alarm module.

[0022] As an optional solution of the present invention, the comprehensive analysis module combines the monitoring results of the structure data acquisition module, environmental data acquisition module, laser data analysis module, image analysis module and infrared data analysis module to comprehensively evaluate the situation of civil engineering structures and comprehensively judge abnormal situations; it includes the following steps:

[0023] 1. Multi-source data collection: Obtain data such as stress and strain, displacement, and vibration of civil engineering structures from the structural data acquisition module. These data reflect the mechanical properties of the structure itself. Collect environmental parameters such as temperature, humidity, wind direction and speed, earthquake, and rainfall provided by the environmental data acquisition module. Extract the 3D point cloud model data of the structure and the deformation measurement results from the laser data analysis module for an intuitive understanding of the structural shape changes. Collect disease information such as surface cracks and spalling of the structure identified by the image analysis module, as well as abnormal internal temperature distribution and defects detected by the infrared data analysis module. Perform format conversion and standardization processing on the collected data to ensure smooth circulation and processing in the comprehensive analysis module.

[0024] 2. Data correlation analysis: including the correlation between structure and environment data and cross-validation of multi-module data;

[0025] Correlation between structure and environment data: Study the influence of environmental factors on the performance of civil engineering structures. Analyze the relationship between temperature changes and the thermal expansion and contraction deformation of the structure, and view the change trends of structural displacement, stress, and strain when the temperature rises or falls. Explore the influence of humidity on the performance of structural materials, such as whether the durability of concrete structures decreases with the increase in humidity, and judge by comparing the changes in mechanical parameters of the structure under different humidity conditions. Analyze the effects of wind direction and speed on structures such as bridges and high-rise buildings, and combine the vibration data of the structure to judge whether the response of the structure is within the normal range under different wind force levels.

[0026] Cross-validation of multi-module data: Use the structural deformation data of the laser data analysis module to verify whether the crack and spalling areas detected by the image analysis module are related to the abnormally deformed parts of the structure. For example, if a crack is found in a certain area by the image analysis, check the deformation of this area in the laser point cloud model to see if there are sudden displacement changes or excessive local deformations. Combine the internal defect information detected by the infrared data analysis module with the stress and strain data of the structural data acquisition module to judge whether the internal defects lead to a decline in the mechanical performance of the structure, such as whether the stress concentration near internal cavities exceeds the normal range.

[0027] 3. Build an evaluation model: Build a comprehensive evaluation index system for civil engineering structures, covering mechanical performance indicators of the structure (such as the maximum, average, and change rate of stress, strain, and displacement), appearance disease indicators (length, width, and density of cracks, area and location of spalling), internal defect indicators (type, size, and location of defects), and environmental impact indicators (degree and duration of environmental parameters exceeding the normal range), etc. Determine the corresponding weights using methods such as the Analytic Hierarchy Process (AHP) according to the importance of each indicator to the structural safety.

[0028] Select the support vector machine (SVM) model of machine learning, and use historical monitoring data and known structural health status labels to train the evaluation model. During the training process, continuously adjust the model parameters to enable the model to accurately evaluate the health status of the structure according to the input multi-source data and output the health level of the structure (such as healthy, sub-healthy, dangerous, etc.).

[0029] 4. Abnormality judgment: Set reasonable warning thresholds for each evaluation index according to the design standards, historical monitoring data, and relevant specifications of the structure. Compare the results obtained by calculating the real-time monitoring data through the evaluation model with the thresholds. For example, if the calculation result of the structural displacement exceeds the maximum displacement threshold allowed by the design, or the evaluation value of the crack width is greater than the preset dangerous crack width threshold, it is initially judged that the structure may have abnormal conditions.

[0030] 5. Comprehensive abnormality judgment: Comprehensively consider the changes of multiple indicators. When multiple indicators show abnormalities at the same time, or the abnormality degree of key indicators reaches a certain level, it is determined that the structure has abnormal conditions. For example, multiple indicators such as structural displacement, stress and strain, and crack width all exceed the normal range, and infrared data analysis also detects serious internal defects. It is comprehensively judged that the structure is in a dangerous state.

[0031] 6. Report generation: Present the comprehensive evaluation results through visual means such as charts and graphs. Draw the deformation nephogram of the structure to intuitively show the deformation size and distribution of different parts; use bar charts to compare the actual values and thresholds of each evaluation index; use time series charts to show the change trend of the structural health level over time. Use a 3D model to show the appearance diseases and internal defect locations of the structure to make the evaluation results more intuitive and understandable. Generate a detailed structural comprehensive evaluation report, and the report content includes the evaluation time, basic information of the structure, overview of the monitoring data of each module, calculation results of the evaluation index, evaluation conclusion of the structural health level, description of abnormal conditions, analysis of abnormal reasons, and corresponding treatment suggestions.

[0032] As an alternative solution of the present invention, the risk prediction module integrates multi-source monitoring data and combines historical data, and uses specific analysis methods to predict potential risks, including the following steps:

[0033] 1. Data collection: Collect real-time data such as stress and strain, displacement, and vibration of civil engineering structures from the structural data acquisition module. Obtain real-time environmental parameters such as temperature, humidity, wind direction and speed, earthquake, and rainfall from the environmental data acquisition module. Extract the three-dimensional point cloud model of the structure and the real-time results of deformation measurement from the laser data analysis module. Collect real-time information on diseases such as surface cracks and spalling of the structure identified by the image analysis module, and the real-time situation of abnormal internal temperature distribution and defects of the structure detected by the infrared data analysis module. Integrate historical data, organize the monitoring data of each module over a past period of time, including structural mechanical property data, environmental data, laser scanning data, image data, and infrared data, etc. Sort the historical data in chronological order to ensure the continuity and integrity of the data.

[0034] 2. Data preprocessing: Clean the collected real-time and historical data to remove obviously incorrect, missing, or duplicate data.

[0035] 3. Time series analysis:

[0036] 3.1 Trend analysis: Use the moving average method to process the monitoring data of the structure. By plotting the time series graph, visually observe the changing trends of data such as structural displacement, stress and strain over time, judge whether there is a gradually increasing or decreasing trend in the structure, and the speed of trend change.

[0037] 3.2 Seasonal analysis: Analyze whether there are seasonal patterns in environmental data (such as temperature, humidity, etc.) and some structural data.

[0038] 3.3 Build a prediction model: Adopt time series prediction models such as autoregressive integrated moving average model (ARIMA) to predict the changes in structural monitoring data over a future period of time based on historical data. Determine the parameters of the model, and continuously adjust the parameters through model training and validation to improve the prediction accuracy.

[0039] 4. Feature engineering: Extract valuable features from multi-source data, such as the change rate of structural displacement, the growth rate of crack width, the extreme values of environmental parameters, etc. Combine these features into feature vectors as the input of the machine learning model. At the same time, screen the features to remove features with too high correlation or small contribution to the prediction result to improve the efficiency and accuracy of the model.

[0040] 5. Model training: Select machine learning models suitable for risk prediction, such as decision trees, random forests, support vector regression, etc. Use historical monitoring data and the corresponding structural health status labels to train the model. During the training process, optimize the model parameters through methods such as cross-validation to improve the generalization ability of the model.

[0041] 6. Model Evaluation and Optimization: Evaluate the trained machine learning model using metrics such as accuracy, recall, and root mean square error. If the model performance does not meet the requirements, optimize the model.

[0042] 7. Risk Assessment and Prediction: Determine the indicators for evaluating the structural risk according to the design standards, relevant specifications, and expert experience of the structure, such as the degree of structural displacement exceeding the allowable value, the possibility of crack width reaching the dangerous standard, and the rate of internal defect expansion. Set different risk levels for each risk indicator, such as low risk, medium risk, and high risk. Use time series analysis and the prediction results of machine learning models, combined with risk indicators and levels, to predict and classify the potential problems and risks of the structure in the future.

[0043] 8. Risk Visualization: Display the risk prediction results of the structure through visual methods such as charts and graphs, such as drawing a risk heat map, where different colors represent different risk levels, and intuitively present the risk distribution of different parts of the structure.

[0044] The present invention provides a structural monitoring method for civil engineering, including the following steps:

[0045] S1. The data collection module collects a large amount of data of civil engineering structures (including geological data, drawings, construction data, maintenance data, etc.);

[0046] S2. The structural data acquisition module reasonably arranges several sensors on the civil engineering structure to collect data such as stress and strain, displacement, vibration, and temperature of the structure in real time.

[0047] S3. The route planning module plans the optimal flight route of the unmanned aerial vehicle according to the civil engineering structure data and the parameter data of the unmanned aerial vehicle;

[0048] S4. The cruise monitoring mechanism collects high-definition images, infrared images, and laser point cloud data of the civil engineering structure according to the planned route; the environmental data acquisition module collects environmental data in real time, including temperature, humidity, wind direction and speed, earthquake, and rainfall;

[0049] S5. The image analysis module preprocesses, extracts features, and performs recognition and analysis on the collected images, and promptly identifies abnormal situations; analyzes diseases such as cracks and spalling on the surface of the structure through image recognition technology;

[0050] S6. The infrared data analysis module analyzes the collected infrared data and promptly identifies abnormal situations; detects internal defects;

[0051] S7. The laser data analysis module collects the laser point cloud data of the civil engineering structure in real time; the laser scanner generates a three-dimensional point cloud model of the structure to accurately measure the deformation of the structure.

[0052] S8. The comprehensive analysis module combines the monitoring results of the structural data acquisition module, the environmental data acquisition module, the laser data analysis module, the image analysis module, and the infrared data analysis module to comprehensively evaluate the condition of the civil engineering structure and comprehensively judge abnormal conditions.

[0053] S9. The risk prediction module integrates the monitoring results of the structural data acquisition module, the environmental data acquisition module, the laser data analysis module, the image analysis module, and the infrared data analysis module, and combines historical data to predict potential problems and risks of the civil engineering structure; uses methods such as time series analysis and machine learning to perform trend prediction on the monitoring data of the structure and anticipate the future development and changes of the structure in advance.

[0054] S10. When abnormal conditions or potential problem risks are detected, the alarm module issues an alarm in a timely manner to remind the user to take corresponding measures to avoid the occurrence of structural safety accidents.

[0055] 3. Beneficial Effects

[0056] One or more technical solutions provided in the technical solution of the present application have at least the following technical effects or advantages:

[0057] 1. The present invention combines sensor monitoring and drone cruise monitoring to achieve comprehensive monitoring of the civil engineering structure; the drone is equipped with a variety of devices to collect high-definition images, infrared images, and laser point cloud data. The high-definition camera can clearly capture subtle diseases on the surface of the structure; the infrared thermal imager detects internal defects; the laser scanner generates a three-dimensional point cloud model to accurately measure deformations, realizing comprehensive and multi-angle monitoring of the structure, and is not limited by terrain and structural complexity, and can cover areas that are difficult for humans to reach.

[0058] 2. The comprehensive analysis module integrates multi-source monitoring data to comprehensively evaluate the structure condition and judge abnormalities. By combining structural data with environmental data, analyzing the influence of environmental factors on the structure, and combining image, infrared, and laser data, the health status of the structure can be judged from different angles, improving the accuracy of abnormal judgment.

[0059] 3. The risk prediction module uses methods such as time series analysis and machine learning to integrate the monitoring results of multiple modules and historical data to predict potential problems and risks of the structure. Anticipate the future development and changes of the structure in advance, providing decision support for long-term management, such as predicting the parts and time when diseases may occur in the structure in the next few years, facilitating the advance planning of maintenance work, and reducing maintenance costs and safety risks. Description of the Drawings

[0060] Figure 1 It is a schematic flowchart of the structural monitoring method for civil engineering disclosed in the present application. Detailed Embodiments

[0061] The present application will be further described in detail below in conjunction with the accompanying drawings of the specification.

[0062] Referring to Figure 1 , an embodiment of the present application provides a structural monitoring system for civil engineering, including:

[0063] Data collection module: Collect a large amount of data of civil engineering structures (including geological data, drawings, construction data, maintenance data, etc.); Obtain a detailed geological exploration report covering information such as soil layer distribution, geotechnical mechanical parameters, and groundwater level. Using Geographic Information System (GIS) technology, digitize and visually display the geological data for convenient subsequent analysis in combination with civil engineering structures. Collect design drawings, construction drawings, and as-built drawings of civil engineering structures, etc. Use professional drawing scanning and recognition software to convert paper drawings into electronic formats, and use Computer-Aided Design (CAD) technology to vectorize the drawings. By establishing a drawing database, achieve efficient management and rapid query of drawings, facilitating comparison with drawing information at any time during the monitoring process to analyze the differences between the design intent and actual status of the structure. Obtain various data during the construction process from the construction unit, including construction progress, material usage records, concrete pouring temperature, prestress application conditions, etc. Use project management software to organize and analyze the construction data to form a digital archive of the construction process. For example, by analyzing the concrete pouring temperature data, the temperature change situation of the concrete during the solidification process can be understood to determine whether structural cracks may be caused by temperature stress. Establish a maintenance data management platform to detail the maintenance conditions of civil engineering structures during use, including information such as maintenance time, maintenance location, maintenance content, and maintenance personnel. Classify, statistically analyze, and summarize the maintenance data to identify the parts and types where the structure is prone to problems, providing a reference basis for subsequent monitoring focuses.

[0064] Structural data acquisition module: Reasonably deploy several sensors on civil engineering structures, including strain sensors, displacement sensors, vibration sensors, temperature sensors, etc., and evenly deploy them at key parts of civil engineering structures, such as bridge piers and beams, and building beam-column joints, etc., to collect data such as stress and strain, displacement, vibration, and temperature of the structure in real time.

[0065] Route planning module: Plan the optimal flight route of the unmanned aerial vehicle according to the civil engineering structure data and the parameter data of the unmanned aerial vehicle;

[0066] Cruise monitoring mechanism: Includes an unmanned aerial vehicle, which is equipped with a high-definition camera, a laser scanner, and an infrared thermal imager on the unmanned aerial vehicle to collect high-definition images, infrared images, and laser point cloud data of civil engineering structures;

[0067] Environmental data acquisition module: Real-time acquisition of environmental data, including temperature, humidity, wind direction and speed, earthquake, rainfall, etc.;

[0068] Image analysis module: Preprocess, extract features and perform recognition and analysis on the collected images, and promptly identify abnormal situations; Analyze diseases such as cracks and spalling on the surface of the structure through image recognition technology;

[0069] Infrared data analysis module: Analyze the collected infrared data and promptly identify abnormal situations; The infrared thermal imager obtains the temperature distribution of the structure and detects internal defects;

[0070] Laser data analysis module: Real-time acquisition of laser point cloud data of civil engineering structures; The laser scanner generates a three-dimensional point cloud model of the structure and accurately measures the deformation of the structure.

[0071] Comprehensive analysis module: Combine the monitoring results of the structure data acquisition module, environmental data acquisition module, laser data analysis module, image analysis module and infrared data analysis module to comprehensively evaluate the situation of civil engineering structures and comprehensively judge abnormal situations;

[0072] Risk prediction module: Integrate the monitoring results of the structure data acquisition module, environmental data acquisition module, laser data analysis module, image analysis module and infrared data analysis module, and combine historical data to predict potential problems and risks of civil engineering structures; Use methods such as time series analysis and machine learning to predict the trend of the monitoring data of the structure, anticipate the future development and changes of the structure in advance, and provide decision-making support for the long-term management of the structure.

[0073] Alarm module: Includes an alarm. When abnormal situations or potential problem risks are detected, an alarm is promptly issued. Set reasonable early warning thresholds according to the design parameters of the structure and historical monitoring data. When the monitoring data exceeds the threshold, the system immediately issues an early warning signal to remind the user to take corresponding measures to avoid the occurrence of structural safety accidents.

[0074] Central control unit: Network-connected to the data collection module, structure data acquisition module, route planning module, cruise monitoring agency, laser data analysis module, environmental data acquisition module, image analysis module, comprehensive analysis module, risk prediction module and alarm module.

[0075] Furthermore, the structure data acquisition module reasonably arranges sensor installation points on the civil engineering structure; Arrange a number of sensors to real-time collect data such as stress and strain, displacement, vibration, temperature of the structure, including the following steps:

[0076] 1. Data collection and analysis: Collect design drawings, completion data, geological exploration reports, etc. of civil engineering structures. For bridges, clarify information such as the structural form, dimensions, material properties, and design loads of bridge piers and beam bodies; for buildings, master the structure of beam-column joints and the overall stress system of the building. Study the mechanical analysis reports of the structure under different working conditions, understand the key stress-bearing parts of the structure and the areas where stress concentration and large deformations may occur, and provide a theoretical basis for determining the sensor installation points.

[0077] 2. Develop an installation plan: Select sensors of appropriate types and specifications according to the structural monitoring requirements and the results of the previous data analysis. Based on the structural characteristics and sensor selection, develop a detailed sensor installation plan. Determine the specific installation positions, installation methods (such as pasting, bolt fixation, welding, etc.) of each type of sensor at key parts of the structure, as well as the required installation tools and materials. Plan the sensor wiring scheme, considering the safety, concealment, and maintainability of the wiring. Try to avoid interference between the wiring and other components of the structure, and at the same time, it should be convenient for subsequent inspection, maintenance, and replacement of sensors and circuits.

[0078] 3. Determine the installation points: Determine the installation points at key parts of bridge and building structures for comprehensive monitoring of strain, displacement, vibration, and temperature. Specifically:

[0079] For bridge structures:

[0080] Bridge piers: Install strain, displacement, and vibration sensors, as well as temperature sensors at the bottom, top, and sides to monitor the strain, displacement, vibration, and temperature changes of the bridge piers under various loads.

[0081] Beam bodies: Arrange strain sensors at the upper and lower edges of the mid-span, 1 / 4-span, and 3 / 4-span cross-sections, install displacement sensors at both ends of the supports, arrange vibration sensors along the length of the beam body, and at the same time, arrange temperature sensors inside and on the surface of the beam body.

[0082] For building structures:

[0083] Beam-column joints: Paste strain sensors on the column side and beam side in the core area of the joint, install displacement sensors at the beam ends and column tops at the joint, arrange vibration sensors near the joint, and bury temperature sensors inside the concrete in the joint area.

[0084] Column structures: Install strain sensors at the bottom, middle, and top of the column, install displacement sensors on the side, arrange vibration sensors along the height of the column, and bury temperature sensors at different positions inside the column.

[0085] Beam structures: Paste strain sensors on the upper and lower surfaces of the mid-span and supports of the beam, install displacement sensors at both ends, arrange vibration sensors at certain intervals along the beam, and arrange temperature sensors inside the beam.

[0086] 4. Sensor Deployment Phase: According to the determined installation points, use a marker pen or other marking tools to accurately mark the installation positions of sensors on the surface of the civil engineering structure. For sensors that need to be installed by drilling, mark the center positions of the drill holes. During the marking process, strictly follow the installation plan to ensure the accuracy of the installation positions. At the same time, pay attention to avoiding obstacles such as steel bars and embedded parts in the structure to prevent affecting the installation and measurement effects of the sensors. Install appropriate sensors according to the planned installation points and conduct debugging. Number each sensor and record information such as the sensor number, type, installation position, and the corresponding acquisition terminal port number for subsequent data acquisition and management.

[0087] 5. Data Acquisition: Start the data acquisition system to begin real-time acquisition of sensor data. The data acquisition system continuously acquires data such as stress-strain, displacement, vibration, and temperature of the structure according to the set sampling frequency and stores the data in the local database or transmits it to the monitoring center through the network. During the data acquisition process, closely monitor the changes in the data and promptly detect abnormal data. If it is found that the data acquired by a certain sensor is significantly abnormal (such as exceeding the range, excessive fluctuations, etc.), immediately check whether there are faults in the sensor, wiring, and data acquisition system and repair them in a timely manner. Regularly back up the acquired data to prevent data loss. The backed-up data can be stored in external storage devices (such as hard disks, optical discs, etc.) or cloud servers. At the same time, establish a perfect data management system to classify, organize, and file the data for subsequent query and analysis.

[0088] Furthermore, the route planning module plans the optimal flight route of the UAV based on the civil engineering structure data and the parameter data of the UAV; including the following steps:

[0089] 1. Data Collection and Sorting: Collect the three-dimensional model data of the civil engineering structure, which can be obtained through Building Information Modeling (BIM), covering the precise geometric shape, dimensions, spatial position relationships, etc. of the structure. Sort out the information of the key parts of the structure, such as the expansion joints and bearings of the bridge, the beam-column joints and weak walls of the building. These parts are the key monitoring points, and it is necessary to ensure that the UAV can comprehensively and closely photograph and monitor them during the flight route planning. Obtain the data of the surrounding environment of the structure, including the terrain and landform (whether there are mountains, ravines, etc.), the distribution of buildings (whether there are tall buildings blocking the flight of the UAV), the position of overhead cables (to avoid the UAV colliding), etc. Define the flight performance parameters of the UAV.

[0090] 2. Define the Flight Area: Determine the boundary of the UAV's flight area according to the actual scope of the civil engineering structure and in combination with the surrounding environment restrictions. Divide the flight area into grids, dividing it into several small grid units, and the size of each unit is determined according to the complexity of the structure and the requirements of monitoring accuracy.

[0091] 3. Define the priority of monitoring tasks: Based on the importance of the structure and the degree of potential risk, determine the priority of monitoring tasks for different parts. For example, set the key stress-bearing parts of the bridge (such as the connection points between the main piers and the main girders) and the load-bearing structures of buildings (core tubes, etc.) as high priority to ensure that these parts are monitored first and with higher monitoring accuracy and frequency. Consider the timeliness of monitoring tasks. For example, for parts showing signs of damage recently or areas under construction, arrange to monitor them at the beginning of the flight route to obtain the latest data in a timely manner.

[0092] 4. Route planning: Select the Dijkstra algorithm, which can find the shortest paths from the starting point to all nodes in the graph structure; set the algorithm parameters according to the UAV parameters and monitoring task requirements. Input the collected civil engineering structure data, UAV parameter data, and flight area and task information into the selected algorithm for calculation. During the calculation process, the algorithm will continuously optimize the path according to factors such as the spatial position of the structure, the distribution of monitoring points, the flight performance of the UAV, and obstacles, and find the optimal flight route starting from the UAV take-off point, traversing all monitoring points and meeting various constraint conditions.

[0093] 5. Route optimization: Check whether there is a risk of collision between the planned flight route and the obstacles around the structure. If so, adjust the route to increase the safety buffer. Evaluate the efficiency of the flight route to see if there are situations where the flight path is too circuitous or repetitive. If the total flight distance of the route is too long or the flight time exceeds the endurance time of the UAV, streamline and optimize the route to reduce unnecessary flight trajectories.

[0094] 6. Simulated flight verification: Use simulated flight software to input the planned flight route for simulated flight. During the simulation process, check whether the attitude of the UAV is stable during flight and whether it can take pictures and monitor the key parts of the structure according to the preset monitoring angles and distances. Collect relevant data through simulated flight and further verify the rationality of the route based on these data.

[0095] Furthermore, the image analysis module preprocesses, extracts features from, and identifies and analyzes the collected images, and promptly identifies abnormal situations; including the following steps:

[0096] 1. Image preprocessing: Preprocess the collected images, including format conversion, image enhancement, and noise removal;

[0097] Format conversion: Use a format conversion tool to convert the collected images into a suitable format, such as TIFF format, to ensure that the software can smoothly read and process the image information.

[0098] Image Enhancement: The histogram equalization technique is used to enhance the overall contrast of the image. By redistributing the pixel gray values of the image, the distribution of each gray level in the image becomes more uniform, highlighting the surface details of the structure for subsequent analysis. For example, for an image of a bridge surface with uneven illumination, histogram equalization can make the details in the originally darker areas more clearly visible. At the same time, an image sharpening algorithm, such as the Laplacian operator, is adopted to enhance the edge and detail features of the image, making the edges of diseases such as cracks and spalling more obvious for feature extraction.

[0099] Noise Removal: The median filtering method is used to remove discrete noises such as salt-and-pepper noise, making the image smoother without affecting the structural disease features.

[0100] 2. Feature Extraction: Feature extraction is performed on the preprocessed image, and the extracted features include color, texture, and shape;

[0101] Crack Feature Extraction: The edge detection algorithm, such as Canny edge detection, is used to accurately extract the edges of cracks on the surface of the structure. The detected edges are subjected to contour tracking to determine the geometric features such as the orientation, length, and width of the cracks.

[0102] Spalling Feature Extraction: The region growing algorithm is adopted. Pixel points with similar features (such as color, gray value) in the image are used as seed points. According to the set growth criteria, adjacent pixel points with similar features are merged into one region. For the spalling area on the surface of the structure, its color and gray level are usually different from the normal area. The region growing algorithm can accurately segment the spalling area, and then calculate the feature parameters such as the spalling area and shape for evaluating the disease degree.

[0103] 3. Recognition and Analysis: The support vector machine (SVM) algorithm in machine learning is used to construct a structural disease recognition model. A large number of image samples containing different disease types such as cracks and spalling and normal structural surfaces are collected, and these samples are labeled (such as labeling information such as crack width and spalling area). Then, the sample features are input into the SVM model for training, enabling the model to learn the mapping relationship between different disease features and categories. The features of the image to be analyzed after preprocessing and feature extraction are input into the trained SVM model. The model classifies and judges the disease types in the image based on the learned knowledge, determining whether it is a crack, spalling, or other diseases. At the same time, the disease degree is quantitatively evaluated in combination with the feature parameters. For example, if the crack width exceeds a certain threshold, it is determined as a severe crack, and if the spalling area ratio reaches a certain value, it is evaluated as severe spalling, providing an accurate basis for the evaluation of the structural health status.

[0104] 4. Report Generation: Compare the recognition and analysis results with the preset structural health standards. If the disease characteristic parameters detected in the image (such as crack length, width, spalling area, etc.) exceed the normal range, it is determined that the structure is abnormal. Once an abnormality is detected, the image analysis module automatically generates a detailed abnormality report. The report content includes the abnormal location, disease type, quantitative data of the disease degree, relevant image comparison, and preliminary treatment suggestions.

[0105] Furthermore, the infrared data analysis module analyzes the collected infrared data and promptly identifies abnormal situations, including the following steps:

[0106] 1. Data Processing: Import the raw data collected by the infrared thermal imager into professional analysis software. Perform format conversion to ensure that the data can be successfully loaded and analyzed.

[0107] 2. Temperature Calibration: Environmental factors such as ambient temperature, humidity, and measurement distance will affect the accuracy of the infrared thermal imager measurement, so temperature calibration is required. According to the thermal imager calibration manual, use a blackbody radiation source with a known temperature, such as a constant temperature furnace with a standard temperature, to carry out the calibration operation. Compare the blackbody temperature measured by the thermal imager with the actual temperature, and construct a temperature correction model. Use this calibration model to correct all subsequent measured temperature data to improve the temperature measurement accuracy.

[0108] 3. Image Registration and Mosaic: First, extract features from each image. For example, use the Scale-Invariant Feature Transform (SIFT) algorithm to extract key points in the image. Then, calculate the relative position and rotation relationship between the images by matching the key points between different images to achieve image registration. On this basis, use an image mosaic algorithm, such as a mosaic method based on weighted average, to seamlessly mosaic multiple images into a complete structural temperature distribution image for overall analysis.

[0109] 4. Internal Defect Detection:

[0110] Temperature Abnormal Region Identification: Adopt the temperature threshold method. Based on the normal operating temperature range of the structural material and historical monitoring data, set a reasonable temperature threshold. Check each pixel point in the infrared image one by one. If the difference between the temperature value corresponding to the pixel point and the average value of the normal temperature range in this area exceeds the set temperature threshold, mark the area where this pixel point is located as a temperature abnormal region. For example, in a concrete structure, the temperature in the internal cavity or debonding area is usually different from that in the normal concrete area. With the temperature threshold method, these potential defect areas can be initially identified.

[0111] Analysis of Defect Types and Degrees: For the identified temperature anomaly regions, combined with the characteristics of structural materials, geometric shapes, and heat transfer principles, analyze the defect types and degrees. Take the internal defects of concrete structures as an example. Use the finite element heat conduction model for simulation. Assume that the defect region is a cavity, whose thermal conductivity is different from that of concrete. Simulate the temperature distributions of cavities with different sizes and positions under given thermal boundary conditions, and then compare them with the temperature distributions of the actually measured infrared images to infer information such as the size and position of the cavities. If a region with abnormally increased temperature is found in a steel structure, it may be due to internal stress concentration or weld defects, resulting in local heating. By analyzing characteristics such as the temperature gradient and the shape of the abnormal region, combined with the mechanical knowledge of steel structures, judge the defect types and severity. The heat conduction equation of concrete structures is:

[0112]

[0113] In the formula, represents the rate of change of temperature T with respect to time t. T represents temperature, t represents time, ρ represents the density of the material, with the unit of kilograms per cubic meter (kg / m³). c p represents the specific heat capacity of the material, with the unit of joules per kilogram per kelvin. k represents the thermal conductivity of the material, with the unit of watts per meter per kelvin. represent the second-order partial derivatives of temperature T in the x, y, and z directions. They describe the rate of change of temperature in space, that is, the distribution of temperature gradient in space.

[0114] Furthermore, the laser data analysis module collects the laser point cloud data of civil engineering structures in real time; generates a three-dimensional point cloud model of the structure, and accurately measures the deformation of the structure, including the following steps:

[0115] 1. Data acquisition: The laser scanner emits laser beams, measures the time from laser emission to reception, calculates the distance between the surface points of the target object and the scanner, and simultaneously records the horizontal and vertical angles of the laser beams to obtain the three-dimensional coordinate information of the target points.

[0116] 2. Generation of three-dimensional point cloud model: Import the original point cloud data into professional point cloud processing software. First, perform denoising processing. Using the statistical filtering method, calculate the distance statistics (such as average distance, standard deviation) between each point and its neighboring points. If the distance statistics of a certain point exceed the set range, determine that point as a noise point and delete it. Then process the outlier points. By setting the spatial range threshold, remove the isolated points outside the main body range of the structure.

[0117] 3. Point cloud registration: Use the iterative closest point (ICP) algorithm. By continuously iterating to find the optimal transformation matrix between two sets of point clouds, make the scanned point clouds accurately aligned. Perform point cloud registration according to the following formula:

[0118] E = Σ N i＝1 [||(R * p i + t 平 ) - q i ||²] 2 ; where E is the error function, representing the sum of the squares of the Euclidean distances between all corresponding points. i is the index of the corresponding points, ranging from 1 to N. N is the total number of corresponding points. p i is the i-th corresponding point in the first point cloud. q i is the i-th point in the second point cloud corresponding to p i . R is the rotation matrix used to rotate the first point cloud to align with the second point cloud. t 平 is the translation vector used to translate the rotated first point cloud to align with the second point cloud. ||·||² is the Euclidean distance (or the norm of the vector), representing the straight-line distance between two points.

[0119] 4. Point cloud stitching and fusion: Based on point cloud registration, use the point cloud stitching algorithm to seamlessly stitch the collected point cloud data into a complete structured point cloud model. For the points in the overlapping area, perform fusion processing according to information such as distance and normal vector to ensure the continuity and smoothness of the surface of the stitched model. 5. Model optimization and refinement: Optimize the generated three-dimensional point cloud model, adopt the mesh simplification algorithm, and reduce the amount of point cloud data while maintaining the main features of the model to improve the model processing efficiency. For defects such as holes and gaps in the model, use the repair algorithm to fill them, making the model more complete and accurately reflect the structural shape.

[0120] 6. Feature point extraction and matching: On the three-dimensional point cloud models collected at different times, extract points with obvious features, such as corner points, edge points, key connection points, etc. of the structure. Algorithms such as Harris corner detection can be used for feature point extraction. Describe the extracted feature points through feature descriptors (such as descriptors constructed based on geometric features and normal vectors of points), and use feature matching algorithms (such as nearest neighbor matching based on Euclidean distance) to find the corresponding feature points in the point cloud models at different times and establish the corresponding relationship of feature points.

[0121] 7. Deformation calculation and analysis: Calculate the deformation amount of the structure according to the coordinate changes of the feature points in the point cloud models at different times. Through the deformation calculation of a large number of feature points, comprehensively understand the deformation conditions of different parts of the structure. Draw the deformation cloud diagram, representing the deformation size with different colors or heights, visually display the deformation distribution of the structure, and judge whether there are abnormal deformations in the structure, such as excessive deformation and abnormal deformation trends, based on the deformation data, and analyze the deformation reasons in combination with structural mechanics knowledge.

[0122] Furthermore, the comprehensive analysis module combines the monitoring results of the structural data acquisition module, the environmental data acquisition module, the laser data analysis module, the image analysis module, and the infrared data analysis module to comprehensively evaluate the situation of civil engineering structures and comprehensively judge abnormal situations. The steps are as follows:

[0123] 1. Multi-source data collection: Obtain data such as stress and strain, displacement, and vibration of civil engineering structures from the structural data acquisition module. These data reflect the mechanical properties of the structure itself. Collect environmental parameters such as temperature, humidity, wind direction and speed, earthquake, and rainfall provided by the environmental data acquisition module. Extract the three-dimensional point cloud model data and deformation measurement results of the structure from the laser data analysis module to visually understand the shape changes of the structure. Collect disease information such as surface cracks and spalling of the structure identified by the image analysis module, as well as abnormal internal temperature distribution and defects detected by the infrared data analysis module. Perform format conversion and standardization processing on the collected data to ensure that the data can flow and be processed smoothly in the comprehensive analysis module.

[0124] 2. Data correlation analysis: It includes the correlation between structure and environmental data and the cross-validation of multi-module data;

[0125] Correlation between structure and environmental data: Study the influence of environmental factors on the performance of civil engineering structures. Analyze the relationship between temperature changes and the thermal expansion and contraction deformation of the structure, and check the change trends of the structure's displacement, stress, and strain when the temperature rises or falls. Explore the influence of humidity on the performance of structural materials. For example, whether an increase in humidity leads to a decrease in the durability of concrete structures, and judge by comparing the changes in the mechanical parameters of the structure under different humidity conditions. Analyze the effects of wind direction and speed on structures such as bridges and high-rise buildings, and combine the vibration data of the structure to judge whether the response of the structure under different wind force levels is within the normal range.

[0126] Cross-validation of multi-module data: Use the structural deformation data of the laser data analysis module to verify whether the crack and spalling areas detected by the image analysis module are related to the abnormally deformed parts of the structure. For example, if a crack is found in a certain area by the image analysis, check the deformation situation of this area in the laser point cloud model to see if there is a sudden change in displacement or excessive local deformation. Combine the internal defect information detected by the infrared data analysis module with the stress and strain data of the structural data acquisition module to judge whether the internal defects cause a decline in the mechanical performance of the structure, such as whether the stress concentration near the internal cavity exceeds the normal range.

[0127] 3. Build an evaluation model: Construct a comprehensive evaluation index system for civil engineering structures, covering mechanical performance indexes of the structure (such as the maximum value, average value and change rate of stress, strain and displacement), appearance disease indexes (the length, width and density of cracks, the area and location of spalling), internal defect indexes (the type, size and location of defects), and environmental impact indexes (the degree and duration of environmental parameters exceeding the normal range), etc. According to the importance of each index to the structural safety, methods such as the Analytic Hierarchy Process (AHP) are used to determine the corresponding weights.

[0128] Select the Support Vector Machine (SVM) model of machine learning, and use historical monitoring data and known structural health status labels to train the evaluation model. During the training process, continuously adjust the model parameters to enable the model to accurately evaluate the health status of the structure according to the input multi-source data and output the health level of the structure (such as healthy, sub-healthy, dangerous, etc.).

[0129] 4. Abnormality judgment: Set reasonable warning thresholds for each evaluation index according to the design standards, historical monitoring data and relevant specifications of the structure. Compare the results obtained by calculating the real-time monitoring data through the evaluation model with the thresholds. For example, if the calculated result of the structural displacement exceeds the maximum displacement threshold allowed by the design, or the evaluated value of the crack width is greater than the preset dangerous crack width threshold, it is preliminarily judged that the structure may have abnormal conditions.

[0130] 5. Comprehensive abnormality judgment: Comprehensively consider the changes of multiple indexes. When multiple indexes show abnormalities at the same time, or the abnormality degree of key indexes reaches a certain level, it is determined that the structure has abnormal conditions. For example, when multiple indexes such as structural displacement, stress and strain, and crack width all exceed the normal range, and infrared data analysis also detects serious internal defects, it is comprehensively judged that the structure is in a dangerous state. The comprehensive abnormality judgment model is:

[0131] S = Σ m i＝1 {w i *[Y i +(Σ m j≠i P ij Y j ) / (m - 1)] 1.8 *(1 + △t i / T ref )}; In the formula, S is the comprehensive abnormality score, which is a scalar value. It is used to measure the overall abnormality degree of all abnormal indexes. The higher the score, the higher the overall abnormality degree. This value comprehensively considers the numerical values, mutual relationships and abnormal duration of multiple abnormal indexes, and can more comprehensively reflect the actual abnormal state of the structure. w i$w_i$ is the weight of the $i$-th anomaly index, which is a scalar value. The weight reflects the importance of this anomaly index in the comprehensive anomaly judgment, and its value can be determined by methods such as expert scoring, analytic hierarchy process, and entropy weight method. $Y$ i $P_i$ is the specific value of the $i$-th anomaly index, which is a scalar value. It is various types of measurement values or statistical data used to quantify a specific anomaly index. $P$ ij $\rho_{ij}$ is the correlation coefficient between the $i$-th index and the $j$-th index ($-1\leq\rho_{ij}\leq1$), which reflects the mutual relationship between different anomaly indexes. The Pearson correlation coefficient can be used. For example, there may be a positive correlation between structural displacement and stress-strain. When the displacement is abnormal, the stress-strain may also be affected. $\Delta t$ ij $\Delta t_i$ is the time that the $i$-th anomaly index continuously exceeds the normal range. For example, the duration of the structural crack width exceeding the allowable value, or the number of hours of abnormal ambient temperature persistence, etc. The longer the anomaly duration, the greater the potential harm to the structure. This value incorporates the time factor of the anomaly into the calculation of the comprehensive anomaly score. $T$ i $T_0$ is a reference time, which needs to be determined according to the structure type, design service life, and relevant specifications, etc. It serves as a reference standard for the time scale. $Y$ ref $P_j$ is the specific value of the $j$-th anomaly index. $m$ represents the total number of anomaly indexes. In the comprehensive assessment of civil engineering structures, it covers various anomaly indexes such as structural mechanical properties, appearance diseases, internal defects, and environmental impacts. j $P_j$ is the specific value of the $j$-th anomaly index. $m$ represents the total number of anomaly indexes. In the comprehensive assessment of civil engineering structures, it covers various anomaly indexes such as structural mechanical properties, appearance diseases, internal defects, and environmental impacts.

[0132] 6. Report Generation: Present the comprehensive assessment results through visual methods such as charts and graphs. Draw the deformation nephogram of the structure to visually display the deformation magnitude and distribution of different parts; use bar charts to compare the actual values and thresholds of various assessment indexes; use time series charts to show the change trend of the structural health level over time. Use 3D models to display the appearance diseases and internal defect locations of the structure to make the assessment results more intuitive and understandable. Generate a detailed structural comprehensive assessment report, and the report content includes the assessment time, basic information of the structure, overview of monitoring data for each module, calculation results of assessment indexes, assessment conclusion of the structural health level, description of abnormal situations, analysis of abnormal reasons, and corresponding treatment suggestions.

[0133] Furthermore, the risk prediction module integrates multi-source monitoring data and combines historical data, and uses specific analysis methods to predict potential risks, including the following steps:

[0134] 1. Data collection: Collect real-time data such as stress and strain, displacement, and vibration of civil engineering structures from the structural data acquisition module. These data intuitively reflect the current mechanical state of the structure. Obtain real-time environmental parameters such as temperature, humidity, wind direction and speed, earthquake, and rainfall from the environmental data acquisition module. Extract the three-dimensional point cloud model of the structure and the real-time results of deformation measurement from the laser data analysis module to understand the real-time changes in the structure's shape. Collect real-time information on diseases such as surface cracks and spalling of the structure identified by the image analysis module, as well as the real-time situation of abnormal internal temperature distribution and defects detected by the infrared data analysis module. Conduct historical data integration, organize the monitoring data of each module over a period of time in the past, including structural mechanical property data, environmental data, laser scanning data, image data, and infrared data, etc. Sort the historical data in chronological order to ensure the continuity and integrity of the data.

[0135] 2. Data preprocessing: Clean the collected real-time and historical data to remove obviously incorrect, missing, or duplicate data. For missing data, according to the characteristics and distribution of the data, use interpolation methods (such as linear interpolation, spline interpolation) or fill it according to the statistical laws of historical data. Standardize the data to convert data of different types and magnitudes into a unified scale for subsequent analysis.

[0136] 3. Time series analysis:

[0137] 3.1 Trend analysis: Use the moving average method to process the monitoring data of the structure. For example, calculate the moving average value of the structure displacement in the past 12 months to smooth the data fluctuations and highlight the long-term trend. By plotting the time series graph, visually observe the change trends of data such as structure displacement, stress and strain over time, and judge whether there is a gradually increasing or decreasing trend in the structure, as well as the speed of trend change.

[0138] 3.2 Seasonal analysis: Analyze whether there are seasonal patterns in environmental data (such as temperature, humidity, etc.) and some structural data. For example, by methods such as Fourier transform, decompose the time series into components of different frequencies to determine whether there are periodic changes on an annual, quarterly, or monthly basis. If it is found that the structure displacement significantly increases during the high temperature in summer every year, the impact of seasonal factors on the structure can be further analyzed based on this.

[0139] 3.3 Build a prediction model: Adopt time series prediction models such as the autoregressive integrated moving average model (ARIMA) to predict the changes in the structure monitoring data in the future period based on historical data. Determine the parameters of the model, and continuously adjust the parameters through model training and verification to improve the prediction accuracy. For example, use the historical stress and strain data of the structure in the past 3 years to train the ARIMA model and predict the stress and strain changes in the next 6 months. The prediction model is:

[0140] Z t = [Σ p i＝1 (ψ i,t Z t-i ) + ε t + [Σ q j＝1 (θ j,t ε t-j )] + [Σ s k＝1 (r k e k,t )];

[0141]

[0142] In the formula, Z t represents the stationary time series value at time t, which is the target variable we want to analyze and predict. It can be the stationary sequence value obtained after processing data such as the displacement, stress and strain of a structure at a certain moment. p is the autoregressive order, which determines how many past moment values the current value Z t depends on. For example, if p = 3, it means that Z t is related to Z t-1 , Z t-2 , and Z t-3 . ψ i,t is the dynamic autoregressive coefficient and changes with time t. It reflects the influence degree of the value Z t-i at the past i-th moment on the current value Z t . Z t-i is the value of the time series Z t at the past i-th moment. ε t is the white noise term, representing the random error at time t. q is the moving average order, which determines how many past moment random errors the current random error ε t depends on. θ j,t is the dynamic moving average coefficient and changes with time t. It reflects the influence degree of the random error ε t-j at the past j-th moment on the current random error ε t . s represents the number of external factors affecting the structural monitoring data. r k is the influence degree of the k-th external factor e k,t on the time series Z t . e k,t is the value of the k-th external influencing factor at time t. β is the learning rate used to adjust the dynamic moving average coefficient θ j,t . a is the learning rate used to adjust the dynamic autoregressive coefficient ψ i,t . L represents the loss

[0143] ^

[0144] Function. Z t is the true value of the time series at time t, and Z t is the model prediction value. T 数据 is the length of the training data.

[0145] 4. Feature engineering: Extract valuable features from multi-source data, such as the change rate of structural displacement, the growth rate of crack width, the extreme values of environmental parameters, etc. Combine these features into feature vectors as the input of the machine learning model. At the same time, screen the features to remove features with too high correlation or small contribution to the prediction result to improve the efficiency and accuracy of the model.

[0146] 5. Model training: Select machine learning models suitable for risk prediction, such as decision trees, random forests, support vector regression, etc. Use historical monitoring data and corresponding structural health status labels (such as whether there have been diseases, whether the structure is safe, etc.) to train the model. During the training process, optimize the model parameters through methods such as cross-validation to improve the generalization ability of the model. For example, use a random forest model to train data such as structural stress and strain, displacement, and cracks to predict whether serious diseases will occur in the future of the structure.

[0147] 6. Model evaluation and optimization: Use indicators such as accuracy, recall rate, and root mean square error to evaluate the trained machine learning model. If the model performance does not meet the requirements, optimize the model, such as adjusting the model structure, increasing the amount of training data, and trying different feature combinations.

[0148] 7. Risk assessment and prediction: According to the design standards, relevant specifications, and expert experience of the structure, determine the indicators for evaluating the structure risk, such as the degree to which the structural displacement exceeds the allowable value, the possibility that the crack width reaches the dangerous standard, the rate of internal defect expansion, etc. Set different risk levels for each risk indicator, such as low risk, medium risk, and high risk. Use the prediction results of time series analysis and machine learning models, combined with risk indicators and levels, to predict and classify the potential problems and risks of the structure in the future.

[0149] 8. Risk Visualization: Display the risk prediction results of the structure through visual methods such as charts and graphs. For example, draw a risk heat map, where different colors represent different risk levels, visually presenting the risk distribution of different parts of the structure. Use a line chart to show the changing trend of the structure risk level over time, facilitating relevant personnel to intuitively understand the dynamic changes of the structure risk. When the risk prediction results show that the structure is in a medium to high risk state, promptly issue risk warning information to relevant engineering and management personnel. The warning information includes the risk type, risk level, the location where problems may occur, and the estimated occurrence time, etc., so that relevant personnel can take corresponding measures in a timely manner. Provide targeted decision-making suggestions based on the risk prediction results and the actual situation of the structure. Based on the risk prediction results, provide decision-making support for the long-term management of the structure, and formulate long-term structure maintenance, monitoring, and upgrade plans.

[0150] The present invention provides a structural monitoring method for civil engineering, comprising the following steps:

[0151] S1. A data collection module collects a large amount of data of civil engineering structures (including geological data, drawings, construction data, maintenance data, etc.);

[0152] S2. A structural data acquisition module reasonably arranges a number of sensors on the civil engineering structure to collect data such as stress and strain, displacement, vibration, and temperature of the structure in real time.

[0153] S3. A route planning module plans the optimal flight route of the unmanned aerial vehicle according to the civil engineering structure data and the parameter data of the unmanned aerial vehicle;

[0154] S4. A cruise monitoring mechanism collects high-definition images, infrared images, and laser point cloud data of the civil engineering structure according to the planned route; an environmental data acquisition module collects environmental data in real time, including temperature, humidity, wind direction and speed, earthquake, and rainfall, etc.;

[0155] S5. An image analysis module preprocesses, extracts features, and performs recognition and analysis on the collected images, and promptly identifies abnormal situations; analyzes diseases such as cracks and spalling on the surface of the structure through image recognition technology;

[0156] S6. An infrared data analysis module analyzes the collected infrared data and promptly identifies abnormal situations; detects internal defects;

[0157] S7. A laser data analysis module collects laser point cloud data of the civil engineering structure in real time; a laser scanner generates a three-dimensional point cloud model of the structure to accurately measure the deformation of the structure.

[0158] S8. The comprehensive analysis module combines the monitoring results of the structural data acquisition module, the environmental data acquisition module, the laser data analysis module, the image analysis module, and the infrared data analysis module to comprehensively evaluate the condition of the civil engineering structure and comprehensively judge abnormal situations.

[0159] S9. The risk prediction module integrates the monitoring results of the structural data acquisition module, the environmental data acquisition module, the laser data analysis module, the image analysis module, and the infrared data analysis module, and combines historical data to predict potential problems and risks of the civil engineering structure; uses methods such as time series analysis and machine learning to perform trend prediction on the monitoring data of the structure and anticipate the future development and changes of the structure in advance.

[0160] S10. When abnormal situations or potential problem risks are detected, the alarm module issues an alarm in a timely manner. Remind users to take corresponding measures to avoid the occurrence of structural safety accidents.

[0161] The working principle of a structural monitoring system for civil engineering according to the present invention is as follows: The data collection module collects a large amount of data of civil engineering structures; the structural data acquisition module reasonably arranges several sensors on the civil engineering structure to collect data such as stress and strain, displacement, vibration, and temperature of the structure in real time. The route planning module plans the optimal flight route of the unmanned aerial vehicle according to the civil engineering structure data and the parameter data of the unmanned aerial vehicle; the cruise monitoring mechanism collects high-definition images, infrared images, and laser point cloud data of the civil engineering structure according to the planned route; the environmental data acquisition module collects environmental data in real time, including temperature, humidity, wind direction and speed, earthquake, and rainfall, etc.; the image analysis module preprocesses, extracts features, and performs recognition and analysis on the collected images, and promptly identifies abnormal situations; analyzes diseases such as cracks and spalling on the surface of the structure through image recognition technology; the infrared data analysis module analyzes the collected infrared data and promptly identifies abnormal situations; detects internal defects; the laser data analysis module collects the laser point cloud data of the civil engineering structure in real time; the laser scanner generates a three-dimensional point cloud model of the structure and accurately measures the deformation of the structure. The comprehensive analysis module combines the monitoring results of the structural data acquisition module, the environmental data acquisition module, the laser data analysis module, the image analysis module, and the infrared data analysis module to comprehensively evaluate the condition of the civil engineering structure and comprehensively judge abnormal situations; the risk prediction module integrates the monitoring results of the structural data acquisition module, the environmental data acquisition module, the laser data analysis module, the image analysis module, and the infrared data analysis module, and combines historical data to predict potential problems and risks of the civil engineering structure; uses methods such as time series analysis and machine learning to perform trend prediction on the monitoring data of the structure. When abnormal situations or potential problem risks are detected, the alarm module issues an alarm in a timely manner. Remind users to take corresponding measures to avoid the occurrence of structural safety accidents.

[0162] The present invention combines sensor monitoring and drone cruise monitoring to achieve comprehensive monitoring of the civil engineering structure; the drone is equipped with a variety of devices to collect high-definition images, infrared images and laser point cloud data. The high-definition camera can clearly capture the subtle diseases on the surface of the structure; the infrared thermal imager detects internal defects; the laser scanner generates a three-dimensional point cloud model to accurately measure deformation, realizing comprehensive and multi-angle monitoring of the structure, and is not limited by the terrain and the complexity of the structure, and can cover areas that are difficult for humans to reach. The comprehensive analysis module integrates multi-source monitoring data to comprehensively evaluate the structure situation and judge anomalies. By combining the structure data with the environmental data, analyzing the influence of environmental factors on the structure, and combining image, infrared and laser data, the health status of the structure can be judged from different angles, improving the accuracy of anomaly judgment. The risk prediction module uses methods such as time series analysis and machine learning to fuse the monitoring results of multiple modules and historical data to predict potential problems and risks of the structure. Anticipate the future development and changes of the structure in advance, provide decision support for long-term management, such as predicting the parts and time when diseases may occur in the structure in the next few years, facilitating the advance planning of maintenance work, and reducing maintenance costs and safety risks.

[0163] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A structural monitoring method for civil engineering, characterized in that: The following steps are involved: S1, data collection module collects data of civil engineering structures; S2, structural data acquisition module collects structural stress strain, displacement, vibration and temperature data in real time; S3, the route planning module plans the optimal flight route of the UAV; S4, the cruise monitoring agency collects high-definition images, infrared images and laser point cloud data according to the planned route; the environmental data acquisition module collects environmental data in real time; S5, the image analysis module performs preprocessing, feature extraction and recognition analysis on the collected images to identify abnormal situations; S6, infrared data analysis module analyzes infrared data to detect internal defects; S7, laser data analysis module accurately measures the deformation of the structure; S8, the comprehensive analysis module combines the monitoring results of the structure data acquisition module, the environmental data acquisition module, the laser data analysis module, the image analysis module and the infrared data analysis module to conduct a comprehensive evaluation and comprehensively judge the abnormal situation; S9, risk prediction module integrates multi-source data and combines historical data to predict potential problems and risks of civil engineering structures; S10. When an abnormal situation or potential problem risk is detected, the alarm module will issue an alarm in time.

2. The structural monitoring method for civil engineering according to claim 1, characterized in that: Step S6 includes the following steps: S61, Data processing: Import the raw data collected by the infrared thermal imager into professional analysis software for format conversion; S62, Temperature calibration: According to the thermal imager calibration manual, use a black body radiation source with a known temperature to perform the calibration operation; S63, Image registration and stitching: Use the scale-invariant feature transform (SIFT) algorithm to extract key points in the image; by matching key points between different images, calculate the relative position and rotation relationship between images to achieve image registration; use the weighted average stitching method to seamlessly stitch multiple images into a complete structural temperature distribution image; S64, internal defect detection: S641, temperature abnormal area identification: adopt the temperature threshold method, by setting a reasonable temperature limit, analyze the temperature value of infrared image pixels one by one, if it exceeds the normal range, mark it as abnormal; S642. Analysis of defect type and degree: For the identified temperature anomaly areas, the defect type and degree are analyzed by combining material properties, geometric shape and heat transfer principles through finite element simulation and actual measurement comparison.

3. The structural monitoring method for civil engineering according to claim 1, characterized in that: Step S7 includes the following steps: S71, data acquisition: measure the time from laser emission to reception, calculate the distance between the surface point of the target object and the scanner, record the horizontal and vertical angles of the laser beam, and obtain the three-dimensional coordinate information of the target point; S72. 3D point cloud model generation: import the original point cloud data into professional point cloud processing software for denoising; then process the outliers and remove the isolated points outside the main structure by setting the spatial range threshold; S73, point cloud registration: using the iterative closest point ICP algorithm, the optimal transformation matrix between two sets of point clouds is found through continuous iteration to accurately align the scanned point clouds; S74, point cloud stitching and fusion: Use the point cloud stitching algorithm to seamlessly stitch the collected point cloud data into a complete structural point cloud model; S75, Model optimization and refinement: Use mesh simplification algorithm to reduce the amount of point cloud data to improve efficiency, and use patching algorithm to fill holes and gaps to make the 3D point cloud model more complete; S76, feature point extraction and matching: extract obvious feature points on the three-dimensional point cloud model, describe these points with feature descriptors, and find corresponding feature points in the point cloud models at different times through feature matching algorithms to establish corresponding relationships; S77. Deformation calculation and analysis: Calculate the structural deformation according to the change of feature point coordinates, fully understand the deformation of different parts, draw deformation cloud map to intuitively display the deformation distribution, judge abnormal deformation and analyze the cause in combination with structural mechanics knowledge.

4. The structural monitoring method for civil engineering according to claim 3, characterized in that: In step S73, point cloud registration is performed according to the following formula: E = Σ N i＝1 [||(R*p i +t 平 )-q i ||2] 2 ; Where E is the error function; i is the index of the corresponding point, from 1 to N; N is the total number of corresponding points; p i is the i-th corresponding point in the first set of point clouds; q i is the second set of point clouds with p i The corresponding i-th point; R is the rotation matrix used to rotate the first set of point clouds to a position aligned with the second set of point clouds; t 平 is the translation vector; ||·||2 is the Euclidean distance, which represents the straight-line distance between two points.

5. The structural monitoring method for civil engineering according to claim 1, characterized in that: Step S8 includes the following steps: S81. Multi-source data collection: collect monitoring data detected by the structural data acquisition module, environmental data acquisition module, laser data analysis module, image analysis module and infrared data analysis module; perform format conversion and standardization on the collected data; S82. Data association analysis: S821, Structural and Environmental Data Correlation: Study the influence of environmental factors on the performance of civil engineering structures, and analyze the relationship between these factors and structural deformation, stress-strain, material durability and vibration response; S822, multi-module data cross-validation: by comparing the structural deformation data of the laser data analysis module, the crack and spalling detection results of the image analysis module, and the internal defect information of the infrared data analysis module, combined with the stress and strain data of the structural data acquisition module, the correlation between structural anomalies and defects and their impact on mechanical properties are comprehensively judged; S83. Construct an assessment model: Construct a comprehensive assessment model for civil engineering structures, use the analytic hierarchy process to determine weights, use a machine learning support vector machine model, and train it in combination with historical monitoring data and structural health status labels to accurately assess the structural health status and output the health level; S84, abnormality judgment: compare the result obtained after the real-time monitoring data is calculated by the evaluation model with the threshold value to determine whether there is an abnormal situation; S85. Comprehensive abnormality judgment: Comprehensively consider the changes of multiple indicators. When multiple indicators are abnormal at the same time, or the abnormality of key indicators reaches a certain level, it is determined that the structure is abnormal. S86, Report generation: Present the comprehensive evaluation results in a visual way and generate a detailed structural comprehensive evaluation report.

6. The structural monitoring method for civil engineering according to claim 5, characterized in that: In step S85, the comprehensive abnormality judgment model is: S=Σ m i＝1 {w i *[Y i +(Σ m j≠i P ij Y j ) / (m-1)] 1.8 *(1+△t i / T ref )}; where S is the comprehensive abnormality score; w i is the weight of the i-th abnormal indicator; Y i is the specific value of the ith abnormal indicator; P ij is the correlation coefficient between the ith indicator and the jth indicator; △t i is the time that the i-th abnormal indicator continues to exceed the normal range; T ref It is a reference time, which needs to be determined according to the structure type, design service life and relevant specifications; Y j is the specific value of the jth abnormal indicator, and m represents the total number of abnormal indicators.

7. The structural monitoring method for civil engineering according to claim 1, characterized in that: Step S9 includes the following steps: S91, data collection: collect monitoring data from the structure data acquisition module, the environment data acquisition module, the laser data analysis module, the image analysis module and the infrared data analysis module, integrate historical data, and sort the historical data in chronological order; S92, data preprocessing: cleaning and standardizing the collected real-time and historical data; S93, time series analysis; S94. Feature Engineering: Extract valuable features from multi-source data, combine these features into feature vectors as input to machine learning models; screen features; S95, Model training: Select the random forest machine learning model for risk prediction training, use historical monitoring data and structural health status labels to optimize model parameters and improve generalization ability to predict whether the structure will have serious diseases in the future; S96. Model evaluation and optimization: Evaluate and optimize the trained machine learning model; S97, Risk assessment and prediction: Use the prediction results of time series analysis and machine learning models, combined with risk indicators and levels, to predict and grade the potential problems and risks of the structure in the future; S98, Risk Visualization: Display the risk prediction results of the structure in a visual way.

8. The structural monitoring method for civil engineering according to claim 7, characterized in that: Step S83 includes the following steps: S93.

1. Trend analysis: Use the moving average method to process the structural monitoring data, observe the displacement and stress-strain change trends through time series graphs, and evaluate the speed and direction of change of the structural state; S93.2, Seasonal analysis: Analyze whether there are seasonal patterns in environmental data and some structural data; analyze the impact of seasonal factors on the structure; S93.

3. Construct a prediction model: Use the autoregressive moving average model (ARIMA) time series prediction model to predict changes in structural monitoring data in the future based on historical data.

9. The structural monitoring method for civil engineering according to claim 8, characterized in that: In step S93.3, the prediction model is: Z t =[Σ p i＝1 (ψ i,t Z t-i )+e t ]+[S q j＝1 (i j,t e t-j )]+[S s k＝1 (r k e k,t )]; In the formula, Z t represents the stationary time series value at time t; p is the autoregressive order; ψ i,t is the dynamic autoregressive coefficient, reflecting the value Z at the i-th moment in the past t-i For the current value Z t The degree of influence of Z t-i is the time series Z t The value at the i-th moment in the past; ε t is the white noise term; q is the moving average order; θ j,t is the dynamic moving average coefficient; s represents the number of external factors affecting the structural monitoring data; r k is the kth external factor e k,t For the time series Z t The degree of influence; k,t is the value of the kth external influencing factor at time t; β is used to adjust the dynamic moving average coefficient θ j,t The learning rate; a is used to adjust the dynamic self-regression coefficient ψ i,t The learning rate; L represents the loss function; Z t is the true value of the time series at time t, Z t is the model prediction value; T 数据 is the length of the training data.

10. A structural monitoring system for civil engineering, comprising: Central control unit, data collection module, structural data acquisition module, route planning module, cruise monitoring mechanism, laser data analysis module, environmental data acquisition module, image analysis module, comprehensive analysis module, risk prediction module and alarm module; characterized in that: Data collection module: collects a large amount of data on civil engineering structures, including geological data, drawings, construction data and maintenance data; Structural data acquisition module: Several sensors are rationally arranged on civil engineering structures to collect stress, strain, displacement, vibration and temperature data of the structure in real time; Route planning module: plans the optimal flight route of the drone based on civil engineering structure data and drone parameter data; Cruise monitoring agencies: collect high-definition images, infrared images and laser point cloud data of civil engineering structures; Environmental data collection module: collect environmental data in real time; Image analysis module: pre-processes, extracts features and performs recognition analysis on the collected images to identify abnormal situations in a timely manner; Infrared data analysis module: analyzes the collected infrared data and detects internal defects; Laser data analysis module: real-time acquisition of laser point cloud data of civil engineering structures; accurate measurement of structural deformation; Comprehensive analysis module: Combines the monitoring results of the structural data acquisition module, environmental data acquisition module, laser data analysis module, image analysis module and infrared data analysis module to comprehensively evaluate the situation of civil engineering structures and comprehensively judge abnormal situations; Risk prediction module: It integrates the monitoring results of the structural data acquisition module, environmental data acquisition module, laser data analysis module, image analysis module and infrared data analysis module, and combines historical data to predict potential problems and risks of civil engineering structures; Alarm module: When abnormal conditions or potential risks are detected, an alarm will be issued in time; Central control unit: network connected with data collection module, structural data collection module, route planning module, cruise monitoring mechanism, laser data analysis module, environmental data collection module, image analysis module, comprehensive analysis module, risk prediction module and alarm module.

Images