A multi-sensor fusion bridge construction integrated health detection and analysis system and method

Abstract

The application discloses a bridge building integrated health detection analysis system based on multi-sensor fusion, which comprises a sensing network layer, an edge gathering layer, a fusion analysis layer and a decision-making and early warning layer; the sensing network layer comprises multi-source sensors which are distributedly arranged at key positions of a bridge structure and are used for collecting physical state parameters of the bridge structure; the edge gathering layer is in communication connection with the sensing network layer, pre-processes and extracts features of the collected physical state parameters, and outputs multi-source feature data; the fusion analysis layer is connected with the edge gathering layer, performs multi-physical quantity coupling analysis based on the multi-source feature data, and identifies abnormal states of the bridge structure; and the decision-making and early warning layer is connected with the fusion analysis layer, and outputs a health evaluation report and graded early warning information according to the abnormal states. Compared with the prior art, the application has the advantages that a bridge building integrated health detection analysis system and method capable of realizing multi-source data coupling analysis and graded dynamic early warning are provided.

Description

Technical Field

[0001] This invention relates to the field of bridge structural health monitoring technology, specifically to a multi-sensor fusion integrated bridge construction health monitoring and analysis system and method. Background Technology

[0002] As an important component of transportation infrastructure, the structural safety of bridges is directly related to public safety.

[0003] As bridges age, their performance gradually degrades due to the combined effects of multiple factors, including vehicle loads, environmental erosion, material aging, and extreme weather.

[0004] Traditional bridge health monitoring mainly relies on regular manual inspections and monitoring with a single type of sensor, which has problems such as long detection cycles, limited data dimensions, significant environmental interference, and delayed early warning.

[0005] In recent years, multi-sensor fusion technology has attracted attention in the field of structural health monitoring. However, existing technologies still have problems such as extensive sensor deployment, insufficient data fusion depth, difficulty in separating environmental interference, and delayed early warning response. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to overcome the above-mentioned technical defects and provide an integrated health monitoring and analysis system and method for bridge construction that can realize multi-source data coupling analysis and hierarchical dynamic early warning.

[0007] To solve the above-mentioned technical problems, the technical solution provided by the present invention is: a multi-sensor fusion integrated bridge construction health detection and analysis system, including a sensor network layer, an edge convergence layer, a fusion analysis layer, and a decision and early warning layer; The sensor network layer includes multi-source sensors distributed in key parts of the bridge structure to collect physical state parameters of the bridge structure. The edge aggregation layer is communicatively connected to the sensor network layer to preprocess and extract features from the collected physical state parameters and output multi-source feature data. The fusion analysis layer is connected to the edge convergence layer, and performs multi-physical quantity coupling analysis based on multi-source feature data to identify abnormal states of the bridge structure. The decision-making and early warning layer is connected to the fusion analysis layer, and outputs health assessment reports and graded early warning information based on abnormal states.

[0008] Preferably, the multi-source sensor includes a combination of two or more of the following: strain sensor, acceleration sensor, temperature and humidity sensor, corrosion potential sensor, and displacement sensor, which respectively collect structural strain, vibration acceleration, ambient temperature and humidity, steel corrosion potential, and structural displacement parameters.

[0009] Preferably, the edge convergence layer includes a data preprocessing unit and a feature extraction unit; The data preprocessing unit filters, denoises, and removes outliers from the physical state parameters; The feature extraction unit extracts multi-source feature data based on time-domain, frequency-domain, and wavelet-domain analysis methods.

[0010] Preferably, the fusion analysis layer includes a multi-physical quantity coupling analysis unit and an anomaly detection unit; The multi-physical quantity coupling analysis unit establishes a coupling correlation model of strain, displacement, and corrosion to analyze the time-delay correlation and causal transmission relationship between different physical quantities; The anomaly detection unit identifies abnormal states of the bridge structure based on a coupled correlation model.

[0011] Preferably, the multi-physical quantity coupling analysis unit further includes an environmental decoupling unit; The environmental decoupling unit separates the coupling interference of environmental temperature and humidity on structural strain and vibration parameters, and extracts the pure structural response characteristics under load.

[0012] Preferably, the decision-making early warning layer includes a health assessment unit and a graded early warning unit; The health assessment unit constructs a multi-indicator comprehensive evaluation model and outputs the overall health index of the bridge structure. The graded early warning unit outputs four levels of early warning information—blue, yellow, orange, and red—based on the degree of deviation between the health index and the preset threshold.

[0013] Another aspect of this invention discloses a multi-sensor fusion integrated health monitoring and analysis method for bridge structures, comprising the following steps: S1: Based on the structural mechanical characteristics of the bridge and historical damage data, identify key parts and optimize the spatial layout of multi-source sensors, and deploy multi-source sensors in a distributed manner at key parts of the bridge structure. S2: Perform time synchronization alignment and spatial coordinate registration on the physical state parameters collected by multi-source sensors, eliminate the dimensional differences and sampling frequency differences between heterogeneous data based on the adaptive filtering algorithm, and output standardized multi-source feature data; S3: Construct a coupling correlation matrix to calculate the time-delay cross-correlation function between different physical quantities. When the peak value of the cross-correlation function exceeds a preset threshold, an abnormal state is determined, and the spatial coordinates of the abnormal source are located. S4: Based on the spatial coordinates and severity of the abnormal state, trigger the corresponding level of early warning response and output a health assessment report containing the abnormal location, abnormal type and handling suggestions.

[0014] Preferably, in S2, the adaptive filtering algorithm dynamically adjusts the filtering weights based on the signal-to-noise ratio of each sensor.

[0015] Preferably, step S4 includes automatically pushing early warning information to the management terminal.

[0016] The advantages of this invention compared to existing technologies are: this solution deploys multi-source sensors such as strain and corrosion sensors in key areas to obtain the internal response of the structure; The images are used to identify the geometric parameters of the cracks through deep learning, and the sensor data are used to extract the pure structural response features through multi-physical quantity coupling analysis. The two data are fused and complemented to achieve a three-dimensional evaluation from the appearance of the disease to the structural performance. Based on time-series evolution analysis and a hierarchical early warning mechanism, a global crack distribution map and four levels of early warning (blue, yellow, orange, and red) are output to improve the comprehensiveness, accuracy, and timeliness of bridge health monitoring. Attached Figure Description

[0017] Figure 1 This is a structural schematic diagram of a multi-sensor fusion integrated health monitoring and analysis system for bridge construction.

[0018] Figure 2 This is a flowchart illustrating a multi-sensor fusion-based integrated health monitoring and analysis method for bridge construction. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings.

[0020] Combined with appendix Figure 1-2 As shown, a multi-sensor fusion integrated health monitoring and analysis system for bridge construction includes a sensor network layer, an edge aggregation layer, a fusion analysis layer, and a decision and early warning layer.

[0021] The sensor network layer includes multi-source sensors distributed across key parts of the bridge structure to collect physical state parameters of the bridge structure. The edge aggregation layer is communicatively connected to the sensor network layer to preprocess and extract features from the collected physical state parameters, outputting multi-source feature data. The fusion analysis layer is connected to the edge aggregation layer to perform multi-physical quantity coupling analysis based on the multi-source feature data to identify abnormal states of the bridge structure. The decision-making and early warning layer is connected to the fusion analysis layer to output health assessment reports and graded early warning information based on the abnormal states.

[0022] The multi-source sensors include two or more combinations of strain sensors, acceleration sensors, temperature and humidity sensors, corrosion potential sensors, and displacement sensors, which respectively collect structural strain, vibration acceleration, ambient temperature and humidity, steel corrosion potential, and structural displacement parameters. The edge convergence layer includes a data preprocessing unit and a feature extraction unit. The data preprocessing unit filters, denoises, and removes outliers from the physical state parameters. The feature extraction unit extracts multi-source feature data based on time domain, frequency domain, and wavelet domain analysis methods.

[0023] The fusion analysis layer includes a multi-physical quantity coupling analysis unit and an anomaly detection unit; the multi-physical quantity coupling analysis unit establishes a coupling correlation model of strain, displacement, and corrosion to analyze the time-delay correlation and causal transmission relationship between different physical quantities; the anomaly detection unit identifies the abnormal state of the bridge structure based on the coupling correlation model.

[0024] In one embodiment, the multi-physical quantity coupling analysis unit further includes an environment decoupling unit; The environmental decoupling unit separates the coupling interference of environmental temperature and humidity on structural strain and vibration parameters, and extracts the pure structural response characteristics under load.

[0025] In one embodiment, the decision-making early warning layer includes a health assessment unit and a graded early warning unit; The health assessment unit constructs a multi-indicator comprehensive evaluation model and outputs the overall health index of the bridge structure. The graded early warning unit outputs four levels of early warning information—blue, yellow, orange, and red—based on the degree of deviation between the health index and the preset threshold.

[0026] In specific implementation, this invention includes the following steps: Based on the structural mechanical properties of the bridge and historical damage data, key parts were identified and the spatial layout of multi-source sensors was optimized. Multi-source sensors were distributed and deployed in key parts of the bridge structure. The structural mechanical properties of the bridge included structural type, span arrangement, cross-sectional shape, material properties and support constraints. The stress distribution, deformation mode and dynamic characteristics of the structure under typical loads were obtained through finite element analysis. Stress concentration areas, deformation sensitive areas and modal vibration mode node areas were identified as the preliminary candidate set of key parts. Historical defect data includes previous bridge inspection reports, maintenance records, and defect statistics of similar bridges. The analysis focuses on the location, development pattern, and causes of defects such as cracks, corrosion, deformation, and abnormal vibration. The candidate set for key parts is supplemented. The spatial layout scheme is optimized by using a sensor optimization layout algorithm based on information entropy or a multi-objective optimization method based on genetic algorithm. The optimization objectives are to maximize the coverage of monitoring information and minimize the number of sensors, and to determine the optimal layout location and number of each type of sensor. The physical state parameters collected by multi-source sensors are synchronized and aligned in time and registered in space. An adaptive filtering algorithm is used to eliminate the differences in dimensions and sampling frequency between heterogeneous data and output standardized multi-source feature data. The time synchronization alignment uses network time protocol or GPS time synchronization to provide a unified time reference for each sensor. The asynchronously acquired data is aligned to a unified time grid by resampling or interpolation methods. The spatial coordinate registration establishes the transformation relationship between the local coordinate system of each sensor and the overall coordinate system of the bridge, and maps the physical state parameters of each sensor to a unified structural space frame. The adaptive filtering algorithm dynamically adjusts the filtering weights based on the signal-to-noise ratio of each sensor. Sensors with high signal-to-noise ratios are given higher weights, and sensors with low signal-to-noise ratios are given lower weights. The weight coefficients are calculated by estimating the noise variance of the output signals of each sensor in real time. The sampling frequency difference is eliminated based on the highest sampling frequency. Interpolation upsampling is used for low-frequency data, and downsampling or moving average filtering is used for high-frequency data. A coupling correlation matrix is ​​constructed to calculate the time-delay cross-correlation function between different physical quantities. When the peak value of the cross-correlation function exceeds a preset threshold, an abnormal state is determined, and the spatial coordinates of the abnormal source are located. The coupling correlation matrix is ​​a symmetric matrix, and the matrix elements are the time-delay cross-correlation function values ​​between different physical quantities. The time-delay cross-correlation function calculates the degree of linear correlation between the time series of two physical quantities under different time delays. The time delay range is from a negative number of sampling periods to a positive number of sampling periods, covering possible lead-lag relationships between physical quantities. The peak value of the cross-correlation function refers to the maximum value of the time-delay cross-correlation function over the entire time-delay range. The time delay corresponding to the peak value is the optimal time delay between the two physical quantities. The preset threshold is determined based on the statistical distribution established by historical normal data. The typical value is the mean of the peak value of the cross-correlation function of normal data plus two or three times the standard deviation. When the peak value of the cross-correlation function exceeds the preset threshold, it indicates that the correlation between the two physical quantities has changed significantly, and an abnormal state is determined to exist. Anomaly source localization is based on column and row sum or vector analysis of the coupling correlation matrix. The location of the sensor with the greatest change in correlation strength with the anomalous physical quantity is the spatial coordinate of the anomaly source. The localization accuracy depends on the sensor deployment density. Based on the spatial coordinates and severity of the abnormal state, a corresponding level of early warning response is triggered, outputting a health assessment report that includes the location of the abnormality, the type of abnormality, and handling recommendations. The severity of the abnormality is comprehensively assessed based on the deviation of the abnormal physical quantity, the duration of the abnormality, and the scope of impact. The deviation refers to the relative deviation between the measured value of the abnormal physical quantity and the normal baseline; the duration refers to the length of time the abnormal state lasts; and the scope of impact refers to the number of sensors and structural areas affected by the abnormality, as determined through coupling correlation analysis. The warning response level is triggered according to the severity and the classification standard of the health assessment unit. The blue to red warnings correspond to mild, moderate, severe and emergency severity, respectively. The health assessment report is automatically generated using a structured template. The report content includes basic bridge information, monitoring period, sensor deployment map, abnormal location marking map, abnormality type judgment basis, time series evolution curve, health index score, warning level and handling suggestions. The handling recommendations are generated from the expert knowledge base based on the type of anomaly. For example, for crack anomalies, it is recommended to detect the crack width and seal it; for rust anomalies, it is recommended to detect the protective layer thickness and remove rust and apply a coating; for deformation anomalies, it is recommended to re-measure the deflection and assess the load-bearing capacity; and for vibration anomalies, it is recommended to conduct modal testing and check the damper. The system automatically pushes early warning information to management terminals, including desktop computers, tablets, and smartphones. It connects to the decision-making and early warning layer via wired networks, WiFi, or mobile data networks. Early warning information is pushed using message queues or push notification services, with different push channels and strategies selected based on the warning level. Blue warnings are pushed via in-system messages on the monitoring platform; yellow warnings are pushed via SMS and in-system messages; orange warnings are pushed via SMS, voice calls, and in-system messages; and red warnings are pushed simultaneously via SMS, voice calls, mobile application push notifications, and the emergency broadcast system. The push notification includes a summary of the alert, a detailed link, and a one-click confirmation button. Once the recipient confirms, the notification status updates to "read." If unconfirmed, the notification is repeated at preset intervals until confirmation or an escalation of the alert level. The management terminal also supports forwarding, commenting on, and updating the handling status of alert information, forming a closed-loop management system for alert handling.

[0027] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

[0028] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0029] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A multi-sensor fusion integrated health monitoring and analysis system for bridge construction, characterized in that: It includes a sensor network layer, an edge aggregation layer, a fusion analysis layer, and a decision and early warning layer; The sensor network layer includes multi-source sensors distributed in key parts of the bridge structure to collect physical state parameters of the bridge structure. The edge aggregation layer is communicatively connected to the sensor network layer to preprocess and extract features from the collected physical state parameters and output multi-source feature data. The fusion analysis layer is connected to the edge convergence layer, and performs multi-physical quantity coupling analysis based on multi-source feature data to identify abnormal states of the bridge structure. The decision-making and early warning layer is connected to the fusion analysis layer, and outputs health assessment reports and graded early warning information based on abnormal states.

2. The multi-sensor fusion integrated health monitoring and analysis system for bridge construction according to claim 1, characterized in that: The multi-source sensor includes two or more of the following: strain sensor, acceleration sensor, temperature and humidity sensor, corrosion potential sensor, and displacement sensor, which respectively collect structural strain, vibration acceleration, ambient temperature and humidity, steel corrosion potential, and structural displacement parameters.

3. The multi-sensor fusion integrated health monitoring and analysis system for bridge construction according to claim 1, characterized in that: The edge convergence layer includes a data preprocessing unit and a feature extraction unit; The data preprocessing unit filters, denoises, and removes outliers from the physical state parameters; The feature extraction unit extracts multi-source feature data based on time-domain, frequency-domain, and wavelet-domain analysis methods.

4. The multi-sensor fusion integrated bridge construction health monitoring and analysis system according to claim 1, characterized in that: The fusion analysis layer includes a multi-physical quantity coupling analysis unit and an anomaly detection unit; The multi-physical quantity coupling analysis unit establishes a coupling correlation model of strain, displacement, and corrosion to analyze the time-delay correlation and causal transmission relationship between different physical quantities; The anomaly detection unit identifies abnormal states of the bridge structure based on a coupled correlation model.

5. The multi-sensor fusion integrated bridge construction health monitoring and analysis system according to claim 4, characterized in that: The multi-physical quantity coupling analysis unit also includes an environment decoupling unit; The environmental decoupling unit separates the coupling interference of environmental temperature and humidity on structural strain and vibration parameters, and extracts the pure structural response characteristics under load.

6. The multi-sensor fusion integrated health monitoring and analysis system for bridge construction according to claim 1, characterized in that: The decision-making and early warning layer includes a health assessment unit and a graded early warning unit; The health assessment unit constructs a multi-indicator comprehensive evaluation model and outputs the overall health index of the bridge structure. The graded early warning unit outputs four levels of early warning information—blue, yellow, orange, and red—based on the degree of deviation between the health index and the preset threshold.

7. A multi-sensor fusion integrated health monitoring and analysis method for bridge structures, using the system as described in any one of claims 1-6, characterized in that: Includes the following steps: S1: Based on the structural mechanical characteristics of the bridge and historical damage data, identify key parts and optimize the spatial layout of multi-source sensors, and deploy multi-source sensors in a distributed manner at key parts of the bridge structure. S2: Perform time synchronization alignment and spatial coordinate registration on the physical state parameters collected by multi-source sensors, eliminate the dimensional differences and sampling frequency differences between heterogeneous data based on the adaptive filtering algorithm, and output standardized multi-source feature data; S3: Construct a coupling correlation matrix to calculate the time-delay cross-correlation function between different physical quantities. When the peak value of the cross-correlation function exceeds a preset threshold, an abnormal state is determined, and the spatial coordinates of the abnormal source are located. S4: Based on the spatial coordinates and severity of the abnormal state, trigger the corresponding level of early warning response and output a health assessment report containing the abnormal location, abnormal type and handling suggestions.

8. The multi-sensor fusion integrated health monitoring and analysis method for bridge construction according to claim 7, characterized in that: The adaptive filtering algorithm in S2 dynamically adjusts the filtering weights based on the signal-to-noise ratio of each sensor.

9. The multi-sensor fusion integrated health monitoring and analysis method for bridge construction according to claim 7, characterized in that: The S4 includes automatically pushing early warning information to the management terminal.

Images