Bridge deformation monitoring system combining contact and non-contact methods

Abstract

The application relates to the technical field of bridge structure health monitoring, and relates to a bridge deformation monitoring system combining contact and non-contact modes, which comprises the following: an intelligent fusion target unit fixed to a monitoring point of a bridge, which is used for providing an optical mark and synchronously collecting state data of the intelligent fusion target unit; a non-contact visual collection unit arranged on a stable reference point outside the bridge, which is used for periodically collecting an optical mark image of the intelligent fusion target unit and generating image data; and a data processing unit connected with the communication modules of the non-contact visual collection unit and the intelligent fusion target unit respectively, which is used for receiving the image data and the state data, and correcting and processing the image data based on the state data to generate a monitoring result of bridge deformation.

Description

Technical Field

[0001] This invention relates to the field of bridge structural health monitoring technology, and in particular to a bridge deformation monitoring system that integrates contact and non-contact methods. Background Technology

[0002] Bridge deformation is a crucial early warning signal for assessing the safety and condition of bridge structures. Currently, there are two methods for monitoring bridge deformation: non-contact monitoring using machine vision and contact monitoring using industrial sensors. Non-contact monitoring using machine vision involves deploying targets on the bridge, capturing images of the targets using industrial cameras fixed at stable reference points, and then calculating the target displacement using image processing algorithms to estimate bridge deformation. Contact monitoring using industrial sensors involves installing sensors such as inclinometers, displacement gauges, and GPS devices at key locations on the bridge for measurement.

[0003] However, while purely non-contact vision methods offer advantages such as non-contact and global monitoring, their measurement accuracy and stability are highly susceptible to environmental factors. For example, changes in lighting can lead to difficulties in target identification or a decrease in accuracy; rain, fog, and temporary obstruction can cause data interruptions; more critically, the targets themselves can expand and contract due to temperature changes, or experience slight loosening due to creep of the mounting material or vibration. These positional drifts and posture changes of the targets themselves cannot be perceived and distinguished by the vision system, resulting in cumulative errors over long-term monitoring and requiring frequent manual calibration and maintenance.

[0004] While the pure contact method provides direct and stable data that is unaffected by ambient light, its drawbacks include high cost, the need for complex wiring and power supply systems, a limited number of measurement points, difficulty in achieving a comprehensive understanding of the overall deformation of the bridge, and a large workload for installation and maintenance. Summary of the Invention

[0005] This application proposes a bridge deformation monitoring system that integrates contact and non-contact methods to solve the technical problems of pure visual monitoring, such as the inability to perceive and correct changes in the target's own state and poor environmental adaptability, as well as the high cost and difficulty in deployment of pure contact monitoring.

[0006] To achieve the above objectives, this application provides the following technical solution: Firstly, this application proposes a bridge deformation monitoring system that integrates contact and non-contact methods, comprising: At least one intelligent fusion target unit, fixed at the monitoring point on the bridge, is used to provide optical identification and synchronously collect its own status data, including: Target body; An optical marking layer is set on the front of the target body; The contact sensor group integrated inside the target body includes at least an angle sensor for monitoring the tilt angle of the target itself and a temperature sensor for monitoring the temperature of the target. A microprocessor connected to the contact sensor array is used to acquire and preprocess status data; A dual-mode communication module connected to the microprocessor is used to send status data; At least one non-contact vision acquisition unit is deployed at a stable reference point outside the bridge to periodically acquire optical identification images of the intelligent fusion target unit and generate image data. At least one data processing unit is connected to the communication modules of the non-contact visual acquisition unit and the intelligent fusion target unit, respectively, for receiving image data and status data, and for correcting the image data based on the status data to generate monitoring results of bridge deformation.

[0007] In conjunction with the first aspect, the contact sensor group also includes one or more of a vibration sensor, a strain sensor, and a humidity sensor, which are fixed to the target body.

[0008] In conjunction with the first aspect, the dual-mode communication module includes a wired communication unit, a wireless communication unit, and an automatic switching circuit; the microprocessor detects the connectivity status of the wired communication link and controls the automatic switching circuit to achieve redundancy backup between wired and wireless communication.

[0009] In conjunction with the first aspect, the automatic switching circuit is a relay or electronic switch that switches to wireless communication when a wired communication failure is detected. When wired communication is restored, switch back to wired communication and control the wireless communication unit to enter sleep mode.

[0010] In conjunction with the first aspect, the intelligent fusion target unit also includes a hybrid power supply module, which includes an AC power interface, a solar power supply unit, an energy storage battery, and a power management IC, for providing primary and backup redundant power supply to the target.

[0011] In conjunction with the first aspect, the power management IC prioritizes mains power supply and automatically switches to energy storage battery power supply when mains power is disconnected; the solar power supply unit charges the energy storage battery.

[0012] In conjunction with the first aspect, the data processing unit is specifically used for: correcting geometric errors in image displacement calculation caused by target tilt based on tilt data collected by the tilt sensor; and compensating for target position drift caused by thermal expansion and contraction based on temperature data collected by the temperature sensor and according to a preset temperature-displacement drift model.

[0013] In conjunction with the first aspect, the data processing unit is also used to: generate an installation loosening warning message when the strain data collected by the strain sensor exceeds a preset threshold.

[0014] In conjunction with the first aspect, in the intelligent fusion target unit: the tilt sensor is fixed inside the target body and parallel to the mounting surface; the temperature sensor is in close contact with the inner wall of the target body and fixed to the target body.

[0015] In conjunction with the first aspect, the data processing unit is further configured to perform the following steps: Receive image data from a non-contact vision acquisition unit; the image data includes the optical identifier of the intelligent fusion target. Receive status data collected by the contact sensor group from the intelligent fusion target unit. The status data includes at least the target's tilt angle data and temperature data. Analyze image data and calculate the visual displacement data of the target; Geometric error correction is performed on visual displacement data based on tilt angle data; Based on temperature data, temperature drift compensation is performed on the corrected visual displacement data according to a preset temperature-displacement drift model. Output the final three-dimensional deformation time history data.

[0016] The beneficial effects of this application are as follows: By integrating contact sensors inside the target, the system gains the ability to sense changes in its own state, successfully solving the long-standing problem of uncorrectable target drift in pure visual monitoring. Through hardware-level sensor integration and data correction, sub-pixel-level monitoring accuracy is achieved, providing more reliable data for bridge safety assessment. This system is not limited by light or weather; even at night or in rainy or foggy weather, the contact sensors can still provide crucial data, ensuring continuous monitoring. The automatic drift compensation function extends the manual maintenance cycle from 3 months to 12 months, reducing maintenance workload. Multi-source data serve as backups and verifications for each other, ensuring that a single sensor failure does not affect the overall system. The strain sensor can also provide real-time warnings of loose installation, avoiding data errors caused by target loosening. Compared to solutions that deploy a large number of contact sensors, this system achieves a balance between high precision and high reliability at a lower cost, reducing deployment costs by approximately 50%.

[0017] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0018] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0019] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0020] In the attached diagram: Figure 1 This is a system composition diagram of a bridge deformation monitoring system that integrates contact and non-contact methods according to an embodiment of the present invention. Figure 2 This is a system scenario architecture diagram of a bridge deformation monitoring system that integrates contact and non-contact methods according to an embodiment of the present invention. Figure 3 This is a structural diagram of the intelligent fusion target unit in an embodiment of the present invention. Detailed Implementation

[0021] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0022] Example 1: As Figure 1 and Figure 2 As shown, this application provides a bridge deformation monitoring system that integrates contact and non-contact methods, including: At least one intelligent fusion target unit is fixed to the monitoring point of the bridge structure to provide optical identification and synchronously collect its own status data.

[0023] In the embodiments of this application, the spacing between multiple intelligent fusion target units is 10-50m; they are fixed to the monitoring points of the bridge, such as the bottom plate or side plate of the box girder, by expansion bolts.

[0024] like Figure 3 As shown, the intelligent fusion target unit includes: the target body; An optical marking layer is set on the front of the target body; The contact sensor group integrated inside the target body includes at least an angle sensor for monitoring the tilt angle of the target itself and a temperature sensor for monitoring the temperature of the target. A microprocessor connected to the contact sensor array is used to acquire and preprocess status data; A dual-mode communication module connected to the microprocessor is used to send status data; In the embodiments of this application, the intelligent fusion target unit includes a target body with an IP68 protection rating. An optical marking layer, which can be a high-intensity reflective film, is provided on the front side of the target body to provide high-contrast recognition features in the visual image. A set of contact sensors is integrated inside the target body, upgrading the traditional passive target into an intelligent node with multimodal sensing capabilities.

[0025] At least one non-contact vision acquisition unit is deployed at a stable reference point outside the bridge, i.e., a bridge-bank stability point, to periodically acquire optical identification images of the intelligent fusion target unit and generate image data. In the embodiments of this application, the non-contact visual acquisition unit is preferably an industrial camera. There are at least two industrial cameras to achieve multi-point monitoring, and the resolution needs to reach 1280×1024 or higher. They are deployed on stable reference points outside the bridge, such as bridge abutments, pier tops or riverside buildings, and their field of view can cover all intelligent fusion target units.

[0026] At least one data processing unit is connected to the communication modules of the non-contact visual acquisition unit and the intelligent fusion target unit, respectively, for receiving image data and status data, and for correcting the image data based on the status data to generate monitoring results of bridge deformation.

[0027] In the embodiments of this application, the data processing unit can be deployed as an edge computing gateway. The edge computing gateway has a built-in data processing and fusion module and is deployed in a chassis near the bridge or in a remote monitoring center. The data processing unit is connected to the dual-mode communication modules of the non-contact visual acquisition unit and all intelligent fusion target units via wired or wireless means, respectively. In actual implementation, the system of this application also includes a power supply unit. The power supply unit is used to provide power to the intelligent fusion target units and the visual acquisition unit.

[0028] In one specific embodiment, the data processing and fusion module includes an image processing module and a sensor data processing module. The image processing module processes image data acquired by industrial cameras to determine bridge deformation data that can be fused. The sensor data processing module processes bridge structure state data, i.e., deformation data, acquired by the intelligent fusion target unit. Then, the bridge deformation data acquired by the industrial cameras and the bridge structure state data acquired by the intelligent fusion target unit are fused by the data fusion module to determine the deformation state of the bridge structure. The deformation state is visualized and an alarm is issued through the monitoring system. The results are displayed visually.

[0029] In one alternative embodiment, in addition to the reflective film, the optical marking layer may also take the form of an infrared LED array (for night vision cameras), a visible light LED (programmable control of flashing frequency to distinguish different targets), a laser dot or a QR code (which can carry target ID information), etc.

[0030] In one alternative embodiment, the dual-mode communication module may be configured with wired communication only (suitable for cable trays with existing pipelines), wireless communication only (suitable for new projects), or other communication technology combinations such as 5G and microwave communication, depending on the requirements.

[0031] In one alternative embodiment, the hybrid power module may use only mains power, only solar power plus batteries, or introduce other new energy sources such as wind power, depending on the site conditions.

[0032] In one alternative embodiment, the sensor in this application may also integrate other sensors, such as a barometric pressure sensor (for monitoring wind load), a wind speed sensor, a corrosion sensor, etc., according to monitoring requirements.

[0033] In one alternative embodiment, the visual acquisition unit can be a fixed-focus camera with a fixed focal length, or a pan-tilt camera with remotely controllable focal length and angle adjustment. For ultra-long bridges, a multi-camera array can also be used for splicing monitoring.

[0034] Example 2: The contact sensor group also includes one or more of a vibration sensor, a strain sensor, and a humidity sensor, which are fixed to the target body.

[0035] In the embodiments of this application, such as Figure 3 As shown, the contact sensor group includes at least: A dual-axis tilt sensor is fixed inside the target body. The dual-axis tilt sensor is strictly parallel to the target's mounting surface and is used to monitor the tilt angle of the target itself caused by loose installation or foundation deformation.

[0036] In a preferred embodiment, the measurement range is ±15° with a resolution as high as 0.001°. A temperature sensor is installed close to the inner wall of the target body to accurately monitor the ambient temperature of the target, providing a data basis for subsequent thermal expansion and contraction drift compensation. Its measurement range is -40°C to +85°C, with an accuracy of ±0.5°C.

[0037] To further enhance the richness of monitoring data and the robustness of the system, such as Figure 1 As shown, the contact sensor array may also optionally include a triaxial vibration sensor, positioned near the connection point between the target body and the bridge, for detecting the dynamic response of the bridge under the influence of vehicles, wind loads, etc. Its measurement range is ±2g, and its frequency range is 0-100Hz.

[0038] The contact sensor array in this application also includes a strain sensor, which is attached to the connecting bolts between the target body and the bridge, i.e., the fixing bolts, for real-time monitoring of installation stress changes at the connection point. When the stress exceeds a threshold, it can provide an early warning of installation loosening. Its measurement range is ±5000με, and its accuracy is ±1με. A humidity sensor, located on the side of the target body and directly exposed to the environment, is used to monitor ambient humidity, assist in determining weather conditions, and provide a reference for adjusting image recognition parameters. Its measurement range is 0–100% RH, and its accuracy is ±2% RH. All sensors are connected to a microprocessor integrated inside the target body.

[0039] The microprocessor is responsible for acquiring data from each sensor and performing preprocessing (such as filtering and noise reduction). The sensors and microprocessor are connected via various hardware interfaces, forming the hardware architecture: the tilt sensor is connected via an SPI interface; the vibration sensor and humidity sensor are connected via an I2C interface. In this application, temperature sensors are installed at the center of the target and on its sides / exposed environment, connected via a single-bus interface; strain sensors are acquired via a 12-bit ADC. The microprocessor preprocessing in this application employs a three-level architecture: the first level is hardware-level processing, including signal conditioning, digital filtering, and analog-to-digital conversion, performed in real-time by the microprocessor; the second level is algorithm-level processing, including multi-sensor time synchronization, adaptive filtering and noise reduction, and outlier data removal; the third level is fusion-level processing, where multi-source data correlation analysis is performed in the data processing unit.

[0040] Example 3: As Figure 3 As shown, the dual-mode communication module includes a wired communication unit, a wireless communication unit, and an automatic switching circuit; the microprocessor detects the connectivity status of the wired communication link and controls the automatic switching circuit to achieve redundancy backup between wired and wireless communication.

[0041] In the embodiments of this application, the dual-mode communication module includes a wired communication unit, a wireless communication unit, and an automatic switching circuit. The wired communication unit can be an Ethernet interface or an RS485 interface. The wireless communication unit can be a LoRa module, an NB-IoT module, or a 4G module. The automatic switching circuit is a relay or an electronic switch.

[0042] Example 4: The automatic switching circuit is a relay or electronic switch. When a wired communication failure is detected, it switches to wireless communication; when wired communication is restored, it switches back to wired communication and controls the wireless communication unit to enter a sleep state. The microprocessor monitors link connectivity by detecting the physical layer status of the wired communication link or by using heartbeat packets. When a wired communication failure is detected, the microprocessor controls the automatic switching circuit to seamlessly switch the communication link to the wireless communication unit. When wired communication is restored, the microprocessor controls the automatic switching circuit to switch back to wired communication and simultaneously controls the wireless communication unit to enter a sleep state to reduce power consumption. This hardware-level redundancy design greatly improves the reliability of monitoring data transmission.

[0043] Example 5: Figure 3 As shown, the intelligent fusion target unit also includes a hybrid power supply module, which includes an AC power interface, a solar power supply unit, an energy storage battery, and a power management IC, used to provide redundant power supply for the target.

[0044] In the embodiments of this application, the hybrid power supply module includes an AC mains interface, a solar power supply unit, an energy storage battery, and a supercapacitor, with the supercapacitor serving as backup power. The energy storage battery is a 12V 20Ah lithium battery and a power management IC. The AC mains interface is connected to 220V AC, which is converted to 12V DC via an AC / DC converter. The solar power supply unit is a 10W monocrystalline silicon solar panel, which charges the energy storage battery via a charging controller.

[0045] Example 6: The power management IC prioritizes mains power supply and automatically switches to energy storage battery power supply when the mains power is disconnected; the solar power supply unit charges the energy storage battery.

[0046] In the embodiments of this application, the power management IC intelligently manages two power sources: under normal operating conditions, it prioritizes mains power; when the mains power is disconnected, it automatically and seamlessly switches to the energy storage battery; when the mains power is restored, it switches back to mains power. Simultaneously, the solar power unit serves as a supplementary energy source, charging the energy storage battery and extending the system's endurance in areas without mains power. The entire system operates with a standby current of less than 10mA and a working current of less than 100mA in low-power mode, and can operate continuously for 7 days relying solely on solar power without sunlight.

[0047] Example 7: The data processing unit is specifically used for: Based on the tilt angle data collected by the tilt sensor, the geometric error in image displacement calculation caused by target tilt is corrected; and based on the temperature data collected by the temperature sensor, the target position drift caused by thermal expansion and contraction is compensated according to the preset temperature-displacement drift model.

[0048] In the embodiments of this application, the data processing unit is used to receive image data from the non-contact visual acquisition unit and state data from all intelligent fusion target units, and perform deep fusion processing. It receives image data containing the optical identifiers of the intelligent fusion targets, as well as state data collected from the contact sensor groups of each target. This state data includes at least the target's own tilt angle data and temperature data, and may also include data such as vibration, strain, and humidity. An image recognition algorithm is used to parse the image data, calculate the pixel displacement of each target in the image coordinate system, and, combined with the camera's intrinsic and extrinsic parameters, convert it into visual displacement data in the physical coordinate system. Based on the real-time tilt angle data collected by the tilt angle sensor, according to the principles of trigonometric geometry, the additional displacement error introduced by the target tilt is calculated and subtracted from the visual displacement data, thereby correcting the measurement error caused by the target attitude change. Based on the real-time temperature data collected by the temperature sensor, a pre-calibrated temperature-displacement drift model (e.g., a linear or polynomial function) is invoked to calculate the target position change caused by thermal expansion and contraction, and secondary compensation is performed on the geometrically corrected visual displacement data.

[0049] In one embodiment of this application, the data processing unit employs a four-layer progressive architecture for fusion processing. The first layer is data-level fusion: a weighted average is calculated from the displacement measured visually, the displacement converted from tilt angle, and the displacement obtained by integrating vibration sensor data. The weights are automatically adjusted based on environmental conditions—visual weights are higher (60%) under normal lighting conditions, while tilt sensor weights are increased in insufficient lighting or rainy / foggy weather. The second layer is state-level fusion: an adaptive extended Kalman filter is used, treating displacement, velocity, and tilt angle as state variables. The optimal estimate is obtained through a prediction-update recursive process, while noise parameters are automatically adjusted based on measurement residuals to improve estimation accuracy. The third layer is feature-level fusion: features such as maximum deformation, deformation rate of change, and dominant frequency are extracted from time-series data, and feature vectors are constructed by combining them with environmental features such as temperature and vibration. The fourth layer is decision-level fusion, used for determining the bridge's health status. In specific implementation, the bridge's health status is assessed based on the feature vectors, outputting a five-level health assessment result, specifically including healthy, alert, warning, alarm, and danger.

[0050] Temperature compensation for thermal expansion and contraction drift employs a multi-factor coupling correction method. The displacement drift caused by total heat is the sum of the material thermal expansion component, the installation structure thermal deformation component, the sensor circuit temperature drift component, and the temperature gradient effect component.

[0051] In one specific embodiment, the temperature-displacement drift model is as follows: ; in, The thermal expansion component of the target body material; This refers to the thermal deformation component of the installation structure. The temperature drift component of the sensor circuit is automatically corrected by the temperature compensation algorithm built into the microprocessor, and its calibration value is provided by the sensor at the factory. This represents the temperature gradient effect component.

[0052] During actual monitoring, the data processing unit acquires the temperature data collected by the temperature sensor in real time, calculates the temperature change, and uses the corresponding coefficients based on the temperature change range to calculate the total thermally induced displacement drift. Then, the drift amount is subtracted from the visual displacement data to obtain the temperature-compensated displacement data.

[0053] The thermal expansion component of the material is used to measure the temperature of the target body and calculate the length change of the target due to temperature changes based on the material's thermal expansion coefficient. The thermal deformation component of the installation structure is the thermal deformation of the installation structure (bolts, connectors), which is corrected by a piecewise linear model; The temperature drift component of the sensor circuit compensates for the temperature drift of the sensor circuit itself; for cases where there is a temperature difference between the upper and lower surfaces of the target, temperature gradient effect correction is also performed.

[0054] The temperature gradient effect component is the total thermally induced displacement drift obtained by superimposing the above four components during correction, and is subtracted from the measured displacement. The system also uses a nighttime benchmark method for adaptive coefficient updates—using data from nighttime periods with no vehicle load and stable temperatures to establish a temperature-displacement benchmark relationship, automatically correcting the compensation coefficients during long-term operation without the need for manual periodic calibration.

[0055] In the specific implementation process of this application, the data processing unit uses an adaptive extended Kalman filter to fuse and estimate multi-source data in order to obtain the optimal three-dimensional deformation time history data.

[0056] Specifically, the single camera acquires 3D data using a monocular 3D measurement method based on a target of known size.

[0057] The intelligent target has four corner feature points. Before leaving the factory, high-precision measuring equipment is used to calibrate the three-dimensional coordinates of the four feature points (such as corner points) in the target coordinate system, with a point-to-point distance accuracy of 0.01 mm. The origin of the target coordinate system is usually set at the center of the target.

[0058] During measurement, the camera captures an image of the target. The system first performs sub-pixel-level feature point detection (accuracy 0.01 pixels). For example, a sub-pixel corner detection algorithm is used to extract the image coordinates of four feature points. Then, the target depth is calculated based on the ratio between the actual size of the target and the size in the image.

[0059] Given the camera intrinsic parameter matrix, the EPNP (Efficient Perspective n-Point) algorithm is further used to solve the rotation matrix and translation vector of the target relative to the camera based on the known world coordinates of the feature points and the detected image coordinates, thereby obtaining the three-dimensional coordinates of the target center point.

[0060] Finally, by combining the data from the target's built-in tilt sensor, the geometric errors caused by the target's tilt are corrected to obtain accurate three-dimensional deformation (horizontal lateral deformation, vertical deflection, and deformation along the bridge axis).

[0061] By establishing a reference coordinate system and calibrating the reference pose of the bridge under no-load conditions, the difference between the real-time measured value and the reference value is the three-dimensional deformation time history data of the bridge.

[0062] In one specific embodiment, the Kalman filter defines three state variables for each intelligent fusion target unit: the current displacement value (in millimeters), the current velocity value (in millimeters per second), and the current acceleration value (in millimeters per square second). These three state variables constitute a three-dimensional state vector, used to describe the target's motion state in a certain direction. For deformation monitoring in three-dimensional space, the system operates three filters independently in three directions: horizontal lateral, vertical deflection, and along the bridge axis. The filters predict the current state based on the optimal estimate from the previous moment. The basic rule for prediction is: assuming the target undergoes uniform acceleration within a very short time interval. Therefore, the predicted displacement value at the current moment equals the displacement at the previous moment plus the velocity at the previous moment multiplied by the sampling time interval, plus half of the acceleration at the previous moment multiplied by the square of the sampling time interval; the predicted velocity value at the current moment equals the velocity at the previous moment plus the acceleration at the previous moment multiplied by the sampling time interval; and the predicted acceleration value at the current moment directly uses the acceleration estimate from the previous moment. The sampling time interval is fixed at 0.1 seconds in this system. The prediction process introduces process noise to represent the uncertainty of the model itself. The magnitude of the process noise is controlled by the process noise covariance matrix, which will be dynamically adjusted in subsequent adaptive steps. The filter acquires three independent types of data at each time step. The first type of observation data comes from the non-contact visual acquisition unit, i.e., the visual displacement value of the target calculated by the image processing algorithm. The second type of observation data comes from the tilt sensor inside the intelligent fusion target unit. Based on the target tilt angle measured by the tilt sensor and the known target installation height, the system calculates the displacement change caused by the tilt using trigonometric relationships, and uses this change as the second displacement observation value.

[0063] The filter's adaptive mechanism can dynamically adjust the process noise covariance matrix based on actual measurement conditions. The specific implementation is as follows: First, the filter calculates a residual vector at each time step, which equals the actual observation minus the predicted observation. Then, a sliding window of length 10 time steps is maintained to store the residual vectors of the most recent 10 time steps. At every other sampling time step, the filter calculates the actual covariance of these 10 residual vectors.

[0064] Meanwhile, the filter calculates a theoretical residual covariance based on the predicted covariance and the observed noise covariance of the current state.

[0065] Next, the filter compares the actual residual covariance with the theoretical residual covariance. Specifically, it calculates the ratio of the sum of the diagonal elements of their matrices. If this ratio is greater than 1.2, it indicates that the actual measurement error is greater than the model's prediction error, meaning the current process noise is underestimated. In this case, the filter multiplies the process noise covariance matrix by this ratio, thereby increasing the process noise. Conversely, if the ratio is less than 0.8, the process noise covariance matrix is ​​appropriately reduced.

[0066] To prevent excessive changes in the process noise covariance matrix during adjustment, which could lead to filter instability, upper and lower limits are set: the minimum value corresponds to a very small noise level, and the maximum value corresponds to a relatively large noise level; the adjusted value will not exceed this range. After obtaining the adaptively adjusted process noise covariance, the filter performs a standard state update: first, the Kalman gain is calculated, which determines the weight ratio of the predicted and observed values ​​in the final estimate. Then, the actual observed values ​​at the current time (i.e., the weighted fusion result of the three types of displacement observations) are used to correct the predicted state, obtaining the optimal state estimate for the current time. Finally, the error covariance matrix of the state estimate is updated to prepare for the prediction at the next time step.

[0067] Example 8: The data processing unit is also used to generate an installation loosening warning message when the strain data collected by the strain sensor exceeds a preset threshold.

[0068] In the embodiments of this application, the data processing unit also monitors the data of the strain sensor in real time, and when the strain value exceeds the preset safety threshold, it immediately generates an installation loosening warning message for the target point.

[0069] Example 9: In the intelligent fusion target unit: the tilt sensor is fixed inside the target body and parallel to the mounting surface; the temperature sensor is in close contact with the inner wall of the target body and fixed to the target body.

[0070] Specifically, the tilt sensor is set strictly parallel to the target mounting surface to ensure that the tilt angle measured by the sensor accurately reflects the target's own attitude changes, thus providing angular data for geometric error correction. The temperature sensor is in close contact with the inner wall of the target body, enabling it to quickly sense the actual temperature changes of the target body, reducing measurement lag caused by air convection or thermal radiation. This allows the temperature compensation model to calculate the amount of thermal expansion and contraction drift in real time and accurately, effectively suppressing the cumulative temperature drift in long-term monitoring. The presence of multiple temperature sensors facilitates the calculation of the temperature difference between the upper and lower surfaces of the target. All sensors are fixed inside the target body, forming an integrated sealed structure, preventing external sensors from being damaged or displaced by environmental factors such as wind, rain, dust, and vibration.

[0071] Example 10: The data processing unit is further configured to perform the following steps: Receive image data from a non-contact vision acquisition unit; the image data includes the optical identifier of the intelligent fusion target. Receive status data collected by the contact sensor group from the intelligent fusion target unit. The status data includes at least the target's tilt angle data and temperature data. Analyze image data and calculate the visual displacement data of the target; Geometric error correction is performed on visual displacement data based on tilt angle data; Based on temperature data, temperature drift compensation is performed on the corrected visual displacement data according to a preset temperature-displacement drift model. Output the final three-dimensional deformation time history data.

[0072] In the embodiments of this application, the data processing unit combines the corrected displacement data with the dynamic response data acquired by the vibration sensor to ultimately output high-precision three-dimensional deformation time history data of key bridge points. This high-precision three-dimensional deformation time history data accurately reflects the structural deformation of the bridge, rather than the changes in the target itself.

[0073] In this system, a temperature sensor inside the target collects real-time data on temperature increases, while a tilt sensor detects minute target tilts caused by temperature differences. This data, along with visual images, is sent to the data processing and fusion unit.

[0074] The data processing unit first processes the visual image to obtain raw displacement data that includes the actual bridge deformation and thermal expansion / contraction drift. Next, it corrects for geometric errors using tilt angle data, and then uses a temperature-displacement drift model to accurately compensate for the thermal expansion / contraction portion. The final output data eliminates spurious displacements caused by temperature, retaining only the actual temperature-induced deformation of the bridge structure. The system's monitoring accuracy reaches ±0.05mm, compared to ±0.1mm for a pure vision system.

[0075] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A bridge deformation monitoring system integrating contact and non-contact methods, characterized in that, include: At least one intelligent fusion target unit, fixed at the monitoring point on the bridge, is used to provide optical identification and synchronously collect its own status data, including: Target body; An optical marking layer is set on the front of the target body; The contact sensor group integrated inside the target body includes at least an angle sensor for monitoring the tilt angle of the target itself and a temperature sensor for monitoring the temperature of the target. A microprocessor connected to the contact sensor array is used to acquire and preprocess status data; A dual-mode communication module connected to the microprocessor is used to send status data; At least one non-contact vision acquisition unit is deployed at a stable reference point outside the bridge to periodically acquire optical identification images of the intelligent fusion target unit and generate image data. At least one data processing unit is connected to the communication modules of the non-contact visual acquisition unit and the intelligent fusion target unit, respectively, for receiving image data and status data, and for correcting the image data based on the status data to generate monitoring results of bridge deformation.

2. The bridge deformation monitoring system integrating contact and non-contact methods as described in claim 1, characterized in that, The contact sensor group also includes one or more of a vibration sensor, a strain sensor, and a humidity sensor, which are fixed to the target body.

3. The bridge deformation monitoring system integrating contact and non-contact methods as described in claim 1, characterized in that, The dual-mode communication module includes a wired communication unit, a wireless communication unit, and an automatic switching circuit; the microprocessor detects the connectivity status of the wired communication link and controls the automatic switching circuit to achieve redundancy backup between wired and wireless communication.

4. The bridge deformation monitoring system integrating contact and non-contact methods as described in claim 3, characterized in that, The automatic switching circuit is a relay or electronic switch that switches to wireless communication when a wired communication failure is detected. When wired communication is restored, switch back to wired communication and control the wireless communication unit to enter sleep mode.

5. The bridge deformation monitoring system integrating contact and non-contact methods as described in claim 1, characterized in that, The intelligent fusion target unit also includes a hybrid power supply module, which includes an AC power interface, a solar power supply unit, an energy storage battery, and a power management IC, used to provide redundant power supply for the target.

6. The bridge deformation monitoring system integrating contact and non-contact methods as described in claim 5, characterized in that, The power management IC prioritizes mains power supply and automatically switches to energy storage battery power supply when mains power is disconnected. The solar power unit charges the energy storage battery.

7. The bridge deformation monitoring system integrating contact and non-contact methods as described in claim 1, characterized in that, The data processing unit is specifically used to: correct the geometric error in image displacement calculation caused by target tilt based on the tilt data collected by the tilt sensor; Furthermore, based on the temperature data collected by the temperature sensor, the target position drift caused by thermal expansion and contraction is compensated according to the preset temperature-displacement drift model.

8. The bridge deformation monitoring system integrating contact and non-contact methods as described in claim 1, characterized in that, The data processing unit is also used for: When the strain data collected by the strain sensor exceeds a preset threshold, an early warning message for loose installation is generated.

9. The bridge deformation monitoring system integrating contact and non-contact methods as described in claim 1, characterized in that, In the intelligent fusion target unit: The tilt sensor is fixed inside the target body and parallel to the mounting surface; The temperature sensor is attached to the inner wall of the target body and fixed to the target body.

10. The bridge deformation monitoring system integrating contact and non-contact methods as described in claim 1, characterized in that, The data processing unit is also configured to perform the following steps: Receive image data from a non-contact vision acquisition unit; the image data includes the optical identifier of the intelligent fusion target. Receive status data collected by the contact sensor group from the intelligent fusion target unit. The status data includes at least the target's tilt angle data and temperature data. Analyze image data and calculate the visual displacement data of the target; Geometric error correction is performed on visual displacement data based on tilt angle data; Based on temperature data, temperature drift compensation is performed on the corrected visual displacement data according to a preset temperature-displacement drift model. Output the final three-dimensional deformation time history data.

Images