Method for monitoring differential settlement of highway subgrade

Abstract

A highway subgrade differential settlement monitoring method belongs to the field of road construction, and comprises the following steps: S1, in a highway subgrade construction and reconstruction stage, according to a preset gradient layout scheme, burying distributed fiber grating sensors and MEMS acceleration and tilt angle sensors at subgrade key nodes, meanwhile, a built-in environment sensor and a reference calibration node are arranged, the environment sensor is used for collecting temperature and humidity environment parameters in real time, and the reference calibration node provides a stable reference to realize sensor self-calibration and environment compensation; according to the scheme, the distributed fiber bragg gratings and the MEMS sensors are arranged at the key nodes of the roadbed, the unmanned aerial vehicle flight technology is combined, continuous, automatic and real-time monitoring of roadbed settlement, flatness, cross slope and center line deviation is achieved, and the problems that a traditional monitoring method is poor in real-time performance and discontinuous in data are fundamentally solved.

Description

Technical Field

[0001] The invention relates to the field of road construction, and more specifically to a method for monitoring uneven settlement of a highway roadbed. Background Art

[0002] With the continuous advancement of highway construction and maintenance in my country, the problem of uneven settlement of highway subgrades has become increasingly prominent, directly affecting the safety and service life of roads. Traditional monitoring methods mainly rely on manual periodic leveling, which has defects such as long measurement cycle, discontinuous data collection, and insufficient real-time performance. In recent years, schemes for real-time monitoring of subgrade settlement using sensor networks and drone flight technology have been gradually proposed. Data collected from different data sources have large differences in sampling frequency, noise characteristics, spatial resolution, etc. How to achieve high-precision spatiotemporal synchronization and data fusion, especially the computational load problem under real-time monitoring requirements, is a difficulty that needs to be solved urgently. Summary of the invention

[0003] To address this problem, this paper proposes a method for monitoring uneven settlement of highway subgrade.

[0004] A method for monitoring uneven settlement of a highway subgrade comprises the following steps: S1. During the construction and reconstruction of the highway subgrade, distributed fiber grating sensors and MEMS acceleration and inclination sensors are buried at key nodes of the subgrade according to the preset gradient layout plan. At the same time, built-in environmental sensors and reference calibration nodes are set. The environmental sensors are used to collect temperature and humidity environmental parameters in real time. The reference calibration nodes provide a stable benchmark to achieve sensor self-calibration and environmental compensation. S2, using drones equipped with laser radar and high-definition camera systems to collect high-density 3D point cloud data, orthophoto data, and simultaneously record GNSS and IMU data according to pre-planned routes and flight parameters; S3. Set up multiple ground base stations in the monitoring area, use differential GPS technology to perform preliminary correction on the GNSS data collected by the drone, and use tight coupling solution in combination with IMU data to achieve preliminary spatiotemporal synchronization; S4: Based on the data from steps S1 and S3, multi-level correction and adaptive spatiotemporal synchronization algorithm are used for data fusion. First, the sensor output is corrected by local linear fitting using the reference calibration node data. Then, the extended Kalman filter is used to adjust the fusion weights of each data source according to the temperature and humidity data collected by the environmental sensor, thereby achieving accurate alignment of each data source in a unified coordinate system and building a real-time updated roadbed digital twin model. S5. Extract the local settlement, flatness, cross slope and centerline deviation parameters of the roadbed from the fused data, and automatically issue a warning signal when the real-time monitoring data exceeds the preset safety threshold.

[0005] Preferably, the sensor network deployment scheme in S1 adopts the method of deploying fiber grating sensors at fixed intervals in the longitudinal direction of the roadbed, and increasing the density of MEMS sensors in the lateral key turning areas to achieve local high-resolution data acquisition.

[0006] Preferably, the UAV flight parameters in S2 include a flight altitude between 100 and 150 meters, a laser radar sampling frequency of not less than 600 kHz, and a cross-coverage strategy is adopted for the route to ensure redundancy and integrity of data collection.

[0007] Preferably, the digital twin model is constructed by first performing geometric correction to map the drone laser point cloud data and sensor data to a unified coordinate system, and then using a multi-model fusion strategy to generate a real-time updated roadbed digital twin model to reflect local settlement and overall deformation characteristics.

[0008] Preferably, the multi-level correction and adaptive spatiotemporal synchronization algorithm further adopts an incremental update mechanism to locally recalculate data segments that have only changed significantly to reduce the overall computational load; the incremental update mechanism includes real-time error monitoring of local data changes and setting a dynamic change threshold. When the local change is lower than the threshold, lightweight processing is used to ensure that the overall data fusion accuracy remains basically unchanged.

[0009] Preferably, the adaptive filtering algorithm in the extended Kalman filter dynamically adjusts the filtering parameters through an error feedback mechanism, and uses the real-time deviation between the sensor output and the reference calibration node data to update the process noise and observation noise covariance. At the same time, a dynamic sampling rate adjustment module is provided, which automatically reduces the sampling frequency to achieve lightweight when the data error remains within a preset tolerance, and immediately increases the sampling frequency to ensure data accuracy when the error exceeds the tolerance.

[0010] Preferably, a sparse sampling strategy is adopted in the data fusion process to reduce the sampling frequency in areas or time periods with low data change amplitudes, but at the same time, a real-time error monitoring and interpolation reconstruction mechanism for key data points is set to ensure that data density is restored in time through local interpolation or supplementary sampling while lightweight processing is performed, thereby reducing the computational complexity while ensuring the overall accuracy of data fusion.

[0011] Compared with the prior art, the advantages of the present invention are: (1) This solution deploys distributed fiber Bragg grating and MEMS sensors at key nodes of the roadbed and combines them with drone flight technology to achieve continuous, automated, and real-time monitoring of roadbed settlement, flatness, cross slope, and centerline deviation, fundamentally solving the problems of poor real-time performance and discontinuous data in traditional monitoring methods. (2) This solution adopts an incremental update mechanism and dynamic sampling strategy, which automatically reduces the sampling frequency in areas with small data changes and only recalculates the data that has changed significantly locally, thereby effectively reducing the real-time computing load and ensuring efficient operation of the system; (3) This scheme further introduces an error feedback mechanism: during the dynamic sampling process, when it is detected that the local data error exceeds the preset threshold, the system automatically increases the sampling frequency of the area and triggers local recalculation, thereby ensuring that the data fusion accuracy is not affected while achieving lightweight processing, thus achieving a balance between real-time and accuracy. DETAILED DESCRIPTION

[0012] The present application is further described in detail below in conjunction with examples, comparative examples and performance testing experiments. These examples should not be construed as limiting the scope of protection claimed in the present application.

[0013] Example A method for monitoring uneven settlement of a highway subgrade comprises the following steps: S1. During the construction and reconstruction of the highway subgrade, distributed fiber grating sensors and MEMS acceleration and inclination sensors are buried at key nodes of the subgrade according to the preset gradient layout plan. The fiber grating sensor measures the strain data inside the subgrade, which is represented by the symbol ε; the MEMS acceleration sensor mainly measures the acceleration, which is represented by the symbol a; the MEMS inclination sensor measures the inclination angle, which is represented by θ; at the same time, a built-in environmental sensor and a reference calibration node are set. The environmental sensor is used to collect temperature and humidity environmental parameters in real time, and the reference calibration node provides a stable benchmark to achieve sensor self-calibration and environmental compensation; The sensor network deployment scheme described in S1 adopts the method of deploying fiber grating sensors at fixed intervals in the longitudinal direction of the roadbed, and increasing the deployment density of MEMS sensors in the lateral key turning areas.

[0014] The key nodes are at the following locations: Construction joints and joints. During the highway roadbed construction process, different batches of filling and compaction may cause construction joints or joints between the new and old roadbeds. These areas are prone to local settlement and uneven deformation due to differences in construction technology and compaction degree.

[0015] Structural transition areas, such as bridges, culverts, tunnel entrances and exits, and the connection areas between the road surface and the roadbed, are often sensitive areas of settlement and deformation due to the obvious changes in structural stiffness and load transfer methods.

[0016] The side slope and slope foot area may experience large local settlement or uneven deformation due to geological conditions (such as soft soil layer, large slope) and external environmental influences (wind, rain, temperature changes, etc.). Therefore, setting sensors here can effectively capture roadbed changes.

[0017] S2. Using a drone equipped with a laser radar and a high-definition camera system, collect high-density three-dimensional point cloud data, orthophoto data, and simultaneously record GNSS and IMU data according to the pre-planned route and flight parameters; the drone flight parameters in S2 include a flight altitude between 100 and 150 meters, a laser radar sampling frequency of not less than 600 kHz, and a cross-coverage strategy for the route.

[0018] S3. Set up multiple ground base stations in the monitoring area, use differential GPS technology to perform preliminary correction on the GNSS data collected by the drone, and use tight coupling solution in combination with IMU data to achieve preliminary spatiotemporal synchronization; Step 1: In the monitoring area, select several high points or locations with wide views to deploy ground base stations according to the coverage of the drone route and terrain conditions. The distance between the base stations should ensure that the drone can receive signals from at least two or more base stations at the same time during flight. Each ground base station is equipped with a high-precision GPS receiver and a differential signal processing module. The base station establishes a stable data communication link with the central control station (for example, through a wireless network or a dedicated frequency band) to transmit correction signals in real time. Step 2: The drone is equipped with a GNSS receiving module to collect raw satellite signal data during flight. The drone also records flight status data (timestamp, preliminary position data). The ground base station uses its known precise position to compare with the satellite data received in the same time period, calculates the position error (i.e., differential correction value), and transmits the correction information to the drone in real time via a wireless link. The GNSS module on the drone combines the correction value with the raw data and outputs precise positioning data after differential GPS correction; Step 3. The IMU sensor on the drone records acceleration and angular velocity data in real time at a high sampling rate. These data reflect the motion state and attitude changes of the drone. Establish a state space model, in which the state variables include position, velocity and attitude. In the prediction stage, the IMU data is used to predict the state of the drone through a discrete time dynamics model. In the update stage, the positioning data corrected by differential GPS is used as the observation value to correct the predicted state. The extended Kalman filter (EKF) algorithm is used to tightly couple the prediction and update processes to achieve high-frequency state estimation, thereby compensating for the uncertainty caused by intermittent loss or noise of GNSS signals.

[0019] Step 4: Ensure that the ground base station and the UAV use a unified time synchronization mechanism (such as using GPS time or network clock synchronization) so that all data have accurate timestamps. Align the differential GPS data and IMU data of the UAV through timestamps to ensure data synchronization during the tightly coupled solution process, forming a continuous and accurate spatiotemporal state estimation.

[0020] Step 5. After tightly coupled solution, the system outputs status data containing precise position, speed and attitude information. This data serves as the basis for preliminary spatiotemporal synchronization and can be used for further data fusion and digital twin construction.

[0021] S4. Based on the data from steps S1 and S3, multi-level correction and adaptive spatiotemporal synchronization algorithm are used for data fusion. First, the sensor output is corrected by local linear fitting using reference calibration node data. Then, the extended Kalman filter is used to adjust the fusion weights of each data source according to the temperature and humidity data collected by the environmental sensor, and a real-time updated roadbed digital twin model is constructed. The digital twin model is constructed by first performing geometric correction to map the drone laser point cloud data and sensor data to a unified coordinate system, and then using a multi-model fusion strategy to generate a real-time updated roadbed digital twin model. The steps for local linear fitting correction using reference calibration node data are as follows: Step 1: Use S1 to collect the original output data of the sensor (strain, acceleration, inclination) and standard data measured at the reference calibration node (True value or high-precision calibration value). Step 2: Establish a linear model Assuming that there is a linear relationship between the sensor output and the true value in the local range, the model can be established: in, is the proportionality coefficient (slope), is the bias (intercept).

[0022] Step 3: Use the least squares method to solve the parameters and , so that the residual sum of squares is minimized, that is, solving: The solution obtained is: in, and They are and Step 4: After obtaining the calibration parameters, for the new sensor output , can be corrected by the following formula: This will adjust the sensor output to be consistent with the reference node data, thereby compensating for sensor drift or system deviation. Step 5: Since sensor drift gradually occurs with environmental changes, repeat the above steps when the error increases, and use the newly collected calibration data to adjust the model parameters. and Update to achieve dynamic self-calibration.

[0023] The specific steps of using extended Kalman filtering to adjust the fusion weights of each data source based on the temperature and humidity data collected by the environmental sensor are as follows: Step 1. Initialization: Define the state vector (strain, acceleration, inclination), and initial state estimation and the initial error covariance matrix . Set the system state transfer function and its Jacobian matrix ; Set the initial process noise covariance matrix ; Establish measurement models for each sensor (mapping state variables to measurements of strain, acceleration, and inclination), and the measurement matrix ; Define the basic measurement noise covariance matrix (corresponding to each sensor), and set the environmental compensation function (This function depends on the temperature and humidity Returns an adjustment factor to reflect the increase in noise level when the environment deteriorates). Step 2: Prediction step: Predict the next state based on the state transition model: Update the forecast error covariance: Step 3: Adaptive adjustment of measurement noise: Read the temperature collected by the environmental sensor and humidity Data; For each data source (strain, acceleration, inclination), use the environmental compensation function Get the adjustment factor (For example, if environmental parameters deviate from ideal values, Indicates that the noise increases); Update the satellite noise covariance matrix: in is the basic noise covariance of each data source under ideal conditions. Step 4: Update step: Collect the measurement data at the current moment (including strain, acceleration, inclination data); Calculate the predicted value of the measurement: Calculate measurement residuals (innovations): Calculate the innovation covariance: Calculate the Kalman gain: Update the state estimate: Update the error covariance matrix: Steps 2 to 4 are executed as a loop to continuously update the state estimation and achieve real-time spatiotemporal synchronization and data fusion.

[0024] In continuous data processing, a local data change threshold is set. When the sensor data change is detected to be less than the threshold, a lower sampling frequency (i.e., sparse sampling) is used; when the change exceeds the threshold, full data fusion calculation is performed. For local areas, only the above prediction and update steps are performed on the part with significant changes, and the remaining areas use the estimation results of the previous moment to reduce repeated calculations, thereby reducing the overall calculation load without affecting the fusion accuracy.

[0025] S5. Extract the local settlement, flatness, cross slope and centerline deviation parameters of the roadbed from the fused data, and automatically issue a warning signal when the real-time monitoring data exceeds the preset safety threshold.

[0026] The multi-level correction and adaptive spatiotemporal synchronization algorithm further adopts an incremental update mechanism to locally recalculate data segments that have only undergone significant changes; the incremental update mechanism includes real-time error monitoring of local data changes and setting a dynamic change threshold, and using lightweight processing when the local change is lower than the threshold.

[0027] The adaptive filtering algorithm in the extended Kalman filter dynamically adjusts the filtering parameters through an error feedback mechanism, and uses the real-time deviation between the sensor output and the reference calibration node data to update the process noise and observation noise covariance. At the same time, a dynamic sampling rate adjustment module is provided, which automatically reduces the sampling frequency to achieve lightweight when the data error remains within a preset tolerance, and immediately increases the sampling frequency to ensure data accuracy when the error exceeds the tolerance.

[0028] In the data fusion process, a sparse sampling strategy is adopted to reduce the sampling frequency in areas or time periods with low data change amplitudes, and a real-time error monitoring and interpolation reconstruction mechanism for key data points is set. While lightweight processing is performed, data density is restored in time through local interpolation and supplementary sampling.

[0029] The basic principles, main features and advantages of the present invention are shown and described above; those skilled in the art should understand that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only preferred examples of the present invention and are not used to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected; the scope of protection claimed by the present invention is determined by the attached claims and their effects.

Claims

1. A method for monitoring uneven settlement of a highway subgrade, characterized in that: The following steps are involved: S1. During the construction and reconstruction of the highway subgrade, distributed fiber grating sensors and MEMS acceleration and inclination sensors are buried at key nodes of the subgrade according to the preset gradient layout plan. At the same time, built-in environmental sensors and reference calibration nodes are set. The environmental sensors are used to collect temperature and humidity environmental parameters in real time. The reference calibration nodes provide a stable benchmark to achieve sensor self-calibration and environmental compensation. S2, using drones equipped with laser radar and high-definition camera systems to collect high-density 3D point cloud data, orthophoto data, and simultaneously record GNSS and IMU data according to pre-planned routes and flight parameters; S3. Set up multiple ground base stations in the monitoring area, use differential GPS technology to perform preliminary correction on the GNSS data collected by the drone, and use tight coupling solution in combination with IMU data to achieve preliminary spatiotemporal synchronization; S4: Based on the data from steps S1 and S3, multi-level correction and adaptive spatiotemporal synchronization algorithm are used for data fusion. First, the sensor output is corrected by local linear fitting using the reference calibration node data. Then, the extended Kalman filter is used to adjust the fusion weights of each data source according to the temperature and humidity data collected by the environmental sensor, and a real-time updated roadbed digital twin model is constructed. S5. Extract the local settlement, flatness, cross slope and centerline deviation parameters of the roadbed from the fused data, and automatically issue a warning signal when the real-time monitoring data exceeds the preset safety threshold.

2. A method for monitoring uneven settlement of a highway subgrade according to claim 1, characterized in that: The sensor network deployment scheme described in S1 adopts the method of deploying fiber grating sensors at fixed intervals in the longitudinal direction of the roadbed, and increasing the deployment density of MEMS sensors in the lateral key turning areas.

3. A method for monitoring uneven settlement of a highway subgrade according to claim 1, characterized in that: The UAV flight parameters in S2 include a flight altitude between 100 and 150 meters, a lidar sampling frequency of not less than 600 kHz, and a cross-coverage strategy for the route.

4. A method for monitoring uneven settlement of a highway subgrade according to claim 1, characterized in that: The digital twin model is constructed by first performing geometric correction to map the drone laser point cloud data and sensor data to a unified coordinate system, and then using a multi-model fusion strategy to generate a real-time updated roadbed digital twin model.

5. A method for monitoring uneven settlement of a highway subgrade according to claim 1, characterized in that: The multi-level correction and adaptive spatiotemporal synchronization algorithm further adopts an incremental update mechanism to locally recalculate data segments that have only undergone significant changes; the incremental update mechanism includes real-time error monitoring of local data changes and setting a dynamic change threshold, and using lightweight processing when the local change is lower than the threshold.

6. A method for monitoring uneven settlement of a highway subgrade according to claim 1, characterized in that: The adaptive filtering algorithm in the extended Kalman filter dynamically adjusts the filtering parameters through an error feedback mechanism, and uses the real-time deviation between the sensor output and the reference calibration node data to update the process noise and observation noise covariance. At the same time, a dynamic sampling rate adjustment module is provided, which automatically reduces the sampling frequency to achieve lightweight when the data error remains within a preset tolerance, and immediately increases the sampling frequency to ensure data accuracy when the error exceeds the tolerance.

7. A method for monitoring uneven settlement of a highway subgrade according to claim 1, characterized in that: In the data fusion process, a sparse sampling strategy is adopted to reduce the sampling frequency in areas or time periods with low data change amplitudes, and a real-time error monitoring and interpolation reconstruction mechanism for key data points is set. While lightweight processing is performed, data density is restored in time through local interpolation and supplementary sampling.

Images