A method and system for early warning of sluice gate subsidence based on UAV lidar and InSAR technology

Abstract

This invention provides a method and system for early warning of sluice gate settlement based on UAV lidar and InSAR technology. The method includes acquiring long-term SAR image data; performing differential interferometry processing on the SAR images using the InSAR method to obtain point-based and regional time-based settlement data for the sluice gate; comparing the annual average settlement value; if the difference in settlement values ​​is within a preset range, then using the point-based time-based settlement data as a sample set; based on the sample set, establishing a statistical regression model and an artificial intelligence prediction model with date as the independent variable and cumulative settlement as the dependent variable; and using the validated artificial intelligence prediction model to predict the settlement of sluice gates without monitoring equipment. This invention utilizes PS-InSAR technology to identify and extract PS point settlement information of the sluice gate, uses an artificial intelligence prediction algorithm to identify sluice gates that may experience settlement, and finally uses UAV lidar to verify the settlement trend, achieving the effect of intelligent monitoring and early warning.

Description

Technical Field

[0001] This invention relates to the fields of time-series InSAR technology, artificial intelligence prediction algorithms, and processing and analysis of UAV lidar images, specifically a sluice gate settlement early warning method and system based on UAV lidar and InSAR technology. Background Technology

[0002] Traditional settlement measurement methods include leveling, trigonometric leveling, and GPS measurement. These all require manual deployment of measuring equipment for on-site measurements, and long-term continuous monitoring incurs significant manpower and material costs, resulting in low efficiency. Furthermore, some older sluice gates have incomplete or no GPS monitoring points, making monitoring impossible.

[0003] Currently, the two most widely used space geodetic surveying methods include GNSS (Global Navigation Satellite System) technology and InSAR (Interferometric Synthetic Aperture Radar) technology. They have solved the problems of traditional surveying and have advantages such as being unaffected by weather, covering large areas, having low cost, and high accuracy.

[0004] GNSS technologies, such as the BDS BeiDou satellite navigation system, use ground displacement monitoring equipment deployed at detection points to receive and analyze BDS data, enabling long-term continuous detection of fixed locations. However, they cannot acquire area data and have low spatial resolution.

[0005] InSAR and time-series InSAR technologies use SAR synthetic aperture radar data from different dates for comprehensive observation and analysis. However, due to satellite revisit periodic imagery, the data acquisition time interval is long, resulting in low temporal resolution. Both methods are based on existing data and lack the ability to predict and verify the future settlement trends of small and medium-sized sluice gates, thus failing to provide timely monitoring and early warning.

[0006] Traditional settlement measurement methods require manual deployment of measuring equipment for on-site measurements. Long-term continuous monitoring incurs significant manpower and material costs, resulting in low efficiency. Furthermore, some older sluice gates have incomplete or no GPS monitoring points, making settlement monitoring and early warning impossible. GNSS technology cannot acquire area data and has low spatial resolution; InSAR and time-series InSAR technologies are affected by satellite revisit periodic imagery, resulting in long data acquisition intervals and low temporal resolution.

[0007] These factors pose significant challenges to the monitoring and early warning of sluice gate settlement, potentially leading to undetected and unaddressed safety hazards. Overcoming the limitations of traditional monitoring methods and establishing a high-precision, wide-range, real-time dynamic sluice gate settlement monitoring and early warning system has become an urgent technical challenge. Summary of the Invention

[0008] The purpose of this invention is to ensure the accuracy and timeliness of monitoring and early warning of sluice gate subsidence along rivers and lakes, and to better prevent the risks caused by sluice gate subsidence. This invention proposes a sluice gate subsidence early warning method and system based on UAV lidar and InSAR technology. This invention uses historical long-term high-resolution SAR imagery for differential interferometry processing. Based on the differential interferogram, it uses temporal InSAR technology to obtain historical deformation information of the sluice gate area. Then, it uses an artificial intelligence prediction algorithm to predict the historical deformation information of the sluice gate, identifying sluice gates that may subside. Finally, it uses UAV lidar to verify the subsidence trend of specific sluice gates to test the effectiveness of the artificial intelligence prediction algorithm, ultimately achieving the goal of intelligent monitoring and early warning.

[0009] The technical solution of this invention is:

[0010] This invention provides a method for early warning of sluice gate subsidence based on UAV lidar and InSAR technology, comprising:

[0011] S1. Acquire long-term SAR image data;

[0012] S2. Differential interferometry processing of SAR images is performed using the PS-InSAR method to obtain the time-series settlement data of the sluice gate locations;

[0013] S3. The SBAS-InSAR method is used to perform differential interferometry processing on SAR images to obtain time-series settlement data of the area where the sluice gate is located. The time-series settlement data of the corresponding point in the SBAS area time-series settlement data is obtained through the PS point coordinates of the sluice gate. The annual average settlement value is compared. If the difference in settlement value is within the preset range, the time-series settlement data of the point is used as the sample set.

[0014] S4. Based on the sample set, establish a statistical regression model with date as the independent variable and cumulative settlement as the dependent variable; and input the sample set into the artificial intelligence prediction model for training. When the accuracy of the artificial intelligence prediction model is higher than that of the statistical regression model, the training is complete.

[0015] S5. Select the target sluice gate in the study area, use the trained artificial intelligence prediction model to predict the future settlement trend, and obtain the predicted settlement data.

[0016] UAV lidar data of the target sluice gates for a future period is obtained. The actual settlement of the sluice gates is obtained based on the changes in lidar digital elevation data. The predicted settlement data is compared with the actual settlement to verify the effectiveness of the prediction model.

[0017] S6. Use a validated artificial intelligence prediction model to predict the settlement of sluice gates in the study area that are not equipped with monitoring devices; based on the comparison between the predicted future settlement and the warning threshold, issue a settlement warning for the sluice gate to the competent authority.

[0018] Furthermore, S1 specifically includes: acquiring long-term synthetic aperture radar image data from a satellite data platform, and processing the image data using an atmospheric correction model to address atmospheric effect errors in the image data, thereby obtaining corrected image data.

[0019] Furthermore, S2 specifically includes:

[0020] SAR image data is acquired, and differential interferometry is performed using the PS-InSAR method to obtain differential interferometry results.

[0021] Extract sluice gate location information from differential interferometry results and generate a time series dataset of the locations;

[0022] Time series analysis was performed on the point time series dataset to calculate the settlement change and the annual average settlement value of each PS point.

[0023] Based on the location information and settlement changes of the sluice gates, a time-series settlement distribution map of the sluice gate locations is obtained, which serves as the time-series settlement data of the sluice gate locations.

[0024] Furthermore, S3 specifically includes:

[0025] The SBAS method was used to perform differential interferometry on the SAR image to obtain the time-series settlement data of the sluice gate area;

[0026] Extract the coordinates of the PS points, match the corresponding point values ​​in the time series settlement data, and if there are missing point values, fill them in using an interpolation algorithm. Also, calculate the average annual settlement value of each PS point.

[0027] If the difference between the annual average settlement value 2 and the annual average settlement value 1 is within the preset range, then the time series settlement data of the point will be used as the sample set.

[0028] Furthermore, the difference between the second annual average settlement value and the first annual average settlement value is expressed as the standard deviation;

[0029] The standard deviation δ of the difference between the annual average settlement value 2 and the annual average settlement value 1 at each PS point is calculated using the following formula;

[0030]

[0031] δ represents the standard deviation of the settlement difference, n represents the total number of PS points, Dij represents the settlement difference between points i and j, and θ represents the average settlement difference between all points.

[0032] If the standard deviation of the difference between the annual average settlement value 2 and the annual average settlement value 1 is within the preset range, then the time series settlement data of the point will be used as the sample set.

[0033] Furthermore, the statistical regression model mentioned in S4 is a linear regression model, a stepwise regression model, a ridge regression model, or a Bayesian regression model; the artificial intelligence prediction model is a random forest regression model or a support vector machine regression model.

[0034] Furthermore, in S4, the sample set is divided into a training set and a validation set, and the performance of various statistical regression models is evaluated to select the optimal model.

[0035] Furthermore, in S4, the sample set is input into the artificial intelligence prediction model for training. When the accuracy of the artificial intelligence prediction model is higher than that of the statistical regression model, the training is completed. Specifically, this includes:

[0036] The training data in the sample set is input into a pre-established artificial intelligence prediction model for training;

[0037] Obtain the prediction accuracy of the artificial intelligence prediction model on the validation set and the prediction accuracy of the statistical regression model on the validation set, respectively;

[0038] If the accuracy of the artificial intelligence prediction model is higher than that of the statistical regression model, then the model training is considered complete.

[0039] Furthermore, in S5, UAV lidar data for the target sluice gates over a future period is acquired, and the actual settlement of the sluice gates is obtained based on the changes in lidar digital elevation data, specifically including:

[0040] Based on the geographical location information of the target sluice gate in the study area, the flight path of the UAV lidar is planned to ensure coverage of the target area;

[0041] The first phase of data collection was carried out by using a drone equipped with a lidar device, following a planned flight path. The three-dimensional point cloud data of the sluice gate area was acquired and preprocessed, including noise reduction, filtering and registration operations, to generate the first phase of digital elevation model.

[0042] After a set time interval, the same UAV lidar equipment and flight parameters are used to conduct a second phase of data acquisition and processing to generate a second phase of digital elevation model;

[0043] Spatial registration of the two digital elevation models is performed to ensure that the coordinates of the same geographical location are consistent;

[0044] Using a differential processing method, the elevation difference between the two digital elevation models is calculated pixel by pixel to generate an elevation change map of the sluice gate area.

[0045] Based on the elevation change map of the sluice gate area, extract the elevation change value of the PS point of the sluice gate and calculate the actual settlement of the sluice gate.

[0046] A system employed in a sluice gate settlement early warning method based on UAV lidar and InSAR technology includes:

[0047] SAR data acquisition module, used to acquire long-term time-series SAR image data;

[0048] A time-series InSAR settlement monitoring module is used to perform differential interferometry processing on SAR image data to obtain historical settlement information of the sluice gate.

[0049] The artificial intelligence settlement prediction module is used to predict and identify sluice gates that may settle based on historical settlement information.

[0050] The lidar verification module is used to verify the settlement trend of the predicted and identified sluice gate using UAV lidar.

[0051] And a settlement early warning module; used to determine whether to trigger an early warning based on the verification results.

[0052] The beneficial effects of this invention are:

[0053] This invention utilizes temporal InSAR technology based on medium-to-high resolution temporal SAR imagery to acquire temporal deformation information data of sluice gates, and employs artificial intelligence algorithms to predict the future settlement trend of the sluice gates. The settlement trend is then verified using UAV lidar. This invention solves both the low spatial resolution problem of GNSS technology and the low temporal resolution problem of InSAR technology. It enables accurate monitoring and early warning for old sluice gates lacking monitoring points, while saving costs and improving monitoring and early warning efficiency.

[0054] This invention integrates multi-source remote sensing data and artificial intelligence algorithms, improving the accuracy and efficiency of sluice gate settlement monitoring. By comparing the predicted settlement with the warning threshold, this invention can promptly issue early warning information to the competent authorities, providing a scientific basis for sluice gate safety management and having significant practical application value.

[0055] Other features and advantages of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0056] The above and other objects, features and advantages of the present invention will become more apparent from the more detailed description of exemplary embodiments of the invention in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments of the invention.

[0057] Figure 1A flowchart of a method for monitoring and providing artificial intelligence-based early warning of settlement in small and medium-sized sluice gates according to an embodiment of the present invention is shown.

[0058] Figure 2 A schematic diagram of the study area of ​​the Sima Bend in the Yangzhou section of the Yangtze River main stream according to an embodiment of the present invention is shown. Detailed Implementation

[0059] Preferred embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While preferred embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0060] Example 1

[0061] Figure 1 A flowchart of a method for monitoring and providing artificial intelligence-based early warning of settlement in small and medium-sized sluice gates according to an embodiment of the present invention is shown.

[0062] like Figure 1 As shown, this invention provides a method for early warning of sluice gate subsidence based on UAV lidar and InSAR technology, comprising:

[0063] S1. Obtain long-term synthetic aperture radar (SAR) image data from the satellite data platform. To address atmospheric effect errors in the image data, use an atmospheric correction model to process the image data and obtain corrected image data.

[0064] S2. Differential interferometry processing of SAR images is performed using the PS-InSAR method to obtain the time-series settlement data of the sluice gate locations;

[0065] Specifically, the process involves: acquiring SAR image data; performing differential interferometry processing using the PS-InSAR method to obtain differential interferometry results; extracting sluice gate location information from the differential interferometry results to generate a time-series dataset of the locations; performing time-series analysis on the time-series dataset to calculate the settlement change and the annual average settlement value for each PS location; and obtaining a time-series settlement distribution map of the sluice gate locations based on the sluice gate location information and the settlement change, which serves as the time-series settlement data for the sluice gate locations.

[0066] S3. The SBAS-InSAR method is used to perform differential interferometry processing on SAR images to obtain time-series settlement data of the area where the sluice gate is located. The time-series settlement data of the corresponding point in the SBAS area time-series settlement data is obtained through the PS point coordinates of the sluice gate. The annual average settlement value is compared. If the difference in settlement value is within the preset range, the time-series settlement data of the point is used as the sample set.

[0067] Specifically, the SBAS method is used to perform differential interferometry processing on the SAR image to obtain the time-series settlement data of the sluice gate area; the coordinate values ​​of the PS points are extracted and matched with the corresponding point values ​​in the time-series settlement data. If the point values ​​are missing, they are supplemented by interpolation algorithm, and the annual average settlement value II for each PS point is calculated; if the standard deviation of the difference between the annual average settlement value II and the annual average settlement value I is within the preset range, the time-series settlement data of the point is used as the sample set.

[0068] The standard deviation δ of the difference between the annual average settlement value 2 and the annual average settlement value 1 at each PS point is calculated using the following formula;

[0069]

[0070] δ represents the standard deviation of the settlement difference, n represents the total number of PS points, Dij represents the settlement difference between points i and j, and θ represents the average settlement difference among all points.

[0071] S4. Based on the sample set, establish a statistical regression model with date as the independent variable and cumulative settlement as the dependent variable; and input the sample set into the artificial intelligence prediction model for training. When the accuracy of the artificial intelligence prediction model is higher than that of the statistical regression model, the training is complete.

[0072] The statistical regression model is a linear regression model, a stepwise regression model, a ridge regression model, or a Bayesian regression model; the artificial intelligence prediction model is a random forest regression model or a support vector machine regression model.

[0073] S5. Select the target sluice gate in the study area, use the trained artificial intelligence prediction model to predict the future settlement trend, and obtain the predicted settlement data.

[0074] UAV lidar data of the target sluice gates for a future period is obtained. The actual settlement of the sluice gates is obtained based on the changes in lidar digital elevation data. The predicted settlement data is compared with the actual settlement to verify the effectiveness of the prediction model.

[0075] S6. Use a validated artificial intelligence prediction model to predict the settlement of sluice gates in the study area that are not equipped with monitoring devices; based on the comparison between the predicted future settlement and the warning threshold, issue a settlement warning for the sluice gate to the competent authority.

[0076] In this embodiment, the method acquires long-term SAR image data, combines PS-InSAR and SBAS-InSAR technologies to extract historical settlement information of sluice gates, establishes statistical regression and artificial intelligence prediction models, and uses UAV lidar data to verify the accuracy of the models, thereby achieving settlement prediction for unmonitored sluice gates.

[0077] In one example, step S3 uses the SBAS method to perform differential interferometry processing on SAR images to obtain temporal settlement data of the sluice gate area. The SBAS method is a time-series analysis method based on multiple SAR images. By selecting multiple SAR images at different time points and performing differential interferometry processing, a series of differential interferograms are generated, which reflect the deformation information of the sluice gate area at different time points. It can effectively reduce atmospheric delay and noise interference, thereby improving the accuracy of deformation monitoring.

[0078] After obtaining the coordinates of the PS (Pressure Point) locations at the sluice gate, it is necessary to find the corresponding point values ​​in the time-series settlement data. For example, if the coordinates of a PS location are (X1, Y1), the corresponding settlement value needs to be found in the time-series settlement data of the area where the sluice gate is located. If the settlement value at a certain time point is missing, it can be supplemented using interpolation algorithms. This can be estimated based on the settlement values ​​of surrounding points, such as using linear interpolation or cubic spline interpolation. In this way, the time-series settlement data for each PS location can be ensured to be complete, providing a reliable data foundation for subsequent analysis.

[0079] After filling in all missing values, the second annual average settlement value is calculated for each PS point. Time-series analysis is then performed on the point dataset to calculate the settlement variation and the first annual average settlement value for each PS point. If the difference between the second and first annual average settlement values ​​is within a preset range, the point's time-series settlement data is used as the sample set. This method allows for the selection of points with stable settlement trends, providing high-quality training data for subsequent AI prediction models.

[0080] In this embodiment, standard deviation is used as the screening criterion, which can effectively eliminate outlier points and improve data reliability. If the settlement value difference at a certain point is significantly higher than that at other points, and the calculated standard deviation exceeds the preset range, it indicates that there may be local settlement or measurement error at that point, requiring further investigation. This method ensures the data quality of the sample set, providing an accurate basis for subsequent settlement prediction and verification. Using the screened time-series settlement data as the sample set helps improve the accuracy of the artificial intelligence prediction model.

[0081] In one example, step S4 inputs the sample set into an artificial intelligence prediction model for training. The AI ​​model can employ algorithms such as Support Vector Machines (SVMs) or Random Forests. Taking Random Forests as an example, predictions are made by constructing multiple decision trees, each trained on a different subset of the sample set, and the final prediction result is determined through a voting mechanism. During training, the model continuously adjusts its parameters to minimize prediction error and gradually improve prediction accuracy. By comparing the model's predictions on the validation set with actual settlement data, accuracy metrics such as mean squared error and coefficient of determination are calculated.

[0082] Simultaneously, the same sample set is needed to train the statistical regression model. Statistical regression models can employ methods such as linear regression and stepwise regression. For example, a linear regression model describes the relationship between settlement and time by fitting a straight line, while a stepwise regression model optimizes the model by progressively introducing or removing variables. The accuracy of the statistical regression model is also evaluated by comparing it with the validation set; training is complete when the accuracy of the AI ​​prediction model is higher than that of the statistical regression model.

[0083] In this embodiment, the mean squared error of the random forest model on the validation set is 2.7604, while the mean squared error of the linear regression model is 4.6248, indicating that the artificial intelligence model has higher prediction accuracy. This demonstrates that the artificial intelligence model can better capture the nonlinear relationships in the data, thus providing more accurate predictions. This comparative analysis helps to further improve the prediction accuracy of the model and provide more reliable technical support for sluice gate settlement monitoring and early warning.

[0084] In one example, step S5 obtains UAV lidar data for the target sluice gate over a future period, and calculates the actual settlement of the sluice gate based on the changes in lidar digital elevation data. Specifically, this includes:

[0085] Based on the geographical location information of the target sluice gate in the study area, the flight path of the UAV lidar is planned to ensure coverage of the target area;

[0086] The first phase of data collection was carried out by using a drone equipped with a lidar device, following a planned flight path. The three-dimensional point cloud data of the sluice gate area was acquired and preprocessed, including noise reduction, filtering and registration operations, to generate the first phase of digital elevation model.

[0087] After a set time interval, the same UAV lidar equipment and flight parameters are used to conduct a second phase of data acquisition and processing to generate a second phase of digital elevation model;

[0088] Spatial registration of the two digital elevation models is performed to ensure that the coordinates of the same geographical location are consistent;

[0089] Using a differential processing method, the elevation difference between the two digital elevation models is calculated pixel by pixel to generate an elevation change map of the sluice gate area.

[0090] Based on the elevation change map of the sluice gate area, extract the elevation change value of the PS point of the sluice gate and calculate the actual settlement of the sluice gate.

[0091] In this embodiment, using UAV lidar to validate the prediction model can compensate for the shortcomings of GNSS and InSAR technologies. GNSS technology has low spatial resolution and cannot obtain detailed settlement information of the sluice gate; InSAR technology has low temporal resolution and cannot reflect settlement changes in a short period of time. UAV lidar, on the other hand, combines the advantages of high spatial resolution and high temporal resolution, and can provide accurate validation data for the prediction model, thereby improving the accuracy and timeliness of settlement monitoring and early warning.

[0092] Example 2

[0093] This invention provides a system for a sluice gate settlement early warning method based on UAV lidar and InSAR technology, comprising:

[0094] SAR data acquisition module, used to acquire long-term time-series SAR image data;

[0095] A time-series InSAR settlement monitoring module is used to perform differential interferometry processing on SAR image data to obtain historical settlement information of the sluice gate.

[0096] The artificial intelligence settlement prediction module is used to predict and identify sluice gates that may settle based on historical settlement information.

[0097] The lidar verification module is used to verify the settlement trend of the predicted and identified sluice gate using UAV lidar.

[0098] And a settlement early warning module; used to determine whether to trigger an early warning based on the verification results.

[0099] In practice:

[0100] This case study selects the Sima Bend section of the Yangtze River's main stream in Yangzhou as the research area to conduct settlement monitoring and early warning of sluice gates along the Yangtze River. The research scope is as follows: Figure 2 As shown.

[0101] (1) Data source: Download long-term Sentinel-1 SAR satellite data covering the study area from 2016 to 2020 as the data source, as shown in Table 1.

[0102] Table 1. Basic parameters of Sentinel-1 data.

[0103] Serial Number Date of shooting polarization mode Observation mode Track direction 1 2016-03-21 VV / VH IW Ascend 2 2016-04-14 VV / VH IW Ascend ...... VV / VH IW Ascend N+1 2020-12-25 VV / VH IW Ascend

[0104] (2) Temporal settlement monitoring: The PS-InSAR method and the SBAS-InSAR method were used to monitor the settlement of the study area and obtain the temporal settlement data of the sluice gates in the study area from 2016 to 2020. The PS-InSAR method obtained the temporal settlement data of the PS point of the sluice gate, while the SBAS-InSAR method obtained the temporal settlement information of the area where the sluice gate is located.

[0105] In this invention, to evaluate the reliability of the monitoring results obtained by the PS-InSAR method, the Small Baseline Subset (SBAS) method is used to verify the PS-InSAR method. The temporal settlement results of corresponding points in the SBAS surface source settlement results are obtained using the PS point coordinates. The annual average deformation of the two methods is compared. If the difference in deformation is within 2 mm / a, it indicates that the PS settlement method is reliable. Table 2 shows the comparison results of the annual average deformation of one of the drastically changing points at the Dongxingang Sluice Gate.

[0106] Table 2 Comparison of annual average deformation between PS-ISAR and SBAS-ISAR

[0107] gate station PS(mm / a) SBAS(m / a) Difference (mm / a) Dongxingang Gate in Economic Development Zone -5.00 -5.21 0.21

[0108] (3) Settlement prediction: Assuming the first data collection date of March 21, 2016 is day 1, March 22, 2016 is day 2, and so on. Using date as the independent variable and cumulative settlement as the dependent variable, conventional statistical regression models (linear regression, stepwise regression) and artificial intelligence algorithm models (support vector machine, random forest) are used to model the existing sluice gate settlement data and predict the settlement trend in the future.

[0109] (4) Verification of airborne lidar: LiDAR UAVs were used to collect lidar image data of the gate station for two future periods. The settlement of the gate station during this period was obtained by analyzing the elevation data of the two images and compared with the previous artificial intelligence prediction results to verify the accuracy of the artificial intelligence prediction model.

[0110] (5) Monitoring and early warning of settlement of old sluice gates: Using verified artificial intelligence prediction algorithms, we can predict and warn of settlement of old sluice gates and pumping stations along the Yangtze River in the future, and detect potential dangers in a timely manner.

[0111] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments.

Claims

1. A method for early warning of sluice gate subsidence based on UAV lidar and InSAR technology, characterized in that, include: S1. Acquire long-term SAR image data; S2. Differential interferometry processing of SAR images is performed using the PS-InSAR method to obtain the time-series settlement data of the sluice gate locations; S3. The SBAS-InSAR method is used to perform differential interferometry processing on SAR images to obtain time-series settlement data of the area where the sluice gate is located. The time-series settlement data of the corresponding point in the SBAS area time-series settlement data is obtained through the PS point coordinates of the sluice gate. The annual average settlement value is compared. If the difference in settlement value is within the preset range, the time-series settlement data of the point is used as the sample set. S4. Based on the sample set, establish a statistical regression model with date as the independent variable and cumulative settlement as the dependent variable; and input the sample set into the artificial intelligence prediction model for training. When the accuracy of the artificial intelligence prediction model is higher than that of the statistical regression model, the training is complete. S5. Select the target sluice gate in the study area, use the trained artificial intelligence prediction model to predict the future settlement trend, and obtain the predicted settlement data. UAV lidar data of the target sluice gates for a future period is obtained. The actual settlement of the sluice gates is obtained based on the changes in lidar digital elevation data. The predicted settlement data is compared with the actual settlement to verify the effectiveness of the prediction model. S6. Use a validated artificial intelligence prediction model to predict the settlement of sluice gates in the study area that have not been equipped with monitoring equipment. Based on the comparison between the predicted future settlement and the warning threshold, a settlement warning for the sluice gate is issued to the relevant authorities. S2 specifically includes: acquiring SAR image data, performing differential interferometry processing using the PS-InSAR method to obtain differential interferometry results; extracting sluice gate location information from the differential interferometry results to generate a time-series dataset of the locations; performing time-series analysis on the time-series dataset of the locations to calculate the settlement change and the annual average settlement value of each PS location; and obtaining a time-series settlement distribution map of the sluice gate locations based on the sluice gate location information and the settlement change, which serves as the time-series settlement data of the sluice gate locations. S3 specifically includes: using the SBAS method to perform differential interferometric processing on the SAR image to obtain the time-series settlement data of the sluice gate area; extracting the coordinate values ​​of the PS points, matching the corresponding point values ​​in the time-series settlement data, and supplementing the missing point values ​​through interpolation algorithms if any are missing, and calculating the second annual average settlement value of each PS point; if the difference between the second annual average settlement value and the first annual average settlement value is within a preset range, then the time-series settlement data of the point is used as a sample set.

2. The sluice gate subsidence early warning method based on UAV lidar and InSAR technology according to claim 1, characterized in that... S1 specifically includes: acquiring long-term synthetic aperture radar image data from a satellite data platform and removing the effects of atmospheric effects.

3. The sluice gate subsidence early warning method based on UAV lidar and InSAR technology according to claim 1, characterized in that, The difference between the second annual average settlement value and the first annual average settlement value is expressed as the standard deviation. The standard deviation of the difference between the annual average settlement value 2 and the annual average settlement value 1 at each PS point is calculated using the following formula. ; ； The standard deviation of the settlement difference is represented by the following: This indicates the total number of PS points. Indicates point and The difference in settlement values ​​between them This represents the average of the settlement differences between all points. If the standard deviation of the difference between the annual average settlement value 2 and the annual average settlement value 1 is within the preset range, then the time series settlement data of the point will be used as the sample set.

4. The sluice gate subsidence early warning method based on UAV lidar and InSAR technology according to claim 1, characterized in that... S4 in; The statistical regression model is a linear regression model, a stepwise regression model, a ridge regression model, or a Bayesian regression model; The artificial intelligence prediction model is either a random forest regression model or a support vector machine regression model.

5. The sluice gate subsidence early warning method based on UAV lidar and InSAR technology according to claim 4, characterized in that... In S4, the sample set is divided into a training set and a validation set, and the performance of various statistical regression models is evaluated to select the optimal model.

6. The sluice gate subsidence early warning method based on UAV lidar and InSAR technology according to claim 5, characterized in that... In S4, the sample set is input into the artificial intelligence prediction model for training. Training is complete when the accuracy of the artificial intelligence prediction model is higher than that of the statistical regression model. Specifically, this includes: The training data in the sample set is input into a pre-established artificial intelligence prediction model for training; Obtain the prediction accuracy of the artificial intelligence prediction model on the validation set and the prediction accuracy of the statistical regression model on the validation set, respectively; If the accuracy of the artificial intelligence prediction model is higher than that of the statistical regression model, then the model training is considered complete.

7. The sluice gate subsidence early warning method based on UAV lidar and InSAR technology according to claim 1, characterized in that... In S5, UAV lidar data for the target sluice gates over a future period is acquired. Based on the changes in lidar digital elevation data, the actual settlement of the sluice gates is obtained, specifically including: Based on the geographical location information of the target sluice gate in the study area, the flight path of the UAV lidar is planned to ensure coverage of the target area; The first phase of data collection was carried out by using a drone equipped with a lidar device, following a planned flight path. The three-dimensional point cloud data of the sluice gate area was acquired and preprocessed, including noise reduction, filtering and registration operations, to generate the first phase of digital elevation model. After a set time interval, the same UAV lidar equipment and flight parameters are used to conduct a second phase of data acquisition and processing to generate a second phase of digital elevation model; Spatial registration of the two digital elevation models is performed to ensure that the coordinates of the same geographical location are consistent; Using a differential processing method, the elevation difference between the two digital elevation models is calculated pixel by pixel to generate an elevation change map of the sluice gate area. Based on the elevation change map of the sluice gate area, extract the elevation change value of the PS point of the sluice gate and calculate the actual settlement of the sluice gate.

8. A system used in the sluice gate settlement early warning method based on UAV lidar and InSAR technology as described in any one of claims 1-7, characterized in that, include: SAR data acquisition module, used to acquire long-term time-series SAR image data; Time-series InSAR settlement monitoring module; Used for differential interferometry processing of SAR image data to obtain historical settlement information of sluice gates; The artificial intelligence settlement prediction module is used to predict and identify sluice gates that may settle based on historical settlement information. The lidar verification module is used to verify the settlement trend of the predicted and identified sluice gate using UAV lidar. And a settlement early warning module; Used to determine whether to trigger an alert based on the verification results.

Images