Icing monitoring method and system for corner reflector deployed based on unmanned aerial vehicle

Abstract

The invention discloses an icing monitoring method and system for a corner reflector deployed based on an unmanned aerial vehicle, and relates to the technical field of environment monitoring, and the method comprises the steps: deploying a corner reflector network, collecting radar cross-section characteristics and environment parameters to generate a data packet, and carrying out the icing monitoring based on the generated data packet; calculating a signal-to-clutter ratio, obtaining a standard deviation of target positioning and phase estimation by using a Cramer-Rao lower bound method, generating a dynamic weight by using the standard deviation, obtaining a corner reflector icing thickness estimation value through data fusion, and generating an optimized signal-to-clutter ratio based on the icing thickness estimation value; and carrying out cooperative judgment by utilizing the generated optimized signal-to-clutter ratio and combining the icing thickness and the attitude deviation, triggering combined adjustment based on a judgment result, and after the combined adjustment is triggered, obtaining the SAR image again and carrying out main and auxiliary image registration and interference pair optimization. According to the invention, the sensing precision and the monitoring robustness of the unmanned aerial vehicle on the state of the corner reflector in an icing environment are improved.

Description

Technical Field

[0001] This invention relates to the field of synthetic aperture radar remote sensing monitoring technology, and in particular to an icing monitoring method and system based on corner reflectors deployed by unmanned aerial vehicles. Background Technology

[0002] Icing poses a serious threat to the safety of power transmission lines, communication facilities, and infrastructure. Traditional monitoring methods, such as manual inspections and fixed sensor networks, suffer from low efficiency and limited coverage when monitoring icing over large areas and complex terrain. Synthetic Aperture Radar (SAR) technology, due to its all-weather and all-time operating characteristics, has been widely used in environmental monitoring, especially UAV-mounted SAR systems. Existing UAV-based SAR monitoring technologies mainly focus on surface deformation monitoring, and research on the specific problem of icing is still insufficient. Although some studies have attempted to combine corner reflectors with UAVs to improve monitoring accuracy, many challenges remain in practical applications. For example, signal distortion caused by corner reflector attitude deviation and changes in electromagnetic properties due to icing affect monitoring accuracy. Existing technologies for icing monitoring typically rely on a single data source or simple data fusion methods, lacking an effective integration and dynamic adjustment mechanism for multi-source information. For instance, while traditional SAR imaging technology can provide high-resolution surface images, it cannot directly reflect the thickness of ice and its impact on the structural stability of targets. Existing technologies struggle to achieve accurate target positioning and phase estimation under complex meteorological conditions and terrain features. In particular, under icing conditions, changes in the attitude of corner reflectors significantly affect the quality of echo signals, thereby reducing the reliability of monitoring results.

[0003] One existing technology, patent CN119437025B, entitled "A New Method for Fusion and Accuracy Evaluation of BeiDou + InSAR Corner Reflector Deformation Monitoring Data," establishes a local coordinate system, decomposes the three-dimensional displacement observed by BeiDou into vertical and horizontal components, and inverts the line-of-sight deformation by combining the geometric relationship of the InSAR ascending and descending orbits. It then uses multi-orbit consistency checks to correct the vertical deformation, achieving complementary fusion and accuracy verification between BeiDou and InSAR at the deformation observation level. However, it does not consider the influence of icing on the electromagnetic characteristics and attitude of the corner reflector, nor does it possess environmental perception and adaptive adjustment capabilities. The first patent does not consider the problems of attitude shift, electromagnetic scattering characteristics change and signal attenuation caused by icing of corner reflectors; another related prior art patent CN114910907B "River Slope Landslide Risk Monitoring Method Based on Corner Reflector and PS-InSAR Technology" uses natural stable points to guide the deployment of artificial corner reflectors and improve the reliability of phase unwrapping, but does not involve the core technology directions of icing perception, multi-source sensor fusion, dynamic signal-to-noise ratio optimization and active attitude adjustment. In particular, it lacks consideration of the performance degradation of corner reflectors under the influence of environmental interference (such as icing) and the closed-loop compensation mechanism. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method and system for monitoring icing of corner reflectors deployed by UAVs, which solves the problems of corner reflector attitude shift, signal attenuation and decreased monitoring accuracy caused by icing.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for monitoring icing based on corner reflectors deployed by unmanned aerial vehicles, comprising, Deploy a corner reflector network, collect radar cross-section characteristics and environmental parameters to generate data packets, calculate the signal-to-clutter ratio based on the generated data packets, and use the Cramer-Rao lower bound method to obtain the standard deviation of target positioning and phase estimation. Use the standard deviation to generate dynamic weights, and obtain the estimated value of corner reflector icing thickness through data fusion. Generate an optimized signal-to-clutter ratio based on the icing thickness estimate. Using the generated optimized signal-to-clutter ratio, combined with ice thickness and attitude deviation, a collaborative judgment is made. Based on the judgment result, a joint adjustment is triggered. After the joint adjustment is triggered, SAR images are reacquired and main and auxiliary image registration, interferometric pair optimization, terrain phase subtraction and phase unwrapping are performed to extract a high-precision continuous line-of-sight phase time series. A state vector is generated based on ice thickness and multidimensional data. The line-of-sight phase, three-dimensional displacement of the corner reflector, and ice thickness are fused by multimodal extended Kalman filtering. The state vector is optimized to generate the optimal state vector. The stability index is calculated using the optimal state vector. The corner reflector coordinates are then re-optimized based on the stability index.

[0007] As a preferred embodiment of the icing monitoring method based on UAV-deployed corner reflectors described in this invention, the method involves: deploying a corner reflector network, collecting radar cross-section characteristics and environmental parameters to generate data packets, deploying a triangular trihedral corner reflector network, marking installation points using a total station, and obtaining the initial coordinates of the corner reflectors. Each corner reflector base integrates multiple sensors and modules to calculate the incident angle. and initial radar cross-section Each corner reflector node generates and uploads a data packet containing key parameters, and a SAR system is deployed on the UAV. The UAV radar echo data is processed by SAR imaging to generate a SAR single-look complex image.

[0008] As a preferred embodiment of the icing monitoring method for corner reflectors deployed by UAVs according to the present invention, the method involves: generating data packets, calculating the signal-to-clutter ratio (SCR), obtaining the standard deviation of target location and phase estimation using the Cramer-Rao lower bound method, generating dynamic weights using the standard deviation, obtaining an estimate of the icing thickness of the corner reflector through data fusion, and generating an optimized SCR index based on the icing thickness estimate to calculate the i-th actual radar cross-section. Hoshin Mix The Cramer-Rao lower bound method is used to evaluate the theoretical minimum standard deviation of target localization and phase estimation of a SAR system under a given signal-to-clutter ratio, including the standard deviation of range-direction localization accuracy. Standard deviation of azimuth positioning accuracy Standard deviation of phase noise of corner reflector ; Obtain the midpoint of the bright spot of the corner reflector in a SAR single-view complex image. Calculate ITRF coordinates The ITRF coordinates of the corner reflector were determined using a zero-Doppler side-looking SAR imaging geometric model. Projecting the image onto the azimuth-range coordinate system of the SAR image yields the theoretical projection position. Calculate coordinate deviation and Calculate the root mean square error ; Set the initial value of the equivalent circular diameter of the SAR antenna beam. =G, using an iterative optimization method for each candidate Values, recalculate the coordinate deviation of each corner reflector and And RMSE, iterative search to minimize RMSE. Introducing the optimal The residual from the initial value G is the confidence factor. ; Calculate the estimated icing thickness on the corner reflector surface. Combined with credibility factor Generate an optimized signal-to-noise ratio The deviation between the actual incident angle and the optimal incident angle of the corner reflector .

[0009] As a preferred embodiment of the icing monitoring method based on corner reflectors deployed by UAVs according to the present invention, the method involves: utilizing the generated optimized signal-to-noise ratio, combining icing thickness and attitude deviation for collaborative judgment, and triggering joint adjustment based on the judgment result and the deviation. and ice thickness Perform trigger judgment, including when deviation >Preset threshold U and When the threshold is 0, the corner reflector is considered to be in an abnormal state. When the preset threshold X is met alone, a check and adjustment are triggered, and a joint adjustment action is performed, such as realigning the corner reflector's reflective surface with the optimal incident angle. If the ice thickness exceeds the set thickness, the resistance wire embedded in the mesh hole will be activated to heat and melt the thin ice. The drone will then hover and fly again, remeasure the RCS, calculate the dynamic SCR, and output the optimized corner reflector state tuple after the adjustment is completed.

[0010] As a preferred embodiment of the icing monitoring method for corner reflectors deployed by UAVs according to the present invention, the method involves: after triggering joint adjustment, re-acquiring SAR images and performing master-slave image registration, interferometric pair optimization, terrain phase subtraction, and phase unwrapping; extracting a high-precision continuous line-of-sight phase time series; after completing corner reflector attitude adjustment and de-icing compensation, the UAV re-acquiring SAR images; and accurately extracting the coordinates of each corner reflector in the image using a sub-pixel-level bright spot localization algorithm. The master and slave images are selected from the registered image sequence. After generating an interferometric pair based on the master image, the flat-ground phase is removed, and the terrain phase is subtracted using a digital elevation model generated by high-precision LiDAR measurement to obtain the interferometric phase. A minimum cost flow algorithm is used to unwrap the interferometric phase, effectively handling low-coherence areas and outputting a continuous, non-jumping line-of-sight phase time series. (t).

[0011] As a preferred embodiment of the icing monitoring method based on a UAV-deployed corner reflector described in this invention, the method involves: generating a state vector based on icing thickness and multidimensional data; fusing line-of-sight phase, three-dimensional displacement of the corner reflector, and icing thickness using multimodal extended Kalman filtering; and optimizing the state vector to generate an optimal state vector index for the time series. (t), coordinates of the corner reflector and ice thickness Together, they constitute the multi-source observation input of EKF. In the EKF framework, the state vector is defined as... The three-dimensional displacement and slip rate of the slope are obtained through a two-stage iterative prediction and update process within a multimodal extended Kalman filter fusion framework. and the state of freezing The joint estimation, obtained during the EKF update phase, uses the observation residuals and Kalman gain to correct the predicted values, resulting in the optimal estimated state vector. .

[0012] As a preferred embodiment of the icing monitoring method based on a UAV-deployed corner reflector described in this invention, the method involves: calculating a stability index using an optimal state vector, and re-optimizing the corner reflector coordinates based on the stability index and the optimal estimated state vector. The stability index is constructed by quantitatively integrating the slope's motion state with environmental disturbance factors. The stability index calculated based on the formula ,when When the preset threshold η is reached, the slope area where the corner reflector is located is determined to be in a potentially unstable state, and a graded early warning mechanism is immediately triggered, including generating a stability assessment report and marking low stability indices. In the designated space region, the UAV mission planning module is automatically activated, and an autonomous reflight is performed at the set time the following day. This re-optimizes the corner reflector coordinates, compensates for attitude and icing, performs InSAR processing, and inverts the latest 3D deformation to verify... Whether it continues to decline, if retested If the level remains below the threshold, the warning level will be upgraded, and emergency response recommendations will be sent to the management platform.

[0013] Secondly, the present invention provides an icing monitoring system based on corner reflectors deployed by UAVs, including a corner reflector network deployment and multi-source sensing module, used to deploy the corner reflector network, collect initial coordinates, attitude, RCS and environmental parameters, and construct a highly coherent monitoring reference point; The dynamic weight generation and icing estimation module is used to calculate the standard deviation of positioning and phase accuracy based on the signal-to-noise ratio and CRLB, fuse multi-source sensor data and generate dynamic weights, invert icing thickness and optimize signal-to-noise ratio; The joint adjustment and data optimization module is used to trigger attitude correction and de-icing actions based on icing and attitude deviation, adjust the re-flight to acquire SAR images, and extract high-precision line-of-sight phase sequences. The multimodal state fusion and deformation inversion module is used to construct a state vector containing three-dimensional displacement and slip rate. It outputs the optimal deformation estimate by fusing InSAR and icing data through extended Kalman filtering. The stability assessment and intelligent early warning module is used to calculate the stability index of the optimal state vector. When the index is lower than the threshold, it triggers a graded early warning and drives the UAV to retest, thereby achieving automated risk identification.

[0014] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein the computer program, when executed by the processor, implements any step of the icing monitoring method based on a corner reflector deployed by an unmanned aerial vehicle as described in the first aspect of the present invention.

[0015] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the icing monitoring method based on a corner reflector deployed by an unmanned aerial vehicle as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: By integrating multi-source sensors into the corner reflector base, attitude, waveguide velocity, and dielectric response environmental parameters are acquired in real time. Combined with SAR image echo characteristics, the theoretical accuracy limit of positioning and phase estimation is evaluated using the Cramer-Rao lower bound theory. Dynamic weights are then generated to achieve physical constraint fusion of ultrasonic, impedance, and thermal infrared multimodal data, significantly improving the accuracy and robustness of icing thickness inversion. A credibility factor is introduced to optimize the geometric model, and an optimized signal-to-noise ratio is generated based on icing thickness and attitude deviation to quantify the degree of signal quality degradation, providing a reliable basis for subsequent monitoring. Through multi-source information collaboration and dynamic weight adjustment, the limitations of single sensors being susceptible to environmental interference are overcome, enhancing the stability and accuracy of icing monitoring under complex meteorological conditions. This represents a leap from passive observation to active sensing and adaptive optimization, demonstrating significant technological progress and practical value. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1This is a flowchart of an icing monitoring method based on a corner reflector deployed by an unmanned aerial vehicle (UAV) in Example 1.

[0019] Figure 2 This is a schematic diagram of an icing monitoring system based on a corner reflector deployed by an unmanned aerial vehicle (UAV) in Example 1.

[0020] Figure 3 This is a flowchart of the joint adjustment and data optimization process in Example 1. Detailed Implementation

[0021] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0022] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0023] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0024] Example 1, referring to Figures 1 to 3 This is the first embodiment of the present invention, which provides a method for monitoring icing based on corner reflectors deployed by unmanned aerial vehicles, including the following steps: S1. Deploy a corner reflector network, collect radar cross-section characteristics and environmental parameters to generate data packets, calculate the signal-to-clutter ratio based on the generated data packets, and use the Cramer-Rao lower bound method to obtain the standard deviation of target positioning and phase estimation. Use the standard deviation to generate dynamic weights, and obtain the estimated value of corner reflector icing thickness through data fusion. Generate an optimized signal-to-clutter ratio based on the icing thickness estimate. Specifically, deploying a corner reflector network and collecting radar cross-section characteristics and environmental parameters to generate data packets involves deploying a triangular trihedral corner reflector (CR) network, using a total station (Leica TS16) to mark installation points, and obtaining the initial coordinates of the corner reflectors. The corner reflectors are oriented towards the main flight path of the UAV and are not obstructed by vegetation, forming an equilateral triangle layout to create a highly coherent space control network. The corner reflectors employ an equilateral triangular aluminum alloy structure with a mesh-like design on the surface to reduce icing load. Each corner reflector base integrates multiple sensors and modules, including a triaxial IMU (BMI088, angular resolution 0.05°) for real-time monitoring of attitude angles. (Roll) Pitch (θ), Yaw ); MCU reads initial attitude angle The data is then uploaded and used as a reference for subsequent attitude changes. The ultrasonic sensor emits pulses to record the guided wave velocity reference value in the ice-free state. The formula is: , Where E is the elastic modulus (e.g., E=70GPa). Density (e.g., 2700 kg / m3). Poisson's ratio (e.g.) =0.33); Impedance sensors record the capacitive response in air, serving as a dielectric reference in ice-free conditions. When icing occurs, the dielectric reference rises significantly, which is used to quickly determine whether icing has occurred. Based on the radar cross-section (RCS) model and the UAV's flight geometry, the incident angle corresponding to the maximum echo response is calculated. and initial radar cross-section The formula for reflecting the echo intensity of a corner reflector is: , in, This is the initial radar cross-section. The side length of the equilateral triangular aluminum plate of the corner reflector affects the RCS (Radio Cross Section). The wavelength of 9.6 GHz determines the penetration capability and resolution. The altitude of the drone, The horizontal distance between the drone and the corner reflector. For the arctangent function, 4π is a mathematical constant derived from the spherical geometric factor of electromagnetic wave scattering, representing the normalization coefficient of the all-directional radiation intensity; Each corner reflector node generates and uploads a data packet containing key parameters. Furthermore, a SAR system is deployed on the UAV, and SAR single-look complex images are generated from the UAV radar echo data through SAR imaging processing (such as the Range-Doppler algorithm).

[0025] By deploying a corner reflector network integrating multiple sensor sources, this method addresses the issue of decreased accuracy in existing icing monitoring technologies due to a lack of real-time environmental awareness. It employs a total station to precisely calibrate the initial position and construct a highly coherent spatial control network. Combined with an aluminum alloy corner reflector design featuring a mesh-like surface, this reduces icing load while ensuring structural stability and radar echo intensity. By integrating IMU, ultrasonic, and impedance sensors, it acquires attitude changes, guided wave velocity, and dielectric properties in real time, enabling rapid identification of icing occurrence and continuous monitoring of physical conditions. Based on the radar cross-section model and flight geometry, it determines the optimal observation angle and generates a data packet containing multi-dimensional information such as position, attitude, and electromagnetic characteristics. This provides accurate prior parameters for subsequent SAR imaging processing and signal-to-clutter ratio optimization, significantly improving the reliability and accuracy of icing monitoring in complex environments and overcoming the response lag and misjudgment problems caused by traditional methods due to a single data source and static configuration.

[0026] Furthermore, based on the generated data packets, the signal-to-clutter ratio (SCR) is calculated, and the standard deviation of target localization and phase estimation is obtained using the Cramer-Rao lower bound method. Dynamic weights are generated using the standard deviation, and the icing thickness estimate of the corner reflector is obtained through data fusion. Based on the icing thickness estimate, an optimized SCR index is generated, and the i-th actual radar cross-section is calculated. : , in, Based on the initial radar cross-section The calibration, the average digital value of the bright spot region of the corner reflector in the SAR image, was obtained by sub-pixel-level centroid extraction. These are the radar system calibration coefficients, used to convert the digital value (DN) of the bright spot of the corner reflector in the SAR image into the actual radar cross section (RCS), obtained through the absolute radiometric calibration method; Calculate the signal-to-noise ratio To evaluate the signal quality of the corner reflector, high The formula indicates that the signal is reliable and can achieve higher precision: , in, The echo intensity of the corner reflector. Mean value of background clutter; The Cramer-Rao lower bound (CRLB) method is used to evaluate the theoretical minimum standard deviation of target localization and phase estimation for a SAR system under a given signal-to-clutter ratio (SCR), including: Standard deviation of distance positioning accuracy The formula is: , , in, Here, represents the range resolution, c represents the speed of light, BW = 2 GHz represents the radar signal bandwidth (2 indicates two-way propagation of the radar signal (outbound + return), A is an empirical factor used to approximate the energy distribution of point targets (e.g., main lobe width), derived from an empirical model of the relationship between radar image resolution and signal-to-noise ratio, obtained from literature on the application of CRLB in SAR positioning (e.g., ESA technical reports), and C is a direct result of the theoretical derivation of CRLB in statistical signal processing. In the direction of distance, This is a mathematical approximation of the main lobe width of the sinc function; Azimuth positioning accuracy standard deviation The formula is: , , in, For azimuth resolution, Where is the drone's speed, and PRF is the pulse repetition frequency; Calculate the standard deviation of the phase noise of the i-th corner reflector. The formula is: , in, Interference phase; Obtain the midpoint of the bright spot of the corner reflector in a SAR single-view complex image. Specific operation: On the SAR image, using initial coordinates... Starting from the initial position, search for pixels within a radius of approximately 1 meter, identify the pixel with the highest intensity (peak value) within the region, and use the centroid method to precisely locate the center of the bright spot at the sub-pixel level, obtaining accurate image coordinates. ; Calculate ITRF coordinates (Global geocentric coordinates under the International Earth Reference Framework (ITRF), fused with SAR and external models (such as atmosphere and orbit), are used for long-term deformation monitoring and belong to the international standard reference system.) The formula is: , in, For solid tidal displacement, calculated using the IERS2010 model. The long-term rotational offset of the ITRF framework was obtained by interpolation based on IGS station data. The ITRF coordinates of the corner reflector are obtained through a zero-Doppler side-looking SAR imaging geometric model (built based on radar physical imaging principles, the model requires no training). Projecting the image onto the azimuth-range coordinate system of the SAR image yields the theoretical projection position. ; Calculate the actual observed location of the bright spot of the corner reflector in the SAR image. With theoretical projection position Coordinate deviation between and The error, reflecting the imaging geometric model, is expressed by the following formula: , Calculate the root mean square error : , in, This represents the total number of corner reflectors involved in the calculation; Based on the flight altitude, antenna beam angle, and imaging resolution of the SAR system platform (UAV), and combined with the zero-Doppler side-looking geometric model, the projection range of the beam on the ground is estimated, and the initial value of the equivalent circular diameter of the SAR antenna beam is set. =G, using an iterative optimization method for each candidate The values ​​are then used to recalculate the coordinate deviation between the actual observed position and the theoretical projected position of each corner reflector (CR). and And RMSE, iterative search to minimize RMSE. Introducing the optimal The residual from the initial value G is the confidence factor. This is used to evaluate the consistency between the optimized geometric parameters and the initial values, and the formula is: , in, Tolerance standard deviation indicates the allowable range of beam diameter deviation; Using a multi-source data fusion method under physical constraints, ultrasonic, impedance, and thermal infrared three-mode data are fused to calculate the estimated ice thickness on the corner reflector surface. The formula is: , , , , in, This is the reference value for guided wave velocity in an ice-free state. The measured wave velocity is (m / s). As an ice-free impedance reference, For the measured impedance, The difference is the thermal infrared temperature (°C). For the dynamic weights of ultrasonic guided wave modes, For the dynamic weights of the impedance sensor modes, For the dynamic weights of the thermal infrared modes, The baseline weighting coefficients were obtained through multimodal sensor calibration and regression analysis. The reference weighting coefficients for the corresponding impedance modes are obtained by fitting the icing response curve under controlled laboratory conditions. The baseline weighting coefficients for the thermal infrared modes are obtained through optimal threshold analysis and weight optimization based on ROC curves. Ice thickness estimate based on corner reflector surface and credibility factor Generate an optimized signal-to-noise ratio The formula is: , , in, The deviation between the actual incident angle and the optimal incident angle of the i-th corner reflector reflects the degree of attitude deviation. This is the maximum effective icing thickness; exceeding this thickness will cause impedance failure. The attitude attenuation coefficient was obtained by fitting laboratory simulation and measured data. The icing attenuation coefficient is determined based on experimental measurements of the ice layer dielectric constant and electromagnetic scattering modeling (such as FDTD), combined with the correspondence between actual ice thickness and SCR. The credibility decay coefficient is obtained through historical data regression analysis. The actual incident angle of the corner reflector, and the real-time attitude angle obtained through the IMU. Calculate the actual angle of incidence.

[0027] This method calculates the signal-to-noise ratio (SNR) by generating data packets and introduces the Cramer-Rao lower bound (CRLB) theory to assess the theoretical accuracy limit of target localization and phase estimation. It addresses the instability of inversion results in existing technologies due to the lack of quantitative assessment of SAR system observation quality. By modeling the standard deviations of range and azimuth positioning accuracy and phase noise, it achieves a refined characterization of the diagonal reflector echo signal quality, overcoming the misjudgment defects caused by relying solely on image intensity in traditional methods. Dynamic weights are constructed using the standard deviations derived from CRLB, and multi-source sensor data such as ultrasonic guided waves, impedance changes, and thermal infrared temperature differences are integrated to achieve [the desired result] under physical mechanism constraints. High-precision inversion of ice thickness enhances measurement robustness in complex environments. Through ITRF coordinate transformation and precise projection of the zero-Doppler side-view geometric model, combined with sub-pixel-level bright spot localization and beam equivalent diameter iterative optimization, a confidence factor is introduced to quantify the geometric consistency of the system, effectively suppressing registration deviations caused by attitude offsets or installation errors. Finally, an optimized signal-to-clutter ratio is generated based on ice thickness and attitude deviation, reflecting the degree of degradation of the corner reflector's working state. It also provides a reliability weighting basis for subsequent InSAR processing, realizing a leap from static observation to dynamic quality assessment, and significantly improving the accuracy, stability, and adaptability of icing monitoring.

[0028] S2. Using the generated optimized signal-to-clutter ratio, combined with ice thickness and attitude deviation, a collaborative judgment is made. Based on the judgment result, a joint adjustment is triggered. After the joint adjustment is triggered, SAR images are reacquired and main and auxiliary image registration, interferometric pair optimization, terrain phase subtraction and phase unwrapping are performed to extract a high-precision continuous line-of-sight phase time series. Specifically, the generated optimized signal-to-noise ratio is used to make a collaborative judgment based on icing thickness and attitude deviation. The judgment result triggers a joint adjustment based on the deviation. and ice thickness Trigger judgment is performed, including setting the attitude deviation threshold U through experimental calibration. The specific operation is as follows: change the incident angle under controlled environment and record the change of SCR (signal-to-noise ratio) until the SCR drops by 10%. Measure the angle deviation at this time as the maximum allowable incident angle deviation. The value range is set to U=2. The basis is that when the incident angle deviation exceeds this value, the SCR drops significantly, affecting the signal quality. Based on the heating response time test, the ice thickness threshold O was set. The specific operation involved applying a standard heating process of 30 seconds to ice layers of different thicknesses, and then observing and recording whether these ice layers could be completely melted without causing any negative impact on the structure. The threshold was set to O=1, which is the minimum ice thickness that the heater can successfully melt without affecting the structure. The signal-to-noise ratio (SCR) threshold X was set through signal quality analysis. Specifically, multiple sets of data were collected and analyzed to identify at what dB the SCR would decrease significantly. Statistical analysis revealed that when the SCR was less than 18 dB, the coherence dropped below 0.9, thus affecting the accuracy of interferometry. Therefore, the value of X was set to X=18. when deviation >Preset threshold U and When the threshold is 0, the corner reflector is considered to be in an abnormal state. When the preset threshold X (signal quality significantly degrades) is triggered alone, a check and adjustment are executed, resulting in a combined adjustment action. For example, the servo motor realigns the corner reflector's reflective surface with the optimal incident angle at a speed of 0.5. If the ice thickness exceeds 1mm, the resistance wire embedded in the mesh holes is activated to heat the ice, melting it within 30 seconds. The drone then hovers and re-flies, re-measuring the RCS and calculating the dynamic SCR to verify the recovery effect. After the adjustment is complete, the optimized corner reflector state tuple is output. , This indicates that the corner reflector has returned to a highly coherent and stable operating state.

[0029] By coordinating the judgment of optimized signal-to-noise ratio, icing thickness, and attitude deviation, and triggering a joint adjustment mechanism integrating attitude correction and active de-icing, the problem of high coherence not being restored in time due to the performance degradation of corner reflectors in existing icing monitoring systems is solved. When attitude deviation or icing impact reaches a critical level, the servo motor is automatically triggered to adjust the orientation of the reflector and the built-in resistance wire is activated to quickly melt the ice, improving the continuous working capability of the corner reflector in harsh environments and avoiding monitoring interruptions caused by signal quality degradation. The recovery effect was also verified by UAV reflight, ensuring the reliability and interferometric accuracy of SAR data, and realizing an intelligent leap from "monitoring-problem detection" to "automatic response-performance recovery".

[0030] Furthermore, after triggering joint adjustment, SAR images are reacquired and primary / secondary image registration, interferometric pair optimization, terrain phase subtraction, and phase unwrapping are performed. High-precision continuous line-of-sight phase time series indexes are extracted. After completing corner reflector attitude adjustment and de-icing compensation, the UAV reacquires SAR images and accurately extracts the position of each corner reflector in the image using sub-pixel-level bright spot localization algorithms (such as two-dimensional Gaussian fitting or phase center offset method). Coordinates are used to select one image from the registered image sequence as the master image, and the rest as slave images. The master image is selected based on its vertical baseline relative to the other images. For images with a depth of <80 m and a time baseline of less than 30 days, to control phase noise caused by spatial and temporal decorrelation, a baseline optimization algorithm (such as the minimum average baseline method or the maximum coherence priority method) is used to automatically select the image with the least spatial and temporal decorrelation impact as the master image, ensuring optimal overall interferometric pair quality. After generating the interferometric pair based on the master image, the flat terrain phase is removed, and then the terrain phase is subtracted using a digital elevation model (DEM) generated by high-precision LiDAR measurements to obtain the interferometric phase. The terrain phase calculation formula is as follows: , in, The phase components caused by topographic relief need to be removed from the interferometric phase. The vertical baseline is the projected distance between the two orbits in the direction perpendicular to the line of sight. It is obtained through satellite / UAV orbital parameters (POS data) and imaging geometry calculations. Common methods are orbital extrapolation or RPC model inversion. z is the ground elevation, which is generated using a DEM (Digital Elevation Model) generated by LiDAR scanning. The DEM is a two-dimensional grid dataset containing ground elevation values. λ is the radar wavelength, corresponding to the X-band (such as TerraSAR-X or a self-developed UAV SAR system), which is determined by the sensor hardware parameters. R is the slant range, which is the straight-line distance from the radar antenna to the ground point, and is obtained by imaging geometry and flight altitude calculations. The Minimum Cost Flow (MCF) algorithm is used to unwrap the interferometric phase, effectively handling the low coherence region and outputting a continuous, non-jumping line-of-sight phase time series. (t), Specific operations: Calculate the phase gradient of the interferogram and detect residual points (i.e., 2π discontinuities) to form nodes of the network flow graph. Use the phase difference between pixels as the flow of the edges. Set the weights of each edge based on coherence or intensity (give high weights to low-coherence regions to reduce error propagation). Construct an integer optimization model. The goal is to minimize the sum of weighted phase gradient residuals in the entire network, that is, to transform the phase jump problem into finding the integer flow distribution that minimizes the total "cost". Calculate the optimal flow field using the minimum cost flow algorithm in graph theory to obtain the 2π integer offset to be added. Integrate this integer offset along the path and superimpose it onto the original wrapped phase to achieve phase unwrapping. Output a continuous, non-jumping line-of-sight phase time series. (t).

[0031] Precise matching of corner reflector coordinates is ensured through sub-pixel-level bright spot localization. High-quality main images are automatically selected by combining vertical and temporal baseline constraints to suppress noise caused by spatiotemporal decorrelation. High-precision LiDAR-DEM is used for accurate modeling and terrain phase subtraction to avoid phase residue caused by digital elevation model errors, thereby improving the purity of the interferometric phase. Phase unwrapping is performed using a minimum cost flow algorithm. A network flow model is constructed based on coherence weighting to overcome the phase jump problem in low coherence regions. Continuous and reliable line-of-sight phase time series are output, achieving high-precision deformation inversion based on dynamic recovery of corner reflector performance. This enhances the system's robustness and monitoring continuity in complex environments and solves the bottleneck problem of obtaining stable temporal deformation data due to signal degradation in existing technologies. By re-flying the UAV and performing a refined InSAR processing workflow, including main and auxiliary image registration, interferometric pair optimization, terrain phase subtraction, and phase unwrapping, the low deformation extraction accuracy caused by phase incoherence, geometric distortion, and terrain interference in traditional monitoring methods is solved.

[0032] S3. Based on the ice thickness and multi-dimensional data, a state vector is generated. The line-of-sight phase, the three-dimensional displacement of the corner reflector, and the ice thickness are fused by multimodal extended Kalman filtering. The state vector is optimized to generate the optimal state vector. The stability index is calculated using the optimal state vector. The corner reflector coordinates are re-optimized based on the stability index. Specifically, a state vector is generated based on icing thickness and multidimensional data. A multimodal extended Kalman filter is used to fuse line-of-sight phase, corner reflector 3D displacement, and icing thickness to optimize the state vector and generate the optimal state vector based on the line-of-sight phase time series. (t) Provides high temporal resolution deformation observations, corner reflector coordinates As a reference position for LiDAR measurements, it is used to calculate the displacement change. The obtained ice thickness As an environmental coupling variable, it is used to correct spurious deformation or sensor response drift caused by surface icing, and to convert the time series... (t), coordinates of the corner reflector and ice thickness The multiple observation inputs that together constitute the EKF ensure that the inversion results are both accurate and environmentally adaptable. In the EKF framework, the state vector is defined as: , in, and This indicates the three-dimensional displacement (east-west direction) of the slope point where the corner reflector is located in the WGS-84 geographic coordinate system. North-South Direction Vertical direction ), Let T be the velocity along the potential sliding surface, and T be the transpose. The three-dimensional displacement and slip rate of the slope are obtained through a two-stage iterative process of "prediction-update" in the multimodal extended Kalman filter (EKF) fusion framework. and the state of freezing The joint estimation, obtained during the EKF update phase, uses the observation residuals and Kalman gain to correct the predicted values, resulting in the optimal estimated state vector. Specifically, in the prediction phase, the prior value of the current state is calculated based on the state at the previous moment and the set motion model (such as constant slip rate). In the update phase, three observations... (InSAR projection) = (LiDAR displacement) = (Ice thickness) of which The unit vector representing the line of sight direction is determined by three observations. The corresponding sub-models together form the overall observation matrix H, which, combined with the noise covariance matrix I, dynamically calculates the Kalman gain. This allows for the correction of predicted values ​​and the acquisition of the optimal estimated state vector. The extended Kalman filter formula is:

[0033] in, The error covariance matrix represents the predicted state.

[0034] By introducing the multimodal extended Kalman filter (EKF) framework and combining multi-source observation data such as line-of-sight phase time series, corner reflector 3D displacement, and icing thickness to generate optimal state vectors, the accuracy and reliability of slope deformation monitoring in complex environments are improved. Addressing the difficulty of distinguishing between true deformation and spurious deformation or sensor response drift caused by surface icing in traditional techniques, this approach utilizes high-resolution InSAR data to provide continuous deformation information. Displacement changes are calculated using LiDAR-measured corner reflector coordinates as a reference position, and icing thickness is incorporated as an environmental coupling variable into the state estimation model to correct for the influence of environmental factors on the monitoring results. The "prediction-update" mechanism under the EKF framework can infer the current state based on the previous state, dynamically adjusting the predicted values ​​through observation residuals and Kalman gain. This achieves joint optimization estimation of slope 3D displacement, slip rate, and icing state, enabling stable and reliable deformation monitoring data even in low-coherence areas or under severe weather conditions. This enhances the ability and adaptability to cope with complex environmental changes, solves the problem of inaccurate deformation monitoring caused by single data source limitations and environmental interference in existing technologies, and provides a more comprehensive and accurate means of slope stability assessment.

[0035] Furthermore, the stability index is calculated using the optimal state vector, and the corner reflector coordinates are re-optimized based on the stability index and the optimal estimated state vector. The slope's motion state (slippage rate) is quantitatively integrated with environmental disturbance factors (ice thickness) to construct a unified risk assessment index with clear physical meaning, and a stability index is then established. It is used to automatically identify potential instability trends, and the formula is: , in, Let be the stability index of the location of the i-th corner reflector, which belongs to the interval [0-1]. A value approaching 0 indicates an imminent instability; this occurs when slippage is small and there is no ice. A value close to 1 indicates stability. This represents the absolute value of the slip velocity, reflecting the severity of the slope movement. As the reference slip rate threshold, To serve as a reference ice thickness scale for normalizing the effect of ice thickness. This is a coefficient for amplifying the risk of icing, controlling the weight of the impact of ice thickness on stability. The larger the ice layer, the stronger its amplifying effect on the risk. Time-series data of three-dimensional displacement, slip velocity, and corresponding instability events of slopes under different working conditions were collected. The critical slip velocity range before instability was statistically analyzed. Combined with slope material properties and geological survey data, the theoretical instability threshold range was determined using the limit equilibrium method. ROC curve analysis was conducted, with early warning sensitivity and false alarm rate as evaluation indicators, and the method that maximizes the AUC was selected. The value is used as the initial threshold η. The threshold η is iteratively corrected based on the actual early warning effect and expert experience, and finally the optimal early warning threshold η that takes into account both early warning capability and reliability is determined. Stability index calculated based on formula ,when When the preset threshold η is reached, the slope area where the corner reflector is located is determined to be in a potentially unstable state, and a graded early warning mechanism is immediately triggered. Specific steps include: generating a stability assessment report and marking low stability indices. In the designated space region, the UAV mission planning module is automatically activated, and an autonomous reflight is performed at the set time the following day. This re-optimizes the corner reflector coordinates, compensates for attitude and icing, performs InSAR processing, and inverts the latest 3D deformation to verify... Whether it continues to decline or deteriorates rapidly, if retested If the level remains below the threshold, the warning level will be upgraded (from yellow to red), and emergency response suggestions will be pushed to the management platform, such as setting up a warning zone or initiating reinforcement projects.

[0036] By constructing a stability index based on the optimal state vector, the slope slip rate is quantitatively integrated with motion states such as icing thickness and environmental disturbance factors to form a unified risk assessment index with clear physical meaning. This addresses the shortcomings of existing monitoring technologies, such as the disconnect between deformation analysis and environmental impact, and the difficulty in achieving early instability warnings. Traditional methods often rely on single thresholds or empirical judgments, lacking a comprehensive consideration of the coupled effects of multiple factors. The stability index can dynamically reflect the gradual process of a slope from stability to instability. By setting warning thresholds to trigger a graded response mechanism, it can achieve closed-loop management from "passive alarm" to "active prediction-verification-upgrade". When a potential instability area is detected, the system automatically drives the UAV to re-fly and re-measure, verifying the deformation trend and updating the corner reflector parameters, improving the level of intelligence and emergency response capabilities. It overcomes the lag and isolation of existing technologies in risk identification, achieving high-precision, interpretable, and adaptive stability assessment, and providing a scientific decision-making basis for the safe operation and maintenance of infrastructure in complex environments.

[0037] This embodiment also provides an icing monitoring system based on corner reflectors deployed by UAVs, including: The corner reflector network deployment and multi-source sensing module is used to deploy a corner reflector network, collect initial coordinates, attitude, RCS and environmental parameters, and construct a highly coherent monitoring reference point; The dynamic weight generation and icing estimation module is used to calculate the standard deviation of positioning and phase accuracy based on the signal-to-noise ratio and CRLB, fuse multi-source sensor data and generate dynamic weights, invert icing thickness and optimize signal-to-noise ratio; The joint adjustment and data optimization module is used to trigger attitude correction and de-icing actions based on icing and attitude deviation, adjust the re-flight to acquire SAR images, and extract high-precision line-of-sight phase sequences. The multimodal state fusion and deformation inversion module is used to construct a state vector containing three-dimensional displacement and slip rate. It outputs the optimal deformation estimate by fusing InSAR and icing data through extended Kalman filtering. The stability assessment and intelligent early warning module is used to calculate the stability index of the optimal state vector. When the index is lower than the threshold, it triggers a graded early warning and drives the UAV to retest, thereby achieving automated risk identification.

[0038] This embodiment also provides a computer device applicable to an icing monitoring method based on a corner reflector deployed by an unmanned aerial vehicle (UAV), comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the icing monitoring method based on a corner reflector deployed by an UAV as proposed in the above embodiment.

[0039] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0040] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the icing monitoring method and system based on a corner reflector deployed by a UAV, as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, disk, or optical disk.

[0041] In summary, this invention integrates multi-source sensors into the corner reflector base to acquire attitude, guided wave velocity, and dielectric response environmental parameters in real time. Combined with SAR image echo characteristics, it utilizes the Cramer-Rao lower bound theory to assess the theoretical accuracy limit of positioning and phase estimation, thereby generating dynamic weights. This achieves physical constraint fusion of ultrasonic, impedance, and thermal infrared multimodal data, significantly improving the accuracy and robustness of icing thickness inversion. A credibility factor is introduced to optimize the geometric model, and an optimized signal-to-noise ratio is generated based on icing thickness and attitude deviation to quantify the degree of signal quality degradation, providing a reliable basis for subsequent monitoring. Through multi-source information synergy and dynamic weight adjustment, it overcomes the limitations of single sensors being susceptible to environmental interference, enhancing the stability and accuracy of icing monitoring under complex meteorological conditions. This represents a leap from passive observation to active sensing and adaptive optimization, demonstrating significant technological advancement and practical value.

[0042] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for monitoring icing based on corner reflectors deployed by unmanned aerial vehicles, characterized in that: include, Deploy a corner reflector network, collect radar cross-section characteristics and environmental parameters to generate data packets, calculate the signal-to-clutter ratio based on the generated data packets, and use the Cramer-Rao lower bound method to obtain the standard deviation of target positioning and phase estimation. Use the standard deviation to generate dynamic weights, and obtain the estimated value of corner reflector icing thickness through data fusion. Generate an optimized signal-to-clutter ratio based on the icing thickness estimate. Using the generated optimized signal-to-clutter ratio, combined with ice thickness and attitude deviation, a collaborative judgment is made. Based on the judgment result, a joint adjustment is triggered. After the joint adjustment is triggered, SAR images are reacquired and main and auxiliary image registration, interferometric pair optimization, terrain phase subtraction and phase unwrapping are performed to extract a high-precision continuous line-of-sight phase time series. A state vector is generated based on ice thickness and multidimensional data. The line-of-sight phase, three-dimensional displacement of the corner reflector, and ice thickness are fused by multimodal extended Kalman filtering. The state vector is optimized to generate the optimal state vector. The stability index is calculated using the optimal state vector. The corner reflector coordinates are then re-optimized based on the stability index.

2. The icing monitoring method based on a corner reflector deployed by an unmanned aerial vehicle as described in claim 1, characterized in that: The deployment of the corner reflector network involves collecting radar cross-section characteristics and environmental parameters to generate data packets. This includes deploying a triangular trihedral corner reflector network, using a total station to mark installation points, and obtaining the initial coordinates of the corner reflectors. Each corner reflector base integrates multiple sensors and modules to calculate the incident angle. and initial radar cross-section Each corner reflector node generates and uploads a data packet containing key parameters, and a SAR system is deployed on the UAV. The UAV radar echo data is processed by SAR imaging to generate a SAR single-look complex image.

3. The icing monitoring method based on a corner reflector deployed by an unmanned aerial vehicle as described in claim 2, characterized in that: Based on the generated data packets, the signal-to-clutter ratio (SCR) is calculated, and the standard deviation of target localization and phase estimation is obtained using the Cramer-Rao lower bound method. Dynamic weights are generated using the standard deviation, and the estimated icing thickness of the corner reflector is obtained through data fusion. Based on the icing thickness estimate, an optimized SCR index is generated, and the i-th actual radar cross-section is calculated. Hoshin Mix The Cramer-Rao lower bound method is used to evaluate the theoretical minimum standard deviation of target localization and phase estimation of a SAR system under a given signal-to-clutter ratio, including the standard deviation of range-direction localization accuracy. Standard deviation of azimuth positioning accuracy Standard deviation of phase noise of corner reflector ; Obtain the midpoint of the bright spot of the corner reflector in a SAR single-view complex image. Calculate ITRF coordinates The ITRF coordinates of the corner reflector were determined using a zero-Doppler side-looking SAR imaging geometric model. Projecting the image onto the azimuth-range coordinate system of the SAR image yields the theoretical projection position. Calculate coordinate deviation and Calculate the root mean square error ; Set the initial value of the equivalent circular diameter of the SAR antenna beam. =G, using an iterative optimization method for each candidate Values, recalculate the coordinate deviation of each corner reflector and And RMSE, iterative search to minimize RMSE. Introducing the optimal The residual from the initial value G is the confidence factor. ; Calculate the estimated icing thickness on the corner reflector surface. Combined with credibility factor Generate an optimized signal-to-noise ratio The deviation between the actual incident angle and the optimal incident angle of the corner reflector .

4. The icing monitoring method based on a corner reflector deployed by an unmanned aerial vehicle as described in claim 3, characterized in that: The generated optimized signal-to-noise ratio is used to perform a collaborative judgment based on icing thickness and attitude deviation. Based on the judgment result, a joint adjustment is triggered based on the deviation. and ice thickness Perform trigger judgment, including when deviation >Preset threshold U and When the threshold is 0, the corner reflector is considered to be in an abnormal state. When the preset threshold X is met alone, a check and adjustment are triggered, and a joint adjustment action is performed, such as realigning the corner reflector's reflective surface with the optimal incident angle. If the ice thickness exceeds the set thickness, the resistance wire embedded in the mesh hole will be activated to heat and melt the thin ice. The drone will then hover and fly again, remeasure the RCS, calculate the dynamic SCR, and output the optimized corner reflector state tuple after the adjustment is completed.

5. The icing monitoring method based on a corner reflector deployed by an unmanned aerial vehicle as described in claim 4, characterized in that: After triggering joint adjustment, SAR images are reacquired and primary and secondary image registration, interferometric pair optimization, terrain phase subtraction, and phase unwrapping are performed. A high-precision continuous line-of-sight phase time series is extracted. After corner reflector attitude adjustment and de-icing compensation are completed, the UAV reacquires SAR images, and the coordinates of each corner reflector in the image are accurately extracted using a sub-pixel-level bright spot localization algorithm. Primary and secondary images are selected from the registered image sequence. After generating interferometric pairs based on the primary image, the flat-ground phase is removed, and the terrain phase is subtracted using a digital elevation model generated by high-precision LiDAR measurements to obtain the interferometric phase. A minimum cost flow algorithm is used to unwrap the interferometric phase, effectively handling low-coherence areas and outputting a continuous, non-jumping line-of-sight phase time series. (t).

6. The icing monitoring method based on a corner reflector deployed by an unmanned aerial vehicle as described in claim 5, characterized in that: The process involves generating a state vector based on ice thickness and multidimensional data. This state vector is then fused with line-of-sight phase, corner reflector 3D displacement, and ice thickness using a multimodal extended Kalman filter to optimize the state vector and generate the optimal state vector pointer for the time series. (t), coordinates of the corner reflector and ice thickness Together, they constitute the multi-source observation input of EKF. In the EKF framework, the state vector is defined as... The three-dimensional displacement and slip rate of the slope are obtained through a two-stage iterative prediction and update process within a multimodal extended Kalman filter fusion framework. and the state of freezing The joint estimation, obtained during the EKF update phase, uses the observation residuals and Kalman gain to correct the predicted values, resulting in the optimal estimated state vector. .

7. The icing monitoring method based on a corner reflector deployed by an unmanned aerial vehicle as described in claim 6, characterized in that: The stability index is calculated using the optimal state vector, and the corner reflector coordinates are re-optimized based on the stability index and the optimal estimated state vector. The stability index is constructed by quantitatively integrating the slope's motion state with environmental disturbance factors. The stability index calculated based on the formula ,when When the preset threshold η is reached, the slope area where the corner reflector is located is determined to be in a potentially unstable state, and a graded early warning mechanism is immediately triggered, including generating a stability assessment report and marking low stability indices. In the designated space region, the UAV mission planning module is automatically activated, and an autonomous reflight is performed at the set time the following day. This reflight optimizes the corner reflector coordinates, compensates for attitude and icing, performs InSAR processing, and inverts the latest 3D deformation to verify... Whether it continues to decline, if retested If the level remains below the threshold, the warning level will be upgraded, and emergency response recommendations will be sent to the management platform.

8. An icing monitoring system based on a corner reflector deployed by an unmanned aerial vehicle (UAV), based on the icing monitoring method based on a corner reflector deployed by an UAV as described in any one of claims 1 to 7, characterized in that: include, The corner reflector network deployment and multi-source sensing module is used to deploy a corner reflector network, collect initial coordinates, attitude, RCS and environmental parameters, and construct a highly coherent monitoring reference point; The dynamic weight generation and icing estimation module is used to calculate the standard deviation of positioning and phase accuracy based on the signal-to-noise ratio and CRLB, fuse multi-source sensor data and generate dynamic weights, invert icing thickness and optimize signal-to-noise ratio; The joint adjustment and data optimization module is used to trigger attitude correction and de-icing actions based on icing and attitude deviation, adjust the re-flight to acquire SAR images, and extract high-precision line-of-sight phase sequences. The multimodal state fusion and deformation inversion module is used to construct a state vector containing three-dimensional displacement and slip rate. It outputs the optimal deformation estimate by fusing InSAR and icing data through extended Kalman filtering. The stability assessment and intelligent early warning module is used to calculate the stability index of the optimal state vector. When the index is lower than the threshold, it triggers a graded early warning and drives the UAV to retest, thereby achieving automated risk identification.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the icing monitoring method based on a corner reflector deployed by an unmanned aerial vehicle as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the icing monitoring method based on a corner reflector deployed by an unmanned aerial vehicle as described in any one of claims 1 to 7.

Images