A high-speed slope monitoring method and system based on InSAR and Beidou ground equipment

Abstract

This invention discloses a method and system for monitoring high-speed slopes based on InSAR and BeiDou ground equipment, belonging to the field of high-speed slope monitoring technology. The method includes the following steps: S1 data acquisition, S2 data preprocessing, S3 multi-dimensional data analysis, S4 error correction and data fusion, and S5 risk warning. During data acquisition, MT-InSAR data for surface deformation monitoring and real-time monitoring data collected by BeiDou ground equipment deployed on the slope are acquired separately. A spatiotemporal dynamic filtering method is used to denoise the two types of raw data, eliminating interference from non-surface deformation factors and unifying the data format and spatiotemporal reference. A spatiotemporal model of surface deformation is established by combining the topographic data of the monitoring area. This invention's method and system for monitoring high-speed slopes based on InSAR and BeiDou ground equipment, through the deep integration of MT-InSAR and BeiDou technologies, constructs a collaborative monitoring system of surface and point areas for global screening and local verification, improving the accuracy, coverage, and reliability of slope deformation monitoring and early warning.

Description

Technical Field

[0001] This invention specifically relates to a method and system for monitoring high-speed slopes based on InSAR and BeiDou ground equipment, belonging to the field of high-speed slope monitoring technology. Background Technology

[0002] As my country's expressway network becomes increasingly complete, a large number of roads have entered the operation and maintenance phase. Expressway slopes are crucial engineering structures for ensuring roadbed stability and traffic safety. Affected by multiple factors such as geological conditions, rainfall erosion, vehicle loads, and foundation settlement, they are prone to roadbed subsidence and slope collapse, seriously threatening traffic safety and causing traffic disruptions. Therefore, monitoring of high slopes is necessary. Existing monitoring methods include a slope displacement monitoring method based on BeiDou positioning disclosed in Chinese Patent Publication No. CN121048551B. This method combines slope geological survey data, geological structure data, and previous monitoring data to select stable monitoring areas and deploy fixed monitoring points, thereby avoiding coordinate distortion caused by the displacement of the reference points themselves and significantly improving the three-dimensional displacement monitoring. The data calculation accuracy and analysis results are reliable; however, it cannot achieve large-scale global monitoring. In addition, existing slope monitoring methods are divided into two categories: one is ground sensor monitoring, which collects data such as deformation, dip angle, and cracks by setting up measuring points on the slope, which can achieve high-precision monitoring of the measuring point location, but the monitoring range is limited to a single point and cannot cover the entire slope area, making it difficult to identify large-scale settlement trends; the other is synthetic aperture radar interferometry (InSAR) technology, which inverts surface deformation through multi-temporal radar images, but is easily affected by spatiotemporal decoherence and atmospheric delay, limiting the monitoring accuracy; and currently, highway slope settlement monitoring mostly adopts a single technical means, which has the defects of single monitoring dimension, high false alarm rate, and difficulty in balancing coverage and monitoring accuracy, increasing maintenance costs and disaster prevention risks. Summary of the Invention

[0003] To address the aforementioned issues, this invention proposes a high-speed slope monitoring method and system based on InSAR and BeiDou ground equipment. Through the deep integration of MT-InSAR and BeiDou technology, a collaborative monitoring system of surface and point monitoring with global screening and local verification is constructed, thereby improving the accuracy, coverage, and reliability of slope deformation monitoring and early warning.

[0004] The high-speed slope monitoring method based on InSAR and BeiDou ground equipment of the present invention includes the following steps: S1 Data Acquisition: Acquire MT-InSAR data for surface deformation monitoring, and real-time monitoring data collected by BeiDou ground equipment deployed on the slope. S2 data preprocessing: Spatiotemporal dynamic filtering method is used to denoise the two types of raw data, remove interference from non-surface deformation factors, and unify the data format and spatiotemporal reference; a spatiotemporal model of surface deformation is established by combining the topographic data of the monitoring area to extract effective monitoring data of the slope area. S3 Multidimensional Data Analysis: Extracts surface deformation information of overall and local displacement, deformation rate, and deformation mode of the entire slope through MT-InSAR data to identify potential risk areas; obtains point-type physical parameters such as dip angle, crack width, and surface strain at measuring points through real-time monitoring data to analyze the refined deformation characteristics of high-risk areas, and conducts slope dynamic behavior and stability assessment by combining the two types of data. S4 Error Correction and Data Fusion: Using the fitting coefficient method, with the BeiDou high-precision observation value as the benchmark, the observation difference between the two types of data at the same point is fitted to correct the deviation of the MT-InSAR data; then, the Kalman filter algorithm is used to fuse the corrected MT-InSAR data with the real-time monitoring data to output high-precision deformation results of highway slopes. S5 Risk Warning: Based on the fused deformation data, a risk warning model for slopes is constructed using cumulative settlement, average settlement rate, and settlement acceleration as indicators. Multi-level warning thresholds are set to monitor slope stability changes in real time and output warning information.

[0005] The risk warning system combines InSAR's large-area survey and long-term trend capture with BeiDou's strength in capturing sudden deformations at single points with high frequency to implement a two-level warning system. For the first-level hazard warning, InSAR monitoring is the primary method for medium- to long-term surveys. Utilizing the global deformation field of InSAR during each period of orbital ascent and descent, areas with abnormal deformation rates are automatically delineated, and a yellow hazard warning is issued. This guides maintenance personnel to densely deploy BeiDou monitoring points in areas of abnormal deformation, completing the transition from area-based scanning to point-based monitoring. For the second-level disaster warning, BeiDou takes the lead in short-term emergency monitoring. For the densely deployed BeiDou monitoring points, displacement acceleration and crack propagation rate are calculated in real time. Once these indicators exceed critical thresholds, orange and red emergency warnings are immediately triggered, initiating slope emergency repair procedures.

[0006] Furthermore, in step S4, the Kalman filter algorithm includes observation equations and state equations: Observation equation: ; Equations of state: ; In the formula, Download wave phase observations for Kalman filtering; Design a matrix for the observations; Δ is the Kalman filter state vector; Δ is the observation noise vector. This is the state transition matrix; This is the original state vector from the previous moment; This is the noise distribution matrix; This is the system noise vector.

[0007] Furthermore, in step S4, the deformation and variance of the slope at the initial monitoring time are set to be 0. The initial deformation rate is calculated by adjusting the real-time monitoring data from BeiDou observation and MT-InSAR data. The system noise variance expression is: ; In the formula, The noise variance between radar and satellite observations; This represents the time interval between adjacent monitoring moments.

[0008] Furthermore, in step S4, the expression for the multi-directional deformation observation matrix after data fusion is: ; In the formula, and These represent the deformations of the satellite in the line-of-sight (LOS) direction during its ascent and descent. , and These represent the deformation of the BeiDou equipment in the east-west, north-south, and vertical directions, respectively. , and The projection vectors of the ascending InSAR data in the east-west, north-south, and vertical directions; , and The projection vectors of the down-orbit InSAR data in the east-west, north-south, and vertical directions; , and These are the original coordinates of the BeiDou measurement points in the east-west, north-south, and vertical directions, respectively.

[0009] Furthermore, in step S5, the risk warning model sets three warning thresholds: yellow, orange, and red. It periodically calculates and updates the slope risk index. When the monitoring index exceeds the corresponding threshold, it automatically pushes graded warning information to the maintenance department.

[0010] Furthermore, due to the atmospheric delay and orbital system deviation of the single-point BeiDou observation value in unidirectional correction of InSAR; under long-term monitoring conditions, the BeiDou reference station will be affected by slow ground settlement and receiver zero-point drift, introducing systematic offsets into its own observation sequence, directly leading to distortion of the InSAR correction reference; therefore, step S2 also includes constructing a regional stable reference set, specifically as follows: forming a regional stable reference set by screening InSAR permanent scatterer points with annual deformation rates below a threshold; before the error correction, a bidirectional error correction step is also performed: First, using the long-term deformation mean of the regional stable reference set, the foundation settlement and instrument drift error of the BeiDou reference station are calibrated in reverse. The drift correction calculation formula is as follows: ; In the formula, This refers to the drift correction amount for the BeiDou reference station, expressed in millimeters per year. The average deformation rate of the globally stable permanent scattering point (using the annual average deformation rate). The original deformation rate observed by the BeiDou reference station is the original annual deformation rate output by the BeiDou reference station before correction; the corrected BeiDou observation sequence (BeiDou observation time series after drift compensation) is as follows: Then, using the calibrated BeiDou data as a reference, the fitting coefficient method is used to positively correct the InSAR systematic error. The permanent scatterer points contained in the regional stable reference set form a large-scale permanent scatterer network of InSAR. The stability of the regional stable reference set enables bidirectional calibration: first, the BeiDou reference drift is corrected in reverse using the globally stable InSAR points, and then the InSAR observation error is positively corrected using the corrected BeiDou data, thus completely cutting off the transmission of systematic error between the two types of data.

[0011] During operation, in the monitoring area of ​​interchanges, 62 solid building foundations were selected as stable permanent scatterers, and the average annual deformation rate was statistically obtained to be 0.3 mm per year. The annual deformation rate of the Beidou reference station without correction was 1.1 mm per year. Substituting these values ​​into the calculation, the drift correction ΔB = 0.3 - 1.1 = -0.8 mm per year was calculated. This correction value was then superimposed on all Beidou time-series observation data to eliminate the system offset caused by the settlement of the reference station itself. After completing the Beidou reference calibration, the fitting coefficient method was used to correct the deviation of the full-domain InSAR line-of-sight deformation data. The calibration reference was more reliable, and the overall systematic error was reduced by more than half compared with the unidirectional calibration.

[0012] Furthermore, in MT-InSAR processing of large deformation sections of slopes, where interference fringes are densely stacked, phase unwrapping is prone to integer jumps, and errors accumulate along the unwrapping path, ultimately causing severe distortion in deformation calculations in large slip regions. Therefore, before the deviation correction, phase unwrapping using BeiDou-assisted InSAR is also included. This converts the high-precision three-dimensional displacement of BeiDou into radar line-of-sight deformation, which is then converted into absolute interferometric phase. This absolute phase is used as a hard control point to constrain the unwrapping path, locking the integer phase number and curbing the spread of unwrapping errors. The specific process is as follows: the three-dimensional deformation of the BeiDou measurement points is projected onto the satellite line-of-sight direction to obtain the line-of-sight deformation. The calculation formula is: In the formula, The eastward, northward, and vertical deformations measured by BeiDou. The projection components of the unit vector of the satellite's line-of-sight direction in the east, north, and vertical directions are calculated from the satellite's incident angle and azimuth angle; the line-of-sight deformation is a one-dimensional deformation of the line-of-sight direction obtained by converting BeiDou observation values, and after obtaining the line-of-sight deformation, it is further converted into the true interferometric phase: In the formula, λ is the true interferometric phase corresponding to the BeiDou point, and λ is the radar electromagnetic wave wavelength. This true interferometric phase is used as a known control point and substituted into the phase unwrapping to constrain the unwrapping path in the large deformation gradient region and correct the unwrapping cumulative error. During operation, in large deformation sections of ramp slopes, without control points, the cumulative deformation obtained from phase unwrapping inversion reached 21.3 mm, significantly deviating from the actual deformation. By projecting the three-dimensional displacement of the BeiDou measurement point at the slope toe into line-of-sight deformation, calculating the absolute phase value, and embedding this phase point into the minimum cost flow unwrapping model as a constraint node, the cumulative deformation after re-unwrapping was 17.9 mm, reducing the deviation from the BeiDou measured value to 0.3 mm. This completely solved the problem of unwrapping failure in dense stripe areas and broadened the applicability of InSAR in medium to large landslide sections.

[0013] Furthermore, since the monitoring process can only calculate the three-dimensional displacement at a limited number of BeiDou measurement points, and the vast area can only output a single line-of-sight deformation, it is impossible to distinguish between settlement and horizontal slippage, making it difficult to determine the slope slippage pattern; therefore, in step S4, after the data fusion, a full-domain three-dimensional deformation inversion step is also included. Using discrete BeiDou three-dimensional observation values ​​(sparse BeiDou constraint) as spatial constraints, relying on the continuity of surface deformation, and through Kriging interpolation, the one-dimensional line-of-sight observation values ​​of each InSAR pixel in the entire slope area are used to calculate the three-dimensional displacement components in the east, north, and vertical directions, upgrading the one-dimensional surface monitoring to a full-domain three-dimensional deformation field; specifically as follows: using the three-dimensional deformation results of BeiDou measurement points as constraints, combined with the assumption of spatial continuity of the deformation field, the InSAR one-dimensional line-of-sight surface deformation data is inverted into a full-domain east, north, and vertical three-dimensional deformation field of the slope through the Kriging spatial interpolation algorithm; wherein the line-of-sight deformation and three-dimensional deformation components of any InSAR pixel satisfy: Under the three-dimensional displacement constraint of discrete BeiDou measurement points, the solution is obtained pixel by pixel for the entire image grid. Generate a continuous three-dimensional deformable mesh; During the work, for example, after three-dimensional inversion of the slope of a road cut in a village or town, two deformations can be clearly distinguished: the vertical settlement rate at the toe of the slope is 3.2 mm per month, while the horizontal slip along the slope is 1.5 mm per month. Relying solely on the original one-dimensional LOS observation, only the overall displacement can be observed, and the settlement and lateral slip cannot be separated. The three-dimensional deformation field can directly determine that the slope belongs to the push-slip failure, providing a key basis for stability analysis.

[0014] Furthermore, because fixed-parameter Kalman filtering in InSAR and BeiDou fusion uses constant observation noise weights and process noise weights throughout, it cannot adapt to the complex and non-uniform surface monitoring environment of highway slopes, i.e., it cannot adapt to the spatial heterogeneity of slopes: InSAR is prone to decoherence in densely vegetated areas, and the observation noise increases sharply; InSAR observation stability is stronger in open, hardened road areas; at the same time, the two types of equipment have huge differences in temporal resolution: BeiDou can achieve hourly high-frequency sampling, while InSAR can only achieve long-term observations of more than ten days. Therefore, by adding an adaptive weight mechanism, the observation covariance of the two types of data in the filtering fusion is dynamically adjusted in terms of space and time period, further improving the reliability of multi-source data fusion; for example, in step S4, the Kalman filtering fusion adopts a spatiotemporal adaptive weight adjustment mechanism, such as dynamically updating the observation noise covariance matrix in real time during the Kalman filtering iteration process, and equivalently dynamically allocating the fusion credibility weight of InSAR area data and BeiDou point data, i.e., the better the coherence and the more the time sequence matches, the higher the corresponding weight. The higher the data source weight, the lower the noise confidence; the worse the coherence and the mismatch in time scale, the lower the corresponding data source weight and the higher the noise confidence. The overall process is divided into spatial adaptive adjustment and temporal adaptive adjustment, which are applied simultaneously to the fusion solution process. Specifically, the InSAR observation noise covariance, BeiDou data weight, and InSAR data weight are adjusted according to the spatial dimension. For example, on slopes with high vegetation cover and poor coherence, the InSAR observation noise covariance is increased, and the weight of BeiDou point observations is raised. In areas with strong scattering and stability, such as hardened roadbeds and buildings, the InSAR noise weight is reduced to give full play to the constraint effect of area monitoring data. The BeiDou data weight and InSAR data weight are adjusted according to the time dimension. Short-term monitoring (hourly time series analysis) mainly uses BeiDou high-frequency observations. In medium- and long-term trend analysis (monthly and quarterly deformation surveys), the InSAR area data weight is increased. After adaptive weight optimization, the random error of the fusion results in incoherent areas is greatly reduced, and the InSAR area field constraint advantage is maintained in stable sections, taking into account both global continuity and point observation accuracy.

[0015] I. The specific implementation of adaptive weight adjustment in space is as follows: Based on the coherence coefficient of InSAR imagery, the entire slope area is divided into three monitoring zones, and fusion weights are dynamically assigned to each zone: (1) High coherence stable region (road surface, hard slope, building): The interference fringes in this region are stable and the degree of decoherence is extremely low. The InSAR surface deformation continuity is strong, the spatial consistency is good, and the long-term trend is reliable. The processing flow is as follows: reduce the InSAR observation noise assignment, increase the InSAR fusion dominant weight, and use the surface continuous deformation field to constrain the small jitter of Beidou single point to ensure that the overall deformation field is smooth, real, and without jumps. (2) Moderate coherence transition area (bare slope, sparse shrub area): There is slight temporal incoherence in this area, and the InSAR is locally small-scale discrete, but the overall deformation trend is reliable; the processing flow is as follows: balanced weight fusion is adopted, InSAR is responsible for spatial trend constraints, and Beidou is responsible for high-precision point correction, with two-way complementary correction. (3) Low coherence and decoherence areas (dense forest, deep shadow, steep slope vegetation area): InSAR pixels in this area have unstable phase, serious time jump, and extremely low reliability of independent pixels. They belong to the areas where traditional InSAR monitoring fails. The processing procedure is as follows: significantly increase the InSAR observation noise value, reduce the InSAR weight, and completely rely on Beidou high-frequency and high-precision three-dimensional observation to dominate the fusion results, avoid decoherence noise pollution, and ensure that the high vegetation slope still has reliable monitoring accuracy.

[0016] II. The specific implementation of the adaptive weight adjustment over time is as follows: The time series weighting is automatically switched according to the monitoring and analysis objectives, distinguishing between short-term dynamic monitoring and medium- to long-term trend monitoring. (1) Short-term instantaneous deformation monitoring (hourly and daily analysis): used to capture short-term slope micro-deformation and instantaneous slippage caused by rainfall, heavy vehicle load, and temperature stress; the processing flow is as follows: the timeliness of Beidou high-frequency sampling is far better than that of satellite revisit, so the weight of Beidou observation is increased and the weight of InSAR long-period data is reduced to ensure that the algorithm is highly sensitive to sudden deformation and local anomalies. (2) Medium and long-term deformation trend analysis (monthly, quarterly and grade-level settlement assessment): used to judge the overall creep trend of slope, long-term settlement evolution and disease development law; the processing flow is as follows: InSAR has the advantages of full-domain spatial continuity, no omissions and long-term stability. Therefore, the InSAR surface field weight is increased, and the massive surface pixels are used to constrain the long-term drift error of Beidou single point to ensure the true continuity of global deformation trend.

[0017] III. The complete work process is as follows: In the S1 preprocessing stage, the InSAR temporal coherence coefficient is calculated pixel by pixel to generate a full-area coherence hierarchical mask, which divides the region into high, medium and low coherence areas. During the initialization of the S2 Kalman filter, the initial observation noise covariance is automatically configured according to the region to which the pixel belongs. During the S3 time-series iteration process, the weight ratio of the two types of data is adjusted twice according to the current analysis mode (short-term / long-term); S4 completes the state-optimal estimation update and outputs a three-dimensional deformation fusion result that balances spatial continuity and high point accuracy.

[0018] A high-speed slope monitoring system based on InSAR and BeiDou ground equipment, employing a high-speed slope monitoring method based on InSAR and BeiDou ground equipment, includes: The data acquisition module is used to acquire InSAR satellite image data and real-time monitoring data from BeiDou ground equipment deployed on the slope; The preprocessing module, which is connected to the data acquisition module, is used to perform spatiotemporal dynamic filtering and denoising, format standardization and spatiotemporal benchmark unification on the two types of raw data, and to construct a spatiotemporal model of surface deformation in combination with the topography of the study area and extract effective monitoring data of slopes. The multi-source data analysis module, which is connected to the preprocessing module, is used to analyze InSAR surface deformation data and identify potential risk areas across the entire slope; analyze BeiDou point monitoring data to obtain refined physical parameters of high-risk areas and conduct a preliminary assessment of slope stability. The fusion calculation module is connected to the multi-source data analysis module. It is used to complete the system deviation correction of Beidou data to InSAR data through the fitting coefficient method, and to achieve the optimal fusion of the two types of data through the Kalman filter algorithm, and output high-precision slope deformation results. The early warning module, connected to the fusion calculation module, is used to construct a slope risk early warning model based on fused deformation data, perform threshold judgment and graded early warning output; at the same time, it verifies the accuracy and effectiveness of the monitoring method by comparing the cumulative settlement and settlement rate indicators with the measured true values; the Beidou ground equipment integrates a GNSS positioning unit, tilt sensor, crack gauge and surface strain sensor, which can simultaneously collect the three-dimensional displacement of the measuring points and the slope physical parameters.

[0019] Compared with existing technologies, the high-speed slope monitoring method and system based on InSAR and BeiDou ground equipment of the present invention has the following advantages: 1. Constructing a two-tiered monitoring system that combines surface and point monitoring: MT-InSAR technology enables comprehensive deformation screening of highway slopes, quickly identifying potential risk areas; BeiDou ground equipment provides precise, point-based monitoring of high-risk areas, forming a hierarchical prevention and control closed loop of InSAR comprehensive hazard scanning and BeiDou point-based hazard monitoring. Simultaneously, a closed loop of global screening and local verification is formed, balancing the needs of large-scale coverage and high-precision monitoring; it avoids missing large-scale potential defects while accurately capturing sudden landslide disasters, significantly improving the economy and reliability of intelligent monitoring of highway slopes.

[0020] 2. Multi-source data fusion improves monitoring accuracy: The fitting coefficient method is used to correct the systematic bias of InSAR data, and then the Kalman filter algorithm is used to achieve optimal fusion of InSAR and Beidou data. This effectively reduces atmospheric delay, satellite orbit error and decoherence interference, and achieves the complementary advantages of InSAR's high spatial resolution and elevation deformation accuracy with Beidou's high temporal resolution and planar positioning accuracy, which significantly improves the accuracy and reliability of slope deformation monitoring.

[0021] 3. After adaptive weight adjustment, the accuracy of the fixed weight model is avoided in vegetated slopes and shaded areas, and the monitoring error in incoherent areas is reduced. It can also fully retain the advantages of InSAR's global surface in stable areas, ensuring that the deformation field is spatially continuous, without faults or voids. In addition, it can ensure that short-term abrupt changes are not missed and long-term trends are not distorted, truly achieving the optimal division of labor between global trends from InSAR and local precision from BeiDou. It has environmental adaptability during the fusion process and can automatically adapt to any high-speed slope terrain.

[0022] 4. When risk warnings are issued, InSAR area monitoring is used to identify key areas in advance. After the key areas are identified, Beidou is deployed to continuously track the deformation development. The deformation rate is stable in the long term and the acceleration does not exceed the limit. Only a yellow patrol warning is maintained, which avoids meaningless emergency rescue and realizes hierarchical management of hidden dangers, which greatly reduces the operation and maintenance cost of slopes.

[0023] 5. Error closed-loop management: The two-way correction mechanism eliminates the systematic error caused by reference drift, not only correcting the InSAR atmospheric and orbital deviations, but also stabilizing the long-term BeiDou observation reference, further reducing the overall monitoring system error.

[0024] 6. Expanding Applicable Scenarios: Relying on BeiDou phase control points, the problem of InSAR unwrapping failure in large deformation sections of landslides is overcome, and the monitoring scenarios are expanded from roadbeds with small settlements to the monitoring of landslides on medium and large road cut slopes.

[0025] 7. The observation dimension has been upgraded from limited three-dimensional observation at a limited number of measuring points to a continuous three-dimensional deformation field across the entire area, which can separate vertical settlement and horizontal slippage and accurately identify slope deformation and failure modes; the spatiotemporal dynamic weights are adapted to different surface coherence conditions and monitoring durations, avoiding the accuracy degradation caused by the loss of coherence in vegetation areas due to fixed weights. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the high-speed slope monitoring method based on InSAR and BeiDou ground equipment according to the present invention.

[0027] Figure 2 This is a schematic diagram showing the distribution of InSAR permanent scatterer monitoring points in the highway interchange section of the present invention.

[0028] Figure 3 This is a schematic diagram showing the distribution of InSAR permanent scatterer monitoring points on the cut slopes of highway and rural road sections according to the present invention.

[0029] Figure 4 This is a schematic diagram illustrating the process of constructing a regionally stable reference set and bidirectional error correction according to the present invention.

[0030] Figure 5 This is a schematic diagram of the high-speed slope monitoring system based on InSAR and BeiDou ground equipment according to the present invention. Detailed Implementation

[0031] Example 1: like Figures 1 to 5 The high-speed slope monitoring method based on InSAR and BeiDou ground equipment shown includes the following steps: S1 Data Acquisition: Acquire MT-InSAR data for surface deformation monitoring, as well as real-time monitoring data collected by BeiDou ground equipment deployed on the slope; periodically acquire rising and falling orbit SAR satellite images covering the study area, and process them using MT-InSAR technology to obtain full-area surface deformation monitoring data; simultaneously deploy BeiDou ground monitoring equipment at key locations on the slope, the equipment integrating GNSS positioning units, tilt sensors, crack gauges and strain sensors, to collect real-time data such as three-dimensional displacement, tilt angle, crack width, and surface strain at the measuring points; S2 Data Preprocessing: Spatiotemporal dynamic filtering was used to denoise the two types of raw data, eliminating interference from non-surface deformation factors and unifying the data format and spatiotemporal reference. A spatiotemporal model of surface deformation was established by combining the topographic data of the monitoring area to extract effective monitoring data from the slope area. Spatiotemporal dynamic filtering was used to denoise the MT-InSAR data and BeiDou monitoring data respectively, eliminating interference from non-deformation factors such as atmospheric disturbance, thermal noise, and acquisition anomalies. The spatiotemporal coordinate system and data format of the two types of data were unified, and a spatiotemporal model of surface deformation was established by combining the digital elevation model (DEM) of the study area to extract effective monitoring data within the highway slope area. S3 Multidimensional Data Analysis: This system extracts area deformation information of the entire slope's overall and local displacement, deformation rate, and deformation pattern from MT-InSAR data to identify potential risk areas. It also acquires point-based physical parameters such as dip angle, crack width, and surface strain at monitoring points using real-time monitoring data to analyze the refined deformation characteristics of high-risk areas. Combining these two types of data, it conducts dynamic behavior and stability assessments of the slope. Based on MT-InSAR area data, it extracts the overall settlement distribution, local displacement, annual average deformation rate, and deformation pattern of the slope, identifying high-risk sections with abnormal deformation rates. Using BeiDou monitoring point data, it acquires real-time three-dimensional deformation, dip angle changes, crack propagation rate, and surface strain parameters at monitoring points within high-risk sections, analyzing local slippage trends and crack development status. Finally, combining these two types of data, it comprehensively assesses the slope's dynamic stability from both overall trend and local detail perspectives. S4 Error Correction and Data Fusion: Using the fitting coefficient method and based on high-precision BeiDou observations, the difference between the two types of data at the same location is fitted to correct the deviation of the MT-InSAR data. Then, a Kalman filter algorithm is used to fuse the corrected MT-InSAR data with real-time monitoring data, outputting high-precision deformation results for highway slopes. Specifically: First, monitoring points where BeiDou and InSAR overlap are selected, and the difference sequence between the two types of observations at the same location is calculated. The fitting coefficient method is used to fit the systematic variation law of the difference, correcting the deviation of the MT-InSAR monitoring values ​​across the entire area and reducing the systematic deviation caused by atmospheric delay and orbital errors. Based on this, a Kalman filter algorithm is used to optimally fuse the corrected InSAR data with the BeiDou data. S5 Risk Early Warning: Based on fused deformation data, a slope risk early warning model is constructed with cumulative settlement, average settlement rate, and settlement acceleration as core indicators. According to the relevant requirements of the "Technical Specifications for Highway Slope Monitoring," three levels of early warning thresholds (yellow, orange, and red) are set. The slope risk index is calculated periodically and the stability status is updated. When an indicator exceeds a threshold, the corresponding level of early warning information is automatically pushed to the maintenance department. To verify the effectiveness of the method, measured values ​​from 20 benchmark observation points in the study area are compared and verified: Cumulative Settlement Dimension: The settlement trend of the fused monitoring results is highly consistent with the measured values, with a coherent spatiotemporal distribution and overall deviation controlled within millimeters; Settlement Rate Dimension: The average settlement rate retrieved after fusion completely matches the measured value, and the monitoring accuracy is significantly better than that of a single InSAR or single BeiDou monitoring scheme. The risk warning system combines InSAR's large-area survey and long-term trend capture with BeiDou's strength in capturing sudden deformations at single points with high frequency to implement a two-level warning system. For the first-level hazard warning, InSAR monitoring is the primary method for medium- to long-term surveys. Utilizing the global deformation field of InSAR during each period of orbital ascent and descent, areas with abnormal deformation rates are automatically delineated, and a yellow hazard warning is issued. This guides maintenance personnel to densely deploy BeiDou monitoring points in areas of abnormal deformation, completing the transition from area-based scanning to point-based monitoring. For the second-level disaster warning, BeiDou takes the lead in short-term emergency monitoring. For the densely deployed BeiDou monitoring points, displacement acceleration and crack propagation rate are calculated in real time. Once these indicators exceed critical thresholds, orange and red emergency warnings are immediately triggered, initiating slope emergency repair procedures.

[0032] In step S4, the Kalman filter algorithm includes observation equations and state equations: Observation equation: ; Equations of state: ; In the formula, Download wave phase observations for Kalman filtering: The observation vector at time moment has a dimension of 5×1, and the elements are, in order, the ascending orbit InSAR LOS directional deformation, the descending orbit InSAR LOS directional deformation, the BeiDou eastward deformation, the BeiDou northward deformation, and the BeiDou vertical deformation, with the unit being mm; A matrix was designed for the observations: with a dimension of 5×3, establishing a linear mapping relationship between the three-dimensional real deformation state and the five types of observations, which was jointly determined by the satellite orbit parameters and the BeiDou observation characteristics. For the Kalman filter state vector: The system state vector at time t is 3×1 in dimension, corresponding to the true three-dimensional deformation of the measuring point in the east (E), north (N), and vertical (U) directions, in mm, and is the quantity to be estimated for filtering; Δ is the observation noise vector: it has a dimension of 5×1, follows a zero-mean Gaussian distribution, and includes inSAR decoherence noise and BeiDou observation noise. Its covariance matrix is ​​denoted as... The accuracy is determined by the nominal accuracy of the two types of equipment; The original state vector from the previous moment: The optimal estimation vector of the system state at time t, with dimensions 3×1 and unit mm; The state transition matrix has a dimension of 3×3 and represents the evolution of the deformation state between adjacent time points. This invention adopts a uniform deformation model and takes a 3rd order identity matrix, which means that it is assumed that the deformation rate is constant within adjacent monitoring periods. The noise distribution matrix is ​​3×3, which maps process noise to state vectors. Under the uniform velocity model, it is a 3rd order identity matrix. The system noise vector has a dimension of 3×1 and follows a zero-mean Gaussian distribution, simulating random deformation disturbances caused by rainfall and load fluctuations. Its covariance matrix is ​​denoted as... It is calculated using the noise variance formula.

[0033] Furthermore, in step S4, the deformation and variance of the slope at the initial monitoring time are set to be 0. The initial deformation rate is calculated by adjusting the real-time monitoring data from BeiDou observation and MT-InSAR data. The system noise variance expression is: ; In the formula, The noise variance between radar and satellite observations: corresponding covariance matrix The diagonal elements, in mm²; The time interval between adjacent monitoring moments: in days (d); To monitor the initial moment: the initial deformation rate of the slope obtained by adjusting the BeiDou and MT-InSAR data, in mm / d; The sum of the absolute values ​​of the initial deformation rate in the three components of the east, north, and vertical directions is used to calibrate the noise intensity benchmark. This formula is derived based on the random walk deformation model. The shorter the time interval, the greater the variance of the process noise, which conforms to the physical law that short-period deformation disturbances are more random. In step S4, the expression for the multi-directional deformation observation matrix after data fusion is: ;

[0034] In the formula, and These represent the deformations of the satellite in the line-of-sight (LOS) direction during its ascent and descent. , and These represent the deformation of the BeiDou equipment in the east-west, north-south, and vertical directions, respectively. , and The projection vectors of the ascending InSAR data in the east-west, north-south, and vertical directions; , and The projection vectors of the down-orbit InSAR data in the east-west, north-south, and vertical directions; , and These are the original coordinates of the BeiDou measurement points in the east-west, north-south, and vertical directions, respectively.

[0035] In step S5, the risk warning model sets three warning thresholds: yellow, orange, and red. It periodically calculates and updates the slope risk index. When the monitoring index exceeds the corresponding threshold, it automatically pushes graded warning information to the maintenance department.

[0036] Because BeiDou single-point observations cause unidirectional correction of InSAR atmospheric delay and orbital system deviation; under long-term monitoring conditions, BeiDou reference stations are affected by slow ground settlement and receiver zero-point drift, introducing systematic offsets into their own observation sequences, directly leading to InSAR correction benchmark distortion; therefore, step S2 also includes constructing a regional stable benchmark set, specifically as follows: forming a regional stable benchmark set by screening InSAR permanent scatterer points with annual deformation rates below a threshold; before the error correction, a bidirectional error correction step is also performed: First, using the long-term deformation mean of the regional stable reference set, the foundation settlement and instrument drift error of the BeiDou reference station are calibrated in reverse. The drift correction calculation formula is as follows: ; In the formula, This refers to the drift correction amount for the BeiDou reference station, expressed in millimeters per year. The average deformation rate of the globally stable permanent scattering point (using the annual average deformation rate). The original deformation rate observed by the BeiDou reference station is the original annual deformation rate output by the BeiDou reference station before correction; the corrected BeiDou observation sequence (BeiDou observation time series after drift compensation) is as follows: Then, using the calibrated BeiDou data as a reference, the fitting coefficient method is used to positively correct the InSAR systematic error. The permanent scatterer points contained in the regional stable reference set form a large-scale permanent scatterer network of InSAR. The stability of the regional stable reference set enables bidirectional calibration: first, the BeiDou reference drift is corrected in reverse using the globally stable InSAR points, and then the InSAR observation error is positively corrected using the corrected BeiDou data, thus completely cutting off the transmission of systematic error between the two types of data.

[0037] During operation, in the monitoring area of ​​interchanges, 62 solid building foundations were selected as stable permanent scatterers, and the average annual deformation rate was statistically obtained to be 0.3 mm per year. The annual deformation rate of the Beidou reference station without correction was 1.1 mm per year. Substituting these values ​​into the calculation, the drift correction ΔB = 0.3 - 1.1 = -0.8 mm per year was calculated. This correction value was then superimposed on all Beidou time-series observation data to eliminate the system offset caused by the settlement of the reference station itself. After completing the Beidou reference calibration, the fitting coefficient method was used to correct the deviation of the full-domain InSAR line-of-sight deformation data. The calibration reference was more reliable, and the overall systematic error was reduced by more than half compared with the unidirectional calibration.

[0038] Because MT-InSAR processing is performed in large deformation sections of slopes, where interference fringes are densely stacked, phase unwrapping is prone to integer jumps. Errors accumulate along the unwrapping path, ultimately causing severe distortion in deformation calculations in large slip regions. Therefore, before the deviation correction, phase unwrapping with BeiDou-assisted InSAR is also included. This converts the high-precision three-dimensional displacement of BeiDou into radar line-of-sight deformation, which is then converted into absolute interferometric phase. This absolute interferometric phase is used as a hard control point to constrain the unwrapping path, lock the phase integer number, and curb the spread of unwrapping errors. The specific process is as follows: the three-dimensional deformation of the BeiDou measurement point is projected onto the satellite line-of-sight direction to obtain the line-of-sight deformation. The calculation formula is: In the formula, The eastward, northward, and vertical deformations measured by BeiDou. The projection components of the unit vector of the satellite's line-of-sight direction in the east, north, and vertical directions are calculated from the satellite's incident angle and azimuth angle; the line-of-sight deformation is a one-dimensional deformation of the line-of-sight direction obtained by converting BeiDou observation values, and after obtaining the line-of-sight deformation, it is further converted into the true interferometric phase: In the formula, λ is the true interferometric phase corresponding to the BeiDou point, and λ is the radar electromagnetic wave wavelength. This true interferometric phase is used as a known control point and substituted into the phase unwrapping to constrain the unwrapping path in the large deformation gradient region and correct the unwrapping cumulative error. During operation, in large deformation sections of ramp slopes, without control points, the cumulative deformation obtained from phase unwrapping inversion reached 21.3 mm, significantly deviating from the actual deformation. By projecting the three-dimensional displacement of the BeiDou measurement point at the slope toe into line-of-sight deformation, calculating the absolute phase value, and embedding this phase point into the minimum cost flow unwrapping model as a constraint node, the cumulative deformation after re-unwrapping was 17.9 mm, reducing the deviation from the BeiDou measured value to 0.3 mm. This completely solved the problem of unwrapping failure in dense stripe areas and broadened the applicability of InSAR in medium to large landslide sections.

[0039] Because the monitoring process can only calculate the three-dimensional displacement at a limited number of BeiDou measurement points, and can only output the deformation along a single line of sight over a large area, it is impossible to distinguish between settlement and horizontal slip, making it difficult to determine the slope slip pattern; therefore, in step S4, after the data fusion, a full-domain three-dimensional deformation inversion step is also included. Using discrete BeiDou three-dimensional observations (sparse BeiDou constraints) as spatial constraints, and relying on the continuity of surface deformation, Kriging interpolation is used to calculate the three-dimensional displacement components in the east, north, and vertical directions from the one-dimensional line-of-sight observations of each InSAR pixel across the entire slope area, upgrading the one-dimensional surface monitoring to a full-domain three-dimensional deformation field; specifically as follows: using the three-dimensional deformation results of the BeiDou measurement points as constraints, combined with the assumption of spatial continuity of the deformation field, the InSAR one-dimensional line-of-sight surface deformation data is inverted into a full-domain east, north, and vertical three-dimensional deformation field of the slope using the Kriging spatial interpolation algorithm; where the line-of-sight deformation and three-dimensional deformation components of any InSAR pixel satisfy: Under the three-dimensional displacement constraint of discrete BeiDou measurement points, the solution is obtained pixel by pixel for the entire image grid. Generate a continuous three-dimensional deformable mesh; During the work, for example, after three-dimensional inversion of the slope of a road cut in a village or town, two deformations can be clearly distinguished: the vertical settlement rate at the toe of the slope is 3.2 mm per month, while the horizontal slip along the slope is 1.5 mm per month. Relying solely on the original one-dimensional LOS observation, only the overall displacement can be observed, and the settlement and lateral slip cannot be separated. The three-dimensional deformation field can directly determine that the slope belongs to the push-slip failure, providing a key basis for stability analysis.

[0040] Because fixed-parameter Kalman filtering in InSAR and BeiDou fusion uses constant observation noise weights and process noise weights throughout, it cannot adapt to the complex and non-uniform surface monitoring environment of highway slopes, i.e., it cannot adapt to the spatial heterogeneity of slopes: InSAR is prone to decoherence in densely vegetated areas, and the observation noise increases sharply; InSAR observations are more stable in open, hardened road areas. At the same time, the two types of equipment have huge differences in temporal resolution: BeiDou can achieve hourly high-frequency sampling, while InSAR can only achieve long-term observations of more than ten days. To address this, an adaptive weighting mechanism is added to dynamically adjust the observation covariance of the two types of data in the filtering fusion according to space and time period, further improving the reliability of multi-source data fusion. For example, in step S4, the Kalman filtering fusion adopts a spatiotemporal adaptive weighting adjustment mechanism. During the Kalman filtering iteration process, the observation noise covariance matrix is ​​dynamically updated in real time, which is equivalent to dynamically allocating the fusion reliability weights of InSAR area data and BeiDou point data. That is, the better the coherence and the more the time series matches, the higher the reliability of the corresponding data source. Higher weights correspond to lower noise confidence; poorer coherence and mismatched time scales result in lower data source weights and higher noise confidence. The overall approach is divided into spatial adaptive adjustment and temporal adaptive adjustment, which work synchronously in the fusion solution process. Specifically, the InSAR observation noise covariance, BeiDou data weight, and InSAR data weight are adjusted according to the spatial dimension. For example, on slopes with high vegetation cover and poor coherence, the InSAR observation noise covariance is increased, raising the weight of BeiDou point observations. In areas with strong scattering and stability, such as hardened roadbeds and buildings, the InSAR noise weight is reduced to fully leverage the constraint effect of area monitoring data. The BeiDou data weight and InSAR data weight are adjusted according to the time dimension. Short-term monitoring (hourly time series analysis) mainly uses BeiDou high-frequency observations. Medium- and long-term trend analysis (monthly and quarterly deformation surveys) increases the InSAR area data weight. After adaptive weight optimization, the random error of the fusion results in incoherent areas is greatly reduced, and the InSAR area field constraint advantage is maintained in stable sections, balancing global continuity and point observation accuracy.

[0041] I. The specific implementation of adaptive weight adjustment in space is as follows: Based on the coherence coefficient of InSAR imagery, the entire slope area is divided into three monitoring zones, and fusion weights are dynamically assigned to each zone: (1) High coherence stable region (road surface, hard slope, building): The interference fringes in this region are stable and the degree of decoherence is extremely low. The InSAR surface deformation continuity is strong, the spatial consistency is good, and the long-term trend is reliable. The processing flow is as follows: reduce the InSAR observation noise assignment, increase the InSAR fusion dominant weight, and use the surface continuous deformation field to constrain the small jitter of Beidou single point to ensure that the overall deformation field is smooth, real, and without jumps. (2) Moderate coherence transition area (bare slope, sparse shrub area): There is slight temporal incoherence in this area, and the InSAR is locally small-scale discrete, but the overall deformation trend is reliable; the processing flow is as follows: balanced weight fusion is adopted, InSAR is responsible for spatial trend constraints, and Beidou is responsible for high-precision point correction, with two-way complementary correction. (3) Low coherence and decoherence areas (dense forest, deep shadow, steep slope vegetation area): InSAR pixels in this area have unstable phase, serious time jump, and extremely low reliability of independent pixels. They belong to the areas where traditional InSAR monitoring fails. The processing procedure is as follows: significantly increase the InSAR observation noise value, reduce the InSAR weight, and completely rely on Beidou high-frequency and high-precision three-dimensional observation to dominate the fusion results, avoid decoherence noise pollution, and ensure that the high vegetation slope still has reliable monitoring accuracy.

[0042] II. The specific implementation of the adaptive weight adjustment over time is as follows: The time series weighting is automatically switched according to the monitoring and analysis objectives, distinguishing between short-term dynamic monitoring and medium- to long-term trend monitoring. (1) Short-term instantaneous deformation monitoring (hourly and daily analysis): used to capture short-term slope micro-deformation and instantaneous slippage caused by rainfall, heavy vehicle load, and temperature stress; the processing flow is as follows: the timeliness of Beidou high-frequency sampling is far better than that of satellite revisit, so the weight of Beidou observation is increased and the weight of InSAR long-period data is reduced to ensure that the algorithm is highly sensitive to sudden deformation and local anomalies. (2) Medium and long-term deformation trend analysis (monthly, quarterly and grade-level settlement assessment): used to judge the overall creep trend of slope, long-term settlement evolution and disease development law; the processing flow is as follows: InSAR has the advantages of full-domain spatial continuity, no omissions and long-term stability. Therefore, the InSAR surface field weight is increased, and the massive surface pixels are used to constrain the long-term drift error of Beidou single point to ensure the true continuity of global deformation trend.

[0043] III. The complete work process is as follows: In the S1 preprocessing stage, the InSAR temporal coherence coefficient is calculated pixel by pixel to generate a full-area coherence hierarchical mask, which divides the region into high, medium and low coherence areas. During the initialization of the S2 Kalman filter, the initial observation noise covariance is automatically configured according to the region to which the pixel belongs. During the S3 time-series iteration process, the weight ratio of the two types of data is adjusted twice according to the current analysis mode (short-term / long-term); S4 completes the state-optimal estimation update and outputs a three-dimensional deformation fusion result that balances spatial continuity and high point accuracy.

[0044] A high-speed slope monitoring system based on InSAR and BeiDou ground equipment, employing a high-speed slope monitoring method based on InSAR and BeiDou ground equipment, includes: The data acquisition module is used to acquire InSAR satellite image data and real-time monitoring data from BeiDou ground equipment deployed on the slope; The preprocessing module, which is connected to the data acquisition module, is used to perform spatiotemporal dynamic filtering and denoising, format standardization and spatiotemporal benchmark unification on the two types of raw data, and to construct a spatiotemporal model of surface deformation in combination with the topography of the study area and extract effective monitoring data of slopes. The multi-source data analysis module, which is connected to the preprocessing module, is used to analyze InSAR surface deformation data and identify potential risk areas across the entire slope; analyze BeiDou point monitoring data to obtain refined physical parameters of high-risk areas and conduct a preliminary assessment of slope stability. The fusion calculation module is connected to the multi-source data analysis module. It is used to complete the system deviation correction of Beidou data to InSAR data through the fitting coefficient method, and to achieve the optimal fusion of the two types of data through the Kalman filter algorithm, and output high-precision slope deformation results. The early warning module, connected to the fusion calculation module, is used to construct a slope risk early warning model based on fused deformation data, perform threshold judgment and graded early warning output; at the same time, it verifies the accuracy and effectiveness of the monitoring method by comparing the cumulative settlement and settlement rate indicators with the measured true values; the Beidou ground equipment integrates a GNSS positioning unit, tilt sensor, crack gauge and surface strain sensor, which can simultaneously collect the three-dimensional displacement of the measuring points and the slope physical parameters.

[0045] Example 2: like Figure 2 As shown in the figure, the high-speed slope monitoring method based on InSAR and BeiDou ground equipment in this embodiment is applied to the InSAR and BeiDou fusion monitoring of highway interchange sections; as Figure 2 As shown, the monitoring object in this embodiment is a section of a highway interchange, with geographical longitudes of 118°53'10″E-118°53'30″E and latitudes of approximately 36°22'59″N-36°23'07″N. The area includes the two-way mainline of the highway, the ring ramps, and surrounding industrial plants and farmland. The scattered points in the figure are the effective permanent scatterer (PS) monitoring points extracted by MT-InSAR, totaling 186. They are densely distributed along the mainline of the road, ramps, factory buildings, and other stable artificial features, while the points are sparse in the farmland area, which can realize the screening of the entire road surface deformation. I. The monitoring deployment process is as follows: First, area monitoring is carried out, using MT-InSAR technology to cover the entire interconnected area. 186 effective PS points constitute the area monitoring network, with a revisit cycle of 12 days and a monitoring cycle of 12 months. Fixed-point verification: BeiDou ground monitoring equipment is deployed at 3 stable locations, namely the main roadbed, the ramp roadbed, and the roof of a stable building in the factory area, as deviation correction reference points, with a sampling frequency of 1 time / hour. Verification measurement points: 20 PS points are randomly selected as independent verification measurement points, which do not participate in the deviation correction fitting and are used to verify the fusion accuracy. II. Data Acquisition and Preprocessing: Simultaneously acquire 12 phases of rising-orbit SAR satellite imagery and BeiDou synchronous monitoring data. After spatiotemporal dynamic filtering preprocessing: remove atmospheric phase and orbital error system interference from InSAR data, and remove cycle slips and coarse difference anomalies from BeiDou data; unify the two types of data to the CGCS2000 coordinate system and UTC time reference to obtain a standardized initial deformation dataset; as shown in Table 1. Table 1: Concurrent Vertical Cumulative Deformation Observations from 3 BeiDou Reference Points (Unit: mm):

[0046] III. Deviation Correction Using Fitting Coefficients: A univariate linear correction model is adopted. The correction coefficients are solved using the least squares method: 3.1 Calculate the mean: ; ; 3.2 Calculate the sum of the products of deviations from the mean: ; Calculate the sum of squared deviations from the mean: ; Solve for the coefficients: ; ; The calibration formula for this calibration area is obtained as follows: ; Taking a verification PS point at a ramp curve as an example, the original InSAR vertical deformation is 15.8 mm, and after correction: The deviation before correction was -3.5 mm, and the deviation after correction was reduced to 0.26 mm, significantly weakening the systematic deviation.

[0047] IV. Kalman Filter Data Fusion: The verification measurement point at the ramp curve is selected as the object, and the corrected ascending and descending orbit InSAR data are input and fused with BeiDou data for solution; Given conditions: time interval initial deformation rate components and ; Time-state estimation (East, North, Vertical, unit mm); Covariance matrix: ; Observation design matrix: ; Observation noise covariance: The state transition matrix is ​​the identity matrix. The fusion steps are as follows: Process noise variance: ; ;in, The variance of the single-valued system noise calculated earlier is approximately equal to... Among them, the three in the diagonal matrix To place the same noise value in three positions on the diagonal of the matrix, corresponding to the three deformation components in the east, north, and vertical directions respectively; State prediction: ; Covariance prediction: ; Calculate Kalman gain (The matrix inversion process is omitted), and the final gain matrix components are obtained; Observation residuals: Time observation vector Predicted observations residual ; Status Update: ; Final fusion results: Eastward ≈ 6.65mm, Northward ≈ 3.32mm, Vertical ≈ 17.62mm.

[0048] V. Accuracy Verification and Result Analysis: The accuracy was verified by calculating and analyzing each of the 20 verification measurement points. Cumulative settlement: The fusion results are completely consistent with the settlement trend of Beidou measured values. The main line has uniform settlement, while the settlement at the ramp curves is slightly larger. The average absolute error is 0.75 mm. Settlement rate: The average settlement rate of 20 measuring points deviates from the measured value by less than 0.1 mm / month, and the accuracy meets the requirements for monitoring settlement of highway subgrade.

[0049] Stability assessment: The interchange is generally stable, with an average settlement rate of 1.2 mm / month for the main line, posing no risk; however, the settlement rate at the bend of the roundabout reaches 2.8 mm / month, with a cumulative settlement of nearly 18 mm, triggering a yellow alert. It is recommended to strengthen the inspection and settlement tracking monitoring at the ramp connection points.

[0050] Example 3: like Figure 3As shown in the figure, the high-speed slope monitoring method based on InSAR and BeiDou ground equipment in this embodiment is applied to the integrated monitoring of slopes in rural sections of highways; as Figure 3 As shown, the monitoring object in this embodiment is the slope monitoring of villages and towns; I. Monitoring areas and monitoring deployment, such as Figure 3 As shown, the monitoring object in this embodiment is the road cutting slope of a highway in a rural section, with a geographical range of longitude 118°33'30″E-118°33'50″E and latitude approximately 36°32'40″N-36°32'47″N; the baseline elevation of the slope bottom is 729m, with a relative elevation difference of 24m. The southern side is adjacent to a rural residential area, and the northern side is a farmland slope. The scattered points in the figure are 124 effective MT-InSAR monitoring points, densely distributed along the main road and residential buildings, and also effectively covered in the roadside slope area, which can support the inversion of the entire slope deformation; the monitoring layout is as follows: Area monitoring: 124 valid InSAR points cover the entire slope section and surrounding buildings, with a revisit period of 12 days and a monitoring period of 12 months; Fixed-point verification: Three Beidou ground monitoring devices were deployed along the central axis of the slope, located at the top, middle and bottom of the slope, respectively, to simultaneously collect three-dimensional displacement, tilt angle and crack width parameters as the deviation correction benchmark; Verification measuring points: Seventeen PS points in the slope area were selected as verification measuring points and were not included in the fitting correction.

[0051] II. Data Acquisition and Preprocessing: 12 phases of ascending and descending orbit SAR imagery and BeiDou monitoring data were acquired simultaneously. After spatiotemporal dynamic filtering and noise reduction, and unification of coordinates and time references, a standardized initial deformation dataset was obtained, as shown in Table 2. Table 2: Concurrent Vertical Cumulative Deformation Observations from 3 BeiDou Reference Points (Unit: mm):

[0052] III. Deviation Correction Using Fitting Coefficients: Using a univariate linear correction model, the coefficients are solved by least squares: Mean: , The sum of the products of deviations from the mean is approximately 308.63; the sum of the squares of deviations from the mean is 290.54; the coefficients are: slope a ≈ 1.062, intercept b ≈ 1.678. Slope correction formula: ; Taking a typical measuring point at the toe of a slope as an example, the original InSAR vertical deformation was 36.2 mm, and after correction: The system deviation before correction was -3.8 mm, and after correction, the deviation was reduced to 0.22 mm. IV. Kalman filter data fusion: First, select the slope toe measuring point for fusion calculation: Given conditions: time interval initial deformation rate and ; Time-state estimation: Covariance The observation design matrix and observation noise covariance are the same as in Example 2.

[0053] Calculation steps: Process noise variance: ; State prediction: ; Covariance prediction: ; Observation residuals: Time observation vector The predicted observations correspond to: The residual is ; Final fusion results after state update: Eastward ≈ 12.01mm, Northward ≈ 6.10mm, Vertical ≈ 39.05mm.

[0054] V. Accuracy Verification and Slope Stability Analysis: Verification was performed on 17 slope verification measurement points. Cumulative settlement: The fusion results and the measured values ​​show a settlement pattern of large settlement at the toe and small settlement at the top of the slope, with good spatiotemporal consistency and an average absolute error of 0.78 mm. Settlement rate: The average settlement rate of 17 measuring points deviated from the measured value by less than 0.12 mm / month, and the accuracy met the requirements of the slope monitoring specification. Slope stability assessment: The slope is in a basically stable state overall, with a monthly settlement rate of less than 2 mm in 82% of the area, and no orange or higher risk. The cumulative settlement in the slope toe section (longitude 118°33'35″E~118°33'45″E) reached 35-40mm, with a monthly settlement rate close to 3.2mm, triggering a yellow alert. It is recommended to increase the frequency of inspections, focusing on monitoring the expansion of slope cracks and drainage at the slope toe. The cumulative settlement at the slope top and on the east and west sides of the slope is less than 15mm, with a settlement rate of less than 1mm / month, which is a low-risk section. Maintaining the monthly monitoring cycle is sufficient. The settlement of buildings in the residential area on the south side is uniform, with an average settlement of less than 5mm. There is no abnormal settlement, and the slope deformation has not had a significant impact on the surrounding buildings.

[0055] Example 4: This embodiment of the high-speed slope monitoring method based on InSAR and BeiDou ground equipment includes the following steps: S1 Multi-Source Data Acquisition: Simultaneously acquires ascending and descending SAR satellite imagery, and three-dimensional displacement and physical parameter data of BeiDou ground equipment; additionally extracts InSAR permanent scatterers (PS points) of stable ground features (such as rocks and solid buildings) in the study area as a reference for reverse calibration of BeiDou. S2 data preprocessing: Spatiotemporal dynamic filtering is used to denoise the two types of data and unify the spatiotemporal reference; at the same time, InSAR global stable PS points (annual deformation rate < 0.5 mm) are identified and screened to construct a regional stable reference set; S3 bidirectional error correction: (1) Reverse calibration: InSAR stable point calibration of BeiDou reference drift, solving the systematic deviation of BeiDou reference station affected by long-term foundation settlement and instrument drift, and using the long-term deformation consistency advantage of InSAR large-scale stable points for reverse calibration: In the formula: This represents the drift correction amount for the BeiDou reference station, expressed in mm / year. The average deformation rate of the globally stable PS points is expressed in mm / year. Original deformation rate observed by the BeiDou reference station, in mm / year; Corrected BeiDou observation sequence: ; (2) Forward correction: The InSAR systematic error is corrected by BeiDou data. The original fitting coefficient method is retained. The calibrated BeiDou data is used as the benchmark to fit the observation difference of the same point and correct the systematic deviation caused by InSAR atmospheric delay and orbit error to obtain the corrected InSAR LOS deformation sequence. S4 BeiDou-assisted InSAR phase unwrapping: Addressing the pain points of dense interferometric fringes and large errors in traditional phase unwrapping in large deformation gradient regions of slopes, this method utilizes high-precision deformation values ​​from BeiDou as unwrapping control points to constrain the unwrapping path and avoid error accumulation. The three-dimensional deformation of the BeiDou measurement points is projected onto the satellite's LOS direction to obtain the corresponding true LOS deformation. ; Convert to InSAR true phase value: In the formula: The true interferometric phase corresponding to the BeiDou point is λ, which is the radar wavelength (e.g., 5.6 cm for Sentinel-1 C-band). This phase is used as a known control point and substituted into the minimum cost flow phase unwrapping algorithm to constrain the unwrapping results across the entire domain, thereby improving the monitoring accuracy of InSAR large deformation areas from the source. S5 Sparse BeiDou Constraints for Global 3D Deformation Inversion: Utilizing sparse BeiDou 3D observation constraints and combining the assumption of spatial continuity of the deformation field, the global 3D deformation field is inverted. For any InSAR pixel, its LOS-direction deformation satisfies: The three-dimensional deformation of the surrounding BeiDou measurement points ( Using ) as constraints, the Kriging space interpolation method is used to solve the three-dimensional deformation components of each pixel in the entire domain, thereby upgrading the InSAR planar one-dimensional data into planar three-dimensional data and giving full play to the advantages of InSAR planar coverage; S6 Spatiotemporal Adaptive Kalman Filter Fusion: Based on the original Kalman filter model, a dynamic weight adjustment mechanism is added: Spatially: In densely vegetated areas with severe decoherence, the InSAR observation noise covariance is increased to enhance the weight of BeiDou data; in open and stable areas, the InSAR weight is enhanced; Temporally: Short-cycle monitoring (hourly level) primarily uses BeiDou high-frequency data, while long-cycle trend analysis (monthly / quarterly level) enhances the weight of InSAR area data; Kalman filter calculation, system noise variance calculation, and multi-source data fusion observation matrix calculation are then performed; Finally, a graded linkage risk warning is implemented, specifically as follows: A two-level system is constructed: a medium-to-long-term hidden danger warning and a short-term disaster warning; Level 1 warning (medium-to-long-term hidden danger warning, InSAR-led): Based on the monthly-quarterly scale area deformation trend of InSAR, areas with abnormal deformation rates are identified, and a yellow hidden danger warning is issued to guide maintenance departments to increase the density of BeiDou deployment and the frequency of inspections; Level 2 warning (short-term disaster warning, BeiDou-led): For BeiDou high-frequency data in high-risk areas, orange / red disaster warnings are triggered using deformation acceleration and crack propagation rate as thresholds, and an emergency response is initiated.

[0056] Example 5: This embodiment is a high-speed slope monitoring method based on InSAR and BeiDou ground equipment. It integrates monitoring of interchange sections based on embodiment 2. First, the monitoring area is set up: Figure 2 In the interchange section, a total of 186 valid InSAR PS points were extracted, and 3 Beidou reference devices were deployed; 20 verification points were selected, including 2 large deformation ramp connection sections. Secondly, bidirectional error correction calculations are performed, as follows: (1) InSAR reverse calibration of BeiDou drift: 62 stable PS points are selected across the entire domain, and the average deformation rate is calculated. =0.3 mm / year; Original deformation rate observed by Beidou reference station =1.1mm / year, then the drift correction amount is: =0.3-1.1=-0.8mm / year; The deformation rate of the Beidou reference station after correction is 0.3mm / year, eliminating the systematic error caused by the settlement of the reference station itself; (2) Beidou positive correction of InSAR deviation: Based on the calibrated 3 Beidou reference points, the correction formula is obtained by least squares fitting: The system deviation was reduced from -3.5mm to 0.26mm by calibrating each of the 186 PS points individually. Next, BeiDou-assisted phase unwrapping: For the large deformation section of the ramp connection, the deformation obtained by traditional unwrapping is 21.3mm. After adding BeiDou control points to assist unwrapping, the deformation is 17.9mm. The deviation from the BeiDou measured value of 18.2mm is reduced from 3.1mm to 0.3mm, and the unwrapping accuracy is improved by more than 90%, which solves the problem of unwrapping failure caused by dense large deformation stripes. Next, the full-domain 3D inversion results: With 3 BeiDou points as constraints, the 3D deformation fields in the east, north, and vertical directions of the entire interconnected area were obtained through inversion: The vertical settlement in the main line area is uniform, averaging 1.2 mm / month, and the east-west horizontal displacement is <0.3 mm / month; the vertical settlement at the bend of the ring ramp is 2.8 mm / month, and the southward horizontal displacement is 0.9 mm / month, showing a lateral slip trend. The original 1D LOS data could not identify this horizontal deformation, but it can be accurately captured after 3D inversion; Finally, a tiered early warning system was implemented. InSAR surface screening triggered a Level 1 yellow hazard warning: the settlement rate at the ramp curve exceeded the standard, and it was recommended to increase patrols. BeiDou high-frequency monitoring showed that the deformation acceleration at this location was stable below 0.02 mm / h², and a Level 2 disaster warning was not triggered. The overall situation was under control.

[0057] Example 6: This embodiment is based on the high-speed slope monitoring method using InSAR and BeiDou ground equipment, and integrates the monitoring of slopes in village and town sections based on Embodiment 3; First, the monitoring area is set up: Figure 3 A total of 124 valid InSAR PS points were extracted from the slopes of the village and town road cuts. Three Beidou monitoring devices were deployed at the top, middle and bottom of the slope, and 17 slope verification points were selected. Secondly, two-way error correction calculation: (1) InSAR reverse calibration of Beidou drift: 35 stable PS points around the slope were selected, with an average deformation rate of 0.2 mm / year; the original deformation rate of the Beidou station at the top of the slope was 0.9 mm / year, and the correction amount was 0.9 mm / year. =-0.7mm / year, eliminating the settlement deviation of the base station itself; (2) Beidou positive correction of InSAR deviation: the slope area correction formula is obtained by fitting: After correction of the slope toe measuring point, the deviation was reduced from -3.8mm to 0.22mm; Next, BeiDou-assisted large deformation unwrapping: In the large deformation gradient region at the toe of the slope, the traditional unwrapping result was 45.2 mm, and after adding BeiDou control points to assist unwrapping, the result was 39.1 mm, which is almost consistent with the BeiDou measured value of 39.05 mm, thus solving the problem of large InSAR deviation in the large deformation gradient region of Example 3. Next, the full-domain 3D inversion results: The inversion yielded a full-domain 3D deformation field of the slope, which identified the downward sliding trend along the slope: the vertical settlement at the toe of the slope was 3.2 mm / month, and the horizontal displacement along the slope was 1.5 mm / month, indicating an overall risk of sliding; the vertical settlement at the top of the slope was 1.0 mm / month, and the horizontal displacement was <0.4 mm / month, indicating a stable state; the original 1D LOS data could only reflect the overall deformation along the line of sight and could not distinguish between vertical settlement and along-slope sliding. After 3D inversion, the slope deformation mode can be directly determined, providing a basis for stability assessment; Finally, the tiered early warning system is as follows: InSAR surface screening triggered a Level 1 Yellow Hazard Warning: the settlement and slippage rate in the slope toe section exceeded the standard, and it is recommended to increase the inspection frequency to once every half month; Beidou real-time monitoring shows that the crack propagation rate at the slope toe is stable within 0.1 mm / d, and the deformation acceleration has not exceeded the limit. The Level 2 Disaster Warning has not yet been triggered, but continuous monitoring is required.

[0058] The above embodiments are merely preferred embodiments of the present invention. Therefore, all equivalent changes or modifications made to the structure, features and principles described in the claims of the present invention are included within the scope of the present invention.

Claims

1. A high-speed slope monitoring method based on InSAR and Beidou ground equipment, characterized by: Includes the following steps: S1 Data Acquisition: Acquire MT-InSAR data for surface deformation monitoring, and real-time monitoring data collected by BeiDou ground equipment deployed on the slope. S2 data preprocessing: Spatiotemporal dynamic filtering method is used to denoise the two types of raw data, remove interference from non-surface deformation factors, and unify the data format and spatiotemporal reference; a spatiotemporal model of surface deformation is established by combining the topographic data of the monitoring area to extract effective monitoring data of the slope area. S3 Multidimensional Data Analysis: Extracts surface deformation information of overall and local displacement, deformation rate, and deformation mode of the entire slope through MT-InSAR data to identify potential risk areas; By acquiring point-based physical parameters such as dip angle, crack width, and surface strain at the measuring points through real-time monitoring data, we can analyze the refined deformation characteristics of high-risk areas and conduct dynamic behavior and stability assessments of slopes by combining the two types of data. S4 Error Correction and Data Fusion: Using the fitting coefficient method, with the BeiDou high-precision observation value as the benchmark, the observation difference between the two types of data at the same point is fitted to correct the deviation of the MT-InSAR data; then, the Kalman filter algorithm is used to fuse the corrected MT-InSAR data with the real-time monitoring data to output high-precision deformation results of highway slopes. S5 Risk Warning: Based on the fused deformation data, a risk warning model for slopes is constructed using cumulative settlement, average settlement rate, and settlement acceleration as indicators. Multi-level warning thresholds are set to monitor slope stability changes in real time and output warning information.

2. The InSAR and Beidou ground equipment based high-speed slope monitoring method according to claim 1, characterized in that: In step S4, the Kalman filter algorithm includes observation equations and state equations: Observation equation: ; Equations of state: ; In the formula, Download wave phase observations for Kalman filtering; Design a matrix for the observations; Δ is the Kalman filter state vector; Δ is the observation noise vector. This is the state transition matrix; This is the original state vector from the previous moment; This is the noise distribution matrix; This is the system noise vector.

3. The high-speed slope monitoring method based on InSAR and BeiDou ground equipment according to claim 2, characterized in that: In step S4, the deformation and variance of the slope at the initial monitoring time are set to 0. The initial deformation rate is calculated by adjusting the real-time monitoring data from BeiDou observation and MT-InSAR data. The system noise variance expression is: ； In the formula, The noise variance between radar and satellite observations; This represents the time interval between adjacent monitoring moments.

4. The high-speed slope monitoring method based on InSAR and BeiDou ground equipment according to claim 3, characterized in that: In step S4, the expression for the multi-directional deformation observation matrix after data fusion is: ； In the formula, and These represent the deformations of the satellite in the line-of-sight (LOS) direction during its ascent and descent. , and These represent the deformation of the BeiDou equipment in the east-west, north-south, and vertical directions, respectively. , and The projection vectors of the ascending InSAR data in the east-west, north-south, and vertical directions; , and The projection vectors of the down-orbit InSAR data in the east-west, north-south, and vertical directions; , and These are the original coordinates of the BeiDou measurement points in the east-west, north-south, and vertical directions, respectively.

5. The high-speed slope monitoring method based on InSAR and BeiDou ground equipment according to claim 4, characterized in that: In step S5, the risk warning model sets three warning thresholds: yellow, orange, and red. It periodically calculates and updates the slope risk index. When the monitoring index exceeds the corresponding threshold, it automatically pushes graded warning information to the maintenance department.

6. The high-speed slope monitoring method based on InSAR and BeiDou ground equipment according to claim 5, characterized in that: Step S2 also includes constructing a regional stable reference set, specifically as follows: A regional stable reference set is formed by selecting InSAR permanent scatterer points with annual deformation rates below a threshold; before the error correction, a two-way error correction step is also performed: First, using the long-term deformation mean of the regional stable reference set, the foundation settlement and instrument drift error of the BeiDou reference station are calibrated in reverse. The drift correction calculation formula is as follows: ; In the formula, This is the drift correction amount for the BeiDou reference station. The average deformation rate of a globally stable permanent scattering point. The original deformation rate observed by the BeiDou reference station; The corrected BeiDou observation sequence is as follows: Then, using the calibrated BeiDou data as a benchmark, the InSAR system error is positively corrected using the fitting coefficient method.

7. The high-speed slope monitoring method based on InSAR and BeiDou ground equipment according to claim 6, characterized in that: Before the deviation correction, phase unwrapping of BeiDou-assisted InSAR is also included, the specific process of which is as follows: the three-dimensional deformation of the BeiDou measurement point is projected onto the satellite line-of-sight direction to obtain the line-of-sight deformation. The calculation formula is: ; In the formula, The eastward, northward, and vertical deformations measured by BeiDou. The projection components of the satellite line-of-sight unit vector in the three geographic directions; After obtaining the line-of-sight deformation, it is then converted into the true interference phase: ; In the formula, Let λ be the true interferometric phase corresponding to the BeiDou point, and λ be the radar wavelength. Substitute this true interferometric phase as a known control point into the phase unwrapping to constrain the unwrapping path in the large deformation gradient region and correct the unwrapping cumulative error.

8. The high-speed slope monitoring method based on InSAR and BeiDou ground equipment according to claim 7, characterized in that: In step S4, after data fusion, a full-domain three-dimensional deformation inversion step is also included: using the three-dimensional deformation results of BeiDou measurement points as constraints, and combining the assumption of spatial continuity of the deformation field, the InSAR one-dimensional line-of-sight surface deformation data is inverted into a full-domain eastward, northward, and vertical three-dimensional deformation field of the slope using the Kriging spatial interpolation algorithm; wherein the line-of-sight deformation and three-dimensional deformation components of any InSAR pixel satisfy: .

9. The high-speed slope monitoring method based on InSAR and BeiDou ground equipment according to claim 8, characterized in that: In step S4, the Kalman filter fusion adopts a spatiotemporal adaptive weight adjustment mechanism, specifically: adjusting the InSAR observation noise covariance, BeiDou data weight, and InSAR data weight according to the spatial dimension; and adjusting the BeiDou data weight and InSAR data weight according to the temporal dimension.

10. A high-speed slope monitoring system based on InSAR and BeiDou ground equipment, employing the high-speed slope monitoring method based on InSAR and BeiDou ground equipment as described in any one of claims 1 to 9, characterized in that: include: The data acquisition module is used to acquire InSAR satellite image data and real-time monitoring data from BeiDou ground equipment deployed on the slope; The preprocessing module, which is connected to the data acquisition module, is used to perform spatiotemporal dynamic filtering and denoising, format standardization and spatiotemporal benchmark unification on the two types of raw data, and to construct a spatiotemporal model of surface deformation in combination with the topography of the study area and extract effective monitoring data of slopes. A multi-source data analysis module, wherein the preprocessing module is connected to the multi-source data analysis module, is used to analyze InSAR surface deformation data and identify potential risk areas across the entire slope area; Analyze BeiDou point-based monitoring data to obtain refined physical parameters of high-risk areas and conduct preliminary assessments of slope stability; The fusion calculation module is connected to the multi-source data analysis module. It is used to complete the system deviation correction of Beidou data to InSAR data through the fitting coefficient method, and to achieve the optimal fusion of the two types of data through the Kalman filter algorithm, and output high-precision slope deformation results. The early warning module, which is connected to the fusion calculation module, is used to construct a slope risk early warning model based on the fused deformation data, and to perform threshold judgment and graded early warning output. Meanwhile, by comparing the cumulative settlement and settlement rate indicators with the measured values, the accuracy and effectiveness of the monitoring method are verified. The Beidou ground equipment integrates a GNSS positioning unit, tilt sensor, crack gauge and surface strain sensor, which can simultaneously collect the three-dimensional displacement of the measuring points and the physical parameters of the slope.

Images