Method for cooperatively monitoring slope deformation of hydropower station through SBAS-InSAR and TLS

Abstract

The invention relates to a method for cooperatively monitoring hydropower station slope deformation through SBAS-InSAR and TLS, and the method comprises the steps: extracting hydropower station slope deformation information through employing a short baseline set technology based on GB-InSAR time sequence radar data; intercepting two groups of point cloud data in the abnormal deformation area; sampling and denoising under the two groups of point clouds; point cloud registration: aligning point clouds in a three-dimensional space to effectively reduce pose errors so as to ensure the consistency of geometric relationships; carrying out point cloud difference and carrying out interpolation on a difference result; and comparing and analyzing the two deformation monitoring results of the landslide abnormal area. According to the embodiment of the invention, comprehensive monitoring with higher precision can be realized through time sequence analysis and multi-dimensional data analysis, the reliability and accuracy of slope disaster early warning are remarkably improved, and the complexity of earth surface movement and deformation modes is revealed.

Description

Technical Field

[0001] The present invention relates to the technical field of ground-based interferometric synthetic aperture radar slope deformation monitoring and ground-based laser scanning slope deformation monitoring, and in particular to a method for collaboratively monitoring the slope deformation of a hydropower station by using SBAS-InSAR and TLS. Background Art

[0002] There are many drawbacks to using a single method for deformation monitoring. First, the monitoring dimension is limited. For example, although ground-based interferometric synthetic aperture radar (GB-InSAR) can provide high-precision displacement monitoring, it can usually only be performed on a two-dimensional plane and cannot obtain three-dimensional deformation information. Laser scanning (TLS), which is good at three-dimensional modeling, lacks real-time performance and has difficulty capturing instantaneous changes. In addition, many high-precision monitoring technologies, such as TLS, rely on periodic data collection, making it difficult to achieve continuous and real-time deformation monitoring and unable to cope with rapidly developing geological disasters such as landslides.

[0003] A single technology may also have poor environmental adaptability. For example, optical and laser technologies are easily affected by bad weather, and although radar technology is not affected by climate, it may produce data distortion or omissions in complex terrain. At the same time, different technologies also have limitations in accuracy and resolution. A single type of data generated by a single technology is also difficult to fully reflect the complexity of deformation, limiting the possibility of multi-dimensional analysis.

[0004] Therefore, a single monitoring method has deficiencies in accuracy, real-time performance, environmental adaptability and data integrity. The coordinated monitoring of multiple technologies (such as the combination of GB-InSAR and TLS) can better make up for these shortcomings and improve the accuracy and applicability of deformation monitoring. Summary of the invention

[0005] The present disclosure intends to provide a method for collaboratively monitoring the slope deformation of a hydropower station by using SBAS-InSAR and TLS. Through time series analysis and multi-dimensional data analysis, more accurate comprehensive monitoring can be achieved, which significantly improves the reliability and accuracy of slope disaster warnings and reveals the complexity of surface movement and deformation patterns.

[0006] According to one of the solutions disclosed herein, a method for collaboratively monitoring slope deformation of a hydropower station using SBAS-InSAR and TLS is provided, comprising:

[0007] Based on GB-InSAR time series radar data, short baseline set technology is used to extract the slope deformation information of the hydropower station;

[0008] Intercept two sets of point cloud data in the abnormal deformation area;

[0009] Downsampling and denoising of two sets of point clouds;

[0010] Point cloud registration, including aligning point clouds in three-dimensional space, effectively reducing pose errors and ensuring the consistency of geometric relationships;

[0011] Point cloud difference and interpolation of difference results;

[0012] A comparative analysis was conducted on the two deformation monitoring results in the abnormal landslide area.

[0013] In some embodiments, based on GB-InSAR time series radar data, short baseline set technology is used to extract slope deformation information of a hydropower station, including:

[0014] Acquire multiple SAR images at different times, and generate multiple pairs of small baseline differential interferograms by selecting image pairs with smaller temporal and spatial baselines. i and t j At a certain point between , the interference phase change can be expressed as:

[0015]

[0016] where φ Geom represents the geometric phase difference, φ Atmo_j and Respectively represent the time t i and t j Atmospheric phase, φ Noise represents the noise phase, 2kπ is the fuzziness of the unwrapped phase, φ Defo Reflecting the surface at t i and t j The phase change of the real deformation between moments.

[0017] In some embodiments, the two sets of point cloud data in the abnormal deformation area are intercepted, including:

[0018] Based on the large-scale deformation monitoring results, the abnormal deformation area is selected to locally intercept the two groups of point clouds, and the point cloud data intercepted at different times are recorded as point cloud P and point cloud Q.

[0019] In some embodiments, wherein two sets of point clouds are downsampled and denoised, comprising:

[0020] The point cloud P is downsampled to match the density of the point cloud Q, and the nearest neighbor distance method based on the kd-tree is used to identify and remove outliers.

[0021] In some embodiments, point cloud registration further comprises:

[0022] The iterative closest point algorithm is used for registration. Through iterative optimization, the source point cloud is mapped to the coordinate system of the target point cloud to minimize the distance error between the two sets of point clouds.

[0023] The root mean square error is used as the key indicator to measure the square root of the Euclidean distance between the corresponding points of the source point cloud and the target point cloud.

[0024] In some embodiments, point cloud differentiation and interpolation of the difference results include:

[0025] The nearest neighbor algorithm is used to find the corresponding points between the two registered point clouds. In the registered point clouds, the nearest neighbor point refers to the closest point in the target point cloud to each point in the given source point cloud.

[0026] Calculate the displacement of each corresponding point in the X, Y, and Z directions;

[0027] The natural neighbor interpolation method is used for interpolation;

[0028] The displacement heat maps in three directions generated by the interpolated displacement are analyzed and compared with the ground truth images of the area.

[0029] In some embodiments, extracting slope deformation information of a hydropower station based on GB-InSAR time series radar data includes:

[0030] Using arc-scanning ground-based synthetic aperture radar technology, the antenna moves at a constant speed along an arc path to collect multi-perspective echo signals, and generates high-resolution two-dimensional or three-dimensional images through the synthetic effect of antenna movement and echo signals.

[0031] In some embodiments, the short baseline set technology is used to extract the slope deformation information of the hydropower station, including:

[0032] The short baseline set technique is used to unwrap the phase of the interferogram, and the least squares method and singular value decomposition method are used to solve the deformation time series of the image subset. Finally, the deformation rate map and cumulative deformation map of the study area are generated by combining atmospheric phase correction and geocoding.

[0033] In some embodiments, the iterative closest point algorithm includes:

[0034] Find a rotation matrix R and a translation vector t so that point p in the source point cloud i The nearest neighbor point q in the target point cloud i Align, minimize the following objective function:

[0035]

[0036] By repeatedly iteratively updating R and t, the algorithm can gradually minimize the distance error between the source point cloud and the target point cloud, and finally align the two in three-dimensional space.

[0037] In some embodiments, calculating the displacement of each corresponding point in the three directions of X, Y, and Z includes:

[0038] For points in the source point cloud and target point cloud The three-dimensional displacement can be calculated by the following formula:

[0039]

[0040] To describe the movement of corresponding points in the X, Y, and Z directions.

[0041] The method of SBAS-InSAR and TLS collaborative monitoring of hydropower station slope deformation in various embodiments of the present disclosure is at least based on GB-InSAR time series radar data, and uses short baseline set technology to extract hydropower station slope deformation information; intercepts two sets of point cloud data in the abnormal deformation area; downsamples and denoises the two sets of point clouds; and aligns the point clouds, including aligning the point clouds in three-dimensional space, effectively reducing the posture error. Compared with the use of a single deformation monitoring method, this method effectively responds to the monitoring challenges under complex terrain, and can achieve more accurate comprehensive monitoring through time series analysis and multi-dimensional data analysis, significantly improving the reliability and accuracy of slope disaster warning, and revealing the complexity of surface movement and deformation patterns. The method disclosed in this disclosure not only enhances the understanding of the mechanism of geological disasters, but also provides a more reliable scientific basis for regional geological stability assessment and future geological disaster risk prediction. The embodiments of the present disclosure show that in the hydropower station slope area with complex terrain, combined with a variety of technical means for joint monitoring, a more comprehensive and detailed deformation analysis can be provided, providing important technical support for risk management and disaster warning of hydropower station slopes.

[0042] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure, as claimed. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In the drawings, which are not necessarily drawn to scale, similar reference numerals in different views may represent similar components. Similar reference numerals with letter suffixes or similar reference numerals with different letter suffixes may represent different instances of similar components. The drawings generally illustrate various embodiments by way of example and not limitation, and together with the description and claims, serve to explain the disclosed embodiments.

[0044] Figure 1 A schematic diagram of an implementation process of a monitoring method according to an embodiment of the present disclosure is shown;

[0045] Figure 2 The cumulative radar deformation results and deformation rate distribution of the monitoring area of ​​an embodiment of the present disclosure are shown;

[0046] Figure 3 It shows an embodiment of the present disclosure that points are selected according to the daily cumulative deformation process to analyze and draw a time series curve;

[0047] Figure 4 A front view and a side view of two phases of point clouds according to an embodiment of the present disclosure are shown;

[0048] Figure 5 A comparison of an embodiment of the present disclosure is shown, showing that outliers and noise are effectively removed;

[0049] Figure 6 A three-dimensional image of a point cloud after registration according to an embodiment of the present disclosure is shown;

[0050] Figure 7 The displacement thermal diagrams in three directions of an embodiment of the present disclosure are shown. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution of the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings of the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. Based on the described embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.

[0052] Landslide disasters have a great impact on the world, highlighting the importance of deformation monitoring and disaster early warning. At present, commonly used monitoring methods include traditional geodetic techniques (such as levels and total stations), GPS differential technology, as well as laser scanning (TLS) technology and interferometric synthetic aperture radar (InSAR) technology. However, a single technology is difficult to fully reflect the real terrain changes and potential disaster risks in complex environments. Joint monitoring can improve monitoring accuracy and enhance the timeliness of disaster early warning. Therefore, the combined application of multiple technologies has gradually been adopted in engineering practice, especially the combination of TLS and InSAR has become a research hotspot. These two technologies have their own advantages, as shown in Table 1.

[0053] Table 1 Characteristics of GB-InSAR and TLS in deformation monitoring

[0054]

[0055] Although there have been some studies on the combined deformation monitoring of TLS and InSAR technology, there are still relatively few applications for hydropower station slopes. For this reason, this paper selects the highway and roadbed slopes next to a hydropower station as the research object. Deformation monitoring of highways and roadbed slopes next to hydropower stations is extremely important, which can effectively ensure traffic safety, prevent damage to hydropower station facilities, cope with complex geological and climatic conditions, ensure the normal operation of hydropower stations, and meet the needs of engineering maintenance and optimization. Regular monitoring not only helps to timely discover potential geological disaster risks and prevent problems such as landslides or foundation settlement, but also provides data support for the maintenance and future design of infrastructure, ensuring the long-term safety and stability of hydropower stations and their surrounding areas.

[0056] In combination with the above-mentioned background technology, the present disclosure records the corresponding solutions in the form of embodiments to solve the defects in the prior art, but it is not intended to limit the scope of protection of the patent right claimed by the present disclosure.

[0057] As one of the solutions, the embodiment of the present disclosure provides a method for collaboratively monitoring slope deformation of a hydropower station using SBAS-InSAR and TLS, including:

[0058] Based on GB-InSAR time series radar data, short baseline set technology is used to extract the slope deformation information of the hydropower station;

[0059] Intercept two sets of point cloud data in the abnormal deformation area;

[0060] Downsampling and denoising of two sets of point clouds;

[0061] Point cloud registration, including aligning point clouds in three-dimensional space, effectively reducing pose errors and ensuring the consistency of geometric relationships;

[0062] Point cloud difference and interpolation of difference results;

[0063] A comparative analysis was conducted on the two deformation monitoring results in the abnormal landslide area.

[0064] In view of the above content, for deformation monitoring of hydropower station slopes under complex geological and harsh environmental conditions, this paper proposes a research method for collaborative monitoring of the deformation of a hydropower station slope using SBAS-InSAR and TLS. Compared with the use of a single deformation monitoring method, this method effectively copes with the monitoring challenges under complex terrain. Through time series analysis and multi-dimensional data analysis, more accurate comprehensive monitoring can be achieved, which significantly improves the reliability and accuracy of slope disaster warnings and reveals the complexity of surface movement and deformation patterns. With its high resolution, TLS can accurately capture small or local deformation details that are often difficult to detect through InSAR. Considering the complex terrain of the hydropower station slope, landslides usually exhibit multi-dimensional dynamic characteristics, and relying solely on InSAR may not be enough to fully capture their deformation characteristics. Therefore, combining high-resolution three-dimensional point cloud deformation data can effectively make up for the shortcomings of InSAR in monitoring.

[0065] The steps of the embodiment of the present disclosure are as follows: Figure 1 A schematic diagram of the implementation process of a monitoring method according to an embodiment of the present disclosure is shown. By way of example, the following takes steps S1 to S6 as an example to further illustrate the method of collaboratively monitoring the slope deformation of a hydropower station by SBAS-InSAR and TLS according to the present disclosure.

[0066] In some specific embodiments, the present disclosure may be, based on GB-InSAR time series radar data, using short baseline set technology to extract slope deformation information of a hydropower station, including:

[0067] Acquire SAR images from multiple different periods, and generate multiple pairs of small baseline differential interferograms by selecting image pairs with smaller temporal and spatial baselines.

[0068] S1: Extract deformation information using GB-InSAR time series radar data;

[0069] This embodiment uses arc scanning ground-based synthetic aperture radar (GB-SAR) technology, which combines the advantages of traditional GB-SAR with the geometric characteristics of arc paths, and is particularly suitable for fine monitoring of complex structures and terrains. Compared with linear trajectory scanning, the arc path can provide more viewing angle information and generate finer image details. The radar system is installed on a fixed arc track of 1-3 meters in length. The antenna moves at a constant speed along the arc path to collect multi-view echo signals. Through the synthesis effect of antenna movement and echo signals, a high-resolution two-dimensional or three-dimensional image is finally generated.

[0070] Table 2 GB-InSAR system and laser scanner parameters

[0071]

[0072] The GB-InSAR system continuously monitored the monitoring area for 38 days, starting from XX / XX / XX to XX / XX / XX. The interferometric phase data was acquired by the Ku-band GB-InSAR system. At the same time, the TLS technology scanned the same monitoring area twice on XX / XX / XX and XX / XX / XX using a long-range laser scanner, acquiring point cloud data for surface deformation analysis. Table 2 lists the main technical parameters of the GB-InSAR and TLS systems in detail.

[0073] In view of the complexity of the terrain of the hydropower station, the present invention adopts the short baseline set (SBAS-InSAR) technology to obtain surface deformation information. SBAS-InSAR technology is a time-series InSAR analysis technology method. By selecting image pairs with short time baselines and spatial baselines, it effectively suppresses the errors caused by temporal and spatial decoherence and atmospheric delays, so that the deformation information of the observation area can be obtained more accurately. The process of SBAS-InSAR technology includes acquiring SAR images of multiple different periods and accurately registering these images. By selecting image pairs with smaller time and space baselines, multiple pairs of small baseline differential interferograms are generated. For any two time points t i and t j At a certain point between , the interference phase change can be expressed as:

[0074]

[0075] This formula shows that the interference phase is the result of the combined effects of multiple factors such as geometric path difference, surface deformation, atmospheric effects and noise. Geom represents the geometric phase difference, φ Atmo_j and Respectively represent the time t i and t j Atmospheric phase, φ Noise represents the noise phase, 2kπ is the fuzziness of the unwrapped phase, and φ Defo It reflects the surface i and t j Phase changes of real deformation between moments. In order to extract the real deformation information of the surface, SBAS-InSAR technology unwraps the phase of the interferogram, and uses the least squares method and singular value decomposition method to solve the deformation time series of the image subset, and finally combines atmospheric phase correction and geocoding to generate deformation rate maps and cumulative deformation maps of the study area, providing detailed surface deformation monitoring results.

[0076] In some specific embodiments, the present disclosure may include intercepting two sets of point cloud data in the abnormal deformation area, including:

[0077] Based on the large-scale deformation monitoring results, the abnormal deformation area is selected to locally intercept the two groups of point clouds, and the point cloud data intercepted at different times are recorded as point cloud P and point cloud Q.

[0078] The embodiment of the present disclosure may further include step S2 based on the above step S1.

[0079] Step S2: intercept two sets of point cloud data in the abnormal deformation area

[0080] Based on the large-scale deformation monitoring results of SBAS-InSAR in step S1, the abnormal deformation area is selected to locally intercept the two groups of point clouds, and the point cloud data intercepted on XX / XX and XX / XX are recorded as point cloud P and point cloud Q.

[0081] The image data of the study area were processed by time-series small baseline integrated interferometry (SBAS). Figure 2 The cumulative radar deformation results and deformation rate distribution of the monitored area from X:00:00 on X ...

[0082] These images clearly reflect the spatial distribution of surface deformation displacement and deformation rate, where the deformation ranges from -100mm to +100mm and the deformation rate ranges from -5mm to 5mm. The blue area represents surface subsidence (negative deformation), while the red area represents surface uplift (positive deformation). In several significant deformation areas near the highway, the display patterns of the cumulative deformation distribution map and the deformation rate distribution map are basically the same. Figure 2 In the red frame 1 area, the dark red area may reflect the surface uplift caused by the landslide accumulation, showing the accumulation effect of the landslide material. The orange-yellow area in the red frame 2 indicates that the deformation of the area is intensifying, suggesting potential geological instability. It is recommended to strengthen monitoring to prevent possible risks.

[0083] Select two feature points in the red frame area in the figure and draw a time series curve to analyze the trend of surface deformation. Figure 3 As shown, point 1 shows significant surface uplift, with a cumulative deformation of about 120 mm and a large rate fluctuation, up to 10 mm / day, indicating that the area is greatly affected by geological activities or human activities. The cumulative deformation of point 2 is about 20 mm, and the rate fluctuation is between -5 mm / day and 5 mm / day, showing a relatively stable geological structure. After analyzing the large-scale deformation monitoring results of SBAS-InSAR, the triangular abnormal deformation area in the red frame is selected for TLS local refinement analysis to further analyze the detailed information of the deformation area.

[0084] The analysis results show that the number and density of the intercepted point cloud P are four times that of the point cloud Q. This resolution difference may introduce errors in the registration and comparison analysis. In addition, there is a lot of noise in the point cloud data, which may further affect the registration accuracy.

[0085] In this embodiment, the three-dimensional laser scanner cannot directly measure the coordinates of the object being measured, but can only obtain the round-trip propagation time Δt of the laser beam from emission to reception, the vertical angle θ provided by the angle measurement system, and the horizontal angle By calculating the distance between the device and the target object, the three-dimensional coordinates (x, y, z) of the target point can be determined. The specific formula is as follows:

[0086]

[0087] Where C is the speed of light. The scanner uses a local coordinate system, with the origin O being the center of the laser emission point of the instrument. The X-axis and Y-axis are located in the horizontal scanning plane of the local coordinate system, and the Z-axis is located in the longitudinal scanning plane and is perpendicular to the XY axis, with the vertical upward direction being the positive direction.

[0088] Before calculating the point cloud deformation, the point cloud preprocessing step improves the quality and consistency of the data through methods such as denoising, registration and interpolation to ensure the accuracy and reliability of deformation analysis. The core of 3D laser point cloud change detection is to identify the difference in point clouds of the same target at different time points, that is, to calculate the Euclidean distance between them. In three-dimensional space, the KD tree nearest neighbor algorithm is used to find the nearest neighbor point P2 of point P1 in the target point cloud in the reference point cloud. Assuming that there are two points P1 (x1, y1, z1) and P2 (x2, y2, z2) located in the target point cloud and the reference point cloud respectively, the Euclidean distance between them is calculated as:

[0089]

[0090] In some specific embodiments, the present disclosure may be, downsampling and denoising of two sets of point clouds, including:

[0091] The point cloud P is downsampled to match the density of the point cloud Q, and the nearest neighbor distance method based on the kd-tree is used to identify and remove outliers.

[0092] The embodiment of the present disclosure may further include step S3 based on the above step S2.

[0093] Step S3: Downsampling and denoising of two sets of point clouds

[0094] To improve data quality, the present disclosure processes and optimizes the two groups of point clouds intercepted in step 2, including downsampling the point cloud P to match the density of the point cloud Q, and using the nearest neighbor distance method based on the kd tree to identify and remove outliers.

[0095] The present disclosure uses the nearest neighbor search to detect abnormal points (outliers) by finding the "nearest neighbor points" of a point in a point cloud. The kd tree algorithm accelerates the nearest neighbor search by recursively building a tree structure each time the space is divided.

[0096] Formula: For a point p, the distance d(p) found by the nearest neighbor search is expressed as:

[0097] d(p)=min{||pp i ||}

[0098] Among them, p i are other points in the point cloud, ||pp i || represents the Euclidean distance between two points.

[0099] In some specific embodiments, the present disclosure may be point cloud registration, further comprising:

[0100] The iterative closest point algorithm is used for registration. Through iterative optimization, the source point cloud is mapped to the coordinate system of the target point cloud to minimize the distance error between the two sets of point clouds.

[0101] The root mean square error is used as the key indicator to measure the square root of the Euclidean distance between the corresponding points of the source point cloud and the target point cloud.

[0102] The embodiment of the present disclosure may further include step S4 based on the above step S3.

[0103] Step S4: point cloud registration;

[0104] In order to further reduce the error during deformation differentiation, the point cloud is registered based on the downsampling and denoising in step S3. Point cloud registration can align the point cloud in three-dimensional space, effectively reducing the position error, thereby ensuring the consistency of the geometric relationship.

[0105] 1) Iterative Closest Point (ICP) algorithm

[0106] The present invention adopts the Iterative Closest Point (ICP) algorithm for registration, which maps the source point cloud to the coordinate system of the target point cloud through iterative optimization to minimize the distance error between the two sets of point clouds. The goal of the ICP algorithm is to find a rotation matrix R and a translation vector t so that the point p in the source point cloud is i The nearest neighbor point q in the target point cloud i Align, minimize the following objective function:

[0107]

[0108] By repeatedly iteratively updating R and t, the algorithm can gradually minimize the distance error between the source point cloud and the target point cloud, and finally align the two in three-dimensional space.

[0109] 2) Root Mean Square Error (RMSE)

[0110] In the registration process, the root mean square error (RMSE) is used as a key indicator to measure the square root of the Euclidean distance between the corresponding points of the source point cloud and the target point cloud. After registration, the root mean square error calculation formula is:

[0111]

[0112] Where: N is the number of corresponding point pairs in the source point cloud and the target point cloud. i ′ is the i-th point in the source point cloud after rotation and translation (registration). i is the corresponding i-th point in the target point cloud. The reduction of RMSE indicates that the alignment accuracy of the source point cloud and the target point cloud is improved and the error is reduced.

[0113] This example uses a laser scanner to scan the study area at two different time points. To better understand the distribution of point clouds in space, Figure 4 The front and side views of the two phases of point clouds are shown. The number of points collected in both phases is 6,000,000, but it is visible to the naked eye that the second phase of the point cloud has a wider scanning coverage and shows more terrain details. This reflects the uneven distribution of the first phase of the point cloud, which is dense in some areas and sparse in other areas. It may also contain more noise, which affects the performance of details.

[0114] In order to facilitate the analysis of deformation data, this embodiment carried out subsequent experiments in MATLAB. Based on the abnormal deformation area identified in the previous article, the two groups of point clouds were locally intercepted, and the interception range was [xMin, xMax] = [462973, 463125] and [yMin, yMax] = [3577429, 3577573]. The analysis results show that the number of points and density of point cloud 1 after interception are four times that of point cloud 2. This resolution difference may introduce errors in registration and comparative analysis. In addition, there is a lot of noise in the point cloud data, which may further affect the registration accuracy. In order to improve the data quality, the two groups of point clouds were processed and optimized, including downsampling point cloud 1 to match the density of point cloud 2, and using the nearest neighbor distance method based on the kd tree to identify and remove outliers. Figure 5 The comparison shows that outliers and noise are effectively removed.

[0115] Registration is a crucial step in calculating the deformation of point clouds at different time points. By aligning point clouds in three-dimensional space, the pose error can be effectively reduced, thereby ensuring the consistency of geometric relationships. This embodiment uses the iterative closest point (ICP) algorithm for registration, which maps the source point cloud to the coordinate system of the target point cloud through iterative optimization to minimize the distance error between the two sets of point clouds. Figure 6 The 3D image of the registered point cloud is shown. During the registration process, the root mean square error (RMSE) is used as a key indicator to measure the square root of the Euclidean distance between the corresponding points of the source point cloud and the target point cloud. After the ICP algorithm optimization, the RMSE after registration was reduced by 0.09015, further proving the effectiveness of the ICP algorithm in improving the registration accuracy.

[0116] In some specific embodiments, the present disclosure may be to perform point cloud differentiation and interpolation on the differential results, including:

[0117] The nearest neighbor algorithm is used to find the corresponding points between the two registered point clouds. In the registered point clouds, the nearest neighbor point refers to the closest point in the target point cloud to each point in the given source point cloud.

[0118] Calculate the displacement of each corresponding point in the X, Y, and Z directions;

[0119] The natural neighbor interpolation method is used for interpolation;

[0120] The displacement heat maps in three directions generated by the interpolated displacement are analyzed and compared with the ground truth images of the area.

[0121] The embodiment of the present disclosure may further include step S5 based on the above step S4.

[0122] Step S5: point cloud difference and interpolation of the difference result;

[0123] 1) Nearest neighbor algorithm to find corresponding points

[0124] The present invention uses the nearest neighbor algorithm to find the corresponding points between the two groups of point clouds after the registration in step 4. In the registered point cloud, the nearest neighbor point refers to the point in the target point cloud that is closest to each point in the given source point cloud. For a point in point cloud P, its nearest neighbor point q in point cloud Q is i It is expressed by the following formula:

[0125]

[0126] where ||p i -q j || represents point p in the source point cloud i and point q in the target point cloud j The Euclidean distance between .

[0127] 2) Calculate the three-dimensional displacement;

[0128] Furthermore, the displacement of each corresponding point in the X, Y, and Z directions is calculated. and target point cloud The three-dimensional displacement can be calculated by the following formula:

[0129]

[0130] These displacement values ​​describe the amount of movement of the corresponding points in the X, Y, and Z directions.

[0131] 3) Natural neighbor interpolation fits the complete displacement;

[0132] The present invention uses the Natural Neighbor Interpolation method for interpolation, which is an interpolation technique based on spatial relationships and is used to map irregular point cloud data to a regular two-dimensional grid. Interpolation is to estimate the value on the grid based on the neighborhood distribution of the input point. Natural neighbor interpolation is usually based on the Voronoi diagram. Given a point p i , which is at the target interpolation point p j The interpolation value f(p j ) is the weighted average of its neighborhood points:

[0133]

[0134] Among them, w i is the weight determined by the Voronoi cell, f(p j ) is point p j The original data value at point p, N is j The number of neighborhood points.

[0135] 4) Heat map generation;

[0136] The displacement heat maps in three directions generated by the interpolated displacement are analyzed and compared with the ground truth images of the area. The landslide direction in the abnormal area and more detailed three-dimensional landslide dynamic characteristics are determined through analysis.

[0137] After the registration is completed, the nearest neighbor algorithm is used to find the corresponding points between the two point clouds and calculate the displacement of each point in the three directions of X, Y, and Z. These displacements are mapped to a regular two-dimensional grid through the natural neighbor interpolation method to generate displacement heat maps in three directions, such as Figure 7 The deformation data after interpolation are more aligned in space, which is convenient for subsequent comparison and analysis. Part (d) is the ground truth image in the same area, and it can be clearly seen that a landslide occurred in the red frame area in the figure.

[0138] In the displacement thermodynamic diagrams in three directions, the color of the central triangle area changes dramatically, indicating that the area has undergone significant deformation. The specific analysis is as follows: The deformation in the X and Y directions can not only reflect the sliding amplitude of the landslide and its affected area, but also indicate the sliding trend of the material in the plane. In parts (a) and (b), the rapid transition of red and blue colors shows significant horizontal displacement, which is speculated to be consistent with the main sliding direction of the landslide. According to the color coding, blue indicates surface subsidence and red indicates surface uplift. In part (a), the displacement in the X direction is from the negative semi-axis to the positive semi-axis, that is, from west to east; in part (b), the displacement in the Y direction is from the negative semi-axis to the positive semi-axis, that is, from south to north. Therefore, it is speculated that the deformation direction is northeast. The direction of the real geographic coordinates in part (d) verifies the accuracy of this inference.

[0139] In addition, the Y-direction and Z-direction displacements in parts (b) and (c) are consistent, indicating that the vertical surface deformation is closely related to the horizontal displacement. Specifically, when the landslide body slides down in the Y direction, some soil and rock blocks accumulate at the front edge or below the landslide, causing the uplift of the area (upward in the Z direction) to be consistent with the displacement in the Y direction, indicating that the vertical movement and horizontal movement occur synchronously, reflecting the three-dimensional dynamic characteristics of the landslide.

[0140] Part (c) shows that the displacement of the landslide is about 0.2 to 0.3 meters. The Z-direction thermal map reveals important details of the vertical surface deformation. The deformation from blue to red (from top to bottom) in the area corresponds to the sinking and uplifting of the surface, reflecting the process of soil and rock blocks falling from high places and accumulating during the landslide, further confirming the landslide phenomenon in this area.

[0141] The embodiment of the present disclosure may further include step S6 based on the above step S5.

[0142] Step S6: Comparative analysis of the two deformation monitoring results of the landslide abnormal area;

[0143] The results show that, in landslide deformation monitoring, although there are differences in the measurement results of the two methods, this difference is mainly attributed to the differences in measurement principles, spatial resolution, data processing methods, observation environment and data acquisition time. However, the errors caused by these factors do not affect the consistency of the magnitude of the deformation values ​​of the two methods. This consistency further verifies the reliability of the two monitoring technologies and the mutual verification of the results.

[0144] The collaborative monitoring method based on SBAS-InSAR and TLS in various embodiments of the present disclosure is used to study the deformation of the slope of a hydropower station. This method combines the all-weather, real-time, sub-millimeter precision monitoring advantages of GB-SAR with the expertise of TLS in high-precision three-dimensional modeling deformation extraction, and can effectively cope with the monitoring challenges of complex terrain of hydropower stations. With its high resolution, TLS can accurately capture tiny or local deformation details, which are often difficult to detect through InSAR. Considering the complex terrain of the slope of the hydropower station, landslides usually exhibit multi-dimensional dynamic characteristics, and relying solely on InSAR may not be enough to fully capture their deformation characteristics. Therefore, combined with high-resolution three-dimensional point cloud deformation data, it can effectively make up for the shortcomings of InSAR in monitoring.

[0145] This comprehensive approach not only enhances the understanding of the mechanism of geological disasters, but also provides a more reliable scientific basis for regional geological stability assessment and future geological disaster risk prediction. This experiment shows that in the hydropower station slope area with complex terrain, combined with a variety of technical means for joint monitoring, can provide a more comprehensive and detailed deformation analysis, and provide important technical support for risk management and disaster warning of hydropower station slopes.

[0146] In some specific embodiments, the device for collaboratively monitoring slope deformation of a hydropower station by SBAS-InSAR and TLS disclosed in the present invention includes one or more processing modules configured to execute the specific implementation methods of steps S1 to S6 in the foregoing text.

[0147] Specifically, one of the inventive concepts of the present disclosure is the method of SBAS-InSAR and TLS collaborative monitoring of hydropower station slope deformation in various embodiments of the present disclosure, which is based on at least GB-InSAR time series radar data, and uses short baseline set technology to extract hydropower station slope deformation information; intercepts two sets of point cloud data in the abnormal deformation area; downsamples and denoises the two sets of point clouds; and aligns the point clouds, including aligning the point clouds in three-dimensional space, effectively reducing the position error. Compared with the use of a single deformation monitoring method, this method effectively responds to the monitoring challenges under complex terrain, and can achieve more accurate comprehensive monitoring through time series analysis and multi-dimensional data analysis, significantly improving the reliability and accuracy of slope disaster warning, and revealing the complexity of surface movement and deformation patterns. The method of the present disclosure not only enhances the understanding of the mechanism of geological disasters, but also provides a more reliable scientific basis for regional geological stability assessment and future geological disaster risk prediction. The embodiments of the present disclosure show that in the hydropower station slope area with complex terrain, combined with a variety of technical means for joint monitoring, it can provide a more comprehensive and detailed deformation analysis, and provide important technical support for risk management and disaster warning of hydropower station slopes. .

[0148] The present disclosure also provides a computer-readable storage medium having computer executable instructions stored thereon. When the computer executable instructions are executed by a processor, the method for collaboratively monitoring slope deformation of a hydropower station using SBAS-InSAR and TLS is mainly implemented, which at least includes:

[0149] Based on GB-InSAR time series radar data, short baseline set technology is used to extract the slope deformation information of the hydropower station;

[0150] Intercept two sets of point cloud data in the abnormal deformation area;

[0151] Downsampling and denoising of two sets of point clouds;

[0152] Point cloud registration, including aligning point clouds in three-dimensional space, effectively reducing pose errors and ensuring the consistency of geometric relationships;

[0153] Point cloud difference and interpolation of difference results;

[0154] A comparative analysis was conducted on the two deformation monitoring results in the abnormal landslide area.

[0155] The above embodiments are only exemplary embodiments of the present disclosure and are not intended to limit the present disclosure. The protection scope of the present disclosure is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present disclosure within the essence and protection scope of the present disclosure, and such modifications or equivalent substitutions shall also be deemed to fall within the protection scope of the present disclosure.

Claims

1. The method of SBAS-InSAR and TLS collaboratively monitoring the slope deformation of hydropower stations includes: Based on GB-InSAR time series radar data, short baseline set technology is used to extract the slope deformation information of the hydropower station; Intercept two sets of point cloud data in the abnormal deformation area; Downsampling and denoising of two sets of point clouds; Point cloud registration, including aligning point clouds in three-dimensional space, effectively reducing pose errors and ensuring the consistency of geometric relationships; Point cloud difference and interpolation of difference results; A comparative analysis was conducted on the two deformation monitoring results in the abnormal landslide area.

2. The method according to claim 1, wherein: Based on GB-InSAR time series radar data, short baseline set technology is used to extract the slope deformation information of the hydropower station, including: Acquire multiple SAR images at different times, and generate multiple pairs of small baseline differential interferograms by selecting image pairs with smaller temporal and spatial baselines. i and t j At a certain point between , the interference phase change can be expressed as: where φ Geom represents the geometric phase difference, φ Atmo_j and Respectively represent the time t i and t j Atmospheric phase, φ Noise represents the noise phase, 2kπ is the fuzziness of the unwrapped phase, φ Defo Reflecting the surface at t i and t j The phase change of the real deformation between moments.

3. The method according to claim 2, wherein: Intercept two sets of point cloud data in the abnormal deformation area, including: Based on the large-scale deformation monitoring results, the abnormal deformation area is selected to locally intercept the two groups of point clouds, and the point cloud data intercepted at different times are recorded as point cloud P and point cloud Q.

4. The method according to claim 3, wherein: Two sets of point cloud downsampling and denoising, including: The point cloud P is downsampled to match the density of the point cloud Q, and the nearest neighbor distance method based on the kd-tree is used to identify and remove outliers.

5. The method according to claim 4, wherein: Point cloud registration, further including: The iterative closest point algorithm is used for registration. Through iterative optimization, the source point cloud is mapped to the coordinate system of the target point cloud to minimize the distance error between the two sets of point clouds. The root mean square error is used as the key indicator to measure the square root of the Euclidean distance between the corresponding points of the source point cloud and the target point cloud.

6. The method according to claim 5, wherein: Point cloud difference and interpolation of difference results, including: The nearest neighbor algorithm is used to find the corresponding points between the two registered point clouds. In the registered point clouds, the nearest neighbor point refers to the closest point in the target point cloud to each point in the given source point cloud. Calculate the displacement of each corresponding point in the X, Y, and Z directions; The natural neighbor interpolation method is used for interpolation; The displacement heat maps in three directions generated by the interpolated displacement are analyzed and compared with the ground truth images of the area.

7. The method according to claim 1, wherein: Extract the slope deformation information of the hydropower station based on GB-InSAR time series radar data, including: Using arc-scanning ground-based synthetic aperture radar technology, the antenna moves at a constant speed along an arc path to collect multi-perspective echo signals, and generates high-resolution two-dimensional or three-dimensional images through the synthetic effect of antenna movement and echo signals.

8. The method according to claim 1, wherein: The short baseline set technology is used to extract the slope deformation information of the hydropower station, including: The short baseline set technique is used to unwrap the phase of the interferogram, and the least squares method and singular value decomposition method are used to solve the deformation time series of the image subset. Finally, the deformation rate map and cumulative deformation map of the study area are generated by combining atmospheric phase correction and geocoding.

9. The method according to claim 5, iterative closest point algorithm, comprising: Find a rotation matrix R and a translation vector t so that point p in the source point cloud i The nearest neighbor point q in the target point cloud i Align, minimize the following objective function: By repeatedly iteratively updating R and t, the algorithm can gradually minimize the distance error between the source point cloud and the target point cloud, and finally align the two in three-dimensional space.

10. The method according to claim 1, wherein the displacement of each corresponding point in the three directions of X, Y and Z is calculated, comprising: For points in the source point cloud and target point cloud The three-dimensional displacement can be calculated by the following formula: To describe the movement of corresponding points in the X, Y, and Z directions.

Images