Multi-frequency sar satellite reservoir dam three-dimensional deformation reconstruction method and electronic device

Abstract

The application belongs to the technical field of radar remote sensing monitoring and water conservancy project safety evaluation, and relates to a multi-frequency SAR satellite reservoir dam three-dimensional deformation reconstruction method and an electronic device, the method comprising: acquiring ascending and descending track high-resolution synthetic aperture radar image data of a target reservoir dam in the same period, and extracting bimodal features of the aperture radar image data; performing partitioning processing on a three-dimensional geometric model of the target reservoir dam, establishing a local differential coordinate system at a dam surface node, and using the local differential coordinate system to construct a conversion matrix from a geographic coordinate system to a dam local coordinate system; using the bimodal features as a joint observation vector to establish a three-dimensional deformation inversion model through the conversion matrix, and using the three-dimensional deformation inversion model to output a three-dimensional digital twin deformation map. The application overcomes the dimension loss of single-view SAR, and realizes three-dimensional high-precision monitoring of reservoir dam deformation in a whole period.

Description

Technical Field

[0001] This invention relates to the field of radar remote sensing monitoring and water conservancy project safety assessment technology, and in particular to a multi-frequency SAR satellite method for three-dimensional deformation reconstruction of reservoir dams and electronic equipment. Background Technology

[0002] During the impoundment and operation of large-scale water conservancy facilities such as arch dams and high earth-rock dams, their deformation has complex three-dimensional spatial characteristics, including the transverse dam axis direction (radial thrust), the longitudinal dam axis direction (tangential slip), and the vertical direction (vertical settlement).

[0003] Synthetic Aperture Radar Interferometry (InSAR) technology, with its advantages of all-weather operation, high precision, and high spatial resolution, has been gradually applied to the monitoring of water conservancy facilities. However, existing InSAR technology for detailed dam monitoring is limited by the geometry of "side-view imaging." Since SAR satellites are mostly in near-polar orbits (flying north-south), their side-view geometry results in extremely low sensitivity to deformation in the north-south direction (i.e., radar azimuth). If the dam experiences slippage or creep along the dam axis (usually including a significant north-south direction), the LOS deformation value obtained by traditional InSAR technology may be extremely small, creating a monitoring blind spot and easily leading to misjudgments in safety assessments.

[0004] Although high-precision (sub-meter level) SAR satellite imagery is widely available, current mainstream algorithms still rely solely on phase information and fail to effectively detect the intensity information hidden within satellite images. There is currently a lack of a method that can effectively fuse InSAR phase with azimuth information from pixel offset tracking and combine it with a complex geometric model of a dam for rigorous three-dimensional calculation. Summary of the Invention

[0005] The purpose of this invention is to provide a method and electronic equipment for reconstructing the three-dimensional deformation of reservoir dams using multi-frequency SAR satellites, aiming to solve the problems that existing InSAR technology cannot effectively monitor the azimuth deformation of dams and is difficult to obtain the three-dimensional deformation in the engineering coordinate system.

[0006] To achieve the above objectives, the present invention provides the following solution: A multi-frequency SAR satellite method for reconstructing the three-dimensional deformation of reservoir dams includes: Acquire high-resolution synthetic aperture radar image data of the target reservoir dam during the same cycle of ascending and descending orbits, and extract the dual-mode features of the aperture radar image data; The three-dimensional geometric model of the target reservoir dam is divided into different regions, and a local differential coordinate system is established at the nodes on the dam surface to construct a transformation matrix from the geographic coordinate system to the dam's local coordinate system. The dual-modal features are used as joint observation vectors to establish a three-dimensional deformation inversion model through the transformation matrix, which is used to output a three-dimensional digital twin deformation map.

[0007] Optionally, extracting the dual-mode features of the aperture radar image data includes: The aperture radar image data is sequentially subjected to denoising, interferometry, adaptive filtering, and phase unwrapping to extract the phase information of the ascending and descending orbit image data in the aperture radar image data, and to calculate the line-of-sight deformation field. The target reservoir dam is divided into high-density search windows. The intensity correlation coefficient between the main image and the auxiliary image in the aperture radar image data in the radar azimuth direction is calculated, and the pixel offset in the radar azimuth and range directions is calculated to compensate for the deficiency of insensitivity to north-south deformation in the line-of-sight deformation field.

[0008] Optionally, establishing a local differential coordinate system at the nodes on the dam surface includes: Extract the normal and surface curvature information of the target reservoir dam surface, and divide the target reservoir dam surface into different zones based on the normal and surface curvature information. Establish a local differential coordinate system at each node on the dam surface. The direction along the dam axis of the local differential coordinate system is along the tangent direction of the dam surface, the direction across the dam axis is perpendicular to the dam surface and points downstream, and the vertical direction is along the gravity vertical direction.

[0009] Optionally, constructing the transformation matrix includes: Obtain the satellite radar incident angle and heading angle, and project the radar line of sight and azimuth direction onto the geographic coordinate system based on the satellite radar incident angle and heading angle to generate the first transformation relationship; Obtain the tangent line on the dam surface, and project the geographic coordinate system to the local differential coordinate system based on the tangent line and the normal line on the dam surface to generate a second transformation relationship; Combine the first transformation relation and the second transformation relation to generate the transformation matrix.

[0010] Optionally, establishing the three-dimensional deformation inversion model includes: ; in, This is the observation vector, containing phase observations and intensity shift observations for both ascending and descending orbits. Three-dimensional deformation parameters in engineering coordinate system , These are the deformation parameters along the dam axis. These are the deformation parameters along the transverse dam axis. The vertical deformation parameter. For the transformation matrix, This is the coefficient matrix, used to describe the logical constraints of different observation types on the unknown parameters and the linear structure of the observation equations. This is a random error vector.

[0011] Optionally, the three-dimensional deformation inversion model is established using an adaptive weighted fusion strategy. That is, if the coherence of the current region is greater than the first target threshold, the phase observation value is given a high weight; if the deformation gradient of the current region is greater than the second target threshold, the phase weight is reduced, and the intensity shift observation value is used to dominate the solution.

[0012] Optionally, the method further includes: The three-dimensional deformation inversion model is introduced with the Laplace space smoothing constraint of deformation continuity to perform regularization processing on the three-dimensional digital twin deformation map and eliminate noise points; The three-dimensional digital twin deformation map includes: deformation cloud map along the transverse dam axis, deformation cloud map along the longitudinal dam axis, and deformation cloud map in the vertical direction.

[0013] To achieve the above objectives, the present invention also provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the various steps of the reconfiguration method.

[0014] The beneficial effects of this invention are as follows: Eliminating blind spots: This invention effectively monitors dam deformation in the longitudinal (tangential) direction, which cannot be acquired by traditional InSAR, by fusing azimuth data from offset tracking.

[0015] Complementary accuracy: This invention combines the millimeter-level accuracy of InSAR with the large range and anti-phase loss advantages of migration tracking, making it suitable for monitoring large deformations in water-filled or geologically unstable areas.

[0016] Application in surveying: This invention directly outputs the deformation along the dam axis and across the dam axis without the need for manual secondary calculation, thus achieving seamless integration of remote sensing data with water conservancy operations. Attached Figure Description

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

[0018] Figure 1 This is a flowchart of the multi-frequency SAR satellite reservoir dam three-dimensional deformation reconstruction method according to an embodiment of the present invention; Figure 2This is a schematic diagram of the geometric mapping between the dam's local coordinate system and the radar coordinate system according to an embodiment of the present invention; Figure 3 This is a logic block diagram of the adaptive weighted fusion inversion algorithm according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the three-dimensional deformation reconstruction effect of the arch dam according to an embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0021] like Figure 1 As shown in the figure, this embodiment discloses a three-dimensional deformation reconstruction method for a reservoir dam using multi-frequency SAR satellites, including the following steps: Step S1 (Acquisition): Acquire high-resolution synthetic aperture radar (SAR) image data of the target reservoir dam area during the same period, including both ascending and descending orbits.

[0022] Step S2 (Feature Extraction): Dual-modal feature description of the constructed data image: Using Interferometric Synthetic Aperture Radar (InSAR) technology, the phase information of the ascending and descending orbit images are extracted respectively, and the line of sight (LOS) deformation field is calculated; using satellite offset tracking technology, the intensity information of the ascending and descending orbit images are extracted respectively, and the pixel offsets in the radar azimuth and range directions are calculated.

[0023] Specifically, the process of extracting the radar azimuth offset using pixel offset tracking technology in step S2 includes: performing sub-pixel-level precise registration of the high-resolution SAR images of ascending and descending orbits; dividing the dam area into high-density search windows and using the normalized cross-correlation (NCC) algorithm to calculate the intensity correlation coefficient between the main image and the auxiliary image in the radar azimuth direction; and using differential oversampling or parabolic symmetry technology to form a sub-pixel-precision azimuth offset to compensate for the InSAR technology's insensitivity to north-south deformation.

[0024] Step S3 (Logical Constraints): Import the 3D geometric model (CAD) of the dam, perform zoning processing on the dam surface, establish local differential coordinate systems at the nodes of the dam surface, and construct a transformation matrix from the geographic coordinate system to the dam local coordinate system. The dam local coordinate system includes the longitudinal dam axis, the transverse dam axis, and the vertical direction.

[0025] Specifically, step S3, which involves constructing the transformation matrix, includes: defining the local coordinate system of the dam: the direction along the dam axis is the tangent direction of the dam surface, the direction across the dam axis is perpendicular to the dam surface and points downstream, and the vertical direction is along the gravity vertical direction; based on the incident angle and heading angle of the satellite radar, constructing the first transformation relationship for projecting the radar line of sight and azimuth direction onto the geographic coordinate system; based on the normal and tangent of the dam surface, projecting the geographic coordinate system onto the local coordinate system of the dam; and combining the first and second transformation relationships to generate the final geometric constraint matrix.

[0026] Step S4 (fusion solution): The line-of-sight deformation field and azimuth pixel offset obtained in step S2 are used as joint observation vectors. Combined with the transformation matrix constructed in step S3, a set of three-dimensional deformation inversion equations is established and solved.

[0027] Specifically, the three-dimensional deformation inversion equations in step S4 employ an adaptive weighted fusion strategy, which includes: constructing a hybrid observation model containing phase compensation values ​​and intensity compensation values; dynamically determining the weight matrix of phase compensation values ​​and intensity compensation values ​​based on the Helmholtz Incremental Estimation (HVCE) method or coherence index; and automatically increasing the weight contribution of the azimuth offset in regions of phase decoherence or excessive deformation compensation to constrain the convergence of the inversion results, such as... Figure 3 As shown.

[0028] Step S5 (Result Output): Output the deformation fields of the dam surface in the longitudinal (tangential), transverse (radial) and vertical directions.

[0029] Furthermore, the method also includes a step of constraining and optimizing the solution results using the physical geometric characteristics of the dam: for arch dam structures, a Laplace space smoothing constraint for deformation continuity is introduced into the inversion equation to regularize the solution results in order to exclude physical solution points.

[0030] In one embodiment, three-dimensional deformation monitoring during the impoundment period of an arch dam is taken as an example: Step S1: Data Acquisition and Registration. This embodiment uses a batch of SAR image data covering the target dam: one set is ascending orbit, and the other is descending orbit. The images are in spotlight mode with a spatial resolution of 1 meter. Image registration is performed first. Based on traditional coarse registration, this system uses fine registration based on normalized cross-correlation (NCC) in the dam area. By setting a search window in the target area, the registration accuracy is improved to the 0.001 pixel level. Simultaneously, the GACOS external atmospheric correction model is introduced to remove long-wave atmospheric delay, and phase elevation regression is performed using the stable bedrock at the dam abutment to suppress interference from nearby water vapor turbulence unique to the reservoir area.

[0031] Step S2: Dual-modal feature extraction. This step constructs a dual-channel data stream: Channel A (phase): The denoised image undergoes interferometric processing, adaptive filtering, and phase unwrapping to obtain the line-of-sight (LOS) deformation fields for ascending and descending orbits. This data has high precision (mm level) but mainly reflects one-dimensional projection.

[0032] Channel B (Intensity): Using pixel offset tracking technology, a high-density grid (e.g., a 32×32 pixel window) is set on the dam surface to calculate texture offsets at different time phases. Sub-pixel offsets in the radar azimuth direction are extracted using oversampling techniques. This data directly reflects the deformation in the satellite's flight direction (approximately north-south), filling the blind spot in Channel A.

[0033] Step S3: Geometric Transformation Module. To convert radar data into engineering language, this invention introduces a dam CAD model: extracting the dam surface normal and surface curvature information, and partitioning the dam surface into point clouds or grids. A local coordinate system is set at each discrete point: the direction along the dam axis (tangential) is along the tangent direction of the dam surface at that point; the direction across the dam axis (radial) is perpendicular to the dam surface at that point and points downstream; and the vertical direction is along the gravity vertical direction. A transformation matrix is ​​constructed. ,like Figure 2 As shown. The observed values ​​in the radar coordinate system... For the perspective to change shape, The azimuth offset is directly related to the unknowns in the engineering coordinate system. .

[0034] Step S4: Constraint Optimization and Solution - Establishing the Joint Observation Equations: (1); in, The observation vector contains the phase observations (line-of-sight deformation) and intensity shift observations (azimuth deformation) for both ascending and descending orbits. The three-dimensional deformation parameters in the engineering coordinate system to be determined ; The transformation matrix (or projection matrix) constructed in step S3 for projecting the radar placement geometry back from the engineering coordinate system; To design a matrix (or coefficient matrix) to describe the logical constraints of different observation types on unknown parameters and the linear structure of the observation equations, This is a random error vector.

[0035] To address the weighting issue of different types of observations (phase and intensity), this invention constructs an adaptive weighted fusion inversion algorithm. The specific technical means are as follows: First, the phase observations corresponding to the line-of-sight (LOS) deformation field and the intensity shift observations corresponding to the azimuth field are obtained separately; second, the Helmert variance component estimation (HVCE) method is introduced, and the observation area is classified and weighted based on coherence indices: 1) In areas with clear interference fringes and high coherence... In regions where the deformation gradient is too large and phase unwrapping is unreliable, the weight of phase observations is automatically reduced. In regions with severe phase decoherence, the weight of azimuth intensity migration observations is significantly increased, and the intensity migration is used to dominate the solution to prevent divergence in the inversion results. Subsequently, based on the dynamically determined weight matrix, an adaptive weighted fusion hybrid observation model is constructed to perform preliminary three-dimensional deformation calculation. Finally, a Laplace spatial smoothing operator is introduced into the inversion equation as a regularization constraint term to ensure the physical continuity of dam deformation in space, prevent abnormal abrupt changes, effectively eliminate noise points, and ultimately output a stable and reliable three-dimensional deformation field.

[0036] Step S5: Output and Display The system finally outputs a three-dimensional digital twin deformation map of the arch dam during the water impoundment process, such as... Figure 4 As shown: Transverse dam axis direction (displacement) deformation cloud diagram: to analyze the elastic deformation of the dam body under water pressure.

[0037] Deformation cloud map along the dam axis (tangential): reveals the compression trend of the dam abutment towards the center of the valley (which is difficult to detect with traditional InSAR).

[0038] Vertical deformation cloud map: shows the settlement distribution of the dam foundation.

[0039] The above results are presented through a 4D visualization platform, providing comprehensive quantitative data support for dam safety assessment.

[0040] One embodiment discloses an electronic device including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the various steps of the reconfiguration method.

[0041] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for reconstructing three-dimensional deformation of reservoir dams using multi-frequency SAR satellites, characterized in that, include: Acquire high-resolution synthetic aperture radar image data of the target reservoir dam during the same period of ascent and descent, and extract the dual-mode features of the aperture radar image data; Extracting the dual-mode features of the aperture radar image data includes: The aperture radar image data is sequentially subjected to denoising, interferometry, adaptive filtering, and phase unwrapping to extract the phase information of the ascending and descending orbit image data in the aperture radar image data, and to calculate the line-of-sight deformation field. The target reservoir dam is divided into high-density search windows. The intensity correlation coefficient between the main image and the auxiliary image in the aperture radar image data in the radar azimuth direction is calculated, and the pixel offset in the radar azimuth and range directions is calculated to make up for the deficiency of insensitivity to north-south deformation in the line-of-sight deformation field. The three-dimensional geometric model of the target reservoir dam is divided into different regions, and a local differential coordinate system is established at the nodes on the dam surface to construct a transformation matrix from the geographic coordinate system to the dam's local coordinate system. The dual-modal features are used as joint observation vectors to establish a three-dimensional deformation inversion model through the transformation matrix, which is used to output a three-dimensional digital twin deformation map.

2. The method for reconstructing three-dimensional deformation of a reservoir dam using multi-frequency SAR satellites according to claim 1, characterized in that, Establishing the local differential coordinate system at the nodes on the dam surface includes: Extract the normal and surface curvature information of the target reservoir dam surface, and divide the target reservoir dam surface into different zones based on the normal and surface curvature information. Establish a local differential coordinate system at each node on the dam surface. The direction along the dam axis of the local differential coordinate system is along the tangent direction of the dam surface, the direction across the dam axis is perpendicular to the dam surface and points downstream, and the vertical direction is along the gravity vertical direction.

3. The method for reconstructing three-dimensional deformation of a reservoir dam using multi-frequency SAR satellites according to claim 1, characterized in that, Constructing the transformation matrix includes: Obtain the satellite radar incident angle and heading angle, and project the radar line of sight and azimuth direction onto the geographic coordinate system based on the satellite radar incident angle and heading angle to generate the first transformation relationship; Obtain the tangent line on the dam surface, and project the geographic coordinate system onto the local differential coordinate system based on the tangent line and the normal line on the dam surface to generate a second transformation relationship; Combine the first transformation relation and the second transformation relation to generate the transformation matrix.

4. The method for reconstructing three-dimensional deformation of a reservoir dam using multi-frequency SAR satellites according to claim 1, characterized in that, Establishing the three-dimensional deformation inversion model includes: ； in, This is the observation vector, containing phase observations and intensity shift observations for both ascending and descending orbits. Three-dimensional deformation parameters in engineering coordinate system , These are the deformation parameters along the dam axis. These are the deformation parameters along the transverse dam axis. The vertical deformation parameter. For the transformation matrix, This is the coefficient matrix, used to describe the logical constraints of different observation types on the unknown parameters and the linear structure of the observation equations. This is a random error vector.

5. The method for reconstructing three-dimensional deformation of a reservoir dam using multi-frequency SAR satellites according to claim 4, characterized in that, The three-dimensional deformation inversion model is established using an adaptive weighted fusion strategy. If the coherence of the current region is greater than the first target threshold, the phase observation value is given a high weight; if the deformation gradient of the current region is greater than the second target threshold, the phase weight is reduced, and the intensity shift observation value is used to dominate the solution.

6. The method for reconstructing three-dimensional deformation of a reservoir dam using multi-frequency SAR satellites according to claim 1, characterized in that, The method also includes: The three-dimensional deformation inversion model is introduced with the Laplace space smoothing constraint of deformation continuity to perform regularization processing on the three-dimensional digital twin deformation map and eliminate noise points; The three-dimensional digital twin deformation map includes: deformation cloud map along the transverse dam axis, deformation cloud map along the longitudinal dam axis, and deformation cloud map in the vertical direction.

7. An electronic device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, when the processor executes the computer program, it implements each step of the reconstruction method according to any one of claims 1 to 6.

Images