Slope deformation sequential monitoring method, device and equipment based on high-speed rail-mounted synthetic aperture radar and storage medium
By using a non-contact monitoring method based on synthetic aperture radar, single-view complex images are generated and differential interferometry is performed, solving the problem of refined monitoring of high-speed railway slopes in all weather conditions, over a wide area, and at high frequency, and achieving high-precision deformation extraction and real-time early warning.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN UNIV
- Filing Date
- 2026-05-11
- Publication Date
- 2026-06-09
Smart Images
Figure CN122172192A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the technical field of slope deformation monitoring for high-speed railways, and more specifically, to a method, device, equipment, and storage medium for sequential monitoring of slope deformation based on synthetic aperture radar mounted on a high-speed railway. Background Technology
[0002] my country has a large-scale high-speed transportation system, but the slopes along high-speed railways are affected by a combination of factors such as geology, climate and train vibration, which can easily lead to gradual deformation and geological disasters such as landslides and rockfalls, seriously threatening the safety of railway operations. Therefore, long-term and efficient monitoring of slopes is of utmost importance.
[0003] Among related technologies, high-speed railway slope monitoring mainly relies on manual inspection, contact sensors, and spaceborne InSAR technology. However, these technologies have inherent limitations in terms of cost, coverage, and timeliness, making it difficult to meet the needs of high-speed railway slope monitoring in a refined manner that is available around the clock, over a wide area, and at high frequency. Summary of the Invention
[0004] In view of the above problems, this application proposes a method, device, equipment and storage medium for sequential monitoring of slope deformation based on synthetic aperture radar mounted on a high-speed railway, which can solve the above problems.
[0005] In a first aspect, embodiments of this application provide a method for sequential monitoring of slope deformation based on synthetic aperture radar mounted on a high-speed railway. The method includes: acquiring raw echo data using synthetic aperture radar and generating single-view complex images based on the raw echo data; performing image registration on the single-view complex images and differential interferometry processing based on the registered single-view complex images to construct a differential interferometric dataset; constructing a coherence matrix based on the differential interferometric dataset, identifying and correcting error components based on spatial independent component analysis, and performing phase optimization on the differential interferometric dataset using the coherence matrix to obtain a corrected differential interferometric dataset; performing phase unwrapping based on the corrected differential interferometric dataset, and incrementally updating and calculating the newly added echo data based on a sequential estimation algorithm to obtain the slope deformation.
[0006] Secondly, embodiments of this application also provide a slope deformation sequential monitoring device based on synthetic aperture radar mounted on a high-speed railway. The device includes: a generation module for acquiring raw echo data via synthetic aperture radar and generating single-view complex images based on the raw echo data; a construction module for image registration of the single-view complex images and differential interferometry processing based on the registered single-view complex images to construct a differential interferometric dataset; a correction module for constructing a coherence matrix based on the differential interferometric dataset, identifying and correcting error components based on spatial independent component analysis, and performing phase optimization on the differential interferometric dataset using the coherence matrix to obtain a corrected differential interferometric dataset; and a determination module for performing phase unwrapping based on the corrected differential interferometric dataset and incrementally updating newly added echo data based on a sequential estimation algorithm to obtain the slope deformation.
[0007] Thirdly, this application also provides a slope deformation sequential monitoring device based on synthetic aperture radar mounted on a high-speed railway, including a processor, a memory, and one or more application programs; the one or more application programs are stored in the memory and configured to be executed by the processor to implement the above-mentioned slope deformation sequential monitoring method based on synthetic aperture radar mounted on a high-speed railway.
[0008] Fourthly, this application also provides a computer-readable storage medium storing program code, wherein the program code is executed by a processor to perform the above-mentioned sequential monitoring method for slope deformation based on synthetic aperture radar mounted on a high-speed railway.
[0009] The technical solution provided in this application includes the following method: acquiring raw echo data through synthetic aperture radar and generating single-view complex images based on the raw echo data; performing image registration on the single-view complex images and performing differential interferometry processing on the registered single-view complex images to construct a differential interferometric dataset; constructing a coherence matrix based on the differential interferometric dataset, identifying and correcting error components based on spatial independent component analysis, and performing phase optimization on the differential interferometric dataset using the coherence matrix to obtain a corrected differential interferometric dataset; performing phase unwrapping on the corrected differential interferometric dataset, and incrementally updating and solving the newly added echo data based on a sequential estimation algorithm to obtain the slope deformation. Therefore, by utilizing the non-contact active remote sensing characteristics of synthetic aperture radar to acquire raw echo data and generate single-view complex images, and leveraging its all-weather and high-penetration physical advantages, the limitations of manual inspection and contact sensors due to geographical environment and climate conditions, such as limited coverage and inability to monitor in all weather conditions, can be effectively solved. Furthermore, by constructing a differential interferometric dataset, using coherence matrix to perform phase optimization on the data and completing phase unwrapping and deformation inversion, high-precision extraction of minute deformations of slopes can be achieved. This data processing-based non-contact monitoring method avoids the hardware deployment and manual maintenance of a large number of sensors, significantly reducing monitoring costs. At the same time, the application of phase optimization and efficient inversion algorithms ensures the timeliness of data processing and the refinement of results, thereby meeting the needs of wide-area, high-frequency, and refined real-time monitoring and early warning of slopes along high-speed railways. Attached Figure Description
[0010] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments and drawings obtained by those skilled in the art without creative effort are within the scope of protection of this invention.
[0011] Figure 1 The illustration shows a flowchart of a sequential monitoring method for slope deformation based on synthetic aperture radar mounted on a high-speed railway, according to an embodiment of this application.
[0012] Figure 2 This document presents a schematic diagram of a high-speed rail-mounted synthetic aperture radar (SAR) for slope monitoring, as provided in an embodiment of this application.
[0013] Figure 3 This illustration shows a structural schematic diagram of a slope deformation sequential monitoring device based on synthetic aperture radar mounted on a high-speed railway, according to an embodiment of this application.
[0014] Figure 4This illustration shows a structural schematic diagram of a slope deformation sequential monitoring device based on synthetic aperture radar mounted on a high-speed railway, according to an embodiment of this application.
[0015] Figure 5 This illustration shows a schematic diagram of the structure of a computer-readable storage medium provided in an embodiment of this application. Detailed Implementation
[0016] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention as detailed in the appended claims.
[0017] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
[0018] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.
[0019] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.
[0020] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0021] References to "one embodiment" or "some embodiments" in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized.
[0022] my country has a large-scale high-speed transportation system, but the slopes along high-speed railways are affected by a combination of factors such as geology, climate and train vibration, which can easily lead to gradual deformation and geological disasters such as landslides and rockfalls, seriously threatening the safety of railway operations. Therefore, long-term and efficient monitoring of slopes is of utmost importance.
[0023] Among related technologies, high-speed railway slope monitoring mainly relies on manual inspection, contact sensors, and spaceborne InSAR technology. However, these technologies have inherent limitations in terms of cost, coverage, and timeliness, making it difficult to meet the needs of high-speed railway slope monitoring in a refined manner that is available around the clock, over a wide area, and at high frequency.
[0024] Specifically, manual inspections are limited by labor costs, inspection frequency, and the closed management of high-speed railways, making it difficult to achieve continuous quantitative monitoring. Furthermore, accessibility is poor at night and in inclement weather. While contact sensors (such as GNSS and total stations) have high accuracy, they are costly to lay over long distances, difficult to maintain, and limited by terrain, making it difficult to cover all potentially risky slopes. Although spaceborne InSAR has the advantage of wide-area observation, it is easily obstructed by complex terrain such as mountains, valleys, and steep slopes along high-speed railways, resulting in missing observations. Moreover, the revisit period and spatial resolution are difficult to balance in terms of monitoring timeliness.
[0025] To address the aforementioned issues, this application provides a method, apparatus, equipment, and storage medium for sequential monitoring of slope deformation based on synthetic aperture radar (SAR) mounted on a high-speed railway. The method includes: acquiring raw echo data using SAR and generating single-view complex images based on the raw echo data; performing image registration on the single-view complex images and differential interferometry processing based on the registered single-view complex images to construct a differential interferometric dataset; constructing a coherence matrix based on the differential interferometric dataset, identifying and correcting error components based on spatial independent component analysis, and performing phase optimization on the differential interferometric dataset using the coherence matrix to obtain a corrected differential interferometric dataset; performing phase unwrapping on the corrected differential interferometric dataset, and incrementally updating and calculating the newly added echo data based on a sequential estimation algorithm to obtain the slope deformation.
[0026] Therefore, by utilizing the non-contact active remote sensing characteristics of synthetic aperture radar to acquire raw echo data and generate single-view complex images, and leveraging its all-weather and high-penetration physical advantages, the limitations of manual inspection and contact sensors due to geographical environment and climate conditions, such as limited coverage and inability to monitor in all weather conditions, can be effectively solved. Furthermore, by constructing a differential interferometric dataset, using coherence matrix to perform phase optimization on the data and completing phase unwrapping and deformation inversion, high-precision extraction of minute deformations of slopes can be achieved. This data processing-based non-contact monitoring method avoids the hardware deployment and manual maintenance of a large number of sensors, significantly reducing monitoring costs. At the same time, the application of phase optimization and efficient inversion algorithms ensures the timeliness of data processing and the refinement of results, thereby meeting the needs of wide-area, high-frequency, and refined real-time monitoring and early warning of slopes along high-speed railways.
[0027] This invention provides a method for sequential monitoring of slope deformation based on synthetic aperture radar (SAR) mounted on a high-speed railway. The execution entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, this method for sequential monitoring of slope deformation based on SAR mounted on a high-speed railway can be executed by software or hardware installed on a terminal device or a server device. The software can be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cluster of cloud servers. The server can be an independent server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.
[0028] Please see Figure 1 , Figure 1 This document illustrates a flowchart of a sequential monitoring method for slope deformation based on synthetic aperture radar mounted on a high-speed railway, as provided in an embodiment of this application. Figure 1 As shown, the method may include steps 110 to 140.
[0029] In step 110, raw echo data is acquired by synthetic aperture radar, and single-view complex images are generated based on the raw echo data.
[0030] In some implementations, synthetic aperture radar (SAR) can be an active microwave remote sensing imaging system installed on high-speed trains. Its working principle involves actively transmitting electromagnetic pulses towards the slope areas along the high-speed rail line and receiving the echo signals reflected back from surface targets. Unlike optical remote sensing, which is limited by lighting conditions, SAR has all-weather, day-and-night observation capabilities, can penetrate clouds and rain / fog, and is particularly suitable for the complex and variable meteorological environment along high-speed rail lines.
[0031] In some implementations, the raw echo data can be baseband signals directly acquired by synthetic aperture radar without focusing processing. The raw echo data records the time sequence information of the electromagnetic wave from its emission to its return to the antenna after reflection from the ground target. In the embodiments of this application, the raw echo data may contain amplitude information of the ground object scattering characteristics and phase information reflecting range accuracy, serving as the foundational data for subsequent generation of high-resolution imagery.
[0032] For example, please refer to Figure 2 , Figure 2 This application provides a schematic diagram of a high-speed rail-mounted synthetic aperture radar (SAR) for slope monitoring, as illustrated in an embodiment of this application. Figure 2 As shown, a synthetic aperture radar (SAR) is mounted on the body of a high-speed train, with its antenna (labeled "SAR antenna") facing the slope along the train line for side-view observation. As the high-speed train moves at high speed along the track, the SAR antenna sequentially emits electromagnetic pulses at different locations along its trajectory and receives raw echo data. For example, Figure 2 (a) and Figure 2 The different time positions are shown in (b) in the figure. That is to say, the raw echo data refers to the unprocessed electromagnetic wave reflection signal received by the SAR antenna.
[0033] In some implementations, single-view complex images can be standard SAR image data generated by imaging focusing processing (e.g., range compression and azimuth compression) of the raw echo data.
[0034] For example, please continue to refer to Figure 2 The train will be in Figure 2 (a) and Figure 2 In (b), the original echo data of the same target received at the corresponding locations are coherently superimposed and focused (including range compression and azimuth compression). This process focuses the originally dispersed echo signals into high-resolution pixels, ultimately generating a single-view complex image containing amplitude and phase information for each pixel.
[0035] In step 120, image registration is performed on the single-view complex image, and differential interferometry is performed on the registered single-view complex image to construct a differential interferometry dataset.
[0036] In one specific implementation, a single-view complex image can be selected as the master image. The offsets of the remaining single-view complex images relative to the master image are calculated, and the remaining single-view complex images are resampled based on the offsets, so that the spatial pixels of all single-view complex images are accurately aligned geographically. Through image registration, the spatial geometric deviation caused by the jitter of the high-speed rail platform's flight trajectory is eliminated, ensuring the coherence between images in subsequent differential interferometric processing.
[0037] In some implementations, the differential interferometric dataset may include topographic phase, deformation phase, orbital error phase, atmospheric phase, and noise phase.
[0038] In one specific implementation, the terrain phase can be a phase component resulting from the change in echo data path length caused by the height difference of the surface target relative to the reference surface. The terrain phase can serve as a major geometric component in the observation signal, and its spatial distribution characteristics are highly correlated with the terrain undulations.
[0039] In one specific implementation, the deformation phase can be a phase component generated due to the physical displacement (i.e. deformation) of the surface target on the slope along the high-speed railway during the two observations, which causes the change in the length of the round-trip propagation path of the radar electromagnetic wave.
[0040] In the embodiments of this application, the deformation phase directly reflects the stability state of the slope structure and is the core observation value for deformation inversion. According to radar observation geometry, there is a definite linear relationship between the deformation phase and the deformation of the slope along the radar line of sight. The ultimate goal of this invention is to extract this component from the complex interferometric phase in order to achieve quantitative monitoring of slope deformation.
[0041] In one specific implementation, the orbital error phase (pose error) may be caused by the deviation between the actual orbital position and the nominal orbital position of the high-speed rail platform during flight (i.e., baseline error), or by the slight jitter of the synthetic aperture radar's attitude (pose), which leads to inaccurate calculation of the geometric distance between the radar and the ground target and introduces the phase error.
[0042] It typically manifests as flat phase fringes or linear phase in interferograms. In long-distance high-speed rail SAR observations, due to the long track and complex platform vibrations, the track error phase often exhibits significant spatial correlation (e.g., a trend term that varies with azimuth or range). If not corrected, it can be misjudged as large-scale slope deformation, severely affecting monitoring accuracy.
[0043] In one specific implementation, the atmospheric phase can be the phase component generated when radar electromagnetic wave signals pass through the atmosphere (especially the troposphere). Due to the changes in meteorological parameters such as temperature, air pressure, and humidity in the environment along the high-speed rail line over time, the propagation speed of electromagnetic waves changes, resulting in an equivalent delay or advance in the propagation path.
[0044] Atmospheric phase is unrelated to the actual deformation of the slope and belongs to spatiotemporally correlated noise signals. In the embodiments of this application, atmospheric phase is a major source of error that is difficult to eliminate in the monitoring of long-distance high-speed railway lines, and it needs to be separated and corrected using methods such as independent component analysis (ICA).
[0045] In one specific implementation, the noise phase can be the sum of random errors in the differential interferometric phase excluding the aforementioned systematic phase components. The noise phase mainly includes two parts: first, temporal decoherence noise introduced by changes in the backscattering characteristics of ground objects over time due to excessively long time baselines or environmental changes (e.g., vegetation growth, rain erosion); and second, thermal noise generated inside the synthetic aperture radar.
[0046] Noise phase typically manifests as a randomly distributed high-frequency signal, which reduces the signal-to-noise ratio of the interferogram and affects the accuracy of phase unwrapping. This application employs a distributed scatterer optimization strategy to suppress the impact of this type of noise.
[0047] Therefore, the slope geometry and deformation information, which are difficult to observe directly, are transformed into computable interferometric phase signals. This makes the phase values contained in the differential interferometric dataset clearly correspond to the linear superposition of the slope's topographic phase, deformation phase, orbital error phase, atmospheric phase affected by meteorological environment, and various random noise phases. This provides the necessary data foundation for subsequent use of algorithms such as independent component analysis to accurately separate and remove error terms and extract true deformation information.
[0048] Furthermore, in some implementations, the step of performing differential interferometric processing on the registered single-view complex image to construct a differential interferometric dataset may include the following steps: (1) Based on the single-view complex images registered at different observation times, calculate the complex phase difference between the first observation time and the second observation time; (2) Use the complex phase difference as an element in the differential interferometric dataset.
[0049] The differential interferometric phase satisfies a linear superposition model consisting of terrain phase, deformation phase, orbital error phase, atmospheric phase, and noise phase.
[0050] In some implementations, the first observation time ( ) and the second observation time ( ( ) can refer to the time points when a synthetic aperture radar mounted on a high-speed train observes the same slope area at different times and locations along the track.
[0051] For example, please continue to refer to Figure 2 When the train reaches the first observation time ( At that time, the SAR antenna is in the first position, transmitting electromagnetic waves and receiving echoes from the slope, recording the phase state at that moment. When the train continued running until the second observation time ( At that time, the SAR antenna moved to the second position and observed the same slope area again, recording the phase state at that moment. .
[0052] In some implementations, the complex phase difference can be achieved by registering a single-view complex image (with phase at the second observation time). The registered single-view complex image (phase is) with the first observation time. The difference obtained after performing conjugate multiplication (or subtraction) on the complex phase difference. .
[0053] Complex phase difference is not a single numerical value, but rather includes the phase change of every pixel in the slope area. It extracts the minute distance changes (deformations) that were originally hidden in two independent images and transforms them into visualized interference fringe information, which forms the basic unit for subsequently constructing the differential interferometric dataset.
[0054] In some implementations, the expression for the complex form data of each pixel in a single-view complex image can be: in, Let be the real part of the complex number. The imaginary part of a complex number. For amplitude, The virtual part, It is a complex phase.
[0055] To more intuitively reflect physical properties, SLC data is usually converted to polar coordinates, containing the following two key physical quantities: amplitude ( ) and phase ( Amplitude represents the intensity of the radar echo signal, reflecting the backscattering characteristics of the monitored slope surface (such as surface roughness and material). Amplitude information is typically used to generate grayscale images, visually displaying the slope's geometry. Phase represents the phase state of the radar wave as it propagates from the SAR antenna to the slope target and returns, recording precise path information.
[0056] In one specific implementation, the amplitude can be solved using the following equation ( ): in, For amplitude, Let be the real part of the complex number. It represents the imaginary part of a complex number.
[0057] In one specific implementation, the phase can be solved using the following equation ( ): in, For phase, Let be the real part of the complex number. It represents the imaginary part of a complex number.
[0058] When a monitored slope undergoes minor deformation (e.g., landslide or settlement) between two observation times, the round-trip path length of the radar waves changes, resulting in a phase shift. Corresponding changes occur. Based on this principle, this embodiment can accurately capture millimeter-level slope deformation by preserving and analyzing the phase information in the SLC data.
[0059] In some implementations, the expression for the linear superposition model can be: in, For terrain phase, For deformation phase, For orbital error phase, Atmospheric phase, This is the noise phase.
[0060] Thus, the differential interferometric phase obtained in the SAR differential interferometry measurement on high-speed rail has been... It is not caused solely by slope deformation, but by topographic phase. Deformation phase Orbital error phase Atmospheric phase and noise phase These factors are combined to form the overall structure. Among them, the orbital error phase and atmospheric phase are the most significant error sources in high-speed rail InSAR data. If not effectively corrected, they will severely interfere with the accuracy of deformation information extraction and may even lead to misjudgments. Therefore, in order to achieve high-precision inversion of slope deformation, it is necessary to model and separate the above error terms.
[0061] In step 130, a coherence matrix is constructed based on the differential interferometric dataset, and error components are identified and corrected based on spatial independent component analysis. The phase of the differential interferometric dataset is then optimized using the coherence matrix to obtain the corrected differential interferometric dataset.
[0062] In some implementations, the coherence matrix can be a multidimensional statistical feature matrix constructed based on a differential interferometric dataset. The coherence matrix is used to quantify the phase stability and amplitude correlation of the same pixel or target region across multiple observation times (or multiple single-view complex images).
[0063] In the embodiments of this application, each element of the coherence matrix represents the degree of cross-correlation (i.e., coherence coefficient) of the complex signals between two images. The closer the value is to 1, the more stable the scattering characteristics of the area at different times (e.g., stable rocks, buildings), and the higher the phase quality; the lower the value, the more the area is affected by decorrelation noise (e.g., vegetation swaying, random noise).
[0064] By constructing a coherence matrix, we can extract statistical patterns of targets in time series from high-dimensional data, providing a mathematical basis for distinguishing real deformation signals from random noise.
[0065] Furthermore, high-coherence points with temporal coherence exceeding a preset threshold are selected based on the coherence matrix. Next, the phase values of these high-coherence points in each differential interferogram are recombined to construct an observation matrix. Subsequently, the observation matrix is decomposed using Spatial Independent Component Analysis (ICA) to separate multiple statistically independent component signals. By analyzing the spatial spectral characteristics and temporal evolution of each component signal, error component signals characterizing atmospheric delay or orbital error are identified and extracted. Finally, the error component signals are subtracted from the original differential interferogram dataset to correct the error components, resulting in a corrected differential interferogram dataset. This will be described in detail later.
[0066] In some implementations, the corrected differential interferometric dataset can be the result of phase optimization processing of the original differential interferometric dataset using the coherence matrix.
[0067] In the corrected differential interferometric dataset, random noise phases and phase errors caused by low-coherence regions in the original differential interferometric phases have been effectively suppressed or filtered out. Compared with the original data, the corrected differential interferometric dataset has a higher signal-to-noise ratio and phase continuity, and can more clearly preserve the phase trend caused by the actual deformation of the slope, while weakening random fluctuations caused by non-deformation factors. It serves as the foundational data for subsequent high-precision deformation inversion and error separation.
[0068] Therefore, by constructing a coherence matrix, the statistical correlation of high-speed rail SAR multi-temporal observation data in the time domain is fully utilized to perform weighted optimization on each pixel in the original differential interferometric dataset. This allows the use of highly coherent (highly stable) observations to correct low-coherence (noise-contaminated) observations, thereby preserving the true deformation information of the slope while filtering out the interference of random noise phases to the greatest extent.
[0069] Furthermore, this achieves the "purification" of differential interferometric data, significantly improving the phase quality of the data and solving the problem of insufficient deformation monitoring accuracy caused by the low signal-to-noise ratio of a single image in complex environments. This lays a solid data foundation for the subsequent accurate separation of pose and atmospheric errors. Specifically, in some implementations, the step of constructing a coherence matrix based on the differential interferometric dataset may include the following steps: (1) Arrange the interferograms in the differential interferometric dataset according to spatial pixels to construct an observation matrix with interferograms as rows and spatial pixels as columns; (2) Using the spatial independent component analysis algorithm, the observation matrix is decomposed to extract mutually independent component signals; (3) Based on the spatial correlation characteristics and temporal correlation characteristics of each component signal, identify and screen out the error components that represent atmospheric phase and orbital error phase from the component signals; (4) Construct a coherence matrix based on the differential interferometric dataset after removing error components.
[0070] In some implementations, constructing the observation matrix can involve unfolding each two-dimensional interferogram in the differential interferometric dataset into a one-dimensional vector, the length of which is equal to the number of effective pixels in the interferogram. Then, the one-dimensional vectors corresponding to all interferograms are arranged sequentially to form a two-dimensional matrix. In this matrix, the number of rows corresponds to the number of interferograms (time dimension), and the number of columns corresponds to the number of spatial pixels (spatial dimension). This observation matrix intuitively reflects the phase variation of each pixel over time, providing a data foundation for subsequent separation of error components using spatially independent component analysis algorithms.
[0071] In some implementations, the component signals can be statistically independent signal components obtained by blind source separation of the observation matrix using a spatially independent component analysis (ICA) algorithm. Since the ICA algorithm assumes that the observation data is a linear mixture of several unknown source signals, each component signal obtained by decomposition corresponds to a physical phenomenon with a specific spatial distribution pattern and temporal evolution law. For example, one component signal may mainly correspond to atmospheric delay, another may correspond to orbital error, and yet another may correspond to actual slope deformation.
[0072] In some implementations, based on the spatial correlation characteristics (e.g., spatial continuity of atmospheric phase, orbital correlation of pose error) and temporal correlation characteristics (e.g., randomness of atmospheric variations, periodicity of pose error) of each component signal, those component signals determined to be dominated by pose phase and atmospheric phase are labeled as error components. These signals are considered as interference items that need to be removed from the raw data.
[0073] In some implementations, the differential interferometric dataset after filtering out error components can be the dataset remaining after subtracting or removing error components from the observation matrix. This dataset mainly retains the deformation phase of the slope and residual random noise, while significantly reducing the positional error caused by high-speed rail track deviation and the atmospheric delay caused by the atmospheric environment.
[0074] For example, in order to separate the error term from the mixed interferometric phase, the differential interferometric dataset is first rearranged according to the first dimension (time series dimension) and the second dimension (spatial flattening dimension) to construct the observation matrix. Among them, the observation matrix Each row represents the spatial unfolding vector of a differential interferogram, and each column represents the time series of a pixel.
[0075] Subsequently, the observation matrix was analyzed using a spatially independent component analysis algorithm. Decompose, that is This allows for the extraction of independent component signals. These component signals These correspond to the statistical characteristics of different source signals in the physical scene.
[0076] Next, error components are identified and filtered out from all components based on spatial correlation characteristics (e.g., spatial smoothness) and temporal correlation characteristics (e.g., abrupt changes or trends in time) of the individual component signals.
[0077] Finally, a coherence matrix is constructed based on the differential interferometric dataset after removing error components. Since the main error sources in the data have been separated at this point, the coherence matrix constructed based on this clean dataset can more realistically reflect the scattering stability of the slope target, thereby improving the accuracy of subsequent phase optimization.
[0078] However, after identifying and removing the main atmospheric phase and pose errors using the aforementioned spatial independent component analysis algorithm, the remaining differential interferometric dataset, although free of systematic error sources, still inevitably contains random noise. This is especially true in regions with widely distributed, low-reflectivity distributed scatterers, where phase quality is often poor and difficult to use directly for high-precision deformation inversion. Therefore, in some implementations, the step of using the coherence matrix to perform phase optimization on the differential interferometric dataset to obtain a corrected differential interferometric dataset may include the following steps: (1) Based on the statistical characteristics of the differential interferometric dataset, identify and select candidate points of the distributed scatterer, and determine the set of homogeneous pixels in the neighborhood of the candidate points of the distributed scatterer; (2) Calculate the sample coherence matrix corresponding to the candidate points of the distributed scatterer using time series data of homogeneous pixel sets; (3) Based on the coherence matrix and the sample coherence matrix, construct the regularized maximum likelihood objective function for the coherence matrix to be estimated, and introduce the coherence matrix regularization constraint term into the objective function to form an optimization model; (4) Solve the optimization model by maximizing the likelihood function of the observed data to obtain the optimal coherence matrix estimate; (5) The phase of the candidate points of the distributed scatterer is re-estimated using the optimal coherence matrix estimate to obtain the corrected differential interferometric dataset containing the optimized phase.
[0079] In some implementations, statistical properties can be the probability distribution patterns of pixels in the differential interferometric dataset in terms of amplitude distribution, phase stability, or temporal correlation. In embodiments of this application, statistical properties (e.g., amplitude deviation index or phase variance) are used to distinguish between distributed scatterers with stable scattering mechanisms and completely random noise points. Through statistical analysis, pixels that exhibit consistent scattering behavior within local regions, even if their individual signal-to-noise ratio is low, can be identified.
[0080] In some implementations, distributed scatterer candidate points can be pixels that are widely distributed within the monitoring area, have moderate reflectivity, and whose scattering mechanism is contributed by multiple scattering sub-targets. Unlike permanent scatterers with extremely high and stable reflectivity, distributed scatterer candidate points (e.g., bare ground surfaces, gravel piles) are easily affected by decorrelation noise, but their large number makes them key to improving the spatial coverage of slope monitoring.
[0081] In some implementations, the set of homogeneous pixels can be a group of pixels within a preset neighborhood window of a candidate point of a distributed scatterer, selected through statistical tests (e.g., KS test) that have similar scattering characteristics or amplitude / phase statistical regularities to the candidate point. These homogeneous pixels are considered to come from the same type of ground object, and their spatial redundancy information can be used to effectively assist in estimating the true phase of the candidate point, thereby suppressing random noise.
[0082] In some implementations, the time-series data of a homogeneous pixel set can be the set of complex phase or amplitude values of each pixel in the homogeneous pixel set at all observation times (or all interference pairs). This data reflects the dynamic characteristics of the target's evolution over time and serves as the fundamental input for calculating the sample coherence matrix.
[0083] For example, setting the candidate point as the center A sliding window is used; the coherence coefficient or spectral correlation coefficient between all pixels within the window and the center pixel is calculated; pixels with a coherence coefficient greater than a set similarity threshold (e.g., 0.8) are included in the set. This ensures that the pixels used for estimation have the same physical scattering mechanism.
[0084] In some implementations, the sample coherence matrix can be a local coherence matrix calculated by statistical averaging of time-series data from a homogeneous set of pixels. The sample coherence matrix initially describes the phase correlation of the candidate points of the distributed scatterer in the time dimension, but due to the limited number of samples, the matrix often contains estimation errors, thus requiring further optimization.
[0085] In one specific implementation, the sample coherence matrix can be represented as: in, The sample coherence matrix, The length of the time series. This is the conjugate transpose. It is understood that this application does not impose restrictions on the specific expression of the sample coherence matrix.
[0086] In some implementations, the objective function can be a mathematical expression used to solve for the optimal coherence matrix, typically based on the maximum likelihood estimation principle. The objective function quantifies the probability of observing the current sample data given the coherence matrix parameters. The optimization process involves finding a set of parameters that maximizes the value of the objective function, thereby obtaining a coherence matrix estimate that best approximates the actual physical situation.
[0087] In one specific implementation, the objective function can be expressed as: in, and Construct the maximum likelihood term, For regularization parameters, This is the Frobenius norm. This term is introduced to utilize the clean data after ICA separation as a guide, ensuring that the optimized coherence matrix conforms to both local observations and global physical laws. It is understood that this application does not impose restrictions on the specific expression of the objective function.
[0088] In some implementations, the coherence matrix regularization constraint can be a penalty term introduced to improve the stability of the estimation. This constraint term uses the coherence matrix constructed based on the entire dataset in the preceding steps as prior information to constrain the objective function, preventing overfitting or estimation bias when the sample size is insufficient, and ensuring that the optimization result maintains a balance between local details and global statistical regularity.
[0089] In some implementations, the optimization model can be a complete mathematical model that incorporates the objective function and the regularization constraint term of the coherence matrix. This model transforms the problem of estimating the coherence matrix into a constrained optimization problem, aiming to find the coherence matrix solution that maximizes the probability of the observed data occurring, while satisfying physical constraints, through numerical computation methods.
[0090] In one specific implementation, the optimization model is specifically described as finding an optimal coherence matrix. This minimizes the objective function. The optimization model can be expressed as: in, This is the optimal coherence matrix estimate. This refers to the determinant operation of a matrix. For the trace operation of a matrix, The inverse of the matrix, " is the constraint matrix It must be a positive definite Hermitian matrix. Understandably, this application does not impose restrictions on the specific expression of the optimization model.
[0091] In some implementations, the likelihood function can be a core component of the optimization model, representing the probability density of observing the current homogeneous pixel set time series data given a specific coherence matrix. Maximizing this likelihood function means finding the coherence matrix that best explains the current observation data, thus achieving the best approximation of the true phase information.
[0092] In one specific implementation, the likelihood function is constructed based on the assumption that the time-series data of a homogeneous pixel set follows a multivariate complex Gaussian distribution. The likelihood function can be expressed as: in, The number of pixels in a set of homogeneous pixels. For matrix The determinant, For matrix The inverse matrix, This refers to the trace operation of a matrix. It is understood that this application does not impose restrictions on the specific expression of the likelihood function.
[0093] In some implementations, the optimal coherence matrix estimate can be the final coherence matrix obtained by solving an optimization model to maximize the likelihood function of the observed data. Compared to the initial sample coherence matrix, this estimate effectively combines the statistical information of locally homogeneous pixels with the prior constraint of global coherence, exhibiting higher accuracy and robustness. It is then used for subsequent re-estimation and correction of the phase of candidate points for distributed scatterers.
[0094] For example, based on the statistical properties of the differential interferometric dataset (e.g., amplitude mean and phase stability), candidate points of distributed scatterers are identified and selected. These points are typically widely distributed areas on a slope but with a generally low signal-to-noise ratio. Subsequently, a set of homogeneous pixels, i.e., a group of pixels that have similar scattering behavior to the candidate point, is searched and determined within the neighborhood of each candidate point.
[0095] Next, using the time-series data of these homogeneous pixels, the sample coherence matrix corresponding to the candidate point is calculated to preliminarily characterize its temporal correlation. To further improve the estimation accuracy, an objective function based on maximum likelihood estimation is constructed, and a coherence matrix regularization constraint term (using global coherence information obtained in the previous steps) is introduced into it, thereby forming an optimization model.
[0096] By maximizing the likelihood function of the observed data using a mathematical algorithm, the optimal model is solved to obtain the optimal coherence matrix estimate. Finally, this high-precision optimal estimate is used to re-estimate the original phase of the candidate points of the distributed scatterer using a weighted method, resulting in a corrected differential interferometric dataset with significantly improved phase quality.
[0097] After obtaining the corrected differential interferometric dataset, systematic errors (e.g., pose errors, atmospheric delays) and random noise in the data are effectively suppressed, and the signal-to-noise ratio is significantly improved. However, the interferometric phase is still contained within the principal value range and includes integer ambiguity, which cannot directly reflect the actual physical displacement.
[0098] In step 140, phase unwrapping is performed based on the corrected differential interferometric dataset, and incremental update calculation is performed on the newly added echo data based on the sequential estimation algorithm to obtain the slope deformation.
[0099] After obtaining the corrected differential interferometric dataset, although the phase values in the dataset are more spatially continuous and have lower noise, their values are still limited to ( Within the principal value interval of [π, π] (i.e., the wrapped phase), there exists an integer ambiguity of 2π, which cannot directly represent the actual physical distance change. Therefore, this step first performs phase unwrapping on the corrected differential interferometric dataset.
[0100] Specifically, a phase unwrapping algorithm (e.g., Minimum Cost Flow (MCF) or branching method) is used to identify the phase gradient between adjacent pixels in the interferogram. The absolute phase value is then recovered through integration, eliminating the 2π periodic jumps, thus obtaining an unwrapped phase map that reflects the true phase changes. Since the corrected dataset has already eliminated atmospheric phase and pose errors through the aforementioned steps, the unwrapping process can more accurately track the true deformation fringes, avoiding unwrapping path errors caused by error accumulation.
[0101] Subsequently, when new SAR echo data is acquired and a new differential interferogram is formed, it is not necessary to reprocess all the data of the historical time series. Instead, the solution results of the previous moment are used as prior information, and the state is updated only for the new data.
[0102] Specifically, a residual vector is constructed based on the newly added differential interferogram and the deformation estimate from the previous time step. This residual vector reflects the difference between the new observation and the predicted value. Subsequently, the gain matrix and covariance matrix are updated using Kalman filtering or recursive least squares method, and the deformation rate increment and residual phase increment at the current time step are calculated based on the residual vector. Finally, the increment is superimposed on the cumulative deformation from the previous time step, thereby quickly obtaining the time series of slope deformation containing the latest data. This incremental update method significantly reduces computational complexity and enables near real-time monitoring of slope deformation.
[0103] In essence, obtaining the slope deformation is a process of converting the phase increment calculated by sequential estimation into physical displacement. According to the interferometric principle of synthetic aperture radar, the phase change is proportional to the surface displacement along the radar line-of-sight (LOS). Therefore, the calculated phase value is converted into a physically meaningful slope deformation using the following relationship: in, This represents the deformation (displacement) of the slope along the radar line of sight. For radar wavelength, This represents the phase value after unwrapping. Through this calculation, abstract phase information is transformed into physically meaningful slope deformation variables (e.g., cumulative settlement or deformation rate), thereby enabling quantitative monitoring of slope stability.
[0104] After efficiently acquiring the time series of slope deformation along the radar line of sight (LOS) using the above technical solution, although the timeliness problem of massive data processing is solved, the results are still mainly limited to one-dimensional projection components. For the complex geological bodies of slopes along high-speed railways, single line-of-sight deformation is difficult to intuitively reflect its true vertical settlement and horizontal slip characteristics, and lacks quantitative assessment of deformation evolution trends, thus failing to directly meet the engineering safety early warning requirements for "precise spatial location" and "rapid risk level determination". Therefore, in order to further explore the engineering value of deformation data, in some implementations, this slope deformation sequential monitoring method based on synthetic aperture radar mounted on high-speed railways may also include the following steps: (1) Based on the corrected differential interferometric dataset, an initial deformation inversion model is established; (2) Solve the initial deformation inversion model by least squares estimation to obtain the initial values of deformation parameters and the first cofactor matrix; (3) When acquiring new single-view complex images, construct a new differential interferometric dataset based on the new single-view complex images and existing single-view complex images according to the preset interferometric combination relationship; (4) Based on the newly added differential interferometric dataset and the initial values of deformation parameters, a sequential estimation model is constructed; (5) Using the sequential estimation model, combined with the weight matrix of the observation equation corresponding to the first cofactor matrix and the newly added differential interferometric dataset, calculate the deformation parameter update increment; (6) Based on the update increment of the deformation parameters and the initial value of the deformation parameters, the updated deformation parameter time series is obtained; the updated deformation parameter time series is determined as the deformation of the slope.
[0105] In some implementations, the initial deformation inversion model can be a set of linear observation equations constructed based on differential interferograms generated from historically archived single-view complex images.
[0106] In some implementations, the initial values of the deformation parameters can be the first batch of deformation parameter estimates obtained after solving the initial deformation inversion model using the least squares method. These represent the cumulative deformation or phase values of the slope at each observation time, calculated based on existing historical data before acquiring new images.
[0107] In some implementations, the preset interferometric combination relationship can be a small baseline set pattern based on spatiotemporal baseline constraints. That is, from existing single-view complex images, images that satisfy preset thresholds on both the temporal and spatial vertical baselines of the newly added single-view complex images are selected as candidate master images, and one or more new differential interferometric pairs are constructed. The preset interferometric combination relationship can maximize interferometric coherence and reduce decoherence noise. In some implementations, the first cofactor matrix can be a statistic characterizing the accuracy of the initial values of the deformation parameters. It is the inverse matrix of the coefficient matrix of the parametric normal equations during the solution process, used to describe the correlation and variance characteristics among the estimated deformation parameters, and will participate in the sequential update as prior information in subsequent steps.
[0108] In some implementations, a newly added single-look complex image refers to the most recently acquired raw SAR image data that has not yet been included in historical deformation calculations. It is a new data source that triggers the sequential update process.
[0109] In some implementations, the new differential interferometric dataset can be a set of differential interferograms generated by interferometric processing of a new single-view complex image with an existing reference image (or master image).
[0110] In some implementations, the sequential estimation model can be an incremental update algorithm model. Instead of reprocessing all historical data, it uses the initial values of deformation parameters obtained in the previous step as prior information, combined with a joint system of equations constructed from the newly added differential interferometric dataset.
[0111] In some implementations, the weight matrix can be a diagonal matrix used to characterize the quality or reliability of the observed data. In the computation, the weight matrix is used to weight interferometric pixels of different qualities, with higher quality (lower noise) data being assigned greater weights, thus dominating the solution.
[0112] In some implementations, the deformation parameter update increment can be a deformation parameter correction due to the introduction of new imagery. It is calculated by multiplying the residuals of the new observations by the gain matrix, reflecting the extent to which the new data adjusts the historical deformation estimates.
[0113] In some implementations, the updated deformation parameter time series can be the latest deformation result obtained by superimposing the initial values of the deformation parameters with the updated values. This series contains the complete slope deformation history up to the latest moment, realizing real-time dynamic updating of deformation monitoring results.
[0114] For example, firstly, an initial deformation inversion model is constructed based on a corrected differential interferometric dataset generated from an initially acquired set of single-view complex images. Assume the number of archived SAR data is... Its observation equation can be expressed as: in, For those with design matrix Sum weight matrix Archived SAR data observation vectors (i.e., differential interferometric phase). The deformation parameter vector to be estimated (i.e., the deformation phase at different times) ).
[0115] The initial deformation inversion model above is solved using the least squares estimation principle to obtain the initial values of the deformation parameters and their corresponding first cofactor matrix. The calculation formula is as follows: in, This is the transpose of the matrix. These are the initial values for the deformation parameters. It is the first cofactor matrix.
[0116] When new single-view multiple images are acquired during the monitoring process (e.g., the first...), When acquiring SAR images (plus two additional images), interferometry is performed between these images and existing single-look complex images to construct a new differential interferometric dataset. This dataset corresponds to unwrapped interferograms dynamically related to the new SAR images.
[0117] Based on the newly added differential interferometric dataset and initial values of deformation parameters, a sequential estimation model is constructed. It is assumed that the observed values of the newly added data are... The weight matrix is The design matrix is and The parameters to be estimated include updated historical deformation parameters. Cumulative deformation corresponding to newly added images Its observation equation can be written as: Based on the least squares Bayes estimation principle, using a sequential estimation model and combining the first cofactor matrix... (As prior information) and the weight matrix of newly added data Calculate the updated values of the deformation parameters. The objective function is to minimize the weighted sum of the residuals and prior errors: Through derivation, the gain matrix is introduced. and intermediate variables The deformation parameter update increment is calculated. The specific recursive formula is as follows: Updated parameter estimates It can be represented as: Based on the calculated update increment and initial values of the deformation parameters, the updated deformation parameter time series is obtained. Simultaneously, the cofactor matrix of the updated parameters is used for the next iteration. The updated cofactor matrix is represented in blocks as follows: The update formula for each block matrix is as follows: Finally, the updated deformation parameter time series (i.e., containing) and The vector is used to determine the latest deformation of the slope. With the continuous accumulation of SAR data on high-speed railways, the above formula can be used to update the deformation parameters in near real-time, thereby obtaining continuous deformation information of the slope structure along the high-speed railway.
[0118] Please see Figure 3 , Figure 3 This illustration shows a structural schematic diagram of a slope deformation sequential monitoring device based on synthetic aperture radar mounted on a high-speed railway, according to an embodiment of this application. The slope deformation sequential monitoring device 200 based on synthetic aperture radar mounted on a high-speed railway includes: a generation module 210, a construction module 220, a correction module 230, and a determination module 240. Specifically: The generation module 210 is used to acquire raw echo data through synthetic aperture radar and generate single-view complex images based on the raw echo data. Module 220 is used to perform image registration on single-view complex images and perform differential interferometric processing on the registered single-view complex images to construct a differential interferometric dataset. The correction module 230 is used to construct a coherence matrix based on the differential interferometric dataset, identify and correct error components based on spatial independent component analysis, and perform phase optimization on the differential interferometric dataset using the coherence matrix to obtain the corrected differential interferometric dataset. The determination module 240 is used to perform phase unwrapping based on the corrected differential interferometric dataset and to perform incremental update calculation on the newly added echo data based on the sequential estimation algorithm to obtain the deformation of the slope.
[0119] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described device and module can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.
[0120] In the several embodiments provided in this application, the coupling or direct coupling or communication connection between the modules shown or discussed may be an indirect coupling or communication connection through some interface, device or module, and may be electrical, mechanical or other forms.
[0121] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.
[0122] Please see Figure 4 , Figure 4 This illustration shows a structural schematic diagram of a slope deformation sequential monitoring device based on synthetic aperture radar mounted on a high-speed railway, according to an embodiment of this application. The slope deformation sequential monitoring device 300 based on synthetic aperture radar mounted on a high-speed railway in this application may include one or more of the following components: a processor 310, a memory 320, and one or more application programs. The one or more application programs may be stored in the memory 320 and configured to be executed by one or more processors 310. The one or more programs are configured to execute the slope deformation sequential monitoring method based on synthetic aperture radar mounted on a high-speed railway as described in the foregoing method embodiments.
[0123] The processor 310 may include one or more processing cores. The processor 310 connects to various parts within the slope deformation sequential monitoring device 300 based on synthetic aperture radar mounted on a high-speed railway using various interfaces and lines. It executes various functions and processes data of the slope deformation sequential monitoring device 300 based on synthetic aperture radar mounted on a high-speed railway by running or executing instructions, programs, code sets, or instruction sets stored in memory 320, and by calling data stored in memory 320. Optionally, the processor 310 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 310 may integrate one or more of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the displayed content; and the modem is used for wireless communication. It is understandable that the aforementioned modem may not be integrated into the processor 310, but may be implemented using a separate communication chip.
[0124] The memory 320 may include random access memory (RAM) or read-only memory (ROM). The memory 320 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 320 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for implementing at least one function, instructions for implementing the various method embodiments described below, etc. The data storage area may also store data created during use by the slope deformation sequential monitoring device 300 based on synthetic aperture radar mounted on a high-speed railway.
[0125] Please see Figure 5 , Figure 5 The diagram shows a computer-readable storage medium 400 provided in an embodiment of this application. The computer-readable storage medium 400 stores program code, which can be called by a processor to execute the slope deformation sequential monitoring method based on synthetic aperture radar mounted on a high-speed railway as described in the above method embodiment.
[0126] The computer-readable storage medium 400 may be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read-Only Memory), EPROM, hard disk, or ROM. Optionally, the computer-readable storage medium 400 includes a non-transitory computer-readable storage medium. The computer-readable storage medium 400 has storage space for program code 410 that performs any of the method steps described above. This program code can be read from or written to one or more computer program devices. The program code 410 may be compressed, for example, in a suitable form.
[0127] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.
Claims
1. A method for sequential monitoring of slope deformation based on synthetic aperture radar mounted on a high-speed railway, characterized in that, The method includes: The raw echo data is acquired through the synthetic aperture radar, and a single-view complex image is generated based on the raw echo data. Image registration is performed on the single-view complex images, and differential interferometric processing is performed on the registered single-view complex images to construct a differential interferometric dataset; A coherence matrix is constructed based on the differential interferometric dataset, and error components are identified and corrected based on spatial independent component analysis. The phase of the differential interferometric dataset is then optimized using the coherence matrix to obtain the corrected differential interferometric dataset. Phase unwrapping is performed based on the corrected differential interferometric dataset, and incremental update calculations are performed on the newly added echo data based on the sequential estimation algorithm to obtain the deformation of the slope.
2. The method for sequential monitoring of slope deformation based on synthetic aperture radar mounted on a high-speed railway, as described in claim 1, is characterized in that... The steps involve performing differential interferometry on the registered single-view complex images to construct a differential interferometry dataset, including: Based on the registered single-view complex images at different observation times, the complex phase difference between the first observation time and the second observation time is calculated; The complex phase difference is used as an element in the differential interferometric dataset; The differential interferometric phase satisfies a linear superposition model consisting of terrain phase, deformation phase, orbital error phase, atmospheric phase, and noise phase.
3. The method for sequential monitoring of slope deformation based on synthetic aperture radar mounted on a high-speed railway, as described in claim 1, is characterized in that... The step of constructing a coherence matrix based on the differential interferometric dataset includes: Arrange the interferograms in the differential interferometric dataset according to spatial pixels to construct an observation matrix with interferograms as rows and spatial pixels as columns; The observation matrix is decomposed using a spatially independent component analysis algorithm to extract mutually independent component signals; Based on the spatial and temporal correlation characteristics of each component signal, error components representing atmospheric phase and orbital error phase are identified and screened from the component signals. The coherence matrix is constructed based on the differential interferometric dataset after removing the error components.
4. The method for sequential monitoring of slope deformation based on synthetic aperture radar mounted on a high-speed railway, as described in claim 1 or 3, is characterized in that... The step involves performing phase optimization on the differential interferometric dataset using the coherence matrix to obtain a corrected differential interferometric dataset, including: Based on the statistical characteristics of the differential interferometric dataset, candidate points of distributed scatterers are identified and selected, and a set of homogeneous pixels in the neighborhood of the candidate points of distributed scatterers is determined. Using the time-series data of the homogeneous pixel set, calculate the sample coherence matrix corresponding to the candidate points of the distributed scatterer; Based on the coherence matrix, a regularized maximum likelihood objective function is constructed for the coherence matrix to be estimated, and a coherence matrix regularization constraint term is introduced into the objective function to form an optimization model. The optimal coherence matrix estimate is obtained by solving the optimization model by maximizing the likelihood function of the observed data. The phase of the candidate points of the distributed scatterer is re-estimated using the optimal coherence matrix estimate to obtain the corrected differential interferometric dataset containing the optimized phase.
5. The method for sequential monitoring of slope deformation based on synthetic aperture radar mounted on a high-speed railway, as described in claim 1, is characterized in that... The method further includes: Based on the corrected differential interferometric dataset, an initial deformation inversion model is established; The initial deformation inversion model is solved by least squares estimation to obtain the initial values of the deformation parameters and the first cofactor matrix. When a new single-view complex image is acquired, a new differential interferometric dataset is constructed based on the new single-view complex image and the existing single-view complex image according to a preset interferometric combination relationship; Based on the newly added differential interferometric dataset and the initial values of the deformation parameters, a sequential estimation model is constructed; Using the sequential estimation model, and combining the weight matrix of the first cofactor matrix with the observation equation corresponding to the newly added differential interferometric dataset, the deformation parameter update increment is calculated; Based on the deformation parameter update increment and the initial value of the deformation parameter, the updated deformation parameter time series is obtained; the updated deformation parameter time series is determined as the deformation of the slope.
6. The method for sequential monitoring of slope deformation based on synthetic aperture radar mounted on a high-speed railway, as described in claim 1 or 2, is characterized in that... The complex form data of each pixel in the single-view complex image is as follows: in, Let be the real part of the complex number. The imaginary part of a complex number. For amplitude, The virtual part, It is a complex phase.
7. The method for sequential monitoring of slope deformation based on synthetic aperture radar mounted on a high-speed railway, as described in claim 2, is characterized in that... The linear superposition model is as follows: in, The terrain phase, The deformation phase, The orbital error phase, The atmospheric phase, The noise phase is described.
8. A sequential monitoring device for slope deformation based on synthetic aperture radar mounted on a high-speed railway, characterized in that, The device includes: The generation module is used to acquire raw echo data through the synthetic aperture radar and generate single-view complex images based on the raw echo data. A construction module is used to perform image registration on the single-view complex image and perform differential interferometric processing on the registered single-view complex image to construct a differential interferometric dataset. The correction module is used to construct a coherence matrix based on the differential interferometric dataset, identify and correct error components based on spatial independent component analysis, and perform phase optimization on the differential interferometric dataset using the coherence matrix to obtain the corrected differential interferometric dataset. The determination module is used to perform phase unwrapping based on the corrected differential interferometric dataset and to perform incremental update calculation on the newly added echo data based on the sequential estimation algorithm to obtain the deformation of the slope.
9. A sequential monitoring device for slope deformation based on synthetic aperture radar mounted on a high-speed railway, characterized in that, include: One or more processors; Memory; One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, the one or more applications being configured to perform the sequential monitoring method for slope deformation based on synthetic aperture radar mounted on high-speed rail as described in any one of claims 1-7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program code, which can be called by a processor to execute the sequential monitoring method for slope deformation based on synthetic aperture radar mounted on a high-speed railway as described in any one of claims 1-7.