Photovoltaic field deformation monitoring method fusing Stacking-InSAR and SBAS-InSAR technologies

Abstract

The invention discloses a photovoltaic field area deformation monitoring method integrating a Stacking-InSAR (Interferometric Synthetic Aperture Radar) technology and an SBAS-InSAR (Spatial Binary Area System-InSAR) technology. The method comprises the following steps: firstly, processing multi-temporal SAR image data by adopting a Stacking-InSAR technology, and acquiring wide-area deformation information (such as linear average deformation rate, deformation distribution and scale) of an area in a set time period; and secondly, in a process of processing the multi-temporal SAR image data by adopting an SBAS-InSAR technology, introducing the wide-area deformation information obtained in the step (a) into a time sequence deformation inversion step by taking the wide-area deformation information as a prior model so as to obtain earth surface time sequence deformation data of the area. According to the method, the wide-area identification capability of the Stacking-InSAR and the time sequence analysis capability of the SBAS-InSAR are fused, so that the technical problem that a single SBAS-InSAR technology fails in a complex mountainous area due to serious spatial-temporal decoherence is solved.

Description

Technical Field

[0001] This invention relates to the field of geological disaster monitoring technology, and more specifically, to a method for monitoring surface deformation in complex environments using synthetic aperture radar interferometry (InSAR) technology, and particularly to a method for monitoring deformation in photovoltaic field areas that integrates Stacking-InSAR and SBAS-InSAR technologies. Background Technology

[0002] Infrastructure construction projects in mountainous areas, such as agricultural photovoltaic fields, power transmission networks, and other critical engineering projects, are vulnerable to threats from complex geological conditions. Particularly in areas affected by underground mining activities, the presence of mined-out areas can trigger large-scale surface movement and subsidence, posing a serious threat to the stability of above-ground structures. Therefore, wide-area, long-term, and detailed surface deformation monitoring in these areas is crucial for disaster early warning and facility safety maintenance.

[0003] Synthetic Aperture Radar Interferometry (InSAR) technology has become an important means of monitoring land surface deformation due to its advantages of all-weather operation, high spatiotemporal resolution, high accuracy, and wide coverage. In this field, temporal InSAR technology, especially Small Baseline Set InSAR (SBAS-InSAR), is widely used to extract temporal series deformation information of the land surface. SBAS-InSAR, by constructing combinations of small baseline interferometric pairs, can effectively analyze nonlinear deformation processes and demonstrates high-precision monitoring capabilities in urban areas and exposed surface regions.

[0004] However, existing SBAS-InSAR technology has serious limitations when applied to specific complex environments. It is well known in the art that InSAR signals are highly susceptible to interference when the monitored area is located in mountainous terrain with significant topographic relief, especially when it is covered by dense vegetation (such as in southwestern China). As relevant research (e.g., a paper in Frontiers in Earth Science) has pointed out, vegetation growth and movement, as well as geometric distortions caused by complex terrain (such as shadows and overlays), can cause radar signals to lose correlation between multiple observations, resulting in a severe spatiotemporal "decoherence" problem.

[0005] This lack of coherence causes SBAS-InSAR technology to fail to identify enough effective coherence points in these critical areas (such as vegetated hillsides), resulting in large data gaps in the monitoring results. For example, Xue et al. (2022), in their study published in the journal *Remote Sensing*, titled "Research on the Applicability of DInSAR, Stacking-InSAR and SBAS-InSAR for Mining Region Subsidence Detection in the Datong Coal field," clearly pointed out that in "densely vegetated mountainous areas" like the Datong coalfield, the number of deformed areas detected by SBAS-InSAR was "fewer" than those detected by Stacking-InSAR. Consequently, SBAS-InSAR technology "fails" in these low-coherence areas, failing to acquire complete deformation information and thus unable to accurately assess the extent and spatiotemporal evolution of geological hazards.

[0006] Another InSAR method known in this field is Stacking-InSAR (interferogram stacking) (e.g., Wu et al. (2025) discusses its application in Geological and Mineral Mapping). This technique effectively suppresses atmospheric delay and noise by weighted averaging or stacking multiple differential interferograms. Stacking-InSAR has relatively high tolerance for decoherence and is better at identifying the overall extent and morphology of deformation in low-coherence regions, and estimating the linear average deformation rate.

[0007] However, an inherent limitation of Stacking-InSAR is that it typically only provides an average linear rate over the entire time span, unlike SBAS-InSAR which offers detailed, nonlinear temporal deformation processes (such as acceleration or deceleration phases). The aforementioned study by Xue et al. (2022) confirms this, concluding that SBAS-InSAR (in high-coherence regions) is "more accurate" and provides "temporally cumulative deformation," while Stacking-InSAR's advantage lies in "detecting the extent and morphology of the deformed region."

[0008] In summary, existing technologies (such as those revealed by Xue et al. (2022)) face an unresolved technical challenge: on the one hand, SBAS-InSAR technology, which can provide detailed temporal deformation information, fails in densely vegetated mountainous areas due to severe decoherence; on the other hand, Stacking-InSAR technology, which has a higher tolerance for decoherence, cannot provide crucial temporal evolution information.

[0009] Therefore, there is an urgent need in this field for a new technical method that can overcome the serious incoherence problem caused by complex terrain and vegetation cover, and accurately extract complete temporal deformation data of the land surface. Summary of the Invention

[0010] The primary objective of this invention is to overcome the deficiencies in the aforementioned background technology, namely, to address the technical problem that existing SBAS-InSAR technology fails due to severe decoherence in areas with complex terrain and vegetation cover, thus failing to obtain complete temporal deformation information, and to provide a photovoltaic field deformation monitoring method that integrates Stacking-InSAR and SBAS-InSAR technologies.

[0011] To achieve the aforementioned first objective, the present invention adopts the following technical solution:

[0012] A method for monitoring deformation in photovoltaic fields that integrates Stacking-InSAR and SBAS-InSAR technologies, characterized by the following steps:

[0013] (a) Using Stacking-InSAR technology, multi-temporal SAR image data is processed to obtain wide-area deformation information of the region within a set time period;

[0014] (b) In the process of processing the multi-temporal SAR image data using SBAS-InSAR technology, the wide-area deformation information obtained in step (a) is used as a priori model and introduced into the temporal deformation inversion step to obtain the temporal deformation data of the region.

[0015] Preferably, the wide-area deformation information mentioned in step (a) includes the linear average deformation rate of the region.

[0016] Preferably, the wide-area deformation information mentioned in step (a) also includes the distribution and scale of the deformation region obtained by phase change diagram analysis.

[0017] Preferably, in step (b), the wide-area deformation information (especially the linear average deformation rate) is modeled as a linear deformation component and introduced as a priori constraint into the time-series deformation inversion step.

[0018] Preferably, the method integrates the wide-area deformation information obtained by Stacking-InSAR technology as a priori constraint term for temporal deformation inversion in SBAS-InSAR technology, thereby effectively reducing the spatiotemporal incoherence caused by the complex terrain or vegetation cover.

[0019] Preferably, the area in question is the area affected by underground mining.

[0020] Preferably, the area affected by underground mining is a mountainous coal mine or an agricultural photovoltaic field.

[0021] Preferably, the method further includes converting the radar line-of-sight (LOS) temporal surface deformation data obtained in step (b) into vertical deformation data.

[0022] Furthermore, the method also includes step (c): (c) for the temporal deformation data of the land surface obtained in step (b), a superimposed dynamic Logistic model is used to perform nonlinear fitting on it, so as to invert and deconstruct the physical parameters of multiple independent subsidence events in the region.

[0023] Preferably, the mathematical expression of the superimposed dynamic Logistic model is: ;

[0024] in: It is a time variable; for The total cumulative deformation calculated by the time-matter model; K represents the number of superpositions of the multiple independent settlement events, where K is an integer greater than or equal to 2. For the first An independent settlement event in The cumulative deformation variable contributed at each moment; For the first The maximum cumulative settlement contributed by an independent settlement event; For the first Rate coefficients of individual settlement events; For the first The inflection point time of an independent settlement event, wherein the inflection point time is the [number]th [event]. The moment when the settlement rate of an independent settlement event reaches its peak.

[0025] Preferably, step (c) includes: (c1) identifying the number K of velocity peaks on the deformation rate curve based on the first derivative (i.e., deformation rate) curve of the temporal deformation data of the land surface; (c2) when the number K of velocity peaks is greater than or equal to 2, using the nonlinear least squares method to iteratively optimize and solve the physical parameters in the superimposed dynamic Logistic model. The goal of the iterative optimization is to minimize the The sum of squared residuals between the data and the temporal deformation data of the land surface.

[0026] Preferably, the nonlinear least squares method is the Levenberg-Marquardt algorithm.

[0027] Preferably, the inversion and deconstruction are used to: quantitatively separate the maximum cumulative settlement contributed by each of the independent settlement events caused by different mining faces or different mining periods. and their respective inflection point times .

[0028] Beneficial effects

[0029] Compared with existing technologies, the "Stacking-InSAR and SBAS-InSAR fusion monitoring technology" method proposed in this invention has the following outstanding substantive features and significant beneficial effects:

[0030] 1. As is known in existing technologies, SBAS-InSAR technology fails to identify sufficient effective coherence points in areas with complex terrain and vegetation cover due to severe spatiotemporal incoherence, leading to unstable or "failed" temporal deformation inversion solutions. This invention addresses this by fusing a workflow: step (a) uses Stacking-InSAR technology to obtain a wide-area linear rate, and in step (b) introduces this rate as a priori model into the SBAS temporal deformation inversion process as a strong constraint term for the linear deformation component. This method effectively avoids the effects of spatiotemporal incoherence and successfully acquires crucial deformation data in low-coherence areas deemed "failed" or "unsuitable" by existing technologies. This invention overcomes the technical challenge of severe spatiotemporal incoherence and solves the "failure" problem of SBAS-InSAR application in complex mountainous areas.

[0031] 2. In existing technologies, Stacking-InSAR and SBAS-InSAR are used in parallel for comparison or as a preferred alternative. It is generally understood by those skilled in the art that Stacking-InSAR excels at identifying the "range and morphology of deformed regions" in low-coherence areas but cannot provide temporal data; while SBAS-InSAR can provide "temporally cumulative deformation," but is limited to "high-coherence areas." Existing technologies teach those skilled in the art to abandon the use of SBAS-InSAR in low-coherence areas (such as densely vegetated mountainous regions). This invention takes the opposite approach, using the wide-area rate results of Stacking-InSAR as a constraint for SBAS-InSAR temporal deformation inversion, achieving a synergistic superposition of the advantages of both technologies, resulting in an unexpected technical effect of "1+1>2." This invention not only identifies wide-area deformation like Stacking-InSAR (as in the embodiments)... Figure 6 and Figure 7 As shown), it also successfully extracted fine temporal processes, similar to SBAS-InSAR (e.g. Figure 8 and Figure 9As shown in the specific embodiment, this method clearly reproduces the complete temporal evolution process of the monitoring point (such as JC5) from the "initial deformation stage" to the "first deformation intensification stage" and then to the "second deformation intensification stage". This ability to obtain "relatively complete deformation information" in the low coherence region is something that cannot be achieved by a single technology or simple combination in the prior art. The present invention achieves unexpected technical effects, namely, simultaneously realizing wide-area identification and complete temporal deformation process analysis in the low coherence region;

[0032] 3. This invention addresses the industry pain point of monitoring surface deformation in complex terrain and densely vegetated areas such as mountainous coal mines and agricultural photovoltaic fields. By overcoming the problem of incoherence and acquiring complete time-series data, this invention can accurately control the spatiotemporal deformation evolution process of surface subsidence basins under the influence of mining, providing a far more reliable and comprehensive scientific basis than existing technologies for the safe operation of important infrastructure (such as booster stations) in mountainous areas, geological disaster early warning, and risk assessment. This invention significantly improves the applicability of monitoring technology and the reliability of monitoring results;

[0033] 4. The deformation deconstruction method based on the Superimposed Dynamic Logistic Model (SDLM) proposed in this invention, based on the aforementioned beneficial effects, can successfully deconstruct complex superimposed signals into two or more independent settlement events with clear physical meaning, and quantitatively deduce the maximum settlement contribution of each event. and peak rate time ; Solve the output data (such as Figure 8 This is a complex phenomenon resulting from a combination of multiple physical events (such as two sampling events), making it difficult to explain. It represents a leap from "phenomenon monitoring" to "physical cause inversion." Attached Figure Description

[0034] Figure 1 This is a schematic diagram of a typical deformation area of ​​a photovoltaic power station in May 2024, according to the present invention.

[0035] Figure 2 This is a schematic diagram of a typical deformation area of ​​a photovoltaic power station in December 2024, according to the present invention.

[0036] Figure 3 This is a flowchart illustrating the fusion monitoring process of Stacking-InSAR and SBAS-InSAR technologies in this invention.

[0037] Figure 4 This is a schematic diagram of the relative differential interferometry solution results for 20240107-20240119 of the present invention;

[0038] Figure 5 This is a Stacking-InSAR identification result map of surface deformation in the study area from 2022 to 2024, based on the findings of this invention.

[0039] Figure 6 This is a Stacking-InSAR identification result map of surface deformation in agricultural photovoltaic field areas from 2022 to 2024, based on the present invention.

[0040] Figure 7 This is a time-series conversion monitoring result of surface deformation SBAS-InSAR in the study area from 2022 to 2024, based on the present invention.

[0041] Figure 8 This is a time-series conversion monitoring result of surface deformation in agricultural photovoltaic field areas from 2022 to 2024, based on the present invention.

[0042] Figure 9 This is a schematic diagram of the monitoring profile and monitoring point layout of the present invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0044] Example 1: Surface Deformation Monitoring Based on Stacking-InSAR and SBAS-InSAR Fusion

[0045] This embodiment takes a mountainous agricultural photovoltaic field in Zhenfeng County, Qianxinan Buyi and Miao Autonomous Prefecture, Guizhou Province, China as an example. The field is located within an underground coal mining area.

[0046] 1.1 Implementation Background (Monitoring Objectives and Environment)

[0047] The study area is a typical mountainous agricultural photovoltaic field. The terrain is very complex, with steep slopes and dense vegetation cover. This environment is a typical scenario that leads to severe spatiotemporal decoherence in traditional temporal InSAR techniques (especially SBAS-InSAR), as described in the background section.

[0048] Meanwhile, the photovoltaic power plant area has been continuously affected by underground coal mining activities. On-site investigations show that since August 2023, the area (especially the substation) has begun to show significant deformation, such as ground cracks in concrete roads, cracks in building walls, outward tilting of retaining walls, and even partial collapses. Figure 1 and Figure 2 (As shown).

[0049] The objective of this embodiment is to accurately acquire temporal deformation data of the surface of agricultural photovoltaic fields caused by underground mining in the aforementioned mountainous environment with high vegetation cover, complex terrain, and susceptibility to incoherence.

[0050] 1.2 Data Preparation

[0051] This embodiment selected Sentinel-1A up-orbit imagery data covering the study area, spanning from January 5, 2022 to November 26, 2024, totaling 80 scenes. Specific data information is shown in Table 1.

[0052] Table 1. Data information for the Sentinel-1A study area (2022-2024, 80 scenes in total)

[0053]

[0054] 1.3 Technical Implementation Process

[0055] The technical process of this embodiment is one of the key steps of the present invention, such as... Figure 3 As shown. This process is the specific implementation of the "fusion" method. The specific execution steps are as follows:

[0056] (1) Data preprocessing: Acquire time-series SAR image data (80 scenes as shown in Table 1) and external DEM data. Select the master image and register the DEM and all SAR images with the master image. Crop, georegister, and (if necessary) stitch strips on the images. Calculate the interferometric phase based on the registered data and use the DEM data to calculate the elevation phase and coherence coefficient.

[0057] (2) Differential interferometry calculations are performed using differential stacking to generate multi-temporal differential interferograms (e.g., Figure 4 (Example shown). Flat phase removal and terrain phase removal are performed on the differential interferogram (using the elevation phase from step (1)).

[0058] (3) Implementation of step (a): Acquisition of Stacking-InSAR wide-area deformation information, such as Figure 3 As shown in the "Stacking-InSAR Technology" module, linear deformation calculation is performed on the differential interferogram produced in step (2). This step obtains wide-area deformation information of the study area during the entire period from 2022 to 2024 by performing differential stacking calculations on differential interferograms from multiple time phases. The wide-area deformation information is specifically manifested as follows:

[0059] • The linear average deformation rate of the study area, such as Figure 5 and Figure 6 As shown.

[0060] • The distribution and scale of the deformation region obtained through phase change diagram analysis, i.e. Figure 3 The image shows "Extracting the influence range of surface moving basins".

[0061] (4) Implementation of step (b): SBAS-InSAR time series transformation extraction based on the results of step (a)

[0062] Fusion basis: The key to this step is to integrate the wide-area deformation information obtained in step (3) (i.e., the linear average deformation rate identified by Stacking-InSAR). ,like Figure 5 , Figure 6 This serves as basic or prior information to guide and optimize SBAS-InSAR processing.

[0063] Timing processing and inversion: Specifically, in the execution Figure 3 When detailing the process of the "SBAS-InSAR technology" module, the linear average deformation rate obtained in step (3) will be used. This is introduced as a priori model into the temporal deformation inversion step of SBAS-InSAR.

[0064] Overcoming decoherence: In this field, time series inversion (e.g., through SVD singular value decomposition) aims to separate deformation, atmospheric, and elevation errors from a series of interferometric phases. The core of this invention lies in separating the total temporal deformation phase ( Modeled as linear and nonlinear components ( ), and utilize step (a) The result is as follows Strong constraints or known terms. This allows the inversion algorithm to solve the nonlinear deformation component more accurately. Even in low-coherence regions (caused by vegetation or complex terrain), the system overcomes the problems of rank deficiency or unstable solutions in the inversion matrix caused by the lack of high-quality coherence points, and achieves accurate and complete temporal deformation monitoring results for the study area. Figure 3 ("Temporal deformation data of the Earth's surface").

[0065] (5) Deformation data processing and analysis

[0066] Coordinate transformation: Correspondence Figure 3 The "geocoding" step encodes the deformation information in the radar coordinate system into the WGS-84 coordinate system.

[0067] Vertical transformation: corresponding Figure 3 The "vertical deformation transformation" step in the process. The radar line-of-sight (LOS) temporal surface deformation data obtained in step (4) will be transformed into... Using radar side-looking observation geometry, it is converted to vertical ( )deformation.

[0068] 1.4 Implementation Results and Effects

[0069] The fusion monitoring method in this embodiment successfully acquired complete time-series deformation data of the photovoltaic field (especially the booster station), such as... Figure 7 and Figure 8 As shown. For quantitative analysis, nine time-series monitoring points (JC1 to JC9) were set up in the substation area, as follows. Figure 9 As shown in Table 2, the monitoring results indicate that during the period from 2022 to 2024, the location of the maximum settlement within the substation area was point JC5, with a maximum cumulative surface settlement of -100.627 mm.

[0070] Table 2. Cumulative Surface Subsidence over Three Years at Each Monitoring Point of the Substation

[0071]

[0072] More importantly, this method successfully extracts temporal evolution processes that cannot be obtained by the single Stacking-InSAR technique. For example... Figure 7 As shown, the spatiotemporal evolution of surface deformation at the booster station is clearly divided into three stages:

[0073] • Initial deformation phase (June 5, 2023 to November 20, 2023);

[0074] • The first phase of intensified deformation (November 20, 2023 to August 22, 2024);

[0075] • The second phase of intensified deformation (August 22, 2024 to November 26, 2024).

[0076] 1.5 Conclusion of Example 1

[0077] This specific embodiment 1 demonstrates that the fusion monitoring method proposed in this invention, by using the wide-area identification results of Stacking-InSAR as a priori constraint for SBAS-InSAR temporal deformation inversion, successfully solves the problem of spatiotemporal decoherence of InSAR in areas with high vegetation cover and complex terrain, such as mountainous areas in Guizhou. It obtains complete temporal deformation data that traditional SBAS-InSAR technology cannot acquire, accurately reproduces the evolution process of surface subsidence under the influence of mining, and verifies the inventiveness and practicality of this invention.

[0078] Specific Implementation Example 2: Deconstruction and Inversion of Settlement Events Based on Superimposed Dynamic Logistic Model (SDLM)

[0079] This embodiment 2 details the further improved technical solution of the present invention based on embodiment 1.

[0080] 2.1 Technical problem to be solved in this embodiment

[0081] As described in the invention summary, specific embodiment 1 successfully solved the decoherence problem and obtained the following: Figure 7 And the precise, complex nonlinear time-series deformation data shown in Table 2.

[0082] However, further research by the inventors revealed that the deformation data obtained in Example 1 (such as the time-series curve of point JC5) exhibited a complex multi-stage accelerated settlement characteristic. Section 1.4 of Example 1 has already qualitatively described it as the "first stage of intensified deformation" and the "second stage of intensified deformation".

[0083] The physical essence of this phenomenon is that the monitoring point (JC5) was simultaneously or sequentially subjected to multiple independent sampling events (such as...). Figure 3 The total deformation (-100.627 mm) observed is a linear superposition of the deformations of these independent events, caused by the successive advancement of different coal mining faces in 2023 and 2024.

[0084] If standard time series models known in the field are used, such as a single Logistic time function model or the Knothe function, it will be impossible to fit this complex "step-like" S-curve, because these standard models all assume that the deformation originates from a single physical process.

[0085] Therefore, although Example 1 obtained monitoring data, how to quantitatively invert and deconstruct the physical parameters of each independent mining event from these superimposed data (e.g., quantitatively separating the settlement contributed by the first mining and the second mining) is a technical problem that Example 1 has not yet solved.

[0086] 2.2 Technical Solution of This Embodiment

[0087] To address the aforementioned technical problems, this invention further proposes a deformation deconstruction method based on the superimposed dynamic logistic model (SDLM), building upon Example 1.

[0088] The core of this method lies in establishing a mathematical model. The model is based on a physical assumption: the observation point Total cumulative vertical deformation at time 1 ,yes An independent settlement event following a Logistic function (S-shaped curve). Linear superposition.

[0089] The mathematical expression for this model is: —(Formula 1)

[0090] Among them, the indivual( Independent Settlement Events The mathematical model in this invention is described using the Logistic time function: —(Formula 2)

[0091] The physical meanings of the symbols in Formulas 1 and 2 above are explained as follows: Time variables (e.g., the number of days since the start of monitoring (January 5, 2022)). The model is in Total cumulative deformation calculated at time step (unit: mm). The number of consecutive independent settlement events (integer, ... This parameter needs to be determined based on the data generated in Example 1. : No. An independent settlement event in The cumulative deformation variable contributed at each moment. : No. The maximum cumulative subsidence (in mm) ultimately contributed by each independent subsidence event. This is one of the key physical parameters that needs to be inverted, representing the total impact of the mining event on the surface. : No. Rate coefficient of an independent settlement event (unit: 1 / day). This parameter reflects the severity of settlement; the higher the value, the faster the settlement. : No. The inflection point time (in days) of an independent settlement event. This is the mathematical inflection point of the Logistic function, which physically corresponds to the moment when the settlement rate of the event reaches its peak, representing the main impact time of the event.

[0092] 2.3 Detailed execution steps of this embodiment (taking point JC5 as an example)

[0093] This embodiment uses the example from embodiment 1. Figure 9 The JC5 monitoring point (which has the largest cumulative settlement, as shown in Table 2) is used as the application object. Eighty time-series deformation data points (denoted as JC5) from January 5, 2022 to November 26, 2024 are analyzed. , Deconstruct it.

[0094] Step (c): Perform SDLM fitting and inversion

[0095] (c1) Step (c1): Determine the number of superpositions K. First, extract the temporal deformation data of monitoring point JC5 in Example 1. .right Numerical differentiation (e.g., using the finite difference method) is performed to obtain the deformation rate curve. Through analysis Curve (its trend can be seen from) Figure 7 Qualitative analysis revealed two distinct rate peaks near November 20, 2023 (“first intensification”) and August 22, 2024 (“second intensification”). These two peaks physically correspond to the impacts of two different underground mining events in 2023 and 2024, respectively. Therefore, this embodiment determines the superimposed number of independent settlement events at this monitoring point. .

[0096] Step (c2): Model building and parameter solving. Decision: Due to the determination in step (c1)... Satisfying The conditions. Execution: Therefore, this embodiment performs a superimposed dynamic Logistic model fitting. When When, the specific form of formula (1) is: —(Formula 3)

[0097] The goal of this embodiment is to solve the model. One unknown physical parameter: To solve for these parameters, this embodiment employs a nonlinear least squares method. The goal of this algorithm is to find a set of optimal parameters that allow the model to predict values... Compared with the observations in Example 1 Sum of squared residuals between To reach the minimum: —(Formula 4)

[0098] In this embodiment, the specific algorithm used is the Levenberg-Marquardt (LM) algorithm. As is known to those skilled in the art, the LM algorithm is a standard iterative algorithm for solving nonlinear least squares problems, which roughly estimates the peak time by setting reasonable initial parameter values ​​(e.g., based on the peak time determined in (c1)). and This allows for rapid convergence to obtain the optimal solution for the six parameters.

[0099] 2.4 Results and Analysis of Example 2

[0100] JC5's 80 data points ( The physical parameter results obtained by inputting the above process are shown in Table 3.

[0101] Table 3. SDLM (K=2) model deconstruction parameters for monitoring point JC5

[0102]

[0103] Results Analysis: The results in Table 3 have clear physical meaning, indicating that the method of this embodiment successfully quantitatively separated the total settlement (approximately -100.5 mm) observed in Example 1 into two independent physical events:

[0104] Event 1 (i=1): Triggered by mining activities in 2023. This event contributed -38.5 mm of settlement, with the maximum settlement rate occurring around December 15, 2023. This coincides perfectly with the "first stage of intensified deformation" (starting November 20, 2023) identified in the qualitative analysis of Example 1 (Section 1.4).

[0105] Event 2 (i=2): Triggered by mining activities in 2024. This event contributed -62.0 mm of settlement, with a rate coefficient... Greater than This indicates that the second settlement was more severe. The maximum settlement rate occurred around September 10, 2024. This coincides perfectly with the "second deformation intensification phase" (starting on August 22, 2024) identified in the qualitative analysis of Example 1 (Section 1.4).

[0106] • Overall fit: The total settlement contributed by the two events is (-38.5) + (-62.0) = -100.5 mm, which is basically consistent with the observed total settlement of -100.627 mm for JC5 in Table 2. Overall goodness of fit The value reached 0.992, indicating that the model (Equation 3) is effective for the observed data ( The fit is extremely high.

[0107] 2.6 Conclusion of Example 2

[0108] In summary, specific embodiment 2 provides a key capability that specific embodiment 1 does not possess:

[0109] It successfully deconstructs the complex superimposed time series data with ambiguous physical meaning produced in Example 1 by introducing a superimposed dynamic logistic model (SDLM) with non-obvious modifications (a creative improvement on the known logistic model).

[0110] It quantitatively separated the contributions of different mining events (Event 1 and Event 2) to the total settlement. and ) and key impact time ( and ).

[0111] This not only solves the technical problem of "interpretability" of the monitoring data in Example 1, but also greatly enhances the application value of the present invention in tracing the causes of disasters, defining engineering responsibilities, and predicting future trends, resulting in unexpected and significant technological progress.

[0112] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for photovoltaic field area deformation monitoring by fusing Stacking-InSAR and SBAS-InSAR techniques, characterized in that, The method comprises the following steps: (a) acquiring wide-area deformation information of the region in a set period of time by processing multi-temporal SAR image data using a Stacking-InSAR technique; (b) introducing the wide-area deformation information obtained in step (a) as a prior model into a time-series deformation inversion step in a process of processing the multi-temporal SAR image data using an SBAS-InSAR technique, to obtain surface time-series deformation data of the region.

2. The method of claim 1, wherein, The wide-area deformation information in step (a) comprises a linear average deformation rate of the region; the wide-area deformation information in step (a) further comprises distribution and scale of deformation regions obtained by phase change map analysis; in step (b), the wide-area deformation information is modeled as a linear deformation part and introduced as a prior constraint term into the time-series deformation inversion step.

3. The method of claim 1, wherein, The method fuses the wide-area deformation information obtained by the Stacking-InSAR technique as a prior constraint term of time-series deformation inversion in the SBAS-InSAR technique.

4. The method of claim 1, wherein, The region is an underground mining influence region, and the underground mining influence region is a mountain coal mine or an agricultural photovoltaic field region.

5. The method of claim 1, wherein, The method further comprises converting radar line-of-sight surface time-series deformation data obtained in step (b) into vertical deformation data.

6. The method of claim 1, wherein, The method further comprises step (c): (c) performing nonlinear fitting on the surface time-series deformation data obtained in step (b) using a superposed dynamic Logistic model, to invert and deconstruct physical parameters of multiple independent subsidence events of the region.

7. The method of claim 6, wherein, The mathematical expression of the superposed dynamic Logistic model is: ； wherein: is a time variable; is the total cumulative deformation variable calculated by the time model; is the superposition of the plurality of independent settlement events, K is an integer greater than or equal to 2; is the cumulative deformation variable contributed by the th independent settlement event at time ; is the maximum cumulative settlement contributed by the th independent settlement event; is the rate coefficient of the th independent settlement event; is the inflection time of the th independent settlement event, which is the time when the settlement rate of the th independent settlement event reaches a peak.

8. The method of claim 7, wherein, The step (c) comprises: (c1) identifying the number K of velocity peaks on the deformation rate curve based on a first derivative curve of the surface time-series deformation data; (c2) when the number of velocity peaks K is greater than or equal to 2, solving the physical parameters in the superposed dynamic Logistic model by iterative optimization using a non-linear least squares method , the objective of the iterative optimization being to minimize the sum of squared residuals between the and the surface time series deformation data.

9. The method of claim 8, wherein, The nonlinear least squares method is a Levenberg-Marquardt algorithm.

10. The method according to any one of claims 6 to 9, characterized in that, The inversion and deconstruction are used to quantitatively separate the respective contributions of the individual subsidence events caused by different mining faces or different mining periods to the maximum cumulative subsidence and the respective inflection times .

Images