A method and system for deformation monitoring of pumped storage power station GBSAR, BDS and LiDAR multi-source cooperation

Abstract

The application discloses a kind of pumped storage power station GBSAR, BDS and LiDAR multi-source coordination's deformation monitoring method and system, belong to hydraulic and hydroelectric engineering deformation monitoring technical field.Method includes: demarcate monitoring subarea and lay out BDS reference station, BDS key monitoring point, GBSAR permanent observation site and unmanned aerial vehicle LiDAR surveying and mapping control network;Synchronous acquisition GBSAR surface area data, BDS three kinds of data and LiDAR point cloud data;Respectively carry out data preprocessing;After time-space registration, multi-source data fusion is solved using three-level fusion algorithm;To the three-dimensional displacement field after fusion, deformation analysis and grading early warning are carried out;Achievement output and visualization.The application considers BDS, and GBSAR station stability monitoring and key position monitoring are considered, introduce real-time displacement correction quantity to improve GBSAR precision, realize seamless three-dimensional displacement field output by three-level fusion, can be adapted to pumped storage power station full working condition, precision is better than 0.1mm, and interference resistance is strong.

Description

Technical Field

[0001] This invention relates to the field of deformation monitoring technology for water conservancy and hydropower projects, specifically to a multi-source collaborative deformation monitoring method and system for pumped storage power stations, which combines ground-based synthetic aperture radar (GBSAR), BeiDou satellite navigation system (BDS), and UAV-borne lidar (LiDAR). It is applicable to full-area, real-time, and high-precision deformation monitoring of key components such as the dam body, reservoir basin, and surrounding slopes of pumped storage power stations. Background Technology

[0002] As the core regulating hub of a new power system, pumped storage power stations' key structures (dam body, reservoir basin, reservoir bank slopes, etc.) are constantly subjected to complex conditions of alternating water storage, power generation, and pumping, bearing multiple loads such as water pressure and geological stress. These conditions make them prone to deformation, landslides, and other safety hazards, seriously threatening the safe and stable operation of the power station. Among these, key areas such as the dam abutment, steep slopes, and surge tank inlets are at higher risk of deformation and require more precise monitoring. Furthermore, the stability of GBSAR stations is easily affected by environmental factors, making it difficult to guarantee the validity of their observation data. Simply removing invalid data cannot fully utilize monitoring resources, and there is still room for improvement in monitoring accuracy. Therefore, implementing comprehensive, real-time, and high-precision monitoring of the key structures of pumped storage power stations, especially strengthening monitoring of key areas, and achieving real-time stability correction of GBSAR stations, is a core requirement for ensuring the safe operation and maintenance of the power station.

[0003] Currently, deformation monitoring of pumped storage power stations mainly employs single monitoring technologies or dual-source fusion technologies, which have many limitations: traditional contact monitoring methods such as leveling and total station measurements are inefficient, labor-intensive, and unable to achieve large-scale area monitoring; while single GBSAR monitoring technology can achieve non-contact area monitoring, it can only acquire line-of-sight displacement and cannot directly obtain three-dimensional real displacement. Moreover, the stability of GBSAR stations is easily affected by environmental factors, and existing technologies mostly only remove invalid observation data without introducing real-time displacement correction, thus limiting monitoring accuracy; single BDS monitoring technology can provide high-precision three-dimensional absolute displacement, but deploying individual BDS measuring points increases monitoring costs and construction difficulty, and it can only achieve discrete point monitoring, failing to cover the entire area deformation and not specifically enhancing monitoring of key areas; UAV LiDAR technology can acquire large-scale terrain point clouds, but its temporal deformation monitoring capability is weak, and its monitoring accuracy is easily affected by environmental interference.

[0004] Existing multi-source data fusion technologies are mostly limited to dual-source fusion (such as GBSAR and BDS, UAV LiDAR and BDS), and often use BDS with separate measurement points to achieve absolute benchmark monitoring. This not only increases the complexity of the monitoring system but also fails to fully utilize the monitoring value of BDS for GBSAR station stability, does not introduce real-time displacement correction to improve GBSAR monitoring accuracy, and does not deploy dedicated BDS monitoring points for key areas, thus failing to meet the high-precision monitoring needs of these areas. Furthermore, existing fusion algorithms are relatively simple and fail to fully leverage the complementary advantages of multi-source data, resulting in insufficient fusion accuracy, weak anti-interference capabilities, and poor spatiotemporal adaptability, making them unsuitable for the high-precision monitoring requirements of pumped storage power stations under complex operating conditions. In addition, existing technologies are not optimized for the alternating water storage-power generation-pumping operation characteristics of pumped storage power stations, and cannot effectively eliminate elastic deformation interference caused by water level changes, leading to decreased deformation analysis accuracy and difficulty in accurately identifying potential safety hazards.

[0005] With the continuous development of remote sensing monitoring technology, multimodal sensor fusion has become a development trend in the field of engineering deformation monitoring. Among them, GBSAR, BDS, and UAV LiDAR each have their advantages. If BDS can be used simultaneously for GBSAR station stability monitoring (with real-time displacement correction) and key area monitoring, optimizing the monitoring layout, achieving efficient collaboration and deep integration of the three technologies, and optimizing the fusion algorithm in combination with the operating characteristics of pumped storage power stations, the shortcomings of existing monitoring technologies can be effectively solved. While ensuring full-area monitoring, the monitoring accuracy of key areas and GBSAR observation accuracy can be improved, and monitoring costs and construction difficulties can be reduced. Therefore, developing a multi-source collaborative deformation monitoring method and system that is suitable for pumped storage power stations, integrates BDS with GBSAR station stability monitoring (real-time displacement correction) and key area monitoring, and combines large-scale high-precision monitoring with real-time monitoring of key areas, has significant engineering application value and practical significance. Summary of the Invention

[0006] To address the aforementioned problems in existing technologies, this invention provides a multi-source collaborative deformation monitoring method and system for pumped storage power stations, utilizing GBSAR, BDS, and LiDAR. The aim is to solve the technical problems of incomplete coverage, insufficient accuracy, low efficiency of multi-source data fusion, weak anti-interference capabilities, and the lack of BDS monitoring points for key areas and the absence of real-time displacement correction for GBSAR station stability, which limits monitoring accuracy. This invention achieves comprehensive, real-time, and high-precision monitoring of key structures in pumped storage power stations, particularly strengthening monitoring of key areas and improving GBSAR observation accuracy, thus providing reliable technical support for the safe operation and maintenance of power stations.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: a deformation monitoring method for a pumped storage power station using a multi-source coordinated approach of ground-based synthetic aperture radar (GBSAR), BeiDou satellite navigation system (BDS), and UAV-borne lidar (LiDAR), comprising the following steps: Step 1: Preliminary preparation and scheme design, delineating key monitoring zones for the pumped storage power station. These zones include the dam body, reservoir bank slopes, inlet and outlet slopes, surge tanks, and road slopes. A BDS base station, key BDS monitoring points, permanent GBSAR observation positions, and a UAV LiDAR aerial survey control network are established. The BDS base station is used for stability monitoring of GBSAR stations, and key BDS monitoring points are located at key locations within the power station (dam shoulders, steep slopes, surge tank openings, etc.) for high-precision monitoring. The observation frequency is determined: daily or weekly observations are used for routine monitoring, hourly intensified observations are used during the flood season / water storage period, and real-time observations are used for extreme conditions.

[0008] Step 2: Multi-source data synchronous acquisition, specifically including: conducting continuous non-contact surface monitoring using GBSAR to acquire millimeter-level two-dimensional surface deformation time-series data and line-of-sight (LOS) displacement in the monitoring area; synchronously acquiring three types of data using BDS: reference coordinate data of the BDS base station, three-dimensional coordinates and time-series displacement data of key BDS monitoring points (for high-precision key monitoring of key areas), and three-dimensional coordinates and time-series displacement data of the GBSAR station (for monitoring the stability of the GBSAR station itself and providing data support for displacement correction); and conducting periodic aerial surveys using UAV LiDAR to acquire high-density point cloud and DEM / DSM data of the monitoring area, capturing large-scale terrain deformation, landslide boundaries, and micro-topographic changes.

[0009] Step 3: Multi-source data preprocessing, specifically including: denoising, atmospheric phase compensation, temporal interferometry calculation, and flat-earth effect removal for GBSAR data; cycle slip detection and repair, tropospheric / ionospheric correction, and relative positioning calculation for BDS data, outputting GBSAR station stability assessment results, real-time displacement data of GBSAR stations (used to calculate displacement correction), and high-precision displacement data of key BDS monitoring points. If the GBSAR station displacement exceeds a preset threshold, the GBSAR observation data for the corresponding time period is removed; if it does not exceed the threshold, the real-time displacement correction is calculated based on the real-time displacement data of the GBSAR station; point cloud denoising, filtering, registration, coordinate unification, and temporal DEM differential processing are performed on UAV LiDAR data.

[0010] Step 4: Spatiotemporal registration. The preprocessed GBSAR data, BDS data (including BDS base station data, BDS key monitoring point data, and GBSAR station displacement data), and UAV LiDAR data are unified to an engineering independent coordinate system or the 2000 National Geodetic Coordinate System. Lagrange interpolation or Kriging interpolation methods are used to unify the temporal sampling rate of different data sources and establish a common timestamp. Combining the GBSAR station stability data and real-time displacement correction obtained from BDS monitoring, and using the BDS base station as a reference, real-time displacement correction is performed on the GBSAR observation data to complete the unification of multi-source data benchmarks. Simultaneously, the BDS key monitoring point data is used as an independent calibration data source to ensure the monitoring accuracy of key areas.

[0011] Step 5: Multi-source data fusion and solution, using a three-level fusion algorithm, specifically including: 5.1 Spatiotemporal coarse fusion: The coordinate system of multi-source data is unified through seven-parameter transformation, and time synchronization is achieved through temporal interpolation. Combined with the GBSAR station stability assessment results and real-time displacement corrections output by BDS, the GBSAR observation data is corrected and invalid GBSAR data is removed to eliminate coordinate and time deviations. At the same time, the data of key monitoring points of BDS are incorporated for preliminary calibration. 5.2 Medium-precision fusion: Based on GBSAR multi-track / multi-angle line-of-sight displacement data (after displacement correction), the true three-dimensional displacement field of GBSAR is inverted using the least squares algorithm; a Kalman filter algorithm is adopted, with GBSAR data as the core, combined with GBSAR station stability parameters and real-time displacement corrections monitored by BDS for noise suppression, to achieve high-frequency continuous deformation data output; an IGGⅢ weight function is introduced, and a robust least squares fusion algorithm is adopted to eliminate gross errors caused by local instability, while the fusion results are calibrated by combining BDS key monitoring point data; 5.3 High-precision deep fusion: Based on the UAV LiDAR high-precision DEM, terrain phase is calculated, and terrain flattening and overlay correction are performed on the displacement-corrected GBSAR data; a two-layer "surface-volume" data fusion model is constructed, and the co-kriging algorithm is adopted. The displacement-corrected GBSAR continuous two-dimensional deformation field is used as the core, the LiDAR terrain gradient is used as the covariate, and the BDS key monitoring point data is used as the constraint condition to achieve seamless three-dimensional displacement field fusion across the entire domain; a temporal differential fusion algorithm is adopted to perform optimal fusion of multi-period UAV LiDAR DEM differential data, displacement-corrected GBSAR temporal deformation data, and BDS key monitoring point temporal data, and adaptively adjust the weights of each data source.

[0012] Step 6: Deformation Analysis and Early Warning. The fused 3D displacement field, deformation rate, and acceleration data, combined with high-precision displacement data from key BDS monitoring points, are used to perform trend analysis, rate analysis, periodic analysis, and abrupt change point identification. A "water level-deformation" response model is established based on the "water storage-power generation-pumping" operating conditions of the pumped storage power station, eliminating elastic deformation caused by water level changes. Based on the deformation data characteristics and changes in data from key BDS monitoring points, tiered early warnings are implemented, and anomaly tracing and safety assessments are conducted.

[0013] Step 7: Output and visualization of results. Generate monitoring reports, deformation curves, displacement cloud maps, and spatiotemporal heat maps. The monitoring report should include GBSAR station stability monitoring results, GBSAR station real-time displacement correction data, and BDS key monitoring point monitoring results. Build a web-based monitoring platform to support real-time viewing, historical backtracking, and automatic report generation of deformation data, GBSAR station stability data, displacement correction data, and BDS key monitoring point data.

[0014] Furthermore, the seven-parameter transformation model described in step 5.1 is as follows:

[0015] in, To transform the three-dimensional coordinate matrix, The transformed 3D coordinate matrix, Here is the translation parameter matrix, and m is the scale parameter. Let ω, φ, and κ be the rotation angles about the X, Y, and Z axes, respectively; Rotation matrix The matrix form is as follows:

[0016] Furthermore, the GBSAR line-of-sight displacement inversion formula and the multi-track combined line-of-sight displacement inversion formula mentioned in step 5.2 are as follows: The formula for single-track line-of-sight displacement inversion is:

[0017] in, This represents the line-of-sight displacement of GBSAR after real-time displacement correction. θ is the matrix of three-dimensional displacement components (eastward, northward, and vertical) of the GBSAR station, where θ is the radar incident angle and α is the radar azimuth angle. Line-of-sight displacement inversion formula for multiple tracks:

[0018] Where n is the number of GBSAR observation orbits, Here are the line-of-sight displacement matrices for each orbit. For each orbital radar incident angle, The azimuth angles of each orbital radar.

[0019] Furthermore, the state vector, state equation, observation equation, and state transition matrix of the Kalman filter described in step 5.2 are as follows: The state vector is:

[0020] Where X, Y, and Z are the three-dimensional coordinates of the target point in the monitoring area. For three-dimensional velocity, For three-dimensional acceleration, the stability parameters of GBSAR stations monitored by BDS and the real-time displacement correction are introduced during the filtering process to dynamically adjust the filtering coefficients; The state equation is:

[0021]

[0022] in, Here is the state transition matrix. Let T be the process noise matrix, and T be the sampling time interval.

[0023] The observation equation is:

[0024] This is the observation vector matrix (containing GBSAR data and BDS key monitoring point data). For the observation matrix, This is the observation noise matrix.

[0025] Meanwhile, the IGGⅢ weight function formula mentioned in step 5.2 is:

[0026] in The weight of the i-th observation. To observe the residuals, The error is the unit weight. The coefficients of the weight function are usually taken as... , .

[0027] Furthermore, the collaborative financial formula described in step 5.3 is as follows: Synergistic Kryptonian Cooperative Formula:

[0028] in, The deformation value after fusion. This is GBSAR surface deformation data after displacement correction. This is LiDAR terrain gradient data. For BDS key monitoring point displacement data, , , These are the fusion weights for the corresponding data sources.

[0029] The weights must satisfy the following constraints:

[0030] LiDAR terrain gradient data The calculation formula is:

[0031] Where h is the LiDAR DEM elevation value, and x and y are the plane coordinates.

[0032] Furthermore, the formula and weights for the optimal fusion of mean squared errors described in step 5.3 are as follows: Optimal fusion formula for mean square error:

[0033] in, The optimal deformation value after fusion. This is the original deformation data from GBSAR. This represents the real-time displacement correction for GBSAR stations. Deformation data for LiDAR on drones, For BDS key monitoring point displacement data, This is an adaptive weight matrix for each data source. The weights are adjusted in real time based on the accuracy, signal-to-noise ratio, GBSAR station stability monitoring results, and the accuracy of BDS key monitoring point data for each data source. The weights satisfy the matrix constraints:

[0034] And the weights are calculated as follows:

[0035] in , , The root mean square errors are BDS key monitoring point data, GBSAR data after displacement correction, and LiDAR data, respectively. Meanwhile, the terrain phase calculation formula described in step 5.3 is as follows:

[0036] in, Where λ represents the terrain phase and λ is the GBSAR radar wavelength. θ represents the terrain elevation difference, and θ is the radar incident angle.

[0037] GBSAR phase value after terrain phase correction:

[0038] in, For GBSAR raw phase, This is the atmospheric phase error (the residual error after preprocessing compensation).

[0039] Furthermore, the graded early warning system described in step 6 is divided into three levels based on deformation rate, acceleration, abrupt change characteristics, and displacement changes at key BDS monitoring points: yellow warning (general), orange warning (relatively severe), and red warning (serious). Different levels correspond to different handling procedures. The quantitative judgment formula for graded early warning is as follows: Yellow alert (general):

[0040] Orange alert (severe):

[0041] Red Alert (Severe):

[0042] Where v is the deformation rate and a is the deformation acceleration. This represents the displacement change of key BDS monitoring points within 24 hours.

[0043] As described in step 6, the water level-deformation response model is:

[0044] Where D is the elastic deformation caused by the water level, and H is the reservoir water level. These are the model fitting coefficients. This represents the model residuals.

[0045] Mutation point identification uses the sliding window method, and its determination formula is as follows:

[0046] in, The deformation values ​​at each time point are: Let w be the average deformation value of all windows, and w be the size of the sliding window. This represents the standard deviation of the deformation data within the corresponding window.

[0047] A multi-source collaborative deformation monitoring system for pumped storage power stations, incorporating GBSAR, BDS, and LiDAR, is used to implement the aforementioned monitoring methods. The system includes: Data acquisition module: includes GBSAR observation equipment, BDS reference station, BDS monitoring terminal, and UAV LiDAR aerial survey equipment, used to simultaneously acquire surface deformation data, GBSAR station displacement data (for calculating displacement correction), BDS key monitoring point displacement data, and terrain point cloud data. Among them, the BDS reference station provides a benchmark for the stability of GBSAR stations and BDS monitoring points.

[0048] Data preprocessing module: Connected to GBSAR observation equipment, BDS reference station, BDS monitoring terminal, and UAV LiDAR aerial survey equipment respectively, it is used to perform noise reduction, correction, calculation and standardization processing on the source data. Among them, the preprocessing of BDS data is used to output GBSAR station stability assessment results, GBSAR station real-time displacement data (and calculate real-time displacement correction) and BDS key monitoring point displacement data.

[0049] GBSAR data denoising employs an adaptive filtering algorithm:

[0050] in, This is the filtered data. The original data, Here, N is the filter coefficient matrix, and N is the radius of the filter window. BDS data cycle slip detection uses the Turbo Edit algorithm, and its cycle slip determination formula is as follows:

[0051] in, The carrier phase difference matrix at each time step. This represents the standard deviation of the phase difference.

[0052] Spatiotemporal registration module: Connected to the data preprocessing module, it is used to realize coordinate system unification, time synchronization and benchmark unification of multi-source data. Combining the GBSAR station stability assessment results and real-time displacement correction, it performs real-time displacement correction on GBSAR observation data. At the same time, it uses BDS key monitoring point data as calibration basis to complete the benchmark unification of multi-source data.

[0053] The time-series interpolation uses the Lagrange interpolation formula:

[0054] in, This is the interpolated time series data matrix. is the original time series data matrix, and n is the number of interpolation nodes.

[0055] The GBSAR real-time displacement correction formula is:

[0056] in, This is the corrected GBSAR data matrix. This is the original GBSAR data matrix. For the real-time displacement correction matrix ( , This is the displacement matrix of the GBSAR station at each time point. (This is the initial reference displacement).

[0057] Multi-source fusion module: Connected to the spatiotemporal registration module, it has a built-in three-level fusion algorithm to complete coarse fusion, medium-precision fusion and high-precision deep fusion of multi-source data. During the fusion process, it incorporates GBSAR station stability parameters, real-time displacement correction and BDS key monitoring point data, outputs three-dimensional displacement field and deformation characteristic parameters, removes invalid data and improves the monitoring accuracy of key parts.

[0058] Deformation Analysis and Early Warning Module: Connected to the multi-source fusion module, it is used to analyze the fused deformation data and BDS key monitoring point data, identify abrupt change points, and combine working condition characteristics to achieve hierarchical early warning and anomaly tracing.

[0059] Results Output and Visualization Module: Connected to the Deformation Analysis and Early Warning Module, it is used to generate monitoring results (including GBSAR station stability monitoring results, GBSAR station real-time displacement correction data, and BDS key monitoring point monitoring results), and realizes visualization display and data query through the Web platform.

[0060] Furthermore, the multi-source fusion module also includes an operating condition adaptive weight adjustment unit, which is used to adjust the weight of each data source according to different operating conditions of the pumped storage power station: increase the GBSAR weight during the water storage period, increase the UAV LiDAR weight during the slope inspection period, increase the weight of BDS key monitoring point data during the key part monitoring period, and dynamically adjust the weight in combination with the GBSAR station stability monitoring results and real-time displacement correction.

[0061] The formula for adaptive weight adjustment based on operating conditions is in matrix form:

[0062] in, The adjusted weight matrix, The initial weight matrix, The working condition weighting coefficient matrix (water storage period) Slope inspection period Key monitoring periods ), GBSAR station stability correction factor ( , (This is the maximum threshold for GBSAR station displacement).

[0063] Furthermore, the GBSAR observation equipment in the data acquisition module is deployed on multiple tracks to achieve multi-angle line-of-sight displacement acquisition and improve the accuracy of three-dimensional displacement inversion; the BDS reference station is deployed in a stable area and works in conjunction with the GBSAR station to achieve stability monitoring and displacement correction calculation; the BDS key monitoring terminal is deployed at key locations such as dam shoulders, steep slopes, and surge tank wellheads, and uses high-precision BDS receiving equipment with a sampling rate consistent with the GBSAR sampling rate to ensure data synchronization; the UAV LiDAR aerial survey equipment uses periodic aerial surveys, preferably once a month or quarter.

[0064] Formula for calculating LiDAR point cloud density for drones:

[0065] in, denoted as point cloud density (points / square meter), f as laser emission frequency (Hz), swath as flight strip width (m), overlap as flight strip overlap (%), v as flight speed (m / s), and h as flight altitude (m).

[0066] Compared with the prior art, the present invention has the following beneficial effects: 1. Comprehensive monitoring coverage with a focus on key areas, achieving a collaborative model of "full-area monitoring + high-precision monitoring of key parts": Using GBSAR as the core of high-density area monitoring, it realizes non-contact continuous monitoring of key structures of pumped storage power stations, covering all key areas such as dams and slopes; BDS key monitoring points are deployed at key parts such as dam shoulders and steep slopes to achieve millimeter-level high-precision monitoring of key parts. At the same time, BDS also takes into account the stability monitoring of GBSAR stations, eliminating the need to deploy redundant BDS monitoring points, reducing monitoring costs and construction difficulty.

[0067] 2. Introducing real-time displacement correction for GBSAR stations significantly improves monitoring accuracy: By monitoring the stability of GBSAR stations in real time using BDS, the real-time displacement correction is calculated and corrected in real time for GBSAR observation data. This eliminates the impact of GBSAR station deformation on the observation results, solving the problem that existing technologies only remove invalid data and cannot fully utilize monitoring resources. This improves GBSAR monitoring accuracy to the 0.1mm level. At the same time, combined with BDS data calibration of key monitoring points, the monitoring accuracy of key areas is further guaranteed.

[0068] 3. Three-level fusion algorithm optimization to fully leverage the complementary advantages of three-source data: A three-level fusion strategy of "spatiotemporal coarse fusion - medium-precision fusion - high-precision deep fusion" is adopted to gradually eliminate spatiotemporal bias, suppress noise, and remove gross errors from multi-source data. Combined with algorithms such as co-kriging and mean square error optimal fusion, deep fusion of GBSAR, BDS, and LiDAR data is achieved, outputting a seamless, high-precision three-dimensional displacement field across the entire domain. The fusion accuracy is superior to existing dual-source fusion technology, and the anti-interference capability is stronger.

[0069] 4. Adaptable to all operating conditions of pumped storage power stations, resulting in more accurate deformation analysis: Combining the alternating operating conditions of pumped storage power stations—water storage, power generation, and pumping—a water level-deformation response model is established to eliminate elastic deformation caused by water level changes and accurately extract the true plastic deformation of the structure. At the same time, the weights of each data source are adaptively adjusted according to different operating conditions. During the water storage period, the weight of GBSAR is increased; during the slope inspection period, the weight of UAV LiDAR is increased; and during key monitoring periods, the weight of BDS key monitoring point data is increased to ensure that the monitoring accuracy meets the requirements under different operating conditions.

[0070] 5. Timely early warning and accurate source tracing provide reliable support for the safe operation and maintenance of power plants: Through the mutation point identification and graded early warning mechanism, potential deformation risks can be quickly identified. Different levels of early warning correspond to clear handling procedures, which facilitates managers to take timely prevention and control measures. Anomaly source tracing, combined with multi-source monitoring data, operating condition data and geological data, can accurately analyze the causes of abnormal deformation and provide a scientific basis for hidden danger management. The web-based monitoring platform enables real-time visualization and sharing of monitoring results, improving the efficiency of monitoring work. Attached Figure Description

[0071] Figure 1 This is a flowchart illustrating the deformation monitoring method of GBSAR, BDS and LiDAR multi-source collaboration for pumped storage power stations according to the present invention. Figure 2 This is a flowchart illustrating the three-level fusion algorithm of the present invention; Figure 3 This is a schematic diagram of the module structure of the multi-source collaborative deformation monitoring system of GBSAR, BDS and LiDAR for pumped storage power stations according to the present invention. Detailed Implementation

[0072] 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. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0073] like Figure 1-2 As shown, the deformation monitoring method for pumped storage power stations using multi-source synergy of GBSAR, BDS, and LiDAR of the present invention includes the following steps: Step 1: Preliminary preparation and scheme design. Delineate key monitoring zones for the pumped storage power station, including the dam body, reservoir bank slopes, inlet and outlet slopes, surge tanks, and road slopes. Deploy BDS base stations, key BDS monitoring points, permanent GBSAR observation positions, and a UAV LiDAR aerial survey control network. The BDS base stations are used for stability monitoring of GBSAR stations. Key BDS monitoring points are deployed at key locations within the power station (dam shoulders, steep slopes, surge tank openings, etc.) for high-precision monitoring of these key areas. Redundant BDS monitoring points are not deployed. Determine the observation frequency: daily or weekly observations for routine monitoring, hourly intensified observations during the flood season / water storage period, and real-time observations for extreme conditions.

[0074] In this step, the delineation of monitoring zones must be based on the engineering layout and safety risk level of the pumped storage power station. High-risk areas (such as steep slopes, dam crests, and dam shoulders) should be given priority coverage. Key BDS monitoring points should be prioritized for these high-risk locations, and their placement should avoid obstructed areas to ensure the reliability of monitoring data. Permanent GBSAR observation sites should be selected in locations with open and unobstructed views to ensure coverage of important monitoring zones and the surrounding areas of key BDS monitoring points. GBSAR stations should maintain linkage with the BDS base station to ensure the acquisition of real-time displacement data, providing accurate data for correcting monitoring results. The UAV LiDAR aerial survey control network must be linked with the BDS base station to ensure coordinate consistency. The BDS base station should be located in a stable area, far from the deformation-affected zone. Step 2: Multi-source data synchronous acquisition, specifically including: conducting continuous non-contact surface monitoring using GBSAR to acquire millimeter-level two-dimensional surface deformation time-series data and line-of-sight (LOS) displacement in the monitoring area; synchronously acquiring three types of data using BDS: reference coordinate data of the BDS base station, three-dimensional coordinates and time-series displacement data of key BDS monitoring points (for high-precision key monitoring of key areas), and three-dimensional coordinates and time-series displacement data of the GBSAR station (for monitoring the stability of the GBSAR station itself and providing data support for displacement correction); and conducting periodic aerial surveys using UAV LiDAR to acquire high-density point cloud and DEM / DSM data of the monitoring area, capturing large-scale terrain deformation, landslide boundaries, and micro-topographic changes.

[0075] In this step, multi-source data acquisition must be synchronized to ensure that the observation times of each data source are consistent or aligned through interpolation methods. GBSAR observations employ temporal interferometry (PS-InSAR / SBAS) to improve the accuracy of surface deformation monitoring. High-precision BDS receiving equipment is used at key monitoring points, with a sampling rate matching the GBSAR sampling rate to ensure temporal synchronization of displacement data at key locations with GBSAR data, for subsequent fusion calibration. BDS sampling of GBSAR stations and key locations is performed simultaneously, ensuring both GBSAR station stability monitoring and accurate monitoring of key locations, thus improving monitoring efficiency. UAV LiDAR aerial surveys require reasonable determination of flight altitude and flight strip overlap based on the size of the monitoring area and accuracy requirements to acquire high-density point cloud data (point cloud density not less than 50 points / square meter). The point cloud density calculation formula is as follows: Point cloud density calculation formula:

[0076] in, denoted as point cloud density (points / square meter), f as laser emission frequency (Hz), swath as flight strip width (m), overlap as flight strip overlap (%), v as flight speed (m / s), and h as flight altitude (m).

[0077] Step 3: Multi-source data preprocessing, specifically including: denoising, atmospheric phase compensation, temporal interferometry calculation, and flat-earth effect removal for GBSAR data; cycle slip detection and repair, tropospheric / ionospheric correction, and precise single-point or relative positioning calculation for BDS data; outputting GBSAR station stability assessment results, real-time displacement data of GBSAR stations (used to calculate displacement correction), and high-precision displacement data of key BDS monitoring points. If the GBSAR station displacement exceeds a preset threshold (the preset threshold is set according to engineering accuracy requirements, generally not exceeding 0.5 mm / d), the GBSAR observation data for the corresponding time period is removed. If it does not exceed the threshold, the real-time displacement correction is calculated based on the real-time displacement data of the GBSAR station. The displacement correction is calculated using the difference method, that is, the real-time displacement correction is equal to the difference between the current displacement of the GBSAR station and the initial reference displacement. The calculation formula is as follows: Point cloud denoising, filtering, registration, coordinate unification, and time-series DEM differential processing are performed on UAV LiDAR data.

[0078] In this step, an adaptive filtering algorithm is used for GBSAR data denoising:

[0079] in, This is the filtered data. The original data, Here, N is the filter coefficient matrix, and N is the radius of the filter window. Atmospheric phase compensation uses meteorological data for auxiliary correction, and the reference point method is used to eliminate the flatland effect; BDS data cycle slip detection uses the Turbo Edit algorithm, and its cycle slip determination formula is as follows:

[0080] in, The carrier phase difference matrix at each time step. This represents the standard deviation of the phase difference.

[0081] Tropospheric correction uses the Saastamoinen model, and ionospheric correction uses a dual-frequency carrier phase combination method to ensure the accuracy of BDS data processing. UAV LiDAR data denoising adopts a combination of statistical filtering and morphological filtering to remove noise points and outliers. After filtering, multi-period point cloud registration is achieved through the ICP registration algorithm, and the coordinates are unified to the same coordinate system as BDS and GBSAR. Temporal DEM differential processing is used to extract large-scale terrain deformation information to provide terrain constraints for subsequent deep fusion.

[0082] Step 4: Spatiotemporal registration. The preprocessed GBSAR data, BDS data (including BDS key monitoring point data, GBSAR station displacement data, and BDS base station data), and UAV LiDAR data are unified to an engineering independent coordinate system or the 2000 National Geodetic Coordinate System. Lagrange interpolation or Kriging interpolation methods are used to unify the temporal sampling rate of different data sources and establish a common timestamp. Combining the GBSAR station stability data and real-time displacement correction obtained from BDS monitoring, and using the BDS base station as a reference, real-time displacement correction is performed on the GBSAR observation data to complete the unification of multi-source data benchmarks. Simultaneously, the BDS key monitoring point data is used as an independent calibration data source to ensure the monitoring accuracy of key areas.

[0083] In this step, the coordinate system is unified using the seven-parameter transformation method. Seven parameters (three translation parameters, three rotation parameters, and one scale parameter) are solved using the common points of the BDS base station and the UAV LiDAR aerial survey control network. The seven-parameter transformation formula is as follows:

[0084] in, To transform the three-dimensional coordinate matrix, The transformed 3D coordinate matrix, Here is the translation parameter matrix, and m is the scale parameter. Let ω, φ, and κ be the rotation angles about the X, Y, and Z axes, respectively; Rotation matrix for:

[0085] During the process of unifying the time-series sampling rate, for GBSAR high-frequency sampling data (minute-level), BDS synchronous sampling data, and UAV LiDAR periodic sampling data (daily / weekly / monthly), Lagrange interpolation was used to fill in missing data and unify the sampling interval. The Lagrange interpolation calculation formula is as follows:

[0086] in, This is the interpolated time series data matrix. is the original time series data matrix, and n is the number of interpolation nodes.

[0087] Ensure that all data sources correspond and match at the same timestamp; GBSAR real-time displacement correction is based on GBSAR station displacement data monitored by BDS, and the real-time displacement correction is calculated by the difference method. The GBSAR line-of-sight displacement data is corrected hourly to eliminate the influence of GBSAR station deformation on the observation results. After correction, the BDS base station is used as the absolute reference to achieve the unification of multi-source data reference; BDS key monitoring point data is used as an independent calibration data source and compared with the corrected GBSAR data. If the deviation exceeds the preset accuracy threshold (generally not exceeding 0.3mm), the registration parameters are readjusted to ensure the consistency and accuracy of monitoring data in key areas.

[0088] Step 5: Multi-source data fusion and solution, using a three-level fusion algorithm, specifically including: 5.1 Spatiotemporal coarse fusion: The coordinate system of multi-source data is unified through seven-parameter transformation, and time synchronization is achieved through temporal interpolation. Combined with the GBSAR station stability assessment results and real-time displacement corrections output by BDS, the GBSAR observation data is corrected and invalid GBSAR data is removed to eliminate coordinate and time deviations. At the same time, the data of key monitoring points of BDS are incorporated for preliminary calibration. The core of this stage is to eliminate spatiotemporal biases in multi-source data, laying the foundation for subsequent high-precision fusion. The criteria for removing invalid GBSAR data are: data from observation periods where GBSAR station displacement exceeds a preset threshold (0.5 mm / d); data with excessive residual error (greater than 0.2 mm) after atmospheric phase compensation; and abnormal data caused by obstruction or interference. The preliminary calibration of key BDS monitoring point data mainly involves comparing discrete high-precision displacement data of key locations with GBSAR area data, marking GBSAR observation areas with significant deviations, and providing a key correction basis for medium-precision fusion.

[0089] 5.2 Medium-precision fusion: Based on GBSAR multi-track / multi-angle line-of-sight displacement data (after displacement correction), the true three-dimensional displacement field of GBSAR is inverted using the least squares algorithm; a Kalman filter algorithm is adopted, with GBSAR data as the core, combined with GBSAR station stability parameters and real-time displacement corrections monitored by BDS for noise suppression, to achieve high-frequency continuous deformation data output; an IGGⅢ weight function is introduced, and a robust least squares fusion algorithm is adopted to eliminate gross errors caused by local instability, while the fusion results are calibrated by combining BDS key monitoring point data; This phase focuses on noise suppression and accuracy improvement of GBSAR area data, while using BDS key monitoring point data for calibration. Specifically, when inverting the 3D displacement field using multi-track GBSAR line-of-sight displacement, at least two mutually perpendicular observation tracks are deployed. The 3D displacement components (eastward, northward, and vertical) are solved using a least-squares algorithm by simultaneously establishing the multi-track line-of-sight displacement equations. The multi-track simultaneous line-of-sight displacement inversion formula is as follows:

[0090] Where n is the number of GBSAR observation orbits, Here are the line-of-sight displacement matrices for each orbit. For each orbital radar incident angle, The azimuth angles of each orbital radar.

[0091] In the Kalman filter algorithm, GBSAR three-dimensional displacement, velocity, and acceleration are used as state vectors. Combined with the GBSAR station stability parameters monitored by BDS and the real-time displacement correction, the filter coefficients are dynamically adjusted to suppress observation noise and output high-frequency continuous (minute-level) three-dimensional deformation data. The core formula of the Kalman filter is as follows: The state vector is:

[0092] Where X, Y, and Z are the three-dimensional coordinates of the target point in the monitoring area. For three-dimensional velocity, For three-dimensional acceleration, the stability parameters of GBSAR stations monitored by BDS and the real-time displacement correction are introduced during the filtering process to dynamically adjust the filtering coefficients; The state equation is:

[0093]

[0094] in, Here is the state transition matrix. Let T be the process noise matrix, and T be the sampling time interval.

[0095] The observation equation is:

[0096] This is the observation vector matrix (containing GBSAR data and BDS key monitoring point data). For the observation matrix, This is the observation noise matrix.

[0097] The IGGⅢ weight function is used to suppress gross errors caused by local instability (such as local slope collapse or temporary equipment disturbance). It assigns smaller weights to observations with large residuals, or even eliminates them, to ensure the robustness of the fusion results. The calculation formula is as follows:

[0098] in The weight of the i-th observation. To observe the residuals, The error is the unit weight. The coefficients of the weight function are usually taken as... , .

[0099] The BDS key monitoring point data calibration adopts a point-by-point comparison method to correct the three-dimensional displacement data of key parts retrieved by GBSAR, thereby further improving the monitoring accuracy of key parts.

[0100] 5.3 High-precision deep fusion: Based on the UAV LiDAR high-precision DEM, terrain phase is calculated, and GBSAR data (after displacement correction) is subjected to terrain flattening and overlay correction; a surface-volume two-layer data fusion model is constructed, and the co-kriging algorithm is adopted, with the GBSAR continuous two-dimensional deformation field (after displacement correction) as the core and the LiDAR terrain gradient as the covariate, combined with the BDS key monitoring point data as the constraint condition, to achieve seamless three-dimensional displacement field fusion across the entire domain; a temporal differential fusion algorithm is adopted to perform optimal fusion of multi-period UAV LiDAR DEM differential data, GBSAR temporal deformation data (after displacement correction), and BDS key monitoring point temporal data, and adaptively adjust the weights of each data source.

[0101] This stage achieves deep fusion of three-source data, outputting a seamless, high-precision 3D displacement field across the entire domain. Specifically, terrain phase calculation is based on a high-precision UAV LiDAR DEM (elevation accuracy better than 0.1m). Terrain phase is calculated using terrain elevation differences and radar incident angles, eliminating terrain interference from GBSAR data and correcting observation biases in overlapping areas (such as areas obscured by steep slopes). The terrain phase calculation formula is as follows:

[0102] in, The terrain phase matrix for all points in the entire region. For GBSAR radar wavelength, Elevation matrix of all points in the global domain obtained from the UAV LiDAR DEM. This is the incident angle of the GBSAR radar.

[0103] In the two-layer fusion model of surface and volume, the "surface" layer consists of continuous surface deformation data from GBSAR, and the "volume" layer consists of UAV LiDAR terrain data. The co-kriging algorithm utilizes the LiDAR terrain gradient to reflect the impact of terrain changes on deformation. Combined with the high-precision constraints of BDS key monitoring point data, it achieves seamless stitching and accuracy improvement of the entire deformation data. The co-kriging fusion formula is as follows:

[0104] in, The deformation value after fusion. This is GBSAR surface deformation data after displacement correction. This is LiDAR terrain gradient data. For BDS key monitoring point displacement data, , , These are the fusion weights for the corresponding data sources.

[0105] The root mean square error optimal fusion adaptively adjusts the weights by calculating the root mean square error of each data source. Its formula and matrix form are as follows:

[0106] in, , ( (Root mean square error of each data source).

[0107] Among them, the BDS key monitoring point data (root mean square error better than 0.05mm) has the highest weight, followed by the displacement-corrected GBSAR data (root mean square error better than 0.1mm), and the UAV LiDAR data (root mean square error better than 0.2mm) has its weight dynamically adjusted according to the terrain complexity to ensure that the root mean square error of the fusion result is minimized, and the final output global three-dimensional displacement field accuracy is better than 0.1mm.

[0108] Step 6: Deformation Analysis and Early Warning. The fused 3D displacement field, deformation rate, and acceleration data, combined with high-precision displacement data from key BDS monitoring points, are used to perform trend analysis, rate analysis, periodic analysis, and abrupt change point identification. A water level-deformation response model is established based on the pumped storage power station's water storage-power generation-pumping operation characteristics, eliminating elastic deformation caused by water level changes. Based on the deformation data characteristics and data changes from key BDS monitoring points, tiered early warnings are implemented, and anomaly tracing and safety assessments are conducted.

[0109] In this step, trend analysis uses a combination of linear and nonlinear fitting to analyze the deformation variation over time and determine whether the deformation is uniform, accelerating, or decelerating. Rate analysis calculates the deformation rate at different time periods (day, week, month) and analyzes the correlation between rate changes and operating conditions in conjunction with the power station's operating conditions (water storage, power generation, pumping, shutdown). Periodic analysis uses Fourier transform to extract the periodic characteristics of deformation and identify periodic signals related to water level changes and geological stress changes. Abrupt change point identification uses the sliding window method. By setting the sliding window size (generally 7-14 days), the mean and standard deviation of deformation within each window are calculated. When the difference between the deformation value at a certain moment and the mean of the corresponding window exceeds 3 times the standard deviation, it is determined to be an abrupt change point. Combined with the displacement changes of key monitoring points of BDS, the authenticity of the abrupt change is confirmed.

[0110] The water level-deformation response model is obtained by collecting deformation data under different water level conditions and fitting the data using multiple regression analysis. The model incorporates linear, quadratic, and logarithmic terms related to water level, which can accurately describe the elastic deformation caused by water level changes. After removing this elastic deformation, the true plastic deformation of the structure can be obtained, improving the accuracy of deformation analysis. The calculation formula is as follows:

[0111] Where D is the elastic deformation caused by the water level, and H is the reservoir water level. These are the model fitting coefficients. This represents the model residuals.

[0112] The sliding window method is used to identify mutation points. By setting the size of the sliding window (usually 7-14 days), the mean and standard deviation of the deformation within each window are calculated to determine mutation points. The calculation formula is as follows:

[0113] in, The deformation values ​​at each time point are: Let w be the average deformation value of all windows, and w be the size of the sliding window. This represents the standard deviation of the deformation data within the corresponding window.

[0114] The tiered early warning system is divided into three levels: Yellow (moderate), Orange (relatively severe), and Red (serious). The quantitative indicators for each level are as follows: Yellow alert (general):

[0115] Orange alert (severe):

[0116] Red Alert (Severe):

[0117] Where v is the deformation rate and a is the deformation acceleration. This represents the displacement change of key BDS monitoring points within 24 hours.

[0118] The response procedures for each warning level are as follows: For a yellow warning (general), increase monitoring frequency and conduct daily inspections of key areas; for an orange warning (severe), initiate special inspections, analyze the causes of anomalies, and implement temporary prevention and control measures; for a red warning (severe), immediately cease relevant operations (such as water storage and power generation), activate the emergency response plan, and organize professional personnel to conduct safety assessments and hazard mitigation. Anomaly tracing combines fused deformation data, BDS key monitoring point data, operating condition data, and geological data to analyze the causes of abnormal deformation (such as landslides, structural damage, sudden water level changes, equipment interference, etc.), providing a scientific basis for safe response.

[0119] Step 7: Output and visualization of results. Generate monitoring reports, deformation curves, displacement cloud maps, and spatiotemporal heat maps. The monitoring report should include GBSAR station stability monitoring results, GBSAR station real-time displacement correction data, and BDS key monitoring point monitoring results. Build a web-based monitoring platform to support real-time viewing, historical backtracking, and automatic report generation of deformation data, GBSAR station stability data, displacement correction data, and BDS key monitoring point data.

[0120] In this step, the monitoring reports are divided into daily, weekly, monthly, and annual reports. Daily reports focus on the deformation data, early warning status, and GBSAR station stability for the day. Weekly and monthly reports analyze the phased deformation trends, operational condition correlations, and early warning handling. The annual report summarizes the year's monitoring results, deformation patterns, and safety assessment conclusions. Deformation curves use time-series graphs to visually display the displacement trends of each monitoring point (key BDS monitoring points and typical GBSAR points) over time. Displacement cloud maps use pseudo-color rendering to visually display the overall deformation distribution, with different colors corresponding to different deformation magnitudes. Spatiotemporal heat maps display the spatiotemporal distribution characteristics of deformation rates, facilitating rapid identification of high deformation rate areas. The web-based monitoring platform supports PC and mobile access and features real-time data push, historical data query, data export, and early warning alerts. Administrators can remotely monitor the monitoring situation in real time, enabling efficient sharing and application of monitoring results.

[0121] like Figure 3 As shown, the multi-source coordinated deformation monitoring system of GBSAR, BDS and LiDAR for pumped storage power stations of the present invention is used to implement the above-mentioned monitoring method, including: Data acquisition module 101 includes GBSAR observation equipment, BDS reference station, BDS monitoring terminal, and UAV LiDAR aerial survey equipment, used to simultaneously acquire surface deformation data, GBSAR station displacement data (used to calculate displacement correction), BDS key monitoring point displacement data, and terrain point cloud data. The BDS reference station provides a benchmark for the stability of GBSAR stations and BDS monitoring points.

[0122] Data preprocessing module 102: It is connected to GBSAR observation equipment, BDS reference station, BDS monitoring terminal and UAV LiDAR aerial survey equipment respectively. It is used to perform noise reduction, correction, calculation and standardization processing on the source data. The preprocessing of BDS data is used to output GBSAR station stability assessment results, GBSAR station real-time displacement data (and calculate real-time displacement correction) and BDS key monitoring point displacement data.

[0123] GBSAR data denoising employs an adaptive filtering algorithm:

[0124] in, This is the filtered data. The original data, Here, N is the filter coefficient matrix, and N is the radius of the filter window. BDS data cycle slip detection uses the Turbo Edit algorithm, and its cycle slip determination formula is as follows:

[0125] in, The carrier phase difference matrix at each time step. This represents the standard deviation of the phase difference.

[0126] Spatiotemporal registration module 103: Connected to the data preprocessing module, it is used to realize coordinate system unification, time synchronization and benchmark unification of multi-source data. Combining the GBSAR station stability assessment results and real-time displacement correction, it performs real-time displacement correction on GBSAR observation data. At the same time, it uses the BDS key monitoring point data as the calibration basis to complete the benchmark unification of multi-source data.

[0127] The time-series interpolation uses the Lagrange interpolation formula:

[0128] in, This is the interpolated time series data matrix. is the original time series data matrix, and n is the number of interpolation nodes.

[0129] The GBSAR real-time displacement correction formula is:

[0130] in, This is the corrected GBSAR data matrix. This is the original GBSAR data matrix. For the real-time displacement correction matrix ( , This is the displacement matrix of the GBSAR station at each time point. (This is the initial reference displacement).

[0131] Multi-source fusion module 104: Connected to the spatiotemporal registration module, it has a built-in three-level fusion algorithm to complete coarse fusion, medium-precision fusion and high-precision deep fusion of multi-source data. During the fusion process, it incorporates GBSAR station stability parameters, real-time displacement correction and BDS key monitoring point data, outputs three-dimensional displacement field and deformation characteristic parameters, removes invalid data and improves the monitoring accuracy of key parts.

[0132] Deformation Analysis and Early Warning Module 105: Connected to the multi-source fusion module, it is used to analyze the fused deformation data and BDS key monitoring point data, identify abrupt change points, and combine working condition characteristics to realize graded early warning and anomaly tracing.

[0133] Output and Visualization Module 106: Connected to the Deformation Analysis and Early Warning Module, it is used to generate monitoring results (including GBSAR station stability monitoring results, GBSAR station real-time displacement correction data, and BDS key monitoring point monitoring results), and realize visualization display and data query through the Web platform.

[0134] Furthermore, the multi-source fusion module also includes an operating condition adaptive weight adjustment unit, which is used to adjust the weight of each data source according to different operating conditions of the pumped storage power station: increase the GBSAR weight during the water storage period, increase the UAV LiDAR weight during the slope inspection period, increase the weight of BDS key monitoring point data during the key part monitoring period, and dynamically adjust the weight in combination with the GBSAR station stability monitoring results and real-time displacement correction.

[0135] The formula for adaptive weight adjustment based on operating conditions is in matrix form:

[0136] in, The adjusted weight matrix, The initial weight matrix, The working condition weighting coefficient matrix (water storage period) Slope inspection period Key monitoring periods ), GBSAR station stability correction factor ( , (This is the maximum threshold for GBSAR station displacement).

[0137] Furthermore, the GBSAR observation equipment in the data acquisition module is deployed on multiple tracks to achieve multi-angle line-of-sight displacement acquisition and improve the accuracy of three-dimensional displacement inversion; the BDS reference station is deployed in a stable area and works in conjunction with the GBSAR station to achieve stability monitoring and displacement correction calculation; the BDS key monitoring terminal is deployed at key locations such as dam shoulders, steep slopes, and surge tank wellheads, and uses high-precision BDS receiving equipment with a sampling rate consistent with the GBSAR sampling rate to ensure data synchronization; the UAV LiDAR aerial survey equipment uses periodic aerial surveys, preferably once a month or quarter.

[0138] Formula for calculating LiDAR point cloud density for drones:

[0139] in, denoted as point cloud density (points / square meter), f as laser emission frequency (Hz), swath as flight strip width (m), overlap as flight strip overlap (%), v as flight speed (m / s), and h as flight altitude (m).

[0140] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for deformation monitoring of a pumped storage power station using multi-source synergy of GBSAR, BDS, and LiDAR, characterized in that, Includes the following steps: Step 1: Preliminary preparation and scheme design, delineate key monitoring zones of pumped storage power station, and deploy BDS base station, BDS key monitoring points, GBSAR permanent observation positions and UAV LiDAR aerial survey control network. Among them, BDS base station is used to monitor the stability of GBSAR station, and BDS key monitoring points are deployed in key parts of power station for high-precision key monitoring of key parts. Determine the observation frequency; Step 2: Multi-source data synchronous acquisition. Continuous non-contact surface monitoring is conducted using GBSAR to acquire millimeter-level two-dimensional surface deformation time-series data and line-of-sight displacement of the monitoring area; reference coordinate data of BDS base station, three-dimensional coordinates and time-series displacement data of key BDS monitoring points, and three-dimensional coordinates and time-series displacement data of GBSAR station are acquired simultaneously using BDS; periodic aerial surveys are conducted using UAV LiDAR to acquire high-density point cloud and DEM / DSM data of the monitoring area. Step 3: Multi-source data preprocessing, including denoising, atmospheric phase compensation, temporal interferometry calculation, and removal of flat-land effect from GBSAR data; The system performs cycle slip detection and repair, tropospheric / ionospheric correction, and relative positioning calculation on BDS data. It outputs GBSAR station stability assessment results, real-time displacement data of GBSAR stations, and high-precision displacement data of key BDS monitoring points. If the displacement of a GBSAR station exceeds a preset threshold, the GBSAR observation data for the corresponding time period is removed. If the displacement does not exceed the threshold, the real-time displacement correction is calculated based on the real-time displacement data of the GBSAR station. The system also performs point cloud denoising, filtering, registration, coordinate unification, and time-series DEM differential processing on UAV LiDAR data. Step 4: Spatiotemporal registration. The preprocessed GBSAR data, BDS data, and UAV LiDAR data are unified to the engineering independent coordinate system or the 2000 National Geodetic Coordinate System. Interpolation methods are used to unify the temporal sampling rate of different data sources and establish a common timestamp. Combining the GBSAR station stability data and real-time displacement correction obtained from BDS monitoring, and using the BDS base station as a reference, real-time displacement correction is performed on the GBSAR observation data to complete the unification of multi-source data benchmarks. At the same time, the data from key BDS monitoring points are used as independent calibration data sources. Step 5: Multi-source data fusion and solution, using a three-level fusion algorithm, specifically including: 5.1 Spatiotemporal coarse fusion: The coordinate system of multi-source data is unified through seven-parameter transformation, and time synchronization is achieved through temporal interpolation. The GBSAR station stability assessment results and real-time displacement corrections output by BDS are combined to correct the GBSAR observation data and remove invalid GBSAR data. At the same time, the data of key monitoring points of BDS are included for preliminary calibration. 5.2 Medium-precision fusion: Based on GBSAR multi-track / multi-angle line-of-sight displacement data, the true three-dimensional displacement field of GBSAR is inverted through the least squares algorithm; Kalman filtering algorithm is used, with GBSAR data as the core, combined with GBSAR station stability parameters and real-time displacement corrections monitored by BDS for noise suppression; IGGIII weighting function is introduced, and robust least squares fusion algorithm is used to remove gross errors, while the fusion results are calibrated by combining BDS key monitoring point data; 5.3 High-precision deep fusion: Based on the UAV LiDAR high-precision DEM, terrain phase is calculated, and terrain flattening and overlay correction are performed on the displacement-corrected GBSAR data; a "surface-volume" two-layer data fusion model is constructed, and the co-kriging algorithm is adopted. The displacement-corrected GBSAR continuous two-dimensional deformation field is used as the core, the LiDAR terrain gradient is used as the covariate, and the BDS key monitoring point data is used as the constraint condition to achieve seamless three-dimensional displacement field fusion across the entire domain; a temporal differential fusion algorithm is adopted to perform optimal fusion of multi-period UAV LiDAR DEM differential data, displacement-corrected GBSAR temporal deformation data, and BDS key monitoring point temporal data, and adaptively adjust the weights of each data source; Step 6: Deformation Analysis and Early Warning. The fused 3D displacement field, deformation rate, and acceleration data, combined with high-precision displacement data from key BDS monitoring points, are used to perform trend analysis, rate analysis, periodic analysis, and abrupt change point identification. A "water level-deformation" response model is established based on the operating characteristics of the pumped storage power station, eliminating elastic deformation caused by water level changes. Based on the deformation data characteristics and changes in data from key BDS monitoring points, tiered early warnings are implemented, and anomaly tracing and safety assessments are conducted. Step 7: Output and visualization of results, generating monitoring reports, deformation curves, displacement cloud maps, and spatiotemporal heat maps; building a web-based monitoring platform to support real-time viewing, historical backtracking, and automatic report generation of deformation data, GBSAR station stability data, displacement correction data, and BDS key monitoring point data.

2. The deformation monitoring method for pumped storage power stations using multi-source synergy of GBSAR, BDS, and LiDAR as described in claim 1, is characterized in that... The monitoring zones mentioned in Step 1 include the dam body, reservoir bank slopes, inlet and outlet slopes, surge tanks, and road slopes; the key parts of the power station include the dam shoulders, steep slopes, and surge tank openings; the observation frequency is as follows: daily or weekly observations are used for routine monitoring, hourly intensified observations are used during the flood season / water storage period, and real-time observations are used for extreme conditions.

3. The deformation monitoring method for pumped storage power stations using multi-source synergy of GBSAR, BDS, and LiDAR as described in claim 1, is characterized in that... The seven-parameter transformation model mentioned in step 5.1 is as follows: in, To transform the three-dimensional coordinate matrix, The transformed 3D coordinate matrix, Here is the translation parameter matrix, and m is the scale parameter. Let be a three-dimensional rotation matrix, where ω, φ, and κ are rotation angles about the X, Y, and Z axes, respectively.

4. The deformation monitoring method for pumped storage power stations using multi-source synergy of GBSAR, BDS, and LiDAR as described in claim 1, is characterized in that... The GBSAR line-of-sight displacement inversion formula and the multi-track combined line-of-sight displacement inversion formula mentioned in step 5.2 are as follows: The formula for single-track line-of-sight displacement inversion is: in, This represents the line-of-sight displacement of GBSAR after real-time displacement correction. Here is the three-dimensional displacement component matrix of the GBSAR station, where θ is the radar incident angle and α is the radar azimuth angle. Line-of-sight displacement inversion formula for multiple tracks: Where n is the number of GBSAR observation orbits, Here are the line-of-sight displacement matrices for each orbit. For each orbital radar incident angle, The azimuth angles of each orbital radar.

5. The deformation monitoring method for pumped storage power stations using multi-source synergy of GBSAR, BDS, and LiDAR as described in claim 1, characterized in that, The state vector, state equation, observation equation, and state transition matrix of the Kalman filter described in step 5.2 are as follows: The state vector is: Where X, Y, and Z are the three-dimensional coordinates of the target point in the monitoring area. For three-dimensional velocity, For three-dimensional acceleration, the stability parameters of GBSAR stations monitored by BDS and the real-time displacement correction are introduced during the filtering process to dynamically adjust the filtering coefficients; The state equation is: in, Here is the state transition matrix. Here, T is the process noise matrix, and T is the sampling time interval. The observation equation is: This is an observation vector matrix containing GBSAR data and BDS key monitoring point data. For the observation matrix, The observation noise matrix; Meanwhile, the IGGⅢ weight function formula mentioned in step 5.2 is: in The weight of the i-th observation. To observe the residuals, The error is the unit weight. Let the coefficients of the weight function be taken as follows: , .

6. The deformation monitoring method for pumped storage power stations using multi-source synergy of GBSAR, BDS, and LiDAR as described in claim 1, characterized in that, The collaborative financial formula described in step 5.3 is as follows: Synergistic Kryptonian Cooperative Formula: in, The deformation value after fusion. This is GBSAR surface deformation data after displacement correction. This is LiDAR terrain gradient data. For BDS key monitoring point displacement data, , , These are the fusion weights for the corresponding data sources; The constraints that the weights must satisfy are: LiDAR terrain gradient data The calculation formula is: Where h is the LiDAR DEM elevation value, and x and y are the plane coordinates.

7. The deformation monitoring method for pumped storage power stations using multi-source synergy of GBSAR, BDS, and LiDAR as described in claim 1, characterized in that, The formula and weights for the optimal fusion of mean squared errors described in step 5.3 are as follows: Optimal fusion formula for mean square error: in, The optimal deformation value after fusion. This is the original deformation data from GBSAR. This represents the real-time displacement correction for GBSAR stations. Deformation data for LiDAR on drones, For BDS key monitoring point displacement data, This is an adaptive weight matrix for each data source. The weights are adjusted in real time based on the accuracy, signal-to-noise ratio, GBSAR station stability monitoring results, and the accuracy of BDS key monitoring point data for each data source. The weights satisfy the matrix constraints: And the weights are calculated as follows: in , , The root mean square errors are BDS key monitoring point data, GBSAR data after displacement correction, and LiDAR data, respectively. Meanwhile, the terrain phase calculation formula described in step 5.3 is as follows: in, Where λ represents the terrain phase and λ is the GBSAR radar wavelength. θ represents the terrain elevation difference, and θ is the radar incident angle. GBSAR phase value after terrain phase correction: in, For GBSAR raw phase, This is the atmospheric phase error, the residual error after preprocessing and compensation.

8. The deformation monitoring method for pumped storage power stations using multi-source synergy of GBSAR, BDS, and LiDAR as described in claim 1, characterized in that, The graded early warning system described in step 6 is divided into three levels—yellow, orange, and red—based on deformation rate, acceleration, abrupt change characteristics, and displacement changes at key BDS monitoring points. Different levels correspond to different handling procedures. The quantitative judgment formula for graded early warning is as follows: Yellow alert: Orange alert: Red Alert: Where v is the deformation rate and a is the deformation acceleration. This refers to the displacement change of key BDS monitoring points within 24 hours. As described in step 6, the water level-deformation response model is: Where D is the elastic deformation caused by the water level, and H is the reservoir water level. These are the model fitting coefficients. For model residuals; Mutation point identification uses the sliding window method, and its determination formula is as follows: in, The deformation values ​​at each time point are: Let w be the average deformation value of all windows, and w be the size of the sliding window. This represents the standard deviation of the deformation data within the corresponding window.

9. A multi-source collaborative deformation monitoring system for pumped storage power stations, incorporating GBSAR, BDS, and LiDAR, characterized in that... To implement the monitoring method according to any one of claims 1-8, comprising: Data acquisition module: includes GBSAR observation equipment, BDS reference station, BDS monitoring terminal, and UAV LiDAR aerial survey equipment, used to simultaneously collect surface deformation data, GBSAR station displacement data, BDS key monitoring point displacement data and terrain point cloud data. Among them, the BDS reference station provides a reference for the stability of GBSAR station and BDS monitoring points. Data preprocessing module: Connected to GBSAR observation equipment, BDS reference station, BDS monitoring terminal and UAV LiDAR aerial survey equipment respectively, it is used to perform noise reduction, correction, calculation and standardization processing on the data from each source. Among them, the preprocessing of BDS data is used to output GBSAR station stability assessment results, GBSAR station real-time displacement data and calculate real-time displacement correction, and output BDS key monitoring point displacement data. Spatiotemporal registration module: Connected to the data preprocessing module, it is used to realize coordinate system unification, time synchronization and benchmark unification of multi-source data. It combines the GBSAR station stability assessment results and real-time displacement correction to perform real-time displacement correction on GBSAR observation data, and uses BDS key monitoring point data as calibration basis. Multi-source fusion module: Connected to the spatiotemporal registration module, it has a built-in three-level fusion algorithm to complete coarse fusion, medium-precision fusion and high-precision deep fusion of multi-source data. During the fusion process, it incorporates GBSAR station stability parameters, real-time displacement correction and BDS key monitoring point data, and outputs three-dimensional displacement field and deformation characteristic parameters. Deformation Analysis and Early Warning Module: Connected to the multi-source fusion module, it is used to analyze the fused deformation data and BDS key monitoring point data, identify abrupt change points, and combine working condition characteristics to achieve hierarchical early warning and anomaly tracing. Results Output and Visualization Module: Connected to the Deformation Analysis and Early Warning Module, it is used to generate monitoring results and enable visualization and data querying through a web platform.

10. The monitoring system according to claim 9, characterized in that, The multi-source fusion module also includes an operating condition adaptive weight adjustment unit, which is used to adjust the weight of each data source according to different operating conditions of the pumped storage power station: increase the weight of GBSAR during the water storage period, increase the weight of UAV LiDAR during the slope inspection period, and increase the weight of BDS key monitoring point data during key monitoring periods. The formula for adaptive weight adjustment based on operating conditions is in matrix form: in, The adjusted weight matrix, The initial weight matrix, The working condition weighting coefficient matrix (water storage period) Slope inspection period Key monitoring periods ), GBSAR station stability correction factor ( , This represents the maximum displacement threshold for GBSAR stations. The GBSAR observation equipment in the data acquisition module is deployed on multiple tracks to achieve multi-angle line-of-sight displacement acquisition and improve the accuracy of three-dimensional displacement inversion. The BDS reference station is deployed in a stable area and works in conjunction with the GBSAR station to achieve stability monitoring and displacement correction calculation. The BDS key monitoring terminals are deployed at key locations such as dam shoulders, steep slopes, and surge tank wellheads, and use high-precision BDS receiving equipment with a sampling rate consistent with the GBSAR sampling rate to ensure data synchronization. The UAV LiDAR aerial survey equipment conducts periodic aerial surveys, with a survey frequency of once a month or quarter. Formula for calculating LiDAR point cloud density for drones: in, denoted as point cloud density (points / square meter); f is the laser emission frequency (Hz); swath is the flight strip width (m); overlap is the flight strip overlap (%); v is the flight speed (m / s); and h is the flight altitude (m).

Images