InSAR + UAV cooperative mine surface deformation monitoring method

Abstract

The invention relates to an InSAR + UAV cooperative mine surface deformation monitoring method. The method comprises the following steps: respectively acquiring radar image data and unmanned aerial vehicle aerial survey data of a target mine area by using an InSAR and a UAV; registering the radar image data and the aerial survey data of the unmanned aerial vehicle to obtain registered data; performing hierarchical processing on the monitoring data of the different deformation gradient regions to obtain a deformation feature region division result; processing the registered data according to the deformation feature region division result to obtain processed data; and performing weighted fusion calculation on the processed data in each deformation feature region to obtain a surface deformation monitoring result. According to the invention, high-precision monitoring and timely early warning of mine surface deformation are realized, powerful support is provided for mine safety management, and the method has important practical application value.

Description

Technical Field

[0001] This invention relates to the field of terrain monitoring technology, and in particular to a method for monitoring surface deformation in mines using a combination of InSAR and UAV. Background Technology

[0002] Monitoring the impact of mining activities on the surface environment is a key technical area for ensuring safe production and environmental protection in mining areas. With the increasing intensity of mineral resource extraction, accurately understanding the patterns of surface deformation in mining areas is of great significance for preventing geological disasters and protecting surrounding infrastructure.

[0003] Current methods for monitoring land surface deformation face significant limitations in practical applications. Traditional single monitoring methods are often constrained by the trade-off between monitoring range and accuracy, making it difficult to simultaneously meet the dual requirements of large-scale coverage and high-precision detection. Existing technologies are prone to incomplete information acquisition when processing monitoring data under complex terrain conditions, especially in areas with drastic land surface changes, where the reliability of monitoring results decreases significantly.

[0004] The core challenge in monitoring surface deformation in mines stems from the multi-scale differences in deformation information. Small deformations and large-gradient deformations often coexist within the same mining area, and this significant difference in deformation amplitude means that a single monitoring technology cannot effectively cover the entire deformation range. When a monitoring system fails to accurately capture deformation information at different scales, compatibility issues inevitably arise during data fusion. Monitoring data from different sources exhibit significant differences in spatial resolution, temporal reference, and measurement accuracy. These differences make it difficult to effectively integrate multi-source data, thus affecting the accurate judgment of surface deformation patterns in mines.

[0005] How to achieve comprehensive and accurate monitoring of surface deformation in mines has become a key issue that urgently needs to be addressed in this field. Summary of the Invention

[0006] The purpose of this invention is to provide a method for monitoring surface deformation in mines using InSAR+UAV synergy, so as to achieve high-precision monitoring and timely early warning of surface deformation in mines.

[0007] To achieve the above objectives, the present invention provides the following solution: A method for monitoring surface deformation in mines using InSAR+UAV synergy includes: InSAR and UAV were used to acquire radar imagery data and UAV aerial survey data of the target mining area, respectively. The radar image data and UAV aerial survey data are registered to obtain the registered data; The monitoring data of different deformation gradient regions are processed in layers to obtain the results of deformation feature region division. The registered data is processed based on the deformation feature region division results to obtain the processed data; The processed data within each deformation characteristic region are weighted and fused to obtain the surface deformation monitoring results.

[0008] Optionally, the radar image data and UAV aerial survey data are registered to obtain the registered data, including: The radar image data and UAV aerial survey data were resampled using a bilinear interpolation method to obtain image data with consistent spatial resolution. A time interpolation algorithm is used to perform time correction on the radar image data and UAV aerial survey data at different times to obtain the registered data.

[0009] Optionally, the monitoring data for different deformation gradient regions can be stratified, including: Based on the registered data, the deformation amplitude sequence of each monitoring point is extracted, and the deformation amplitude distribution characteristics are analyzed by kernel density estimation method, and the cumulative probability density curve is calculated. Based on the cumulative probability density curve, determine the threshold for small deformation and the threshold for large gradient deformation; The deformation amplitude of each monitoring point on the surface of the target mine is obtained. If the deformation amplitude of the monitoring point is less than the small deformation threshold, it is divided into a small deformation region. If the deformation amplitude of the monitoring point is greater than the large gradient deformation threshold, it is divided into a large gradient deformation region. If the deformation amplitude of the monitoring point is greater than the small deformation threshold but less than the large gradient deformation threshold, it is divided into a medium deformation region. The deformation feature region division result is obtained.

[0010] Optionally, processing the registered data based on the deformation feature region segmentation result includes: If the monitoring point is located in the small deformation region, the registered UAV aerial survey data is weighted and enhanced; if the monitoring point is located in the large gradient deformation region, the registered radar image data is gradient compensated and corrected. The processed data is obtained based on the enhanced UAV aerial survey data and the corrected radar image data.

[0011] Optionally, gradient compensation correction of the registered radar image data includes: For the monitoring points within the large gradient deformation region, the spatial distribution characteristics within a preset range are obtained, and the compensation correction amount for each data point is calculated using a gradient compensation algorithm. Spatial interpolation is performed on the preset range based on the compensation correction amount to obtain the corrected data distribution matrix, wherein the data distribution matrix contains the correction values ​​of each monitoring point.

[0012] Optionally, weighted fusion calculations are performed on the processed data within each deformation feature region, including: A weighted average algorithm is used to perform preliminary fusion calculations on the processed data to obtain preliminary fused data. If the deviation between the preliminary fused data and the GPS station data exceeds a preset value, the weight value of the radar image data corresponding to each monitoring point will be dynamically adjusted. The surface deformation monitoring results are obtained by recalculating based on the adjusted weight values.

[0013] Optionally, after obtaining the surface deformation monitoring results, the process includes: inputting the surface deformation monitoring results into a deformation time series analysis model to predict the deformation development trend of each monitoring point and identify abnormal surface deformation areas. The deformation time series analysis model uses a time series model to fit and analyze the deformation time series dataset. If the deformation data of the monitoring points exhibits linear change characteristics, a linear regression algorithm is used to calculate the deformation rate parameter. If it exhibits nonlinear change characteristics, a multinomial regression algorithm is used to fit the deformation trend curve.

[0014] Optionally, obtaining the deformation time series dataset includes: obtaining the coordinate information and historical deformation measurement values ​​of each monitoring point, and obtaining the deformation time series dataset of each monitoring point by sorting by time series.

[0015] The beneficial effects of this invention are as follows: By acquiring radar imagery data and UAV aerial survey data and performing spatiotemporal registration, this invention solves the problem of inconsistent spatial resolution and time reference of the data source. Addressing differences in measurement accuracy, this invention performs layered processing of different deformation gradient regions based on deformation amplitude distribution characteristics and uses a gradient compensation algorithm to optimize the data fusion process. By constructing a deformation time-series analysis model, this invention can predict deformation development trends, identify potential anomaly areas, and establish an early warning mechanism. This method achieves high-precision monitoring and timely early warning of surface deformation in mines, providing strong support for mine safety management and possessing significant practical application value. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of an InSAR+UAV collaborative method for monitoring surface deformation in mines, according to an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] This embodiment provides a method for monitoring surface deformation in mines using a combination of InSAR and UAV, such as... Figure 1 As shown, it includes: InSAR and UAV were used to acquire radar imagery data and UAV aerial survey data of the target mining area, respectively. Register radar imagery data and UAV aerial survey data to obtain the registered data; The monitoring data of different deformation gradient regions are processed in layers to obtain the results of deformation feature region division. The registered data is processed based on the deformation feature region division results to obtain the processed data; The processed data within each deformation characteristic region are weighted and fused to obtain the surface deformation monitoring results.

[0021] Furthermore, the radar imagery data and UAV aerial survey data are registered to obtain the registered data, including: Bilinear interpolation was used to resample radar image data and UAV aerial survey data to obtain image data with consistent spatial resolution. A time interpolation algorithm is used to perform time correction on radar image data and UAV aerial survey data at different times to obtain the registered data.

[0022] Specifically, radar imagery uses 10-meter resolution, while UAV aerial data has a resolution of 0.1 meters. When resampling satellite imagery using bilinear interpolation, the 10-meter pixel is decomposed into 100 1-meter pixels. The value of each new pixel is determined by the weighted average of its four surrounding original pixels, thus maintaining spatial resolution consistency with the UAV data. This process effectively eliminates scale differences caused by different sensors. A linear time interpolation algorithm is used, with the radar imagery as the baseline, to overlay the UAV data by 30 minutes, ensuring all data have the same time reference. This process effectively eliminates instantaneous surface deformation errors caused by time differences in acquisition. In the spatiotemporal registration stage, for example, feature points on a mine slope are selected as control points. A point with coordinates (500, 600) in the radar imagery corresponds to a point (502, 598) in the UAV imagery. A pixel-level correspondence is established using a polynomial correction model, achieving a registration accuracy within 0.5 pixels. This fine registration lays the foundation for subsequent deformation analysis.

[0023] Furthermore, the monitoring data for different deformation gradient regions are processed in a stratified manner, including: Based on the registered data, the deformation amplitude sequence of each monitoring point is extracted, and the kernel density estimation method is used to analyze the deformation amplitude distribution characteristics and calculate the cumulative probability density curve. Determine the threshold for small deformation and the threshold for large gradient deformation based on the cumulative probability density curve; The deformation amplitude of each monitoring point on the surface of the target mine is obtained. If the deformation amplitude of the monitoring point is less than the small deformation threshold, it is classified as a small deformation region. If the deformation amplitude of the monitoring point is greater than the large gradient deformation threshold, it is classified as a large gradient deformation region. If the deformation amplitude of the monitoring point is greater than the small deformation threshold but less than the large gradient deformation threshold, it is classified as a medium deformation region. The deformation characteristic region classification results are obtained.

[0024] Specifically, after obtaining the registered data, the deformation amplitude sequence of each monitoring point is extracted, and the deformation amplitude distribution characteristics are analyzed using the kernel density estimation method.

[0025] Raw deformation monitoring data from various monitoring points on the mine surface are acquired. A data preprocessing module is used to filter noise and detect outliers in the monitoring data, resulting in a standardized monitoring dataset. Based on the spatial coordinates of each monitoring point in the standardized dataset, the deformation gradient value for each monitoring point is calculated using a gradient calculation formula, yielding a deformation gradient distribution matrix. If the gradient value in the deformation gradient distribution matrix is ​​less than a preset threshold T1, the region is identified as a region with small deformation; if the gradient value is between T1 and T2, it is identified as a region with medium deformation; and if the gradient value is greater than T2, it is identified as a region with large gradient deformation, resulting in a regional stratification result.

[0026] In one embodiment, it is assumed that deformation sequences from 1000 monitoring points in a mining area were extracted. The probability density distribution curve was calculated using kernel density estimation, revealing that the deformation amplitude was mainly concentrated in the range of -5mm to 5mm. The cumulative probability density curve shows that the 10% quantile corresponds to a deformation amplitude of -8mm, and the 90% quantile corresponds to a deformation amplitude of 10mm. If a monitoring point deforms to -9mm, it falls into the region of small deformation; if the deformation is 12mm, it is marked as a region of large gradient deformation.

[0027] Furthermore, the registered data is processed based on the deformation feature region segmentation results, including: If the monitoring point is located in a region of slight deformation, the registered UAV aerial survey data will be weighted and enhanced; if the monitoring point is located in a region of large gradient deformation, the registered radar image data will be corrected by gradient compensation. The processed data is obtained based on the enhanced UAV aerial survey data and the corrected radar image data.

[0028] Furthermore, gradient compensation correction is performed on the registered radar image data, including: For monitoring points within a large gradient deformation region, obtain the spatial distribution characteristics within a preset range, and calculate the compensation correction amount for each data point through a gradient compensation algorithm; Spatial interpolation is performed on the preset range based on the compensation correction amount to obtain the corrected data distribution matrix, which contains the correction values ​​for each monitoring point.

[0029] Specifically, this embodiment also employs a multi-scale adaptive fusion algorithm to perform spatial continuity analysis on the preliminary regional classification results. By calculating the gradient correlation coefficient between adjacent monitoring points, spatially continuous deformation regions of the same type are identified, resulting in a region division result with optimized continuity. Based on the optimized region division result, a clustering algorithm is used to refine the boundaries of each deformation region. By iteratively calculating the gradient mean and variance of monitoring points within a region, the precise boundary range of each region is determined, resulting in refined regional boundary data. The spatial topological relationship of the mine surface deformation characteristic regions is constructed using the refined regional boundary data. An adjacency matrix is ​​used to record the spatial adjacency relationship and gradient change trend between regions, resulting in a complete hierarchical deformation characteristic region division result. Based on the hierarchical deformation characteristic region division result, a deformation characteristic parameter table for each region is generated, including region number, deformation type, gradient range, number of monitoring points, and spatial area information, forming a hierarchical management database for mine surface deformation monitoring.

[0030] For monitoring points within areas of slight deformation, high-precision data are extracted, and a weighted enhancement method is used to amplify the contribution of these high-precision values ​​in the data fusion process. For monitoring points within areas of large-gradient deformation, the spatial distribution characteristics of the surrounding data are obtained, and the compensation correction amount for each data point is calculated using a gradient compensation algorithm. Based on the compensation correction amount, spatial interpolation is performed on the surrounding data to obtain a corrected data distribution matrix, which contains the corrected values ​​for each monitoring point. A weighted average method is used to combine the weighted enhancement results for areas of slight deformation and the compensation correction results for areas of large-gradient deformation to calculate the fusion coefficient for each region. The original monitoring data for each deformation region is then weighted using the fusion coefficient to obtain the optimized regional deformation monitoring results.

[0031] For example, in a region with slight deformation, there are 15 monitoring points with an original weight coefficient of 0.3. After weight enhancement processing, the weight coefficient is increased to 0.8, allowing these high-precision data to play a greater role in the final fusion result. Gradient compensation algorithms mainly address the data distortion problem in regions with large gradient deformation.

[0032] In one possible implementation, when the deformation in a large gradient deformation region reaches 50 mm, the excessive gradient may lead to data measurement deviation. The system analyzes the spatial distribution characteristics within a 500-meter radius of this region, calculates a compensation correction of -8 mm, and adjusts the original measurement value accordingly. Spatial interpolation fills in data gaps using mathematical modeling methods. For example, if there are three missing data points within a deformation region, the system uses data from eight surrounding valid monitoring points and employs inverse distance weighted interpolation to calculate the deformation value at the missing locations. After interpolation, a complete data distribution matrix is ​​formed, ensuring data continuity across the entire region. The calculation of the fusion coefficient comprehensively considers the data characteristics of different regions.

[0033] Specifically, the fusion coefficient for regions with small deformations is typically set between 0.7 and 0.9 to reflect their high precision. The fusion coefficient for regions with large deformation gradients is generally in the range of 0.4 to 0.6, balancing data reliability and deformation significance. Through this differentiated fusion strategy, the resulting regional deformation monitoring results accurately reflect the true deformation state of the mine surface, providing reliable data support for safety early warning and engineering decision-making.

[0034] Furthermore, the weighted fusion calculation of the processed data within each deformation feature region includes: A weighted average algorithm is used to perform preliminary fusion calculations on the processed data to obtain preliminary fused data. If the deviation between the initial fused data and the GPS station data exceeds the preset value, the weight value of the radar image data corresponding to each monitoring point will be dynamically adjusted. The surface deformation monitoring results are obtained by recalculating based on the adjusted weight values.

[0035] Specifically, if the deviation between the preliminary fused data and the GPS station data exceeds a preset value, the weight value of the radar image data corresponding to each monitoring point will be dynamically adjusted.

[0036] Specifically, when a significant deviation is detected between the data from a GPS station and the initial fused data, the fusion coefficient will be automatically adjusted.

[0037] Furthermore, after obtaining the surface deformation monitoring results, the process includes: inputting the surface deformation monitoring results into the deformation time series analysis model to predict the deformation development trend of each monitoring point and identify abnormal surface deformation areas. The deformation time series analysis model uses a time series model to fit and analyze the deformation time series dataset. If the deformation data of the monitoring points shows linear change characteristics, a linear regression algorithm is used to calculate the deformation rate parameter. If it shows nonlinear change characteristics, a multinomial regression algorithm is used to fit the deformation trend curve.

[0038] Furthermore, obtaining the deformation time series dataset includes: acquiring the coordinate information and historical deformation measurement values ​​of each monitoring point, and sorting them by time series to obtain the deformation time series dataset for each monitoring point.

[0039] Specifically, this embodiment can calculate the deformation rate change gradient value of each monitoring point based on the fitting results, and determine whether the rate change exceeds the normal range by setting a preset threshold, thus obtaining rate anomaly identification data. The rate anomaly identification data of the monitoring points is extended to the entire mining area using spatial interpolation methods, generating continuous spatially distributed anomaly intensity field data. A clustering algorithm is used to divide the spatially distributed anomaly intensity field data into regions. If the anomaly intensity value is greater than a preset anomaly threshold and is spatially continuous, the region is determined to be a deformation region. Based on the spatial range and anomaly intensity level of the deformation region, combined with the predictive analysis results of the monitoring points, comprehensive evaluation data including anomaly identification results and risk level assessment is generated. Data visualization technology is used to convert the comprehensive evaluation data into charts and text formats, automatically generating an assessment report document containing mine monitoring result analysis.

[0040] For example, within the mining impact area, if the deformation rate at a monitoring point gradually increases from an initial 2 mm per month to 8 mm per month, with a rate change gradient of 0.5 mm squared per month, the system will automatically mark the point as having an abnormal rate when this gradient exceeds a preset threshold of 0.3 mm squared per month. This dynamic monitoring mechanism can promptly detect dangerous signals of accelerated deformation, providing crucial early warning information for mine safety management.

[0041] In one possible implementation, the spatial interpolation method uses the Kriging interpolation algorithm to extend discrete monitoring point data into a continuous spatial field.

[0042] Specifically, the method in this embodiment calculates the estimated anomaly intensity of each grid point within the entire mining area based on the known anomaly intensity values ​​of the monitoring points through spatial correlation analysis. For example, when the anomaly intensity of monitoring point A is 0.8 and the anomaly intensity of the adjacent monitoring point B is 0.6, the anomaly intensity of the area between the two points will be interpolated according to distance weights.

[0043] Clustering algorithms consider both the magnitude of abnormal intensity values ​​and spatial continuity when identifying deformable regions. When the abnormal intensity values ​​of continuously distributed grid points within a region all exceed a preset threshold of 0.7, and the area of ​​that region reaches 500 square meters or more, it is identified as a high-risk deformable region. This spatial clustering-based identification method effectively distinguishes between local anomalies and true deformable regions, avoiding false alarms. In this embodiment, the risk level is divided into low, medium, and high levels based on factors such as the spatial extent of the deformable region, the level of abnormal intensity, and the trend of deformation rate changes.

[0044] In one embodiment, data visualization technology can automatically generate a comprehensive assessment report including contour maps, heat maps, and trend analysis charts. This visualization method can intuitively reflect the spatial distribution characteristics and temporal evolution patterns of surface deformation in mines, providing clear technical support for management personnel's decision-making.

[0045] In one embodiment, a deformation early warning mechanism can be established based on the comprehensive monitoring and assessment report of mine surface deformation. If the deformation rate of a certain area exceeds the safety threshold or the cumulative deformation reaches a dangerous level, an early warning signal will be automatically triggered and detailed risk assessment data will be generated, resulting in the complete output of the mine surface deformation monitoring and early warning system.

[0046] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Any modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for monitoring surface deformation in mines using InSAR+UAV synergy, characterized in that, include: InSAR and UAV were used to acquire radar imagery data and UAV aerial survey data of the target mining area, respectively. The radar image data and UAV aerial survey data are registered to obtain the registered data; The monitoring data of different deformation gradient regions are processed in layers to obtain the results of deformation feature region division. The registered data is processed based on the deformation feature region division results to obtain the processed data; The processed data within each deformation characteristic region are weighted and fused to obtain the surface deformation monitoring results.

2. The InSAR+UAV synergistic method for monitoring surface deformation in mines according to claim 1, characterized in that, The radar imagery data and UAV aerial survey data are registered to obtain the registered data, which includes: The radar image data and UAV aerial survey data were resampled using a bilinear interpolation method to obtain image data with consistent spatial resolution. A time interpolation algorithm is used to perform time correction on the radar image data and UAV aerial survey data at different times to obtain the registered data.

3. The InSAR+UAV synergistic method for monitoring surface deformation in mines according to claim 1, characterized in that, Layered processing of monitoring data from different deformation gradient regions includes: Based on the registered data, the deformation amplitude sequence of each monitoring point is extracted, and the deformation amplitude distribution characteristics are analyzed by kernel density estimation method, and the cumulative probability density curve is calculated. Based on the cumulative probability density curve, determine the threshold for small deformation and the threshold for large gradient deformation; The deformation amplitude of each monitoring point on the surface of the target mine is obtained. If the deformation amplitude of the monitoring point is less than the small deformation threshold, it is divided into a small deformation region. If the deformation amplitude of the monitoring point is greater than the large gradient deformation threshold, it is divided into a large gradient deformation region. If the deformation amplitude of the monitoring point is greater than the small deformation threshold but less than the large gradient deformation threshold, it is divided into a medium deformation region. The deformation feature region division result is obtained.

4. The InSAR+UAV synergistic method for monitoring surface deformation in mines according to claim 3, characterized in that, The processing of the registered data based on the deformation feature region segmentation results includes: If the monitoring point is located in the small deformation region, the registered UAV aerial survey data is weighted and enhanced; if the monitoring point is located in the large gradient deformation region, the registered radar image data is gradient compensated and corrected. The processed data is obtained based on the enhanced UAV aerial survey data and the corrected radar image data.

5. The InSAR+UAV synergistic method for monitoring surface deformation in mines according to claim 4, characterized in that, Gradient compensation correction of the registered radar image data includes: For the monitoring points within the large gradient deformation region, the spatial distribution characteristics within a preset range are obtained, and the compensation correction amount for each data point is calculated using a gradient compensation algorithm. Spatial interpolation is performed on the preset range based on the compensation correction amount to obtain the corrected data distribution matrix, wherein the data distribution matrix contains the correction values ​​of each monitoring point.

6. The InSAR+UAV synergistic method for monitoring surface deformation in mines according to claim 4, characterized in that, The weighted fusion calculation of the processed data within each deformation feature region includes: A weighted average algorithm is used to perform preliminary fusion calculations on the processed data to obtain preliminary fused data. If the deviation between the preliminary fused data and the GPS station data exceeds a preset value, the weight value of the radar image data corresponding to each monitoring point will be dynamically adjusted. The surface deformation monitoring results are obtained by recalculating based on the adjusted weight values.

7. The InSAR+UAV synergistic method for monitoring surface deformation in mines according to claim 1, characterized in that, After obtaining the surface deformation monitoring results, the process includes: inputting the surface deformation monitoring results into a deformation time series analysis model to predict the deformation development trend of each monitoring point and identify abnormal surface deformation areas. The deformation time series analysis model uses a time series model to fit and analyze the deformation time series dataset. If the deformation data of the monitoring points shows linear change characteristics, a linear regression algorithm is used to calculate the deformation rate parameter. If it shows nonlinear change characteristics, a polynomial regression algorithm is used to fit the deformation trend curve.

8. The InSAR+UAV synergistic method for monitoring surface deformation in mines according to claim 7, characterized in that, Obtaining the deformation time series dataset includes: obtaining the coordinate information and historical deformation measurement values ​​of each monitoring point, and sorting them by time series to obtain the deformation time series dataset for each monitoring point.

Images