A reservoir capacity dynamic monitoring method and system of an unmanned aerial vehicle laser radar

Abstract

The application relates to the technical field of water resource management and ecological environment protection, and discloses a reservoir capacity dynamic monitoring method and system of an unmanned aerial vehicle laser radar, the method comprising the following steps: obtaining point cloud data of a reservoir area and constructing a digital elevation model through the unmanned aerial vehicle laser radar; obtaining a slope distribution map and a concave-convex change layer based on gridding analysis; extracting local irregular terrain parameters and optimizing a water storage distribution model; integrating real-time water level data to calculate a dynamic water body volume and obtain the current reservoir capacity; comparing and analyzing the capacity change trend in combination with historical data; and updating the digital elevation model to generate accurate monitoring results. The method realizes high-precision real-time dynamic monitoring of the reservoir capacity.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of water resources management and ecological environment protection, and particularly relates to a reservoir capacity dynamic monitoring method and system of an unmanned aerial vehicle laser radar. BACKGROUND

[0002] At present, in the field of modern water resources management and ecological environment protection, dynamic monitoring of reservoir capacity is a key basis for flood control scheduling and rational allocation of water resources.

[0003] In one prior art, reservoir capacity monitoring relies on fixed-point intelligent sensors or periodic manual surveying, for example, water depth data of limited points is obtained through water level scales or pressure type water level sensors arranged on the reservoir bank, and then the overall capacity is estimated in combination with a pre-drawn static reservoir capacity curve; the terrain data acquisition is often based on historical surveying results or low-resolution remote sensing images, and lacks the ability of continuous perception and updating of terrain changes in the reservoir bottom and surrounding areas.

[0004] Therefore, in the prior art, there is a problem that the reservoir capacity monitoring result under complex terrain is inaccurate and cannot meet the dynamic scheduling requirement. SUMMARY

[0005] The present application provides a reservoir capacity dynamic monitoring method and system of an unmanned aerial vehicle laser radar to solve the problem that the reservoir capacity monitoring result under complex terrain is inaccurate and cannot meet the dynamic scheduling requirement.

[0006] In a first aspect, to solve the above technical problem, the present application provides a reservoir capacity dynamic monitoring method of an unmanned aerial vehicle laser radar, comprising:

[0007] acquiring point cloud data of a reservoir area by an unmanned aerial vehicle laser radar, constructing an initial digital elevation model according to the point cloud data, extracting concave-convex change features from the initial digital elevation model and performing filtering processing to obtain high-precision terrain data;

[0008] performing grid unit division according to the high-precision terrain data, calculating a slope value for the grid unit to obtain a slope distribution map, performing curvature calculation according to the slope distribution map, and performing smoothing correction if the curvature appears abnormal fluctuation to obtain a concave-convex change layer;

[0009] extracting a high-slope area from the slope distribution map, spatially matching the high-slope area with the concave-convex change layer to obtain terrain feature data, performing edge detection and terrain parameter quantization processing for parts with significant concave-convex changes in the terrain feature data to obtain terrain parameters of local terrain;

[0010] The water storage space is processed in layers and integrated with the terrain parameters, the area proportion of the water accumulation region is calculated according to the integrated data, the initial water storage distribution model is locally adjusted according to the area proportion of the water accumulation region, and an optimized water storage distribution model is obtained;

[0011] Real-time water level data is obtained, and the gradient change of the water level is calculated in combination with the optimized water storage distribution model to determine the water body change trend, the specific deviation position of the real-time water body volume compared with historical data is obtained through the water body change trend, the optimized water storage distribution model is corrected according to the specific deviation position, and the current reservoir capacity is calculated according to the corrected model;

[0012] The historical capacity data is obtained, and the current reservoir capacity is compared and analyzed with the historical capacity data to obtain a capacity change trend;

[0013] The real-time monitoring data is obtained and compared with the predicted value obtained according to the capacity change trend, the initial digital elevation model is updated according to the comparison result, the current reservoir capacity is calibrated according to the updated elevation model, and a precise monitoring result of the final reservoir capacity is generated.

[0014] In an optional implementation, the point cloud data of the reservoir area is obtained by a drone laser radar, and an initial digital elevation model is constructed according to the point cloud data, the concave-convex change feature is extracted from the initial digital elevation model and is filtered to obtain high-precision terrain data, including:

[0015] Complete point cloud data of the reservoir area is obtained by a drone laser radar, and is subjected to gridding processing and interpolation filling of data missing areas to obtain a terrain height map;

[0016] According to the terrain height map, three-dimensional reconstruction processing is performed to obtain an initial digital elevation model;

[0017] According to the initial digital elevation model, the concave-convex change feature is extracted and filtered to obtain high-precision terrain data.

[0018] In an optional implementation, the high-precision terrain data is subjected to grid unit division, the slope value is calculated for the grid unit to obtain a slope distribution map, the curvature is calculated according to the slope distribution map, and if the curvature abnormally fluctuates, it is subjected to smoothing correction to obtain a concave-convex change layer, including:

[0019] The high-precision terrain data is subjected to gridding unit division and interpolation filling to obtain a grid unit data set;

[0020] The slope distribution is calculated for the grid unit data set, and the area where the slope exceeds a preset slope threshold is marked to obtain a slope distribution map.

[0021] According to the slope distribution map, curvature is calculated, and if the curvature abnormally fluctuates, smoothing correction is performed to obtain a concave-convex change layer.

[0022] In an optional embodiment, a high-slope region is extracted from the slope distribution map, the high-slope region is spatially matched with the concave-convex change layer to obtain terrain feature data, edge detection and terrain parameter quantization processing are performed on a part with significant concave-convex change in the terrain feature data to obtain terrain parameters of a local terrain, including:

[0023] A region of grid cells exceeding a preset slope threshold is extracted from the slope distribution map, and is marked to obtain a marked high-slope region;

[0024] The high-slope region is spatially matched with the concave-convex change layer to obtain superimposed terrain feature data;

[0025] If significant concave-convex change occurs in the terrain feature data, edge detection is performed on the part with significant concave-convex change to obtain a local terrain irregular region;

[0026] According to the local terrain irregular region, terrain parameter quantization is performed to extract fluctuation amplitude and boundary sharpness indexes to obtain terrain parameters of a local terrain.

[0027] In an optional embodiment, the water storage space is subjected to hierarchical processing and integrated with the terrain parameters, the integrated data is used to calculate an area proportion of a water accumulation region, the initial water storage distribution model is locally adjusted according to the area proportion of the water accumulation region to obtain an optimized water storage distribution model, including:

[0028] The water storage space is subjected to preliminary hierarchical processing and integrated with the terrain parameters to obtain hierarchical basic data;

[0029] The hierarchical basic data is spatially matched with the concave-convex change layer, and regions with concave-convex change exceeding a first preset threshold are marked as potential water accumulation regions to obtain a potential water accumulation range;

[0030] The terrain parameter difference value of the potential water accumulation range is analyzed, and if the difference value exceeds a second preset threshold, the potential water accumulation range is determined as a water accumulation influence region to obtain a specific water accumulation position;

[0031] According to the specific water accumulation position, an area proportion of a water accumulation region is calculated, and the initial water storage distribution model is locally adjusted according to a preset influence factor to obtain an optimized water storage distribution model.

[0032] In an alternative embodiment, the real-time water level data is obtained, and the gradient change of the water level is calculated in combination with the optimized water storage distribution model to determine the water body change trend, the specific deviation position of the real-time water body volume compared with the historical data is obtained through the water body change trend, the optimized water storage distribution model is corrected according to the specific deviation position, and the current reservoir capacity is calculated according to the corrected model, comprising:

[0033] The real-time water level data is obtained, and the data is arranged to obtain the water level fluctuation value. If the water level fluctuation value exceeds the preset fluctuation threshold, it is marked as an abnormal water level point to obtain an abnormal distribution range.

[0034] According to the abnormal distribution range, the real-time water level data is matched with the optimized water storage distribution model in coordinates and the gradient change of the water level is calculated to determine the water body change trend.

[0035] Through the water body change trend, the real-time water body volume is estimated in combination with the optimized water storage distribution model and the real-time water level data. If the difference between the estimation result and the historical data exceeds the preset deviation threshold, it is marked as a volume deviation area to obtain a specific deviation position.

[0036] According to the specific deviation position, the optimized water storage distribution model is locally corrected, and the reservoir capacity is calculated according to the corrected model to obtain the current reservoir capacity.

[0037] In an alternative embodiment, the historical capacity data is obtained, and the current reservoir capacity is compared and analyzed with the historical capacity data to obtain the capacity change trend, comprising:

[0038] The historical capacity data is obtained from a preset topographic database, and is preliminarily compared with the current reservoir capacity to calculate the capacity deviation and obtain the specific distribution range of the capacity deviation.

[0039] The historical topographic data is obtained from the topographic database and compared with the high-precision topographic data to obtain topographic change data.

[0040] The correlation degree of the specific distribution range and the topographic change data is calculated. If the correlation degree exceeds the preset correlation threshold, the corresponding area is marked as an abnormal position.

[0041] For the abnormal position, the convolutional neural network is used to extract features from the historical topographic data to obtain dynamic trend features of the topographic change, and the mutual information between the dynamic trend features and the current reservoir capacity is calculated.

[0042] The correlation degree, the dynamic trend features, the topographic change data and the capacity deviation are integrated to generate a change trend report to obtain the capacity change trend.

[0043] In an alternative embodiment, the real-time monitoring data is obtained and compared with the predicted value obtained according to the capacity change trend, the initial digital elevation model is updated according to the comparison result, the current reservoir capacity is calibrated according to the updated elevation model, and the accurate monitoring result of the final reservoir capacity is generated, including:

[0044] The real-time monitoring data is obtained and compared with the predicted value obtained according to the capacity change trend, and the change distribution characteristics of the real-time monitoring data are obtained;

[0045] According to the change distribution characteristics, the key area of topographic dynamic adjustment is determined;

[0046] For the key area, the initial elevation model is updated combined with the real-time monitoring data, and the current reservoir capacity is calibrated according to the updated elevation model to obtain the accurate monitoring result of the final reservoir capacity.

[0047] In a second aspect, the present application provides a reservoir capacity dynamic monitoring system of unmanned aerial vehicle laser radar, comprising:

[0048] The data acquisition module is used for acquiring point cloud data of the reservoir area through the unmanned aerial vehicle laser radar, constructing an initial digital elevation model according to the point cloud data, extracting concave-convex change characteristics from the initial digital elevation model and performing filtering processing to obtain high-precision topographic data;

[0049] The feature analysis module is used for dividing grid units according to the high-precision topographic data, calculating slope values for the grid units to obtain a slope distribution map, calculating curvature according to the slope distribution map, and performing smoothing correction if the curvature appears abnormal fluctuation to obtain a concave-convex change layer;

[0050] The topographic parameter acquisition module is used for extracting a high-slope area from the slope distribution map, spatially matching the high-slope area with the concave-convex change layer to obtain topographic feature data, performing edge detection and topographic parameter quantization processing for the part with significant concave-convex change in the topographic feature data to obtain topographic parameters of local topography;

[0051] The model optimization module is used for layering the water storage space and integrating with the topographic parameters, calculating the area proportion of the accumulated water area according to the integrated data, and locally adjusting the preset initial water storage distribution model according to the area proportion of the accumulated water area to obtain an optimized water storage distribution model;

[0052] The capacity calculation module is configured to acquire real-time water level data, calculate gradient change of the water level in combination with the optimized water storage distribution model, determine water body change trend, acquire specific deviation position of real-time water body volume compared with historical data through the water body change trend, correct the optimized water storage distribution model according to the specific deviation position, and calculate current reservoir capacity according to the corrected model.

[0053] The trend analysis module is configured to acquire historical capacity data, compare and analyze the current reservoir capacity with the historical capacity data, and obtain capacity change trend.

[0054] The update output module is configured to acquire real-time monitoring data and compare the real-time monitoring data with a predicted value obtained according to the capacity change trend, update the initial digital elevation model according to a comparison result, calibrate the current reservoir capacity according to the updated elevation model, and generate precise monitoring result of the final reservoir capacity.

[0055] In a third aspect, the present application further provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the water reservoir capacity dynamic monitoring method of the unmanned aerial vehicle laser radar according to any one of the above.

[0056] In a fourth aspect, the present application further provides a computer readable storage medium, comprising a stored computer program, wherein the computer program controls a device where the computer readable storage medium is located to execute the water reservoir capacity dynamic monitoring method of the unmanned aerial vehicle laser radar according to any one of the above when the computer program is running.

[0057] Compared with the prior art, the present application has the following beneficial effects:

[0058] (1) The present application collects reservoir area images by the unmanned aerial vehicle laser radar, and performs three-dimensional reconstruction on the images by using the digital elevation model processing to obtain high-precision topographic data to capture slope distribution and concave-convex change characteristics; this technical means breaks through the limitation of the traditional monitoring method which relies on fixed point measurement or manual survey, realizes comprehensive and accurate coverage of the reservoir area under complex terrain, solves the problem that the prior art cannot accurately reflect the subtle changes of the terrain due to limited data acquisition range and insufficient precision, and lays a reliable data foundation for subsequent capacity calculation.

[0059] (2) The application extracts features from the area exceeding the preset slope threshold based on the slope distribution map and the concave-convex change layer obtained by raster analysis, obtains the irregularity parameters of the local terrain, and judges the water storage area of the whole reservoir according to the water storage distribution model, and generates the adjusted water storage distribution model; the processing mechanism combining quantitative analysis of terrain features and artificial intelligence recognition can accurately describe the influence mechanism of slope distribution and concave-convex change on reservoir capacity, solve the systematic deviation problem of capacity estimation caused by ignoring the complexity of terrain or insufficient analysis means in the prior art, and effectively improve the scientificity and accuracy of capacity calculation under complex terrain conditions.

[0060] (3) The application integrates the adjusted water storage distribution model and real-time water level data, uses interpolation calculation to obtain the current water reservoir capacity, and compares and analyzes the historical terrain data through convolutional neural network when the capacity deviation exceeds the threshold, and then updates the digital elevation model to optimize the influence mechanism; the closed-loop feedback optimization process of "data collection-model construction-dynamic calibration-trend prediction" makes the whole monitoring system have the real-time perception, accurate calculation and forward prediction ability of the dynamic change of the reservoir (such as water level fluctuation and terrain evolution), solves the problem that the monitoring result lags behind and cannot effectively support flood control scheduling and water resource allocation caused by the lack of real-time data fusion and dynamic model updating mechanism in the prior art, and realizes high-precision real-time dynamic monitoring and decision support of reservoir capacity. BRIEF DESCRIPTION OF DRAWINGS

[0061] Figure 1 is a reservoir capacity dynamic monitoring method flow diagram of a UAV laser radar provided by the first embodiment of the application;

[0062] Figure 2 is a reservoir capacity dynamic monitoring system structure diagram of a UAV laser radar provided by the second embodiment of the application. DETAILED DESCRIPTION

[0063] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the application.

[0064] With reference to Figure 1 The first embodiment of the application provides a reservoir capacity dynamic monitoring method of a UAV laser radar, including the following steps:

[0065] S11, obtain point cloud data of a reservoir area by a UAV laser radar, construct an initial digital elevation model according to the point cloud data, extract concave-convex change features from the initial digital elevation model and perform filtering processing to obtain high-precision terrain data;

[0066] S12, perform grid unit division according to the high-precision terrain data, calculate a slope value for a grid unit to obtain a slope distribution map, perform curvature calculation according to the slope distribution map, and perform smoothing correction if the curvature appears abnormal fluctuation to obtain a concave-convex change layer;

[0067] S13, extract a high-slope area from the slope distribution map, perform spatial matching of the high-slope area and the concave-convex change layer to obtain terrain feature data, perform edge detection and terrain parameter quantization processing for a part with significant concave-convex change in the terrain feature data to obtain terrain parameters of a local terrain;

[0068] S14, perform layering processing on a water storage space and integrate the terrain parameters, calculate an accumulated water area proportion according to the integrated data, perform local adjustment on a preset initial water storage distribution model according to the accumulated water area proportion to obtain an optimized water storage distribution model;

[0069] S15, obtain real-time water level data, calculate gradient change of the water level in combination with the optimized water storage distribution model, determine a water body change trend, obtain a specific deviation position of real-time water body volume compared with historical data through the water body change trend, correct the optimized water storage distribution model according to the specific deviation position, and calculate a current reservoir capacity according to the corrected model;

[0070] S16, obtain historical capacity data, compare and analyze the current reservoir capacity and the historical capacity data to obtain a capacity change trend;

[0071] S17, compare real-time monitoring data with a predicted value obtained according to the capacity change trend, update the initial digital elevation model according to a comparison result, calibrate the current reservoir capacity according to the updated elevation model, and generate a precise monitoring result of a final reservoir capacity.

[0072] In step S11, point cloud data of a reservoir area is obtained by a UAV laser radar, an initial digital elevation model is constructed according to the point cloud data, concave-convex change features are extracted from the initial digital elevation model and filtering processing is performed to obtain high-precision terrain data, including:

[0073] Complete point cloud data of a reservoir area is obtained by a UAV laser radar, gridding processing is performed and interpolation filling is performed on a data missing area to obtain a terrain height map;

[0074] According to the terrain height map, a three-dimensional reconstruction process is performed to obtain an initial digital elevation model;

[0075] According to the initial digital elevation model, a concave-convex change feature extraction is performed, and a filtering process is performed to obtain high-precision terrain data.

[0076] It should be noted that the unmanned aerial vehicle laser radar obtains high-density three-dimensional point cloud data of the reservoir area by emitting laser pulses and receiving echoes, and these data contain surface elevation information. However, due to flight path, vegetation obstruction or device error, the point cloud data may have uneven density or local missing problems, so it is necessary to generate a continuous and complete terrain height map through gridding processing and interpolation filling to provide reliable input for subsequent three-dimensional reconstruction; the three-dimensional reconstruction process converts the two-dimensional height map into an initial digital elevation model based on the digital elevation model algorithm, which accurately reflects the three-dimensional terrain structure of the reservoir area in the form of a regular grid to accurately reflect the slope distribution and concave-convex change characteristics of the reservoir area. The concave-convex change feature extraction identifies the surface undulation through terrain analysis and eliminates noise interference by combining the mean filtering method to ensure the high precision and reliability of the terrain data, which is the basis for realizing accurate monitoring of the reservoir capacity.

[0077] In one implementation, the present embodiment deploys an unmanned aerial vehicle carrying a laser radar device (such as RIEGL VUX-1LR) to perform a flight task above the reservoir area to collect point cloud data in real time; the density threshold of the point cloud data is preset to 10 points per square meter, which is determined based on statistical analysis of typical reservoir terrain monitoring requirements. Specifically, by comparing the accuracy of terrain models generated by point clouds with different densities in historical flight data, it is found that when the density is less than 10 points per square meter, the model error rate increases significantly (for example, more than 5%), so the threshold is set to 10 to balance data quality and collection efficiency. If the data density collected by the initial flight is only 5 points per square meter, adjust the flight parameters, such as reducing the flight height from 100 meters to 50 meters or slowing down the flight speed from 10 meters per second to 5 meters per second, and re-collect until the threshold requirement is met to obtain complete point cloud data. Subsequently, a digital elevation processing (such as Global Mapper) is used to perform gridding processing on the point cloud data, divide the point cloud into 1 meter x 1 meter grids, calculate the average height value in each grid, and generate a preliminary terrain height map; for the data missing area (such as the void caused by water reflection or vegetation coverage) in the height map, the present embodiment fills it by the inverse distance weighted interpolation method, and the interpolation formula is

[0078]

[0079] where z is the elevation value of the point to be filled, is the elevation value of the surrounding known point, The Euclidean distance is used to measure the distance between the to-be-filled point and the known point, and p is a power parameter (usually set to 2).

[0080] Exemplarily, the height values of the grids around a certain cavity are 50.1 meters, 50.3 meters, 49.8 meters, and 50.0 meters, respectively, and the filling value obtained after distance weight calculation is 50.05 meters, thereby forming a continuous terrain height map. According to the terrain height map, the initial digital elevation model of the reservoir area is constructed by three-dimensional reconstruction (such as ArcGIS Pro) in this embodiment, the height map is converted into a grid surface by using a triangular mesh generation, and if the slope distribution deviation after reconstruction exceeds a preset range, such as the standard slope change should be within 5 degrees, and the actual slope of part of the area suddenly changes to 10 degrees, the reconstruction parameters are adjusted, such as the grid resolution is optimized from 1 meter to 0.5 meter or the height data is corrected, and the initial digital elevation model is recalculated until the slope distribution is reasonable, thereby obtaining an accurate initial digital elevation model.

[0081] Finally, according to the initial digital elevation model, the concave-convex change feature is extracted by terrain analysis (such as SAGA GIS) in this embodiment, the surface concave-convex is identified by calculating the curvature index, the mean filter method is used for smoothing processing for the noise (such as height sudden change caused by abnormal laser radar reflection) in the feature data, exemplarily, the original height of a certain noise point is 55 meters, the height of the surrounding points is 52 meters, 53 meters, and 54 meters, and the height is updated to 53 meters after filtering, and the boundary detection is used to determine the position of the terrain edge, identify the junction of the reservoir bank line and the water surface, and finally obtain high-precision terrain data as the basis for subsequent capacity analysis.

[0082] In step S12, according to the high-precision terrain data, the grid unit is divided, the slope value is calculated for the grid unit, the slope distribution map is obtained, the curvature is calculated according to the slope distribution map, the smoothing correction is performed if the curvature appears abnormal fluctuation, and the concave-convex change layer is obtained, including:

[0083] The high-precision terrain data is rasterized and unit-divided and interpolated to obtain a grid unit data set;

[0084] The slope distribution is calculated for the grid unit data set, the area with a slope exceeding a preset slope threshold is marked, and a slope distribution map is obtained;

[0085] According to the slope distribution map, the curvature is calculated, and the smoothing correction is performed if the curvature appears abnormal fluctuation, and a concave-convex change layer is obtained.

[0086] In an implementation manner, the embodiment divides high-precision terrain data into 2-meter-by-2-meter grid cells by rasterization, which is based on the demand analysis of reservoir terrain monitoring accuracy. Specifically, by comparing the detail retention of terrain models generated by different grid sizes, it is found that when the cell size is greater than 2 meters, the recognition rate of key terrain features (such as cliffs and valleys) decreases by more than 10%. Therefore, 2 meters are selected to balance the calculation efficiency and feature fidelity. In view of the problem of incomplete data in the boundary area, the embodiment is supplemented by the same inverse distance weighted interpolation idea as step S11. For example, the elevation values of the four cells around a certain boundary cell are 102.5 meters, 103.1 meters, 101.8 meters, and 102.9 meters. After distance weight calculation, the filling value is 102.6 meters, thereby forming a complete grid cell data set.

[0087] For the grid cell data set, the embodiment uses slope calculation (such as GDAL) to analyze the slope distribution and calculate the slope value of each cell. The area with a slope exceeding a preset slope threshold is marked. The Horn calculation formula is

[0088]

[0089] wherein and are the elevation change rates in the x and y directions, respectively, is the calculated slope value. If the calculated slope value exceeds the preset slope threshold, it is marked as a key area, and a slope distribution map is generated.

[0090] The preset slope threshold is determined based on statistical analysis of reservoir safety monitoring standards. Historical data shows that the probability of terrain change in areas with a slope exceeding 15 degrees increases by about 30%. Therefore, the slope threshold is set to 15 degrees.

[0091] According to the slope distribution map, the embodiment uses the following formula to calculate the curvature:

[0092]

[0093] wherein z represents the elevation value of the grid cell, and represent the second-order partial derivatives of the elevation in the x and y directions, respectively; C is the comprehensive curvature index, and its numerical value reflects the strength of the concave-convex change of the ground surface. A positive value indicates a convex feature of the ground surface, and a negative value indicates a concave feature.

[0094] In a specific implementation, the second-order partial derivative is calculated by a center difference method of a 3*3 moving window; if the curvature index appears abnormal fluctuation (for example, the curvature value of a certain unit reaches 0.8, while the normal range is between-0.3 and 0.3), the Gaussian filter is used for smoothing correction, and the filter formula is

[0095]

[0096] Exemplarily, an abnormal unit curvature 0.8 is corrected to 0.25 after filtering, and finally a concave-convex change layer is obtained, which provides reliable input for subsequent feature extraction.

[0097] In step S13, a high-slope region is extracted from the slope distribution map, the high-slope region is spatially matched with the concave-convex change layer to obtain terrain feature data, edge detection and terrain parameter quantization processing are performed on a part with significant concave-convex change in the terrain feature data, and terrain parameters of a local terrain are obtained, including:

[0098] A grid unit region exceeding a preset slope threshold is extracted from the slope distribution map, and is marked to obtain a marked high-slope region;

[0099] The high-slope region is spatially matched with the concave-convex change layer to obtain superimposed terrain feature data;

[0100] If there is significant concave-convex change in the terrain feature data, edge detection is performed on the part with significant concave-convex change to obtain a local terrain irregular region;

[0101] According to the local terrain irregular region, terrain parameter quantization is performed to extract relief amplitude and boundary sharpness indexes, and terrain parameters of a local terrain are obtained.

[0102] In an implementation manner, the embodiment extracts a grid unit region exceeding a preset slope threshold of 15 degrees from a slope distribution map, marks these units through data screening (such as Python GeoPandas) to obtain a marked high-slope region, and exemplarily, 2000 high-slope regions are screened from 10000 grid units in a certain reservoir region.

[0103] Subsequently, the embodiment superimposes and analyzes the high-slope region and the corresponding concave-convex change layer by spatial matching (such as ArcGIS Spatial Analyst), fuses the slope data and the concave-convex data through coordinate alignment, and generates superimposed terrain feature data.

[0104] If some areas in the terrain feature data show significant concave-convex changes, a Canny edge detection algorithm is applied for boundary recognition to obtain a local terrain irregular area. The judgment conditions of the significant concave-convex changes include that the absolute value of curvature is greater than 0.5 and the concave-convex height difference is more than 1 meter. The conditions are determined based on statistical analysis of historical terrain data. Statistics show that about 85% of the areas that meet the two conditions simultaneously have obvious terrain mutation characteristics. The absolute value of curvature reflects the bending degree of the ground surface, and the concave-convex height difference represents the local terrain fluctuation amplitude. The combination of the two can effectively identify the terrain area with significant changes. The Canny edge detection algorithm extracts the terrain mutation edge through the steps of Gaussian filtering, gradient calculation and non-maximum suppression. For example, in a high slope area with significant concave-convex changes, an irregular polygon area is identified after edge detection, and the boundary point sequence is [(120, 80), (125, 82), (130, 85)].

[0105] According to the local terrain irregular area, the embodiment uses parameter calculation to quantify the terrain parameters, extracts the fluctuation amplitude and the boundary sharpness index. The fluctuation amplitude is the difference between the maximum and minimum values of the elevation value set in the area, and the boundary sharpness calculation formula is

[0106]

[0107] Wherein R is the boundary sharpness, n represents the total number of boundary line segments, is the elevation difference between adjacent boundary points, is the horizontal distance. For example, in a certain irregular area, the elevation value changes from 50 meters to 53 meters, the fluctuation amplitude is 3 meters, and the boundary sharpness index is 0.6. The final terrain parameters of the local terrain are generated for subsequent bias correction in the stratified simulation of the water storage space.

[0108] In step S14, the water storage space is stratified and integrated with the terrain parameters. The integrated data is used to calculate the area proportion of the water accumulation area. According to the area proportion of the water accumulation area, the initial water storage distribution model is locally adjusted to obtain an optimized water storage distribution model, including:

[0109] The water storage space is preliminarily stratified and integrated with the terrain parameters to obtain stratified basic data.

[0110] The stratified basic data is spatially matched with the concave-convex change layer. The areas with concave-convex changes exceeding a first preset threshold are marked as potential water accumulation areas to obtain a potential water accumulation range.

[0111] For the potential water accumulation range, terrain parameter difference value analysis is performed. If the difference value exceeds a second preset threshold, it is determined as a water accumulation affected area to obtain the specific position of the water accumulation.

[0112] According to the specific location of the accumulated water, an area proportion of the accumulated water area is calculated, and a preset initial water storage distribution model is locally adjusted in combination with a preset influence factor to obtain an optimized water storage distribution model.

[0113] In an implementation manner, the embodiment preliminarily divides the water storage space of the reservoir according to the terrain parameters by using data integration, divides the reservoir into three levels of an upper layer (0-5 meters), a middle layer (5-10 meters) and a lower layer (10-15 meters) according to every 5-meter depth interval, and determines the division scheme based on historical statistical analysis of typical reservoir depth distribution. Statistical data shows that about 85% of reservoir capacity changes occur within 15 meters in depth, so this division range is selected to ensure the balance between calculation accuracy and efficiency. After the division, the data is integrated with the terrain parameters to generate layered basic data containing terrain parameters and water depth intervals of each layer.

[0114] According to the layered basic data, the embodiment uses spatial matching (such as ArcGIS Spatial Analyst) to perform overlay analysis on the concave-convex change layer, and realizes accurate matching through coordinate alignment. If the height difference of the concave-convex change of a layer exceeds a first preset threshold value of 1.5 meters, the threshold value is determined based on analysis of historical accumulated water cases, and statistics show that the probability of the area with a height difference exceeding 1.5 meters to form accumulated water is more than 70%, the area corresponding to the layer is marked as a potential accumulated water area. Exemplarily, the height difference of the concave-convex change of a certain area in the middle layer (5-10 meters) of the reservoir reaches 2.0 meters, which exceeds the first preset threshold value. The system automatically marks the area and calculates the distribution range as a 300-square-meter area on the east side of the reservoir to obtain a potential accumulated water range.

[0115] For the potential accumulated water range, the embodiment performs differential analysis on the terrain parameters, extracts the slope and curvature parameters in the area, and performs Min-Max normalization processing. The difference value formula is calculated by comparing with the surrounding non-accumulated water area.

[0116]

[0117] Wherein D is the difference value, n is the total number of terrain parameters participating in the analysis, is the normalized average value of the i-th parameter in the potential accumulated water area, and are the mean and standard deviation of the i-th parameter in the surrounding non-accumulated water area, respectively. If the difference value D exceeds a second preset threshold value of 0.6 (the threshold value is determined based on statistical analysis of the variance of the terrain parameters, and historical data shows that about 80% of the areas with a difference value exceeding 0.6 are confirmed to have a significant impact on capacity calculation), it is determined as an accumulated water influence area, and the specific location of the accumulated water is obtained.

[0118] Exemplarily, the average slope in the potential water accumulation area is 0.62 (normalized value), the average curvature is 0.58, the average slope of the surrounding area is 0.38, the standard deviation is 0.12, the average curvature of the surrounding area is 0.31, and the standard deviation is 0.09. The difference value is calculated as 2.5, which exceeds the threshold value of 0.6, and it is determined that the area is a water accumulation affected area and its specific location is recorded.

[0119] According to the specific location of the water accumulation, the area proportion of the water accumulation area is calculated, and the preset initial water storage distribution model is locally adjusted combined with the preset influence factor to obtain an optimized water storage distribution model. The local adjustment formula is

[0120]

[0121] wherein is the adjusted capacity value, is the original capacity value, is the preset influence factor (0.2 based on historical data calibration), is the water accumulation area proportion. Exemplarily, the original capacity value of a certain area is 5000 cubic meters, the water accumulation area proportion is 10%, and the adjusted capacity value is 5000x(1-0.2x0.1)=4900 cubic meters. The final optimized water storage distribution model is generated, effectively correcting the capacity deviation caused by irregular terrain.

[0122] It should be noted that the preset initial water storage distribution model is constructed based on reservoir design parameters and standard terrain data, with design water depth, theoretical slope, etc. as the core input. According to the principle of uniform water depth distribution, the water storage space is divided into standard units, and the water storage capacity of each unit is calculated by multiplying the unit area by the design water depth. The integrated basic water storage distribution model covers the entire area and is used to preliminarily represent the water storage capacity and distribution law.

[0123] In step S15, real-time water level data is obtained, and the gradient change of the water level is calculated combined with the optimized water storage distribution model to determine the water body change trend. Through the water body change trend, the specific deviation position of the real-time water body volume compared with the historical data is obtained, and the optimized water storage distribution model is corrected according to the specific deviation position. The current reservoir capacity is calculated according to the corrected model, including:

[0124] Real-time water level data is obtained, and data is arranged to obtain water level fluctuation value. If the water level fluctuation value exceeds the preset fluctuation threshold, it is marked as an abnormal water level point to obtain an abnormal distribution range.

[0125] According to the abnormal distribution range, the real-time water level data is matched with the coordinates of the optimized water storage distribution model, and the gradient change of the water level is calculated to determine the water body change trend.

[0126] By analyzing the trend of water body changes, combined with the optimized water storage distribution model and the real-time water level data, the real-time water volume is estimated. If the difference between the estimation result and the historical data exceeds the preset deviation threshold, it is marked as a volume deviation area, and the specific location of the deviation is obtained.

[0127] Based on the specific location of the deviation, the optimized water storage distribution model is locally corrected, and the reservoir capacity is calculated based on the corrected model to obtain the current reservoir capacity.

[0128] In one implementation, this embodiment acquires real-time water level data from water level sensors (such as pressure level gauges) deployed at key locations in the reservoir. The sampling frequency is once per hour. Data processing (such as Python NumPy) is used to clean and format the raw data. The water level fluctuation value at adjacent time points is calculated. If the fluctuation value exceeds a preset fluctuation threshold of 0.3 meters per hour, this threshold is determined based on historical statistical analysis of the normal water level fluctuation range of the reservoir. The data shows that 95% of the normal fluctuations are within 0.3 meters per hour. Then, the point is marked as an abnormal water level point. For example, if a sensor shows that the water level rises from 105.2 meters to 105.8 meters in one hour, with a change rate of 0.6 meters per hour, which exceeds the preset fluctuation threshold, the point is marked and, together with the surrounding sensor data, the abnormal distribution range is determined to be a 150-square-meter area on the north side of the reservoir.

[0129] Based on the aforementioned abnormal distribution range, spatial mapping (such as QGIS Processing) is used to match the real-time water level data with the optimized water storage distribution model to calculate the gradient change of the water level. Discrete water level points are mapped to the water storage distribution grid using bilinear interpolation to obtain the dynamic water body change trend. The formula for calculating the gradient change is as follows:

[0130]

[0131] in The gradient is Δh, the amount of water level change, and Δt, the time interval. For example, the water level in the northern area rises from 105.8 meters to 106.5 meters in 3 hours, with a gradient of 0.23 meters per hour. The water change trend is a rapid upward trend concentrated in this area.

[0132] Based on the water body change trend, combined with the optimized water storage distribution model and the real-time water level data, the real-time water volume is calculated using the trapezoidal integral method formula:

[0133]

[0134] Where V is the water volume and n is the number of aquifers. Let be the water storage area of ​​the i-th layer. is the water storage area of the i+1th layer, and Δh is the water level change. If the estimation result differs from the historical data by more than a preset deviation threshold of 5% (the threshold is determined based on the requirement for the accuracy of capacity calculation, and the industry standard generally allows a capacity error range of ±5%), the area is marked as a volume deviation area, and the specific location of the deviation is obtained.

[0135] For example, the current estimated volume is 85,000 cubic meters, the historical same-period data is 90,000 cubic meters, the difference is 5.6%, which exceeds the threshold, and the area is marked and the specific location is determined to be the 200 square meter range around the coordinates (100, 120) of the center area of the reservoir.

[0136] According to the specific location of the deviation, the embodiment locally corrects the optimized water storage distribution model, adjusts the capacity of the area by using a weighted average method, and the correction formula is

[0137]

[0138] wherein are the capacities before and after correction, w is a weight factor (set to 0.8 based on historical calibration), ΔV is the volume difference, is the historical reference capacity value.

[0139] For example, if there is a capacity deviation in the center area of the reservoir, the original capacity value of the deviation area is 1,000 cubic meters, the current estimated volume is 85,000 cubic meters, and the historical same-period reference capacity is 90,000 cubic meters, and the total volume difference is 5,000 cubic meters. According to the area ratio of 5% of the deviation area, the volume difference of the area is calculated to be 50 cubic meters, and the correction formula is substituted to obtain Vnew as 999.56 cubic meters. Local correction adjusts the overall capacity value from 85,000 cubic meters to 84,999.56 cubic meters. Through successive correction of multiple deviation areas, the final output of the current reservoir capacity value is 84,500 cubic meters.

[0140] In step S16, historical capacity data is obtained, the current reservoir capacity is compared and analyzed with the historical capacity data, and a capacity change trend is obtained, including:

[0141] The historical capacity data is obtained from a preset topographic database, and is preliminarily compared with the current reservoir capacity, the capacity deviation is calculated, and the specific distribution range of the capacity deviation is obtained.

[0142] The historical topographic data is obtained from the topographic database and compared with the high-precision topographic data, and topographic change data is obtained.

[0143] The correlation degree of the specific distribution range and the terrain change data is calculated, and if the correlation degree exceeds a preset correlation threshold, the corresponding region is marked as an abnormal position;

[0144] For the abnormal position, feature extraction is performed on the historical terrain data by a convolutional neural network to obtain dynamic trend features of terrain change, and mutual information of the dynamic trend features and the current reservoir capacity is calculated.

[0145] The correlation degree, the dynamic trend features, the terrain change data, and the capacity deviation are integrated to generate a change trend report to obtain a capacity change trend.

[0146] In an implementation manner, the embodiment obtains historical capacity data in the past 5 years from a preset terrain database, including elevation models and capacity statistics of each year, calculates the difference between the current reservoir capacity 85000 cubic meters and the historical data of the same period by preliminary comparison (such as Python Scikit-learn), and if the difference exceeds a preset capacity deviation threshold 5000 cubic meters, the threshold is determined based on statistical analysis of the interannual capacity fluctuation range of the reservoir, and the historical data shows that 95% of the normal interannual change is within 5000 cubic meters, the specific distribution range of the capacity deviation is marked. Illustratively, the difference between the current reservoir capacity 85000 cubic meters and the historical average value 90000 cubic meters of the same period is spatially distributed by spatial interpolation calculation, and the significant concentrated area of the difference is identified by K-means clustering to obtain the specific distribution range of the capacity deviation as a 400 square meter area on the south side of the reservoir.

[0147] The correlation degree of the specific distribution range and the terrain change data (such as annual elevation change rate) is calculated, the Pearson correlation coefficient is taken as the correlation index, and if the absolute value of the calculated Pearson correlation coefficient exceeds a preset correlation threshold 0.7, the threshold is determined based on historical analysis of the terrain-capacity correlation strength, and statistics show that 85% of the regions with a correlation coefficient exceeding 0.7 have significant capacity abnormalities, and the region coordinates (80, 150) to (120, 180) are determined as the abnormal position. Illustratively, the correlation coefficient of the south side area terrain subsidence rate and the capacity reduction is 0.75, which exceeds the threshold, and the system determines that the region coordinates (80, 150) to (120, 180) are abnormal positions.

[0148] For the abnormal position, the embodiment extracts features from historical terrain data through a convolutional neural network (such as CNN with ResNet architecture), obtains dynamic trend features of terrain changes, and calculates the mutual information between the dynamic trend features and the current reservoir capacity. The convolutional neural network model adopts ResNet-18 architecture, contains 17 convolutional layers and 1 fully connected layer, the input is continuous 5 period elevation image time series data (size is 256*256 pixels), and the residual connection is used to relieve the gradient vanishing problem; the mean square error is used as the loss function in the training process, the Adam optimizer is used for parameter update, the learning rate is set to 0.001, the batch size is 32, and the model converges after 100 training cycles. The output is the dynamic trend feature of terrain change (such as subsidence acceleration, erosion mode).

[0149] For example, the model extracts the dynamic trend feature of the south side area, which shows that the annual average subsidence rate accelerates from 0.1 m / year to 0.3 m / year, and then calculates the correlation degree of the feature and the current reservoir capacity, using the mutual information index formula:

[0150] I(X;Y)=

[0151] Where X is the trend feature, Y is the capacity value, is the data value in the trend feature, is the data value in the capacity value, the mutual information value is 0.8, indicating strong correlation.

[0152] Integrate the correlation degree, the dynamic trend feature, the terrain change data and the capacity deviation to generate a change trend report. Integrate key indicators such as capacity deviation (current 5.6%), terrain change data terrain subsidence rate (0.3 m / year) and correlation degree (mutual information 0.8), compare these indicators with historical benchmarks, for example, the capacity deviation exceeds the threshold of 3% of the historical average value, and the conclusion of the capacity change trend is "continuous decline and acceleration", which provides early warning information for water resources management departments.

[0153] In step S17, real-time monitoring data is obtained and compared with the predicted value obtained according to the capacity change trend, and the initial digital elevation model is updated according to the comparison result, and the current reservoir capacity is calibrated according to the updated elevation model, and the final reservoir capacity is generated. Precise monitoring results, including:

[0154] Obtain real-time monitoring data, and compare it with the predicted value obtained according to the capacity change trend to obtain the change distribution characteristics of the real-time monitoring data;

[0155] According to the change distribution characteristics, determine the key area of terrain dynamic adjustment;

[0156] For the key area, the initial elevation model is updated in combination with the real-time monitoring data, the current reservoir capacity is calibrated according to the updated elevation model, and the accurate monitoring result of the final reservoir capacity is obtained.

[0157] In an implementation manner, the embodiment acquires real-time monitoring data (such as elevation data collected by a GNSS receiver, with a sampling frequency of once a day), calculates the difference between the real-time data and the capacity change trend by preliminary comparison, and the difference calculation formula is

[0158]

[0159] Among them, is the calculated difference value, is the real-time elevation value, is the predicted value, which is obtained by simulating and predicting the elevation value according to the capacity change trend, if the difference value exceeds the preset real-time change threshold value 0.2 meters (the threshold value is determined based on statistical analysis of the accuracy of the monitoring equipment and the natural fluctuation range, the equipment measurement error is ±0.1 meters, and the natural daily fluctuation is usually less than 0.1 meters), it is marked as significant change, and the change distribution characteristics of the real-time monitoring data are obtained.

[0160] Exemplarily, the real-time elevation of a certain node is 102.3 meters, the predicted value is 102.5 meters, the difference is 0.2 meters, which reaches the threshold value, and the change distribution characteristics of the real-time monitoring data are determined as the systematic elevation drop of the west side area of the reservoir.

[0161] According to the change distribution characteristics and the complex terrain characteristics (such as the area with a slope greater than 15 degrees and an absolute curvature greater than 0.3) in the initial digital elevation model, the Jaccard similarity coefficient formula is calculated

[0162]

[0163] Among them, A is the real-time change area, and B is the complex terrain area, if the coefficient exceeds the preset correlation threshold value 0.6 (the threshold value is determined based on experimental analysis of terrain sensitivity, and experiments show that 90% of the areas with a similarity coefficient exceeding 0.6 need model adjustment), it is determined as the key area of terrain dynamic adjustment.

[0164] Exemplarily, the similarity coefficient of the west side change area and the complex terrain area reaches 0.65, and the system determines that the coordinates (50, 100) to (80, 130) are the key area of terrain dynamic adjustment.

[0165] For the key area, real-time change data and complex terrain features are integrated, and Kalman filtering is used for data fusion. The state equation of Kalman filtering is established based on the initial elevation change model, and the state variable is defined as the terrain elevation and its change rate to describe the dynamic process of the evolution of the terrain over time. The observation equation directly uses the real-time monitored elevation data as the observation variable to reflect the direct measurement of the system state. If the deviation of the fused data exceeds the preset integration threshold of 0.15 meters (this threshold is determined based on the model accuracy requirement, and the goal is to control the root mean square error of the digital elevation model within 0.1 meters), the initial digital elevation model in step S11 is updated locally.

[0166] For example, the original elevation of a certain point in the key area is 100.5 meters, the real-time monitoring value is 100.3 meters, and the updated value after Kalman filtering is 100.35 meters, with a deviation of 0.15 meters reaching the threshold. The system obtains the updated terrain dynamic adjustment result, and the overall model accuracy is improved to a root mean square error of 0.08 meters. According to the updated terrain dynamic adjustment result, the embodiment uses an influence mechanism optimization degree comparison analysis method such as a sensitivity analysis method to calculate the difference coefficient formula of the capacity monitoring results before and after optimization

[0167]

[0168] where σ is the standard deviation and μ is the average value. If the difference coefficient decreases from 0.12 before optimization to 0.08 (the preset optimization threshold is 0.10), it is determined that the influence mechanism has been optimized. For example, the difference coefficient of the capacity monitoring value after optimization is 0.08, which is lower than the threshold. The system confirms the final result of the monitoring data calibration as the reservoir capacity of 84500 ± 500 cubic meters (confidence level 95%), which is 40% less than the initial monitoring result, significantly improving the accuracy and reliability of the dynamic monitoring of the reservoir capacity.

[0169] In summary, the application discloses a reservoir capacity dynamic monitoring method of an unmanned aerial vehicle laser radar, which comprises the following steps: acquiring point cloud data of a reservoir area by means of an unmanned aerial vehicle laser radar, constructing an initial digital elevation model according to the point cloud data, extracting concave-convex change features from the initial digital elevation model and performing filtering processing to obtain high-precision terrain data, performing grid unit division according to the high-precision terrain data, calculating a slope value for the grid unit to obtain a slope distribution map, performing curvature calculation according to the slope distribution map, performing smoothing correction if the curvature appears abnormal fluctuation to obtain a concave-convex change layer, extracting a high-slope area from the slope distribution map, performing spatial matching on the high-slope area and the concave-convex change layer to obtain terrain feature data, performing edge detection and terrain parameter quantization processing on the part with significant concave-convex change in the terrain feature data to obtain terrain parameters of a local terrain, performing layered processing on a water storage space and integrating the terrain parameters, calculating an accumulated water area proportion according to the integrated data, performing local adjustment on a preset initial water storage distribution model according to the accumulated water area proportion to obtain an optimized water storage distribution model, acquiring real-time water level data, combining the optimized water storage distribution model to calculate gradient change of the water level, determining a water body change trend, acquiring a specific deviation position of real-time water body volume compared with historical data through the water body change trend, correcting the optimized water storage distribution model according to the specific deviation position, calculating a current reservoir capacity according to the corrected model, acquiring historical capacity data, comparing and analyzing the current reservoir capacity and the historical capacity data to obtain a capacity change trend, acquiring real-time monitoring data and comparing the real-time monitoring data with a predicted value obtained according to the capacity change trend, updating the initial digital elevation model according to a comparison result, calibrating the current reservoir capacity according to the updated elevation model, and generating a precise monitoring result of a final reservoir capacity. Through the closed-loop feedback optimization process of “data acquisition-model construction-dynamic calibration-trend prediction”, the whole monitoring system has the real-time perception, precise calculation and forward prediction ability for dynamic changes (such as water level fluctuation and terrain evolution) of the reservoir, solves the problem that the monitoring result lags behind and cannot effectively support flood control scheduling and water resource allocation due to the lack of real-time data fusion and dynamic model updating mechanism in the prior art, and realizes high-precision real-time dynamic monitoring and decision support of the reservoir capacity.

[0170] Reference Figure 2 The second embodiment of the application provides a reservoir capacity dynamic monitoring system of an unmanned aerial vehicle laser radar, which comprises:

[0171] A data acquisition module acquires point cloud data of a reservoir area by means of an unmanned aerial vehicle laser radar, constructs an initial digital elevation model according to the point cloud data, extracts concave-convex change features from the initial digital elevation model and performs filtering processing to obtain high-precision terrain data.

[0172] The feature analysis module performs grid unit division according to the high-precision terrain data, calculates a slope value for each grid unit to obtain a slope distribution map, performs curvature calculation according to the slope distribution map, and performs smoothing correction if the curvature appears abnormal fluctuation to obtain a concave-convex change layer;

[0173] The terrain parameter acquisition module extracts a high-slope area from the slope distribution map, performs spatial matching on the high-slope area and the concave-convex change layer to obtain terrain feature data, performs edge detection and terrain parameter quantization processing on a part of the terrain feature data that has significant concave-convex changes to obtain terrain parameters of a local terrain;

[0174] The modeling module performs layered processing on the water storage space and integrates the terrain parameters, calculates an accumulated water area proportion according to the integrated data, performs local adjustment on a preset initial water storage distribution model according to the accumulated water area proportion to obtain an optimized water storage distribution model;

[0175] The capacity calculation module obtains real-time water level data, combines the optimized water storage distribution model to calculate a gradient change of the water level, determines a water body change trend, obtains a specific deviation position of real-time water body volume compared with historical data through the water body change trend, corrects the optimized water storage distribution model according to the specific deviation position, and calculates a current reservoir capacity according to the corrected model;

[0176] The trend analysis module obtains historical capacity data, compares and analyzes the current reservoir capacity with the historical capacity data to obtain a capacity change trend;

[0177] The update output module obtains real-time monitoring data and compares the real-time monitoring data with a predicted value obtained according to the capacity change trend, updates the initial digital elevation model according to a comparison result, calibrates the current reservoir capacity according to the updated elevation model, and generates a precise monitoring result of a final reservoir capacity.

[0178] It should be noted that the reservoir capacity dynamic monitoring system of the unmanned aerial vehicle laser radar provided in the embodiments of the present application is used to execute all process steps of the reservoir capacity dynamic monitoring method of the unmanned aerial vehicle laser radar provided in the embodiments, and the working principles and beneficial effects of the two are one-to-one corresponding, and thus will not be described again.

[0179] The embodiments of the present application also provide an electronic device. The electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a reservoir capacity dynamic monitoring method program of an unmanned aerial vehicle laser radar. The processor implements the steps in the above various reservoir capacity dynamic monitoring method embodiments of the unmanned aerial vehicle laser radar when executing the computer program, such as Figure 1The step S11 is shown. Alternatively, the processor implements the functions of the modules / units in each of the above apparatus embodiments when executing the computer program, such as the data acquisition module.

[0180] For example, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present application. The one or more modules / units can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the electronic device.

[0181] The electronic device can be a desktop computer, a notebook computer, a palm computer, a smart tablet and the like. The electronic device can include, but is not limited to, a processor, a memory. Those skilled in the art can understand that the above components are only examples of the electronic device and do not constitute a limitation on the electronic device, and can include more or less components than the above, or combine certain components, or different components, for example, the electronic device can also include an input / output device, a network access device, a bus and the like.

[0182] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor and the like. The processor is the control center of the electronic device, and connects all parts of the electronic device through various interfaces and lines.

[0183] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0184] Wherein, if the modules / units integrated in the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0185] It should be noted that the apparatus embodiments described above are merely illustrative, and the units described as separate units can or can not be physically separate, and the units displayed as units can or can not be physical units, i.e. can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment scheme according to actual needs. In addition, the connection relationship between the modules in the apparatus embodiment provided by the present application indicates that there is a communication connection between them, which can be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement it without creative labor.

[0186] The above specific embodiments further illustrate the purpose, technical scheme and beneficial effects of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the protection scope of the present application. It is particularly pointed out that any modification, equivalent replacement, improvement, etc. made by those skilled in the art within the spirit and principles of the present application shall be included in the protection scope of the present application.​

Claims

1. A method for dynamic monitoring of reservoir capacity using UAV lidar, characterized in that, include: Point cloud data of the reservoir area is acquired by UAV lidar, an initial digital elevation model is constructed based on the point cloud data, and concave and convex variation features are extracted from the initial digital elevation model and filtered to obtain high-precision terrain data. Based on the high-precision terrain data, grid cells are divided, slope values ​​are calculated for each grid cell to obtain a slope distribution map, and curvature is calculated based on the slope distribution map. If abnormal fluctuations in curvature occur, smoothing correction is performed to obtain a concave-convex variation layer. High-slope areas are extracted from the slope distribution map, and spatial matching is performed between the high-slope areas and the unevenness variation layer to obtain terrain feature data. Edge detection and terrain parameter quantization are performed on the parts of the terrain feature data that have significant unevenness variations to obtain the terrain parameters of the local terrain. The water storage space is layered and integrated with the terrain parameters. The area ratio of the water accumulation area is calculated based on the integrated data. The preset initial water storage distribution model is locally adjusted based on the area ratio of the water accumulation area to obtain an optimized water storage distribution model. Real-time water level data is acquired, and the gradient change of water level is calculated in combination with the optimized water storage distribution model to determine the trend of water body change. Based on the trend of water body change, the specific deviation position of real-time water volume compared with historical data is obtained. The optimized water storage distribution model is corrected according to the specific deviation position, and the current reservoir capacity is calculated based on the corrected model. Historical capacity data is obtained, and the current reservoir capacity is compared and analyzed with the historical capacity data to obtain the capacity change trend; Real-time monitoring data is acquired and compared with the predicted value obtained based on the capacity change trend. The initial digital elevation model is updated based on the comparison result. The current reservoir capacity is calibrated based on the updated elevation model to generate the final accurate monitoring result of the reservoir capacity.

2. The method for dynamic monitoring of reservoir capacity using UAV lidar according to claim 1, characterized in that, The process involves acquiring point cloud data of the reservoir area using UAV lidar, constructing an initial digital elevation model (DEM) based on the point cloud data, extracting convexity / concave variation features from the initial DEM and performing filtering to obtain high-precision terrain data, including: Complete point cloud data of the reservoir area was obtained by using drone lidar, and then the data was processed into a grid and interpolated to fill in the missing data areas to obtain a terrain height map. Based on the terrain elevation map, a three-dimensional reconstruction process is performed to obtain an initial digital elevation model; Based on the initial digital elevation model, concave and convex variation features are extracted and filtered to obtain high-precision terrain data.

3. The method for dynamic monitoring of reservoir capacity using UAV lidar according to claim 1, characterized in that, The process involves dividing the high-precision terrain data into raster cells, calculating slope values ​​for each raster cell to obtain a slope distribution map, calculating curvature based on the slope distribution map, and smoothing out any abnormal fluctuations in curvature to obtain a convexity / concave variation layer. The high-precision terrain data is divided into raster cells and interpolated to obtain a raster cell dataset. For the raster cell dataset, the slope distribution is calculated, and areas with slopes exceeding a preset slope threshold are marked to obtain a slope distribution map; Based on the slope distribution map, the curvature is calculated. If abnormal fluctuations in curvature occur, smoothing correction is performed to obtain a concave-convex variation layer.

4. The method for dynamic monitoring of reservoir capacity using UAV lidar according to claim 1, characterized in that, The process involves extracting high-slope areas from the slope distribution map, spatially matching these high-slope areas with the unevenness layer to obtain terrain feature data, and then performing edge detection and terrain parameter quantization on the parts of the terrain feature data that exhibit significant unevenness to obtain local terrain parameters, including: Extract grid cell regions that exceed a preset slope threshold from the slope distribution map and mark them to obtain marked high slope regions. Spatially match the high-slope area with the unevenness layer to obtain superimposed terrain feature data; If significant unevenness or convexity changes occur in the terrain feature data, edge detection is performed on the parts with significant unevenness or convexity changes to obtain local terrain irregularities. Based on the irregular local terrain, terrain parameters are quantified, and undulation amplitude and boundary sharpness indices are extracted to obtain the local terrain parameters.

5. The method for dynamic monitoring of reservoir capacity using UAV lidar according to claim 1, characterized in that, The process of stratifying the water storage space and integrating it with the terrain parameters, calculating the area ratio of the water accumulation zone based on the integrated data, and locally adjusting the preset initial water storage distribution model based on the area ratio of the water accumulation zone to obtain an optimized water storage distribution model includes: The water storage space is initially stratified and integrated with the terrain parameters to obtain stratified basic data; Spatial matching is performed between the layered basic data and the unevenness variation layer. Areas where the unevenness variation exceeds a first preset threshold are selected and marked as potential water accumulation areas to obtain the potential water accumulation range. For the potential water accumulation area, a topographic parameter difference analysis is performed. If the difference value exceeds the second preset threshold, it is determined to be a water accumulation area, and the specific location of the water accumulation is obtained. Based on the specific location of the water accumulation, the area ratio of the water accumulation area is calculated, and the preset initial water storage distribution model is locally adjusted in conjunction with preset influence factors to obtain an optimized water storage distribution model.

6. The method for dynamic monitoring of reservoir capacity using UAV lidar according to claim 1, characterized in that, The process involves acquiring real-time water level data, calculating the gradient change of water level using the optimized water storage distribution model, determining the water body change trend, obtaining the specific deviation location of the real-time water volume compared to historical data based on the water body change trend, correcting the optimized water storage distribution model based on the specific deviation location, and calculating the current reservoir capacity based on the corrected model. This includes: Real-time water level data is acquired and processed to obtain water level fluctuation values. If the water level fluctuation values ​​exceed a preset fluctuation threshold, they are marked as abnormal water level points, and the abnormal distribution range is obtained. Based on the abnormal distribution range, the real-time water level data is matched with the optimized water storage distribution model to calculate the gradient change of the water level and determine the trend of water body change. By analyzing the trend of water body changes, combined with the optimized water storage distribution model and the real-time water level data, the real-time water volume is estimated. If the difference between the estimation result and the historical data exceeds the preset deviation threshold, it is marked as a volume deviation area, and the specific location of the deviation is obtained. Based on the specific location of the deviation, the optimized water storage distribution model is locally corrected, and the reservoir capacity is calculated based on the corrected model to obtain the current reservoir capacity.

7. The method for dynamic monitoring of reservoir capacity using UAV lidar according to claim 1, characterized in that, The step of acquiring historical capacity data and comparing and analyzing the current reservoir capacity with the historical capacity data to obtain the capacity change trend includes: Historical capacity data is obtained from a preset terrain database and compared with the current reservoir capacity to calculate the capacity deviation and obtain the specific distribution range of the capacity deviation. Historical terrain data is obtained from the terrain database and compared with the high-precision terrain data to obtain terrain change data; Calculate the correlation between the specific distribution range and the terrain change data. If the correlation exceeds a preset correlation threshold, mark the corresponding area as an abnormal location. For the abnormal location, a convolutional neural network is used to extract features from the historical terrain data to obtain dynamic trend features of terrain changes, and the mutual information between the dynamic trend features and the current reservoir capacity is calculated. By integrating the correlation degree, the dynamic trend characteristics, the terrain change data, and the capacity deviation, a change trend report is generated to obtain the capacity change trend.

8. The method for dynamic monitoring of reservoir capacity using UAV lidar according to claim 1, characterized in that, The process of acquiring real-time monitoring data and comparing it with predicted values ​​obtained based on the capacity change trend, updating the initial digital elevation model based on the comparison results, calibrating the current reservoir capacity based on the updated elevation model, and generating a final accurate monitoring result for the reservoir capacity includes: Real-time monitoring data is acquired and compared with the predicted values ​​obtained based on the capacity change trend to obtain the change distribution characteristics of the real-time monitoring data. Based on the aforementioned distribution characteristics of the changes, key areas for dynamic terrain adjustment are identified; For the key areas, the initial elevation model is updated based on the real-time monitoring data, and the current reservoir capacity is calibrated according to the updated elevation model to obtain the final accurate monitoring result of the reservoir capacity.

9. A dynamic monitoring system for reservoir capacity using UAV lidar, characterized in that, include: The data acquisition module is used to acquire point cloud data of the reservoir area through UAV lidar, construct an initial digital elevation model based on the point cloud data, extract concave and convex variation features from the initial digital elevation model and perform filtering processing to obtain high-precision terrain data. The feature analysis module is used to divide the high-precision terrain data into grid cells, calculate the slope value for each grid cell to obtain a slope distribution map, calculate the curvature based on the slope distribution map, and perform smoothing correction if the curvature fluctuates abnormally to obtain a concave-convex variation layer. The terrain parameter acquisition module is used to extract high-slope areas from the slope distribution map, spatially match the high-slope areas with the concavity and convexity variation layer to obtain terrain feature data, and perform edge detection and terrain parameter quantization processing on the parts of the terrain feature data with significant concavity and convexity variations to obtain the terrain parameters of the local terrain. The model optimization module is used to perform stratified processing of the water storage space and integrate it with the terrain parameters. Based on the integrated data, the area ratio of the water accumulation area is calculated, and the preset initial water storage distribution model is locally adjusted according to the area ratio of the water accumulation area to obtain an optimized water storage distribution model. The capacity calculation module is used to acquire real-time water level data, and calculate the gradient change of water level in combination with the optimized water storage distribution model to determine the water body change trend. Through the water body change trend, the specific deviation position of the real-time water volume compared with historical data is obtained. The optimized water storage distribution model is corrected according to the specific deviation position, and the current reservoir capacity is calculated according to the corrected model. The trend analysis module is used to acquire historical capacity data, compare and analyze the current reservoir capacity with the historical capacity data, and obtain the capacity change trend. The update output module is used to acquire real-time monitoring data and compare it with the predicted value obtained from the prediction based on the capacity change trend. The initial digital elevation model is updated according to the comparison result. The current reservoir capacity is calibrated according to the updated elevation model to generate the final accurate monitoring result of the reservoir capacity.

Images