Atmospheric ecological environment data synchronous monitoring method and system

Abstract

The invention discloses an atmospheric ecological environment data synchronous monitoring method and system, and relates to the technical field of atmospheric environment monitoring, and the method comprises the steps: building a sky-air-ground global monitoring network, and achieving the unification of global data collection and space-time reference; on the basis of a hybrid interpolation model, performing hybrid interpolation and residual analysis on the atmospheric ecological environment monitoring data collected in a global manner, identifying a monitoring weak region, and performing positioning and labeling; based on monitoring weak area position distribution, starting an unmanned aerial vehicle to execute a weak area inspection task, and collecting weak area monitoring data through field monitoring to realize monitoring-prediction closed loop verification; actual measurement and prediction data are fused to reconstruct a global monitoring network, a digital twin platform is constructed through three-dimensional situation mapping, and global sensing and synchronous monitoring of the atmospheric ecological environment are achieved. According to the invention, by constructing a space-air-ground collaborative three-dimensional monitoring network, total element synchronous acquisition and dynamic integration of atmospheric pollution data are realized, and a seamlessly connected data base is provided for pollution traceability and trend prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of atmospheric environment monitoring, and in particular to a method and system for synchronous monitoring of atmospheric ecological environment data. Background Art

[0002] Atmospheric ecological environment data is the core basis for evaluating air quality and pollution control effects, covering multi-dimensional indicators such as particulate matter (PM2.5, PM), gaseous pollutants (SO, NOx, O, etc.), volatile organic compounds (VOCs), and meteorological parameters (temperature, humidity, wind direction, wind speed). These data are obtained through physical, chemical, remote sensing and other technical means. For example, the β-ray method is used for measuring particulate matter concentration, chemiluminescence method and ultraviolet fluorescence method are used to detect NOx and SO respectively, and satellite remote sensing realizes large-scale pollution distribution monitoring by retrieving aerosol optical depth (AOD). The data not only reflects the spatio-temporal changes of pollutant concentrations, but also can trace the pollution sources through component analysis (such as the proportion of silicon and aluminum elements in PM2.5), supporting targeted measures such as coal combustion control and motor vehicle exhaust control.

[0003] Atmospheric ecological environment data is the core basis for evaluating environmental quality and formulating pollution control strategies. Its dynamics, multi-source nature and spatial heterogeneity pose higher requirements for monitoring technologies. Current atmospheric environment monitoring mainly relies on the combination of ground monitoring stations, satellite remote sensing and mobile monitoring equipment, and obtains data such as pollutant concentrations and meteorological parameters through technical means such as physical and chemical analysis, sensor networks and model prediction. Although traditional ground monitoring stations can achieve fixed-point continuous observation, they are limited by insufficient spatial coverage and are difficult to capture the cross-regional diffusion law of pollutants; although satellite remote sensing technology can provide large-scale periodic scans, there are problems such as low time resolution and cloud interference, resulting in limited real-time monitoring ability.

[0004] With the application of Internet of Things and artificial intelligence technologies, emerging means such as unmanned aerial vehicles and micro sensors have been gradually popularized, but problems such as the inconsistent spatio-temporal benchmarks of multi-source heterogeneous data and the low efficiency of identifying monitoring blind areas still restrict the overall effectiveness of the monitoring system. For example, the coordinate system differences between ground-based sensor networks and satellite data may cause spatial matching deviations, while the sampling frequency differences of different devices result in difficult synchronization and fusion of time series data, affecting the accuracy of pollutant diffusion models. In addition, the spatial prediction model based on a single interpolation algorithm often produces large residuals when facing complex terrains or sudden pollution events due to ignoring the dynamic changes of pollution sources, and traditional residual analysis methods lack the ability to accurately locate weak monitoring areas and are difficult to guide the optimal allocation of monitoring resources.

[0005] Although current technological developments have achieved multi-dimensional data collection, there are still significant bottlenecks in data synchronization, system self-adaptability and real-time decision support. There is an urgent need to design a method and system for synchronous monitoring of atmospheric ecological environment data. Summary of the Invention

[0006] Based on this, in view of the above technical problems, it is necessary to provide a method and system for synchronous monitoring of atmospheric ecological environment data.

[0007] In a first aspect, the present invention provides a method for synchronous monitoring of atmospheric ecological environment data, including: S1. Build a sky-air-ground full-domain monitoring network to achieve full-domain data collection and unified spatio-temporal reference; S2. Based on a hybrid interpolation model, perform hybrid interpolation and residual analysis on the atmospheric ecological environment monitoring data collected in the full domain, identify weak monitoring areas and mark their positions; S3. Based on the position distribution of weak monitoring areas, enable drones to perform patrol tasks in weak areas, collect weak area monitoring data through on-site monitoring, and achieve closed-loop verification of monitoring and prediction; S4. Reconstruct the full-domain monitoring network by integrating measured and predicted data, and build a digital twin platform through three-dimensional situation mapping to achieve full-domain perception and synchronous monitoring of the atmospheric ecological environment.

[0008] Further, building a sky-air-ground full-domain monitoring network to achieve full-domain data collection and unified spatio-temporal reference includes: S11. Build a space-based monitoring network based on remote sensing satellites, and achieve full-domain coverage and periodic scanning through multi-spectral, thermal infrared and synthetic aperture radar payloads; S12. Use a mixed formation of drones to achieve multi-region hierarchical monitoring and form an air-based monitoring network; S13. Deploy distributed ground monitoring stations and form a ground-based monitoring network using ground sensors; S14. Build a hierarchical transmission and multi-source data fusion framework based on the Beidou spatio-temporal reference to achieve unified spatio-temporal reference and dynamic correction of full-domain atmospheric ecological environment monitoring data.

[0009] Further, based on a hybrid interpolation model, performing hybrid interpolation and residual analysis on the atmospheric ecological environment monitoring data collected in the full domain, identifying weak monitoring areas and marking their positions includes: S21. Sparsify the monitoring data, divide it into several spatial grids, retain one optimal monitoring point in each grid, and mark the area not covered by the ground monitoring station as the area to be verified; S22. Optimize the range parameter and smoothness parameter of the Kriging interpolation algorithm based on the adaptive mutation chaotic particle swarm algorithm to construct a hybrid interpolation model; S23. Use the hybrid interpolation model to perform hybrid interpolation calculation on the monitoring data in the area to be verified, and generate the full-domain pollutant concentration field and confidence distribution; S24. Perform residual analysis based on the prediction results of the interpolation data, dynamically identify weak monitoring areas, and mark the position information and corresponding area types of the weak monitoring areas.

[0010] Further, based on the adaptive mutation chaotic particle swarm optimization algorithm to optimize the range parameter and smoothness parameter of the Kriging interpolation algorithm, the construction of the hybrid interpolation model includes: S221. Generate the initial position of the particle swarm using the sine chaotic mapping and initialize the population size; S222. Map the range parameter and smoothness parameter of the Kriging interpolation to the two-dimensional position vector of the particle swarm, and convert the chaotic sequence value to the parameter range through linear scaling; S223. Use the root mean square error of cross-validation as the fitness function to calculate the particle fitness value; S224. Set the adaptive update strategy of the inertia weight to adjust the dynamic weight and learning factor; S225. When the global optimal solution has not been updated for six consecutive generations, perform Gaussian mutation on the global optimal solution; S226. When the maximum number of iterations is reached or the convergence threshold is satisfied, stop the iteration and output the global optimal parameter combination.

[0011] Further, use the hybrid interpolation model to perform hybrid interpolation calculation on the monitoring data in the area to be verified, and generate the global pollutant concentration field and confidence distribution, including: S231. Call the range parameter and smoothness parameter to initialize the hybrid interpolation model, and load the preprocessed data set, including the monitoring data of the ground monitoring station, the inversion value of satellite data, and the covariates; S232. Based on the monitoring data of the ground monitoring station, generate the basic field, calculate the benchmark value of the pollutant concentration at each spatial grid point for interpolation, convert the inversion value of satellite data into the spatial distribution trend of pollutants, superimpose it on the basic field, and use the covariates to perform anisotropic correction on the pollution diffusion direction; S233. Apply Gaussian noise to the range parameter and smoothness parameter to generate parameter combinations, perform interpolation calculations respectively, and statistically calculate the quantile interval of the pollutant concentration value at each grid point to calculate the confidence index.

[0012] Further, based on the prediction results of the interpolation data, perform residual analysis, dynamically identify the weak monitoring areas, and mark the location information and corresponding area types of the weak monitoring areas, including: S241. Extract the predicted concentration of the global interpolation grid to the coordinate system of the ground monitoring station, calculate the residual of the grid point prediction data through the preset differential trigger condition, and mark the potential weak area points; S242. Use the clustering algorithm for spatial clustering, perform mathematical morphological dilation operation on the initial clustering result, synthesize fragmented areas, and output the weak monitoring area including the boundary through terrain constraint; S243. Use a terrain analysis tool to identify the location coordinates and terrain information of each monitoring weak area, and divide the area types of the monitoring weak areas into drone - flyable areas and no - fly areas according to the terrain information.

[0013] Further, by presetting differential triggering conditions, calculate the residual of the grid point prediction data, and mark potential weak area points, including: Set a sliding time window and a residual calculation period, and for non - measured grid points in the area to be verified, calculate the interpolation prediction value sequence and the sliding standard deviation for N consecutive periods. Perform filtering processing on the interpolation prediction value sequence, divide it into a trend term and a periodic term, and set the ratio of the difference between the interpolation prediction value minus the trend term to the trend term as the relative residual. Set differential triggering conditions. When the relative residual is greater than the preset threshold for M consecutive periods and the sliding standard deviation is greater than the preset threshold, mark this grid point as a potential weak area point.

[0014] Further, based on the location distribution of the monitoring weak areas, enable the drone to perform weak area inspection tasks, and collect weak area monitoring data through on - site monitoring to achieve monitoring - prediction closed - loop verification, including: S31. Generate a drone weak area inspection task based on the location coordinates and area types of the monitoring weak areas, and output a set of waypoint coordinates and a time - altitude profile. S32. Dynamically configure the monitoring load according to the pollutant type in the monitoring weak area, start the drone for inspection, and conduct on - site monitoring and data quality verification of the monitoring weak area. S33. Through encrypted transmission of the measured data packet, use the measured weak area monitoring data to correct and replace the interpolation data of the grid points.

[0015] Further, dynamically configure the monitoring load according to the pollutant type in the monitoring weak area, start the drone for inspection, and conduct on - site monitoring and data quality verification of the monitoring weak area, including: S321. Configure a laser scattering sensor and a PID sensor at the drone load end, access the WRF real - time wind speed field, and dynamically adjust the flight altitude. S322. Perform multi - point hovering monitoring within the set monitoring range at the center coordinates of the monitoring weak area, and collect minute - level environmental monitoring data in real time. S323. Compare the measured weak area monitoring data with the interpolation prediction data, calculate the measured relative deviation. When the measured relative deviation exceeds the preset threshold, mark it as a high - deviation area and trigger the lidar vertical profile scan to assist in pollution source tracing.

[0016] In the second aspect, the present invention also provides an atmospheric ecological environment data synchronous monitoring system, which includes: The global monitoring module is used to build a sky-air-ground global monitoring network to achieve global data collection and unified spatio-temporal reference; The interpolation prediction module is used to perform hybrid interpolation and residual analysis on the atmospheric ecological environment monitoring data collected globally based on a hybrid interpolation model, identify weak monitoring areas, and predict the pollutant concentration in the weak areas; The closed-loop verification module is used to enable drones to perform inspection tasks in weak areas based on the location distribution of weak monitoring areas, collect weak area monitoring data through on-site monitoring, and achieve monitoring-prediction closed-loop verification; The synchronous perception module is used to reconstruct the global monitoring network by fusing measured and predicted data, build a digital twin platform through three-dimensional situation mapping, and achieve global perception and synchronous monitoring of the atmospheric ecological environment.

[0017] The beneficial effects of the present invention are as follows: 1. By constructing a three-dimensional collaborative sky-air-ground monitoring network, synchronous collection and dynamic integration of all elements of air pollution data are achieved. Satellite remote sensing provides global periodic scanning data, drone formations perform real-time dynamic verification for weak areas, and the ground-based sensor network completes minute-level high-frequency monitoring. Based on the unified framework of spatio-temporal reference, the time stamp alignment and spatial coordinate correction of multi-source data ensure the spatio-temporal consistency of cross-platform data, breaking through the spatio-temporal fragmentation limitation of traditional single-point monitoring. Thanks to the synchronous mechanism of "wide-area coverage-dynamic verification-fixed-point high-frequency", the core problem of spatio-temporal mismatch of monitoring data during the pollutant diffusion process is effectively solved, providing a seamless data base for pollution source tracing and trend prediction.

[0018] 2. By comparing the prediction results of the hybrid interpolation model with the measured data of drones in real time, a synchronous closed-loop verification system of "prediction-verification-correction" is constructed. Drones dynamically adjust the flight path and payload configuration according to the location of weak areas, perform fixed-point hovering monitoring, and the measured data is transmitted back in real time through an encrypted link to drive the update of interpolation model parameters. For high-deviation areas, laser radar vertical profile scanning and pollution source correlation analysis are synchronously triggered, forming a complete feedback chain of "spatial anomaly identification-on-site verification-model iteration", thereby realizing the two-way synchronous optimization of monitoring data and prediction models, and improving the dynamic prediction accuracy of pollutant concentration fields and the reliability of weak area identification. Description of the Drawings

[0019] The drawings described herein are used to provide a further understanding of the present invention, form a part of the present invention, and the schematic embodiments and descriptions thereof are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings: Figure 1 is a flowchart of a method for synchronously monitoring atmospheric ecological environment data according to an embodiment of the present invention; Figure 2It is a system principle block diagram of an atmospheric ecological environment data synchronous monitoring system according to an embodiment of the present invention.

[0020] Reference numerals in the attached drawings: 1, global monitoring module; 2, interpolation prediction module; 3, closed-loop verification module; 4, synchronous perception module. Specific implementation manners

[0021] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the attached drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0022] Please refer to Figure 1 , and a method for synchronous monitoring of atmospheric ecological environment data is provided, including: S1. Build a sky-air-ground global monitoring network to achieve global data collection and unification of spatio-temporal benchmarks.

[0023] In the description of the present invention, building a sky-air-ground global monitoring network to achieve global data collection and unification of spatio-temporal benchmarks includes: S11. Build a space-based monitoring network based on remote sensing satellites, and through multi-spectral, thermal infrared and aperture radar payloads, achieve global coverage and periodic scanning.

[0024] Specifically, through high-resolution remote sensing satellites equipped with multi-spectral, thermal infrared and synthetic aperture radar (SAR) sensors, multi-dimensional environmental data collection is realized. The satellite orbit parameters are optimized to a sun-synchronous orbit of 500 - 700 km, the revisit period ≤ 12 hours, and the resolution reaches 0.3 m (optical) and 1 m (radar).

[0025] The data collection using remote sensing satellites to build a space-based monitoring network includes: 1. Multi-spectral data: Collect blue (450 nm), green (560 nm), red (650 nm), near-infrared (860 nm) and short-wave infrared (1650 nm) bands for vegetation index (NDVI), land use classification and pollutant diffusion trend analysis; 2. SAR data: Through the microwave cloud penetration characteristics, obtain surface deformation, soil moisture and flood inundation range; 3. Thermal infrared data: Monitor surface temperature anomalies and identify industrial area thermal pollution and nighttime illegal emissions.

[0026] S12. Use a mixed formation of unmanned aerial vehicles to achieve multi-region hierarchical monitoring and form an air-based monitoring network.

[0027] Specifically, the mixed formation of drones can use fixed-wing + multi-rotor drones to achieve hierarchical monitoring. For example, fixed-wing drones equipped with LiDAR can be used to perform grid scanning to generate digital surface models (DSM) and point cloud data for surface settlement and earthwork volume calculations, and to achieve regional wide-area screening. Multi-rotor drones equipped with multispectral cameras (such as MicaSenseRedEdge-P) and PID sensors can then be used to perform sub-meter-level refined monitoring of abnormal areas identified by satellites, and to achieve key verification within the area.

[0028] Drones can collect PM2.5, VOCs and SO2 concentration data through onboard sensors; realize three-dimensional pollution source positioning through RTK / PPK positioning; the onboard Jetson Nano node performs real-time data compression and outlier filtering, and transmits key data back through the 5G-MEC edge node.

[0029] S13. Deploy distributed ground monitoring stations and use ground-based sensors to form a ground-based monitoring network.

[0030] Specifically, the ground-based monitoring network needs to integrate multiple types of sensors, covering air pollution, meteorological elements and auxiliary parameters, including the following aspects: 1. Pollutant monitoring: PM2.5 / PM10 sensor (laser scattering principle), VOCs detector (photoionization PID technology), SO2 / NO2 electrochemical sensor, ozone (O) ultraviolet absorption sensor; 2. Meteorological parameters: temperature and humidity sensor, wind speed and direction meter, atmospheric pressure sensor, precipitation monitor; 3. Auxiliary parameters: ultraviolet radiation sensor, carbon dioxide (CO) infrared absorption sensor.

[0031] The layout of ground monitoring stations needs to be centered on the pollution source and arranged in concentric circles, focusing on monitoring characteristic pollutants such as VOCs and SO2. For ecologically sensitive areas, such as nature reserves and urban green spaces, they should be arranged in a grid pattern, focusing on PM2.5, O3 and meteorological related parameters.

[0032] S14. Build a hierarchical transmission and multi-source data fusion framework based on Beidou space-time benchmark to achieve the unification and dynamic correction of space-time benchmarks for global atmospheric ecological environment monitoring data.

[0033] Specifically, a network-wide NTP synchronization service is established based on BeiDou Time (BDT) to achieve cross-platform data timestamp alignment; a 50km-spaced ground reference station network is deployed to solve atmospheric delays and tropospheric errors in real time, and Kalman filtering is used to achieve sub-meter spatial matching of air-ground data.

[0034] S2. Based on the hybrid interpolation model, hybrid interpolation and residual analysis are performed on the atmospheric ecological environment monitoring data collected over the entire region to identify weak monitoring areas and locate and mark them.

[0035] It should be noted that the hybrid interpolation model is constructed by optimizing the Kriging interpolation algorithm based on the adaptive mutation chaotic particle swarm optimization (AMCPSO). Its core principle lies in improving the accuracy and stability of spatial interpolation through multi-strategy fusion. Kriging interpolation relies on the semi-variogram model to characterize spatial correlation through range parameters and smoothness parameters.

[0036] Traditional Kriging method requires manual parameter adjustment, which is easily restricted by subjective experience and has insufficient adaptability to complex terrain or scenarios of dynamic diffusion of pollutants. By introducing the chaotic particle swarm algorithm, the initial particle swarm is generated using the sine chaotic mapping to enhance the global search ability; the global and local optimization are dynamically balanced through the adaptive inertia weight to avoid premature convergence. When the global optimal solution stagnates, the Gaussian mutation mechanism is triggered to force jumping out of the local optimum and improve the parameter optimization efficiency.

[0037] Through the optimized Kriging interpolation algorithm, sparse ground monitoring data and satellite inversion data are fused to generate a high-resolution pollutant concentration distribution map, covering traditional monitoring blind areas. Based on the residual sequence of the interpolation prediction value and the measured value, combined with the sliding time window analysis (such as the residual exceeding the threshold for 5 consecutive periods), potential weak areas are dynamically marked. By generating multiple groups of interpolation results through parameter perturbation, the quantile interval of the pollutant concentration at each grid point is statistically analyzed to provide a basis for the subsequent priority of UAV inspection.

[0038] In the description of the present invention, based on the hybrid interpolation model, hybrid interpolation and residual analysis are performed on the atmospheric ecological environment monitoring data collected in the whole region, and identifying and positioning the monitoring weak areas include: S21. Sparsify the monitoring data, divide it into several spatial grids, retain one optimal monitoring point in each grid, and mark the area not covered by the ground monitoring station as the area to be verified.

[0039] S22. Optimize the range parameter and smoothness parameter of the Kriging interpolation algorithm based on the adaptive mutation chaotic particle swarm algorithm to construct a hybrid interpolation model.

[0040] In the description of the present invention, optimizing the range parameter and smoothness parameter of the Kriging interpolation algorithm based on the adaptive mutation chaotic particle swarm algorithm to construct a hybrid interpolation model includes: S221. Use the sine chaotic mapping to generate the initial position of the particle swarm and initialize the population size.

[0041] Specifically, use the sine chaotic sequence (formula: x n+1 = μ sin( πxn )) to generate the initial position of the particle, break the local aggregation that may be caused by traditional random initialization, and enhance the population diversity.

[0042] The range and smoothness parameters of Kriging are mapped to the two-dimensional coordinates (X, Y) of the particles, and the chaotic values ​​are converted into the actual range of the parameters through linear scaling to ensure that the parameter search space covers the theoretical feasible domain.

[0043] S222, mapping the range parameter and smoothness parameter of the Kriging interpolation to the two-dimensional position vector of the particle swarm, and converting the chaotic sequence value into the parameter range through linear scaling.

[0044] S223, using the cross-validated root mean square error as the fitness function, calculating the particle fitness value.

[0045] S224. Set an adaptive update strategy for the inertia weight, and adjust the dynamic weight and learning factor.

[0046] Specifically, the weight is adjusted dynamically according to the iteration progress (formula: ω = ω max ( ω max ω min )× t / T , t is the current iteration number, T is the total number of times), the initial focus is on global exploration (high ω), and the later stage turns to local optimization (low ω), balancing the convergence speed and accuracy.

[0047] S225. When the global optimal solution has not been updated for six consecutive generations, perform Gaussian mutation on the global optimal solution.

[0048] S226. When the maximum number of iterations is reached or the convergence threshold is met, the iteration is stopped and the globally optimal parameter combination is output.

[0049] S23. Use the hybrid interpolation model to perform hybrid interpolation calculations on the monitoring data of the verification area to generate the global pollutant concentration field and confidence distribution.

[0050] In the description of the present invention, the hybrid interpolation model is used to perform hybrid interpolation calculation on the monitoring data of the verification area to generate the global pollutant concentration field and confidence distribution, including: S231, calling range parameters and smoothness parameters to initialize the hybrid interpolation model, loading the preprocessed data set, including ground monitoring station monitoring data, satellite data inversion values ​​and covariates.

[0051] Specifically, the Kriging interpolation parameters optimized by the Adaptive Mutation Chaotic Particle Swarm Optimization (AMCPSO) algorithm are called, including the range parameter (Range, determining the range of spatial correlation) and the smoothness parameter (Smoothness, controlling the smoothness of the interpolation result).

[0052] Loading the preprocessed multi-source data set includes the following aspects: 1. Ground monitoring data: Hourly average values of pollutants such as PM2.5 and NO2 after unified spatio-temporal benchmark, and the outlier data (such as interference data when humidity > 85%) have been removed; 2. Satellite data inversion value: Aerosol Optical Thickness (AOD) data processed by radiometric calibration and cloud masking, with a resolution of 1 km; 3. Covariate data: Parameters such as wind speed vector, boundary layer height, and inversion layer thickness output by the WRF meteorological model, which are used to correct the pollution diffusion direction.

[0053] S232. Based on the monitoring data of ground monitoring stations, generate a basic field, calculate the benchmark values of pollutant concentrations at each spatial grid point for interpolation, convert the satellite data inversion value into the spatial distribution trend of pollutants, superimpose it on the basic field, and use covariates to perform anisotropic correction on the pollution diffusion direction.

[0054] Specifically, for the basic field interpolation, ordinary Kriging interpolation needs to be performed based on the ground monitoring station data to generate an initial pollutant concentration field, reflecting the local pollution distribution characteristics. The satellite AOD data is converted into the spatial trend term of PM2.5 concentration through the Geographically Weighted Regression (GWR) model to eliminate the non-linear deviation between AOD and ground concentration; the trend term is superimposed on the basic field to enhance the accuracy of the large-scale spatial distribution. Finally, using wind speed as the weight factor, inverse distance weighting (IDW) correction is performed on the pollution diffusion direction.

[0055] S233. Apply Gaussian noise to the range parameter and the smoothness parameter to generate parameter combinations, perform interpolation calculations respectively, and statistically analyze the quantile intervals of pollutant concentration values at each grid point, and calculate the confidence index.

[0056] For example, apply ±15% Gaussian noise to the range and smoothness parameters to generate 300 parameter combinations, perform interpolation calculations respectively, and simulate the impact of parameter uncertainty on the results. Statistically analyze the 90% quantile interval (i.e., the 5% quantile to the 95% quantile) of the 300 concentration values at each grid point, and calculate the confidence index: Confidence = 1 - Concentration median value / Quantile interval width. Finally, use the density clustering algorithm (such as DBSCAN, Eps = 300 m) to spatially aggregate the grid points with a confidence level < 75%, and output the boundary and area parameters of the weak monitoring area.

[0057] Among them, the confidence index directly reflects the prediction reliability of the hybrid interpolation model at a specific spatial position. The confidence distribution generated through Monte Carlo parameter perturbation simulation reveals the fluctuation range of the predicted pollutant concentration values. For example, if the median value of PM2.5 concentration at a certain grid point is 50 μg / m and the quantile interval is [45, 55], then the confidence level is 0.8 (i.e., 1 - 10 / 50). A high confidence level (≥0.8) indicates that the predicted value at this position is less affected by parameter perturbation and the result is stable; a low confidence level (<0.6) indicates that there is a large uncertainty in the model in this area, which may be due to sparse monitoring data or complex meteorological conditions leading to a decrease in interpolation reliability.

[0058] The confidence distribution map identifies low-confidence areas through spatial clustering (such as when the confidence level of multiple consecutive grids <0.7). Combining terrain data (slope, occlusion) with the distribution of pollution sources, two types of weak areas are accurately marked: 1. Weak areas with insufficient data: Due to insufficient coverage of monitoring stations, interpolation relies on extrapolation, and mobile monitoring equipment needs to be deployed preferentially; 2. Weak areas with model limitations: Areas with complex terrain or sudden meteorological changes, where the interpolation algorithm needs to be adjusted or additional covariates need to be introduced.

[0059] Therefore, the confidence distribution map can also be used to directly drive the path planning of drone inspections. Low-confidence areas are automatically promoted to high-priority verification targets, optimizing the input efficiency of monitoring resources.

[0060] S24. Perform residual analysis based on the prediction results of the interpolation data, dynamically identify weak monitoring areas, and mark the location information of the weak monitoring areas and the corresponding area types.

[0061] In the description of the present invention, performing residual analysis based on the prediction results of the interpolation data, dynamically identifying weak monitoring areas, and marking the location information of the weak monitoring areas and the corresponding area types includes: S241. Extract the predicted concentration of the entire-domain interpolation grid to the coordinate system of the ground monitoring station, calculate the residual of the grid point prediction data through a preset differential triggering condition, and mark potential weak area points.

[0062] In the description of the present invention, calculating the residual of the grid point prediction data through a preset differential triggering condition and marking potential weak area points includes: S2411. Set a sliding time window and a residual calculation period, and for non-observed grid points in the area to be verified, calculate the interpolation prediction value sequence and the sliding standard deviation for N consecutive periods.

[0063] S2412. Perform filtering processing on the interpolation prediction value sequence, divide it into a trend term and a periodic term, and set the ratio of the difference between the interpolation prediction value minus the trend term to the trend term as the relative residual.

[0064] S2413. Set the differential triggering condition. When the relative residuals in M consecutive cycles are greater than the preset threshold and the moving standard deviation is greater than the preset threshold, mark this grid point as a potential weak area point.

[0065] S242. Perform spatial clustering using a clustering algorithm. Conduct a mathematical morphological dilation operation on the initial clustering result to synthesize fragmented regions, and through terrain constraints, output the monitored weak areas including boundaries.

[0066] Specifically, use the density clustering algorithm to spatially aggregate the marked weak area points, set the clustering parameters of the neighborhood radius of 300 meters and the minimum number of points of 3 to identify discrete clusters of abnormal points. Then perform a mathematical morphological dilation operation on the clustering result, horizontally expand the weak area boundary through a 3×3 pixel rectangular structuring element, and fuse isolated regions with an area of less than 0.1 square kilometers into continuous blocks.

[0067] At the same time, overlay Digital Elevation Model (DEM) data, eliminate inaccessible areas (such as steep slopes and canyons) with a slope exceeding 15 degrees. The corrected weak area polygon boundary not only retains the spatial continuity of pollutant diffusion but also conforms to the actual terrain accessibility constraints.

[0068] S243. Use a terrain analysis tool to identify the location coordinates and terrain information of each monitored weak area, and based on the terrain information, divide the region type of the monitored weak area into drone - flyable areas and no - fly areas.

[0069] Specifically, based on the terrain analysis tool, extract terrain parameters such as the elevation standard deviation, average slope, and undulation degree of the weak area, and combine airspace control information (such as airport clearance areas and high - voltage line distribution) to divide the weak areas into drone - flyable areas and no - fly areas.

[0070] The flyable area needs to meet the conditions of slope ≤ 15 degrees, no obstacles, and no permanent no - fly restrictions. Its coordinates and pollutant types are automatically imported into the drone mission system; the no - fly area triggers the deployment of ground - based mobile monitoring equipment or the localization optimization strategy of model parameters. The terrain characteristics and pollution types are synchronously recorded in the weak area attribute table to provide data support for subsequent differential governance.

[0071] S3. Based on the location distribution of the monitored weak areas, enable drones to perform weak area inspection tasks, collect weak area monitoring data through on - site monitoring, and achieve the monitoring - prediction closed - loop verification.

[0072] In the description of the present invention, based on the location distribution of the monitored weak areas, enabling drones to perform weak area inspection tasks and collecting weak area monitoring data through on - site monitoring to achieve the monitoring - prediction closed - loop verification includes: S31. Based on the location coordinates and region type of the monitored weak areas, generate a drone weak area inspection task, and output a set of waypoint coordinates and a time - altitude profile.

[0073] S32. Dynamically configure monitoring loads according to pollutant types in weak monitoring areas, start drone inspections, and conduct on-site monitoring and data quality verification of weak monitoring areas.

[0074] In the description of the present invention, dynamically configuring the monitoring load according to the pollutant type in the weak monitoring area, starting the drone for inspection, and conducting on-site monitoring and data quality verification of the weak monitoring area include: S321. Configure laser scattering sensors and PID sensors on the UAV payload end, access the WRF real-time wind speed field, and dynamically adjust the flight altitude.

[0075] S322. Perform multi-point hovering monitoring within the monitoring range set at the center coordinates of the weak monitoring area, and collect minute-level environmental monitoring data in real time.

[0076] S323. Compare the measured weak zone monitoring data with the interpolation prediction data, calculate the measured relative deviation, and when the measured relative deviation exceeds the preset threshold, mark it as a high deviation area, triggering the laser radar vertical profile scanning to assist in pollution tracing.

[0077] S33, by encrypting and transmitting the measured data packet, the interpolation data of the grid points are corrected and replaced using the measured weak area monitoring data.

[0078] S4. Integrate measured and predicted data to reconstruct the global monitoring network, build a digital twin platform through three-dimensional situation mapping, and achieve global perception and synchronous monitoring of the atmospheric ecological environment.

[0079] Specifically, the real-time data from ground monitoring stations, measured values ​​from drone inspections, satellite remote sensing inversion data, and WRF meteorological model output are integrated, and the multi-source data are mapped to a unified space-time grid using a space-time alignment algorithm through data cleaning and format standardization.

[0080] Build a three-dimensional digital twin platform to integrate geographic information data, pollution concentration field, meteorological field and monitoring equipment status. Use WebGL technology to achieve lightweight rendering on the browser side, support pollutant concentration contour surface sectioning, historical data backtracking and multi-layer overlay.

[0081] See also Figure 2 The present invention also provides an atmospheric ecological environment data synchronization monitoring system, the system comprising: The global monitoring module 1 is used to build a sky-air-ground global monitoring network to achieve global data collection and unification of time and space benchmarks.

[0082] Interpolation prediction module 2 is used to perform mixed interpolation and residual analysis on the atmospheric ecological environment monitoring data collected in the entire domain based on the mixed interpolation model, identify weak monitoring areas, and predict pollutant concentrations in weak areas.

[0083] The closed-loop verification module 3 is used to enable the drone to perform weak area inspection tasks based on the distribution of monitored weak areas, collect weak area monitoring data through on-site monitoring, and achieve monitoring-prediction closed-loop verification.

[0084] The synchronous perception module 4 is used to reconstruct the global monitoring network by fusing measured and predicted data, build a digital twin platform through three-dimensional situation mapping, and achieve global perception and synchronous monitoring of the atmospheric ecological environment.

[0085] In summary, by means of the above technical solutions of the present invention, through the construction of a three-dimensional monitoring network that coordinates the sky, air, and ground, the all-element synchronous acquisition and dynamic integration of air pollution data are achieved. Satellite remote sensing provides global periodic scanning data, the drone formation performs real-time dynamic verification for weak areas, and the ground-based sensor network completes minute-level high-frequency monitoring. Based on the unified framework of the spatio-temporal reference, the time stamp alignment and spatial coordinate correction of multi-source data ensure the spatio-temporal consistency of cross-platform data, breaking through the spatio-temporal fragmentation limitation of traditional single-point monitoring. Thanks to the synchronous mechanism of "wide-area coverage-dynamic verification-fixed-point high-frequency", the core problem of spatio-temporal mismatch of monitoring data during the pollutant diffusion process is effectively solved, providing a seamless data base for pollution source tracing and trend prediction. Through the real-time comparison of the prediction results of the hybrid interpolation model and the measured data of the drone, a synchronous closed-loop verification system of "prediction-verification-correction" is constructed. The drone dynamically adjusts the flight path and payload configuration according to the weak area position, performs fixed-point hovering monitoring, and the measured data is transmitted back in real time through an encrypted link to drive the update of the interpolation model parameters. For high-deviation areas, the vertical profile scanning of the lidar and the correlation analysis of pollution sources are synchronously triggered, forming a complete feedback chain of "spatial anomaly identification-on-site verification-model iteration", thus realizing the two-way synchronous optimization of monitoring data and prediction models, and improving the dynamic prediction accuracy of pollutant concentration fields and the reliability of weak area identification.

[0086] It should be understood that although each step in the flowchart of the accompanying drawings is shown sequentially as indicated by the arrows, these steps do not necessarily have to be executed in the order indicated by the arrows. Unless otherwise clearly stated in this article, there is no strict order restriction for the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages. These sub-steps or stages do not necessarily have to be executed at the same time, but can be executed at different times, and their execution order does not necessarily have to be sequential, but can be executed alternately or alternately with at least a part of other steps or sub-steps or stages of other steps.

Claims

1. An atmospheric ecological environment data synchronous monitoring method, characterized in that, Including: S1. Build a sky-air-ground global monitoring network to achieve global data collection and unified spatio-temporal reference; S2. Based on the hybrid interpolation model, perform hybrid interpolation and residual analysis on the atmospheric ecological environment monitoring data collected globally, identify weak monitoring areas and mark their positions; S3. Based on the location distribution of weak monitoring areas, enable drones to perform patrol tasks in weak areas, collect weak area monitoring data through on-site monitoring, and achieve closed-loop verification of monitoring-prediction; S4. Reconstruct the global monitoring network by integrating measured and predicted data, and build a digital twin platform through three-dimensional situation mapping to achieve global perception and synchronous monitoring of the atmospheric ecological environment.

2. The atmospheric ecological environment data synchronization monitoring method according to claim 1, characterized in that, The building of the sky-air-ground global monitoring network to achieve global data collection and unified spatio-temporal reference includes: S11. Build a space-based monitoring network based on remote sensing satellites, and achieve global coverage and periodic scanning through multi-spectral, thermal infrared and synthetic aperture radar payloads; S12. Use a hybrid formation of drones to achieve multi-region hierarchical monitoring and form an air-based monitoring network; S13. Deploy distributed ground monitoring stations and use ground-based sensors to form a ground-based monitoring network; S14. Build a hierarchical transmission and multi-source data fusion framework based on the Beidou spatio-temporal reference to achieve unified spatio-temporal reference and dynamic correction of global atmospheric ecological environment monitoring data.

3. The method for synchronously monitoring atmospheric ecological environment data according to claim 1, characterized in that, The hybrid interpolation and residual analysis of the atmospheric ecological environment monitoring data collected globally based on the hybrid interpolation model to identify weak monitoring areas and mark their positions includes: S21. Sparsify the monitoring data, divide it into several spatial grids, retain one optimal monitoring point in each grid, and mark the area not covered by the ground monitoring station as the area to be verified; S22. Optimize the range parameter and smoothness parameter of the Kriging interpolation algorithm based on the adaptive mutation chaotic particle swarm algorithm to construct a hybrid interpolation model; S23. Use the hybrid interpolation model to perform hybrid interpolation calculation on the monitoring data in the area to be verified, and generate the global pollutant concentration field and confidence distribution; S24. Perform residual analysis based on the prediction results of the interpolation data, dynamically identify weak monitoring areas, and mark the location information of the weak monitoring areas and the corresponding area types.

4. The atmospheric ecological environment data synchronization monitoring method according to claim 3, wherein, The construction of a hybrid interpolation model by optimizing the range parameter and smoothness parameter of the Kriging interpolation algorithm based on the adaptive mutation chaotic particle swarm algorithm includes: S221. Generate the initial position of the particle swarm using sine chaotic mapping and initialize the population size; S222. Map the range parameter and smoothness parameter of the Kriging interpolation to the two-dimensional position vector of the particle swarm, and convert the chaotic sequence value to the parameter range through linear scaling; S223. Use the root mean square error of cross-validation as the fitness function to calculate the particle fitness value; S224. Set the adaptive update strategy of the inertia weight to adjust the dynamic weight and learning factor; S225. When the global optimal solution has not been updated for six consecutive generations, perform Gaussian mutation on the global optimal solution; S226. When the maximum iteration number is reached or the convergence threshold is met, stop the iteration and output the global optimal parameter combination.

5. The method for synchronously monitoring atmospheric ecological environment data according to claim 3, characterized in that, The use of the hybrid interpolation model to perform hybrid interpolation calculation on the monitoring data in the area to be verified and generate the global pollutant concentration field and confidence distribution includes: S231. Initialize the hybrid interpolation model with range parameters and smoothness parameters, and load the preprocessed dataset, including ground monitoring station data, satellite data inversion values, and covariates. S232. Generate a basic field based on the ground monitoring station data, calculate the benchmark values of pollutant concentrations at each spatial grid point for interpolation, convert the satellite data inversion values into the spatial distribution trend of pollutants, superimpose them on the basic field, and perform anisotropic correction on the pollution diffusion direction using covariates. S233. Apply Gaussian noise to the range parameters and smoothness parameters to generate parameter combinations, perform interpolation calculations respectively, and statistically calculate the quantile intervals of pollutant concentration values at each grid point, and calculate the confidence index.

6. The method for synchronously monitoring atmospheric ecological environment data according to claim 3, characterized in that, The residual analysis based on the interpolation data prediction results, dynamically identify the weak monitoring areas, and mark the location information and corresponding area types of the weak monitoring areas include: S241. Extract the predicted concentration of the global interpolation grid into the coordinate system of the ground monitoring station, calculate the residual of the grid point prediction data through a preset differential trigger condition, and mark the potential weak area points. S242. Use the clustering algorithm for spatial clustering, perform a mathematical morphological dilation operation on the initial clustering result, synthesize fragmented areas, and output the weak monitoring area including the boundary through terrain constraints. S243. Use the terrain analysis tool to identify the location coordinates and terrain information of each weak monitoring area, and divide the area type of the weak monitoring area into drone flyable areas and no-fly areas according to the terrain information.

7. The method for synchronously monitoring atmospheric ecological environment data according to claim 6, wherein The calculation of the residual of the grid point prediction data through a preset differential trigger condition and marking the potential weak area points includes: Set a sliding time window and a residual calculation period, and calculate the interpolation prediction value sequence and the sliding standard deviation for N consecutive periods for the non-measured grid points in the area to be verified. Filter the interpolation prediction value sequence, divide it into a trend term and a periodic term, and set the ratio of the difference between the interpolation prediction value minus the trend term to the trend term as the relative residual. Set a differential trigger condition. When the relative residual is greater than the preset threshold for M consecutive periods and the sliding standard deviation is greater than the preset threshold, mark this grid point as a potential weak area point.

8. A method for synchronous monitoring of atmospheric ecological environment data according to claim 2, characterized in that, Based on the location distribution of the weak monitoring areas, enable the drone to perform weak area inspection tasks, and collect weak area monitoring data through on-site monitoring to achieve monitoring-prediction closed-loop verification, including: S31. Generate a drone weak area inspection task based on the location coordinates and area type of the weak monitoring area, and output the waypoint coordinate set and the time-height profile. S32. Dynamically configure the monitoring payload according to the pollutant type in the weak monitoring area, start the drone for inspection, and perform on-site monitoring and data quality verification on the weak monitoring area. S33. Through encrypted transmission of the measured data packet, use the measured weak area monitoring data to correct and replace the interpolation data of the grid point.

9. The atmospheric ecological environment data synchronization monitoring method according to claim 8, characterized in that, The dynamic configuration of the monitoring payload according to the pollutant type in the weak monitoring area, starting the drone for inspection, and performing on-site monitoring and data quality verification on the weak monitoring area include: S321. Configure a laser scattering sensor and a PID sensor at the drone payload end, access the WRF real-time wind speed field, and dynamically adjust the flight altitude. S322. Perform multi-point hovering monitoring within the monitoring range set by the center coordinates of the weak monitoring area, and collect minute-level environmental monitoring data in real time; S323. Compare the measured weak area monitoring data with the interpolated prediction data, calculate the measured relative deviation, and when the measured relative deviation exceeds the preset threshold, mark it as a high deviation area and trigger the vertical profile scanning of the lidar to assist in pollution source tracing.

10. An atmospheric ecological environment data synchronous monitoring system for implementing the atmospheric ecological environment data synchronous monitoring method according to any one of claims 1-9, characterized in that, The system includes: A global monitoring module for building a sky-air-ground global monitoring network to achieve unified global data collection and spatio-temporal reference; An interpolation prediction module for performing hybrid interpolation and residual analysis on the atmospheric ecological environment monitoring data collected globally based on a hybrid interpolation model, identifying weak monitoring areas, and predicting the pollutant concentration in the weak areas; A closed-loop verification module for enabling drones to perform weak area inspection tasks based on the location distribution of weak monitoring areas, collecting weak area monitoring data through on-site monitoring, and achieving monitoring-prediction closed-loop verification; A synchronous perception module for reconstructing the global monitoring network by fusing measured and predicted data, and building a digital twin platform through three-dimensional situation mapping to achieve global perception and synchronous monitoring of the atmospheric ecological environment.

Images