Method for monitoring and managing three-dimensional space grid situation based on space hash

Abstract

The application provides a three-dimensional space grid situation monitoring and management method based on space hash, and belongs to the technical field of low-altitude traffic monitoring and control, and comprises the following steps: S1, three-dimensional monitoring space time reference calibration and global core parameter initialization; S2, multi-source flight data access and fusion calculation; S3, three-dimensional space grid-box layered modeling and box topology pre-construction; S4, time-space coupling space hash index construction and grid mapping writing; S5, topology retrieval and accurate determination of the grid where the aircraft is located; S6, trajectory prediction and grid sequence mapping; S7, multi-format space layer import analysis and element-grid-box binding mapping; S8, box unit calculation level three threshold dynamic hysteresis activation and deactivation scheduling; S9, space situation warning calculation, closed-loop parameter adaptive optimization and monitoring data archiving. The application can realize the situation monitoring and management goals of efficient retrieval, accurate control, low consumption operation and stable expansion of low-altitude space.

Description

Technical Field

[0001] This invention relates to the fields of low-altitude traffic monitoring and control, three-dimensional airspace grid management, multi-source flight data fusion and spatial index retrieval technology, and specifically to a three-dimensional airspace grid situational monitoring and management method based on spatial hashing. Background Technology

[0002] With the large-scale development of the low-altitude economy, the number of low-altitude aircraft continues to grow, their types become more diverse, and their activity density gradually increases. Flight trajectories exhibit complex characteristics of nonlinearity, strong disturbances, and high density. This places higher technical demands on air traffic control entities for comprehensive situational awareness, refined airspace management, and real-time conflict risk early warning in low-altitude airspace.

[0003] Most existing low-altitude situational awareness monitoring technologies are based on a two-dimensional planar architecture and employ traditional methods such as full-domain traversal retrieval, simple data format splicing, and single-dimensional spatial indexing. These methods have insurmountable technical defects in three-dimensional airspace management scenarios. Existing technologies lack a hierarchical three-dimensional airspace grid system. Neighborhood retrieval requires a full-domain traversal operation, and computational resource consumption increases synchronously with grid size, failing to meet real-time retrieval requirements. Furthermore, the absence of a spatiotemporal coupling index mechanism prevents the synchronous matching of spatial location and temporal dimension information, resulting in insufficient retrieval accuracy and effectiveness. Multi-source heterogeneous flight data undergoes only format unification processing without incorporating signal quality and spatial characteristics for noise reduction and fusion, leading to low data reliability. Trajectory prediction is not adapted to the nonlinear flight characteristics of aircraft, exhibiting poor adaptability to disturbed flight trajectories and insufficient stability of prediction results. The lack of a computational resource-level dynamic scheduling mechanism results in continuous consumption of computational resources across the entire grid, causing significant waste of computing power. The airspace control layer is not deeply bound to the three-dimensional grid, preventing precise linkage between control rules and grid computation. Additionally, the system lacks closed-loop adaptive optimization capabilities, exhibiting weak engineering scalability and scenario adaptability, making it difficult to support the routine, high-precision situational monitoring needs of large-scale low-altitude aircraft. Summary of the Invention

[0004] This invention provides a three-dimensional airspace gridded situation monitoring and management method based on spatial hashing. By establishing a unified spatiotemporal benchmark across the entire domain, a hierarchical grid box structure, and a spatiotemporally coupled spatial hash index, combined with algorithms, it achieves high-quality fusion of multi-source data, accurate grid retrieval, and dynamic trajectory prediction. It also constructs a dynamic scheduling and closed-loop adaptive optimization mechanism for computing resources, ultimately achieving the goals of efficient retrieval, accurate control, low-consumption operation, and stable expansion of low-altitude airspace situation monitoring and management.

[0005] A three-dimensional spatial gridded situational awareness monitoring and management method based on spatial hashing includes:

[0006] S1. Perform spatiotemporal reference calibration and global core parameter initialization on the three-dimensional monitoring airspace to obtain spatiotemporal reference parameters and global core parameters;

[0007] S2. Acquire raw flight data of multi-source low-altitude aircraft, and process the data using a local weighted projection homomorphic filtering algorithm based on spatiotemporal reference parameters and global core parameters to obtain standardized fused flight track data;

[0008] S3. Based on the core parameters of the entire domain and standardized fused track data, construct a three-dimensional airspace basic spatial grid and box unit hierarchical structure, pre-construct the adjacent topological relationship of box units, and obtain the grid index, box index and adjacent topological relationship of box units;

[0009] S4. Based on the global core parameters, grid index and standardized fused track data, construct a spatiotemporal coupled spatial hash index to obtain the spatiotemporal coupled spatial hash index table.

[0010] S5. Based on the spatiotemporal coupled spatial hash index table, standardized fused track data, global core parameters and adjacent topological relationships of container units, the spatial hash nested octree topological retrieval algorithm is used to process the data to obtain the basic spatial grid where the aircraft is located, the set of twenty-six neighboring basic spatial grids and the index of associated container units.

[0011] S6. Based on the basic spatial grid where the aircraft is located, standardized fused track data and grid index, an adaptive fractional Kalman filter algorithm is used to process the data to obtain the basic spatial grid sequence of the predicted trajectory.

[0012] S7. Import multi-format spatial layer files and complete the parsing. Based on the grid index and box index, complete the binding mapping between spatial features and basic spatial grids and box units to obtain the binding relationship table.

[0013] S8. Based on the adjacent topology relationship of the container units, the associated container unit index, the binding relationship table and the global core parameters, perform dynamic hysteresis activation and deactivation scheduling of the container units at the three-level threshold to obtain the activated container unit calculation dataset.

[0014] S9. Based on the aircraft's location in the basic spatial grid, the set of twenty-six neighboring basic spatial grids, the predicted trajectory basic spatial grid sequence, the activated box unit calculation dataset, the binding relationship table, and the standardized fused trajectory data, complete the calculation of airspace situational alarms, adaptive optimization of closed-loop parameters, and archiving of monitoring data.

[0015] In this specification, the local weighted projection homomorphic filtering algorithm first converts the original flight data of multi-source low-altitude aircraft into local geocentric rectangular coordinates, performs multiplicative noise reduction through homomorphic filtering, and then performs coordinate offset fitting through local weighted projection, finally outputting standardized fused flight track data.

[0016] In this specification, during the processing of the local weighted projection homomorphic filtering algorithm, outdated flight data is filtered through an out-of-order tolerance window. Corresponding coefficients are calibrated based on the data source type, signal-to-noise ratio, data update frequency, and data position error, and data quality weight values ​​are calculated based on the corresponding coefficients.

[0017] In this specification, the spatial hash nested octree topology retrieval algorithm first completes the coarse screening of candidate grids through a spatiotemporal coupled spatial hash index table, and then completes the precise positioning of the basic spatial grid where the aircraft is located by traversing the nested octree layer by layer. Combined with the adjacent topological relationship of the box unit, a set of twenty-six neighborhood basic spatial grids is obtained.

[0018] In this specification, the adaptive fractional Kalman filter algorithm constructs a fractional state transition matrix based on a fractional difference operator, calculates process noise using data quality weight values, completes aircraft trajectory prediction, and maps it to a predicted trajectory base space grid sequence.

[0019] In this specification, after the adaptive fractional Kalman filter algorithm is completed, the predicted trajectory basic spatial grid sequence is stored in the memory cache pool with the aircraft's unique identifier as the key. When the trajectory data is updated, only the sequence difference part is updated incrementally.

[0020] In this specification, the local weighted projection homomorphic filtering algorithm, the spatial hash nested octree topology search algorithm, and the adaptive fractional Kalman filtering algorithm form a real-time bidirectional interaction between each other;

[0021] The local weighted projection homomorphic filtering algorithm outputs standardized fused track data in the forward direction to the spatial hash nested octree topology search algorithm, and receives the basic spatial grid index of the aircraft in the reverse direction to correct the data source type coefficient;

[0022] The locally weighted projection homomorphic filtering algorithm outputs data quality weight values ​​in the positive direction to the adaptive fractional Kalman filtering algorithm, and receives prediction bias in the negative direction to correct the filtering projection weight coefficients.

[0023] The spatial hash nested octree topology search algorithm outputs a set of 26 neighborhood basic spatial grids to the adaptive fractional Kalman filter algorithm in the forward direction, and receives the predicted coordinates in the reverse direction to correct the search radius.

[0024] In this specification, the binding relationship table marks the alarm level of the basic spatial grid corresponding to the airspace element. The no-fly zone corresponds to the level 3 alarm level, the restricted flight zone corresponds to the level 2 alarm level, the flight path corresponds to the level 1 alarm level, and the level 3 and level 2 alarm levels are high-level alarm grids.

[0025] The container unit's three-level threshold dynamic hysteresis activation and deactivation scheduling switches the container state based on the three-dimensional distance between the aircraft and the center of the container unit. It executes a forced activation strategy on container units bound to high-level alarm grids and synchronously updates the scheduling state to the memory hash table.

[0026] In this specification, the airspace situation alarm calculation generates neighborhood conflict alarms, no-fly zone intrusion alarms, restricted-fly zone intrusion alarms, and trajectory deviation alarms based on the binding relationship table. All alarms are associated with the aircraft's unique identifier, time information, and grid information.

[0027] In this specification, the closed-loop parameter adaptive optimization dynamically adjusts the core parameters of the entire domain based on the system retrieval time, computing resource utilization, and prediction deviation rate, and feeds them back to the grid modeling, index construction, and spatiotemporal retrieval stages to form a closed-loop collaboration.

[0028] This instruction manual can achieve at least the following beneficial effects:

[0029] Significantly improved retrieval efficiency: By combining spatiotemporal coupled spatial hash index with nested octree topology retrieval, the global traversal retrieval is transformed into hierarchical precise retrieval, greatly reducing the computational complexity of retrieval and improving the response speed and effectiveness of neighborhood grid positioning.

[0030] Reasonable optimization of computing resources: A three-level threshold dynamic hysteresis scheduling mechanism for container units is adopted to activate and deactivate grid computing data on demand, avoid unnecessary computing power consumption, and improve the overall operating efficiency and resource utilization of the system.

[0031] More reliable multi-source data fusion: The local weighted projection homomorphic filtering algorithm is used to complete the noise reduction and quality weighted fusion of multi-source heterogeneous flight data, thereby improving the accuracy, stability and consistency of flight data.

[0032] More robust trajectory prediction: Combining adaptive fractional Kalman filtering with grid space constraints, it adapts to the nonlinear and strongly disturbed flight characteristics of aircraft, improving the accuracy and scene adaptability of trajectory prediction.

[0033] More precise airspace control: Deeply binding the airspace control layer with the 3D grid box enables real-time linkage between control rules and grid calculations, improving the precision and standardization of airspace control.

[0034] The system architecture is more stable and scalable: it adopts a layered decoupled architecture and a closed-loop adaptive optimization mechanism, which can be adapted to monitoring airspaces of different sizes and multiple types of aircraft, significantly enhancing its engineering feasibility and scenario scalability.

[0035] More accurate and unified spatiotemporal calculations: Construct a unified spatiotemporal benchmark and spatiotemporal coupling index across the entire domain, synchronously associate spatial location and time dimension information, eliminate calculation deviations caused by spatiotemporal asynchrony, and ensure accurate and consistent calculation results throughout the entire process. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a schematic diagram of a three-dimensional spatial gridded situational monitoring and management method based on spatial hashing.

[0038] Figure 2 This is a schematic diagram of the multi-source data fusion processing flow involved in this invention.

[0039] Figure 3 This is a schematic diagram of the grid topology retrieval and positioning process involved in the present invention.

[0040] Figure 4 This is a schematic diagram of the closed-loop process for airspace situation control involved in this invention. Detailed Implementation

[0041] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the spirit or scope of the invention. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0042] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0043] like Figure 1 As shown, this embodiment provides a three-dimensional spatial gridded situational awareness monitoring and management method based on spatial hashing, including:

[0044] S1. Perform spatiotemporal reference calibration and global core parameter initialization on the three-dimensional monitoring airspace to obtain spatiotemporal reference parameters and global core parameters;

[0045] S2. Acquire raw flight data of multi-source low-altitude aircraft, and process the data using a local weighted projection homomorphic filtering algorithm based on spatiotemporal reference parameters and global core parameters to obtain standardized fused flight track data;

[0046] S3. Based on the core parameters of the entire domain and standardized fused track data, construct a three-dimensional airspace basic spatial grid and box unit hierarchical structure, pre-construct the adjacent topological relationship of box units, and obtain the grid index, box index and adjacent topological relationship of box units;

[0047] S4. Based on the global core parameters, grid index and standardized fused track data, construct a spatiotemporal coupled spatial hash index to obtain the spatiotemporal coupled spatial hash index table.

[0048] S5. Based on the spatiotemporal coupled spatial hash index table, standardized fused track data, global core parameters and adjacent topological relationships of container units, the spatial hash nested octree topological retrieval algorithm is used to process the data to obtain the basic spatial grid where the aircraft is located, the set of twenty-six neighboring basic spatial grids and the index of associated container units.

[0049] S6. Based on the basic spatial grid where the aircraft is located, standardized fused track data and grid index, an adaptive fractional Kalman filter algorithm is used to process the data to obtain the basic spatial grid sequence of the predicted trajectory.

[0050] S7. Import multi-format spatial layer files and complete the parsing. Based on the grid index and box index, complete the binding mapping between spatial features and basic spatial grids and box units to obtain the binding relationship table.

[0051] S8. Based on the adjacent topology relationship of the container units, the associated container unit index, the binding relationship table and the global core parameters, perform dynamic hysteresis activation and deactivation scheduling of the container units at the three-level threshold to obtain the activated container unit calculation dataset.

[0052] S9. Based on the aircraft's location in the basic spatial grid, the set of twenty-six neighboring basic spatial grids, the predicted trajectory basic spatial grid sequence, the activated box unit calculation dataset, the binding relationship table, and the standardized fused trajectory data, complete the calculation of airspace situational alarms, adaptive optimization of closed-loop parameters, and archiving of monitoring data.

[0053] In some embodiments, the local weighted projection homomorphic filtering algorithm first converts the original flight data of multi-source low-altitude aircraft into local geocentric rectangular coordinates, performs multiplicative noise reduction through homomorphic filtering, and then performs coordinate offset fitting through local weighted projection, finally outputting standardized fused flight track data.

[0054] In some embodiments, during the processing of the local weighted projection homomorphic filtering algorithm, outdated flight data is filtered through an out-of-order tolerance window, and corresponding coefficients are calibrated based on the data source type, signal-to-noise ratio, data update frequency, and data position error, and data quality weight values ​​are calculated based on the corresponding coefficients.

[0055] In some embodiments, the spatial hash nested octree topology retrieval algorithm first completes the coarse screening of candidate grids through a spatiotemporal coupled spatial hash index table, then completes the precise positioning of the basic spatial grid where the aircraft is located by traversing the nested octree layer by layer, and obtains a set of twenty-six neighborhood basic spatial grids by combining the adjacent topological relationships of the box unit.

[0056] In some embodiments, the adaptive fractional Kalman filter algorithm constructs a fractional state transition matrix based on a fractional difference operator, calculates process noise with data quality weight values, completes aircraft trajectory prediction, and maps it to a predicted trajectory base space grid sequence.

[0057] In some embodiments, after the adaptive fractional Kalman filter algorithm has finished processing, the predicted trajectory basic spatial grid sequence is stored in a memory cache pool with the aircraft's unique identifier as the key, and incremental updates are performed only on the sequence differences when the trajectory data is updated.

[0058] In some embodiments, the local weighted projection homomorphic filtering algorithm, the spatial hash nested octree topology search algorithm, and the adaptive fractional Kalman filtering algorithm form a real-time bidirectional interaction between each other;

[0059] The local weighted projection homomorphic filtering algorithm outputs standardized fused track data in the forward direction to the spatial hash nested octree topology search algorithm, and receives the basic spatial grid index of the aircraft in the reverse direction to correct the data source type coefficient;

[0060] The locally weighted projection homomorphic filtering algorithm outputs data quality weight values ​​in the positive direction to the adaptive fractional Kalman filtering algorithm, and receives prediction bias in the negative direction to correct the filtering projection weight coefficients.

[0061] The spatial hash nested octree topology search algorithm outputs a set of 26 neighborhood basic spatial grids to the adaptive fractional Kalman filter algorithm in the forward direction, and receives the predicted coordinates in the reverse direction to correct the search radius.

[0062] In some embodiments, the binding relationship table marks the basic spatial grid corresponding to the airspace element with alarm level, no-fly zone corresponds to level 3 alarm level, restricted flight zone corresponds to level 2 alarm level, flight route corresponds to level 1 alarm level, and level 3 and level 2 alarm levels are high-level alarm grids.

[0063] The container unit's three-level threshold dynamic hysteresis activation and deactivation scheduling switches the container state based on the three-dimensional distance between the aircraft and the center of the container unit. It executes a forced activation strategy on container units bound to high-level alarm grids and synchronously updates the scheduling state to the memory hash table.

[0064] In some embodiments, the airspace situation alarm calculation generates neighborhood conflict alarms, no-fly zone intrusion alarms, restricted-fly zone intrusion alarms, and trajectory deviation alarms based on a binding relationship table. All alarms are associated with the aircraft's unique identifier, time information, and grid information.

[0065] In some embodiments, the closed-loop parameter adaptive optimization dynamically adjusts the core parameters of the entire domain based on the system retrieval time, computing resource utilization rate and prediction deviation rate, and feeds them back to the grid modeling, index construction and spatiotemporal retrieval stages to form a closed-loop collaboration.

[0066] The technical concept of this invention is as follows:

[0067] This invention discloses a three-dimensional airspace gridded situation monitoring and management method based on spatial hashing. The overall process follows a complete logical chain of spatiotemporal benchmark construction, data processing, grid modeling, index construction, retrieval and positioning, trajectory prediction, layer binding, resource scheduling, and alarm optimization. The method first establishes a unified spatiotemporal benchmark across the entire domain and initializes core calculation parameters. After accessing multi-source low-altitude flight data, it completes data denoising and quality-weighted fusion using a local weighted projection homomorphic filtering algorithm. It then constructs a hierarchical discrete model of a three-dimensional basic spatial grid and box units, pre-calculates box topological relationships, and constructs a spatiotemporally coupled spatial hash index. A spatial hash nested octree topology retrieval algorithm is used to accurately determine the aircraft's location within the grid and its neighboring grids. An adaptive fractional-order Kalman filter algorithm is used to complete trajectory prediction and grid sequence mapping. Multi-format airspace layer elements are bound and mapped to grid boxes, and a three-level threshold dynamic hysteresis scheduling of box units is executed to optimize computational resource allocation. Finally, airspace situation alarm calculation and closed-loop adaptive optimization of system parameters are completed, achieving full-process monitoring data archiving and traceability, and realizing efficient, accurate, and low-consumption situation monitoring and management of the three-dimensional airspace.

[0068] S1. Three-dimensional monitoring spatial spatiotemporal reference calibration and global core parameter initialization:

[0069] refer to Figure 2 This step is the foundational step of the entire three-dimensional spatial gridded situational monitoring and management method. It plays a crucial role in establishing a unified spatiotemporal calculation benchmark across the entire domain and solidifying the core calculation parameters of the entire process. Through standardized spatiotemporal benchmark calibration and parameter initialization, it eliminates problems such as coordinate transformation errors and time synchronization deviations caused by multi-source acquisition devices from the root, providing a stable calculation reference for all subsequent data processing, algorithm calculation, grid modeling, index construction, resource scheduling, and alarm calculation, and ensuring the consistency and accuracy of the calculation results throughout the entire process.

[0070] First, the latitude and longitude range of the area to be monitored and the vertical monitoring altitude range are obtained through the airspace management interface. Using the arithmetic mean of the latitude and longitude of the airspace boundary and the midpoint of the altitude range as the calculation basis, the geometric center point of the monitored airspace is selected as the local geocentric rectangular coordinate origin, and the longitude of this origin is accurately recorded. ,latitude With altitude Furthermore, an airspace boundary verification algorithm was used to confirm that the origin was located at the geometric center of the monitored airspace, preventing subsequent calculation errors caused by origin offset. Simultaneously, a unified UTC timestamp benchmark was established. For the time-stamp calibration, the timing information of all subsequently accessed raw flight data, generated intermediate retrieval data, and output trajectory prediction data must be aligned with this timestamp reference. If the time error exceeds one second, a forced calibration operation is performed to completely eliminate the time drift problem of multi-source acquisition devices such as airborne terminals, surveillance radar, and remote control links. After the spatiotemporal reference calibration is completed, the core calculation parameters are initialized to a fixed global domain, and the side length of the basic spatial grid is set. The mesh size is set to fifty meters to ensure a balance between monitoring accuracy and computational efficiency; the number of meshes in the basic space on one side of the box unit is... Setting it to eight allows the number of grid cells in a single bin to adapt to memory scheduling and computational load; it also adjusts the spatiotemporally coupled hash cell size. The distance is set to 200 meters to achieve a balance between coarse-grained hash retrieval and fine-grained grid positioning; the neighborhood grid level parameters are fixed at a standard of 26 neighborhoods to cover all adjacent areas in three-dimensional space; the three-level thresholds of the container computing level scheduling are set as activation threshold of 500 meters, maintenance threshold of 800 meters, and deactivation threshold of 1200 meters, forming a hysteresis scheduling interval to avoid frequent state switching; the spatiotemporal retrieval dual thresholds are set as spatial distance threshold of 100 meters and timestamp deviation threshold of 2 seconds, respectively, to accurately filter effective retrieval data.

[0071] S2. Fusion calculation of multi-source flight data access and local weighted projection homomorphic filtering algorithm:

[0072] refer to Figure 2This step, relying on the spatiotemporal reference parameters and global core parameters calibrated by S1, performs full-process data processing and is the core data preprocessing step of the entire solution. It is responsible for accessing raw flight data from all types of aircraft in the low-altitude domain. Through a local weighted projection homomorphic filtering algorithm, it completes noise reduction, quality weight calibration, and bidirectional interactive fusion of multi-source heterogeneous data, providing high-precision and high-reliability foundational data for subsequent spatiotemporal retrieval and trajectory prediction. This completely solves the technical challenges of strong noise, high heterogeneity, and spatiotemporal asynchrony in low-altitude flight data that traditional data fusion methods cannot address. The raw data types accessed in this step comprehensively cover the entire low-altitude monitoring scenario, including ADS-B broadcast data, airborne BeiDou terminal positioning data, low-altitude surveillance radar detection data, UAV remote control link transmission data, and data collected by airborne sensors on general aviation aircraft. Each type of data is accessed through an independent channel to avoid data crosstalk and loss.

[0073] The core formula for constructing the model of the locally weighted projection homomorphic filtering algorithm is:

[0074] ;

[0075] ;

[0076] in To standardize and integrate local geocentric rectangular coordinates, The local geocentric rectangular coordinates are the converted raw flight data of multi-source low-altitude aircraft. The algorithm's filtering projection weight coefficients have an initial value of 0.6 and a range of 0 to 1. They are used to balance the output weights of noise reduction and spatial fitting. This is a homomorphic filtering noise reduction function, and its complete expression is: , To avoid the meaninglessness of logarithmic calculations for extremely small positive numbers, The Gaussian low-pass filter operator converts multiplicative noise in the original coordinates into additive noise through logarithmic transformation, then removes high-frequency noise through Gaussian low-pass filtering, and finally restores the denoised coordinate signal through exponential transformation. This is a locally weighted projection fitting function, and its complete expression is: Constructing a least-squares objective function based on local spatial features Complete coordinate offset correction. The weights are local spatial distances, and the calculation method is as follows: , For fitting bandwidth parameters in the local spatial domain, This represents the number of effective sampling points within the local spatial domain. For the first in the local airspace One original sampling coordinate, To fit the output coordinates; This is the data quality weight value output by the algorithm, with a value range from zero to one. The larger the value, the higher the data quality. The data source type coefficient is 0.9 for ADS-B data, 0.95 for airborne Beidou terminal data, 0.85 for low-altitude surveillance radar data, 0.8 for UAV remote control link data, and 0.9 for general aviation aircraft airborne sensor data. The signal carrier-to-noise ratio coefficient is used to map the carrier-to-noise ratio in decibels and hertz to the range of zero to one using a formula. The data update frequency coefficient is set to 1 for update frequencies greater than or equal to 1 Hz, 0.7 for update frequencies between 0.5 Hz and 1 Hz, and 0.4 for update frequencies less than 0.5 Hz. The data location error coefficient is calculated using the circular probability error, which is directly extracted from the ADS-B NIC and NAC parameters, radar accuracy parameters, and BeiDou positioning accuracy parameters.

[0077] The algorithm model training focuses on minimizing coordinate fusion error. A training sample set is constructed by collecting measured aircraft data from different data sources, locations, and flight states within the monitored airspace. A gradient descent optimization algorithm is used to iteratively optimize the filter projection weight coefficients, using the mean square error between the fused and true coordinates as the loss function. Convergence is achieved when the fusion error is less than 0.5 meters. Simultaneously, baseline values ​​for data source type coefficients, carrier-to-noise ratio coefficients, update frequency coefficients, and position error coefficients are calibrated. After training, the core model coefficients are fixed, and dynamic adjustments are only made based on feedback parameters from subsequent algorithms. When applying the algorithm model, all multi-source raw data are first converted to local geocentric rectangular coordinates. Signal denoising is performed using a homomorphic filtering function, followed by spatial fitting using a local weighted projection function. The final standardized fused coordinates are calculated using the filter projection weight coefficients. Simultaneously, data quality weights are calculated according to the quality weight formula. A three-second out-of-order tolerance window is set, and outdated data exceeding the window is discarded based on timestamp differences to ensure data real-time performance and validity.

[0078] This algorithm forms a real-time bidirectional interaction with the subsequent two core algorithms. The interaction with the spatial hash nested octree topology retrieval algorithm is positive, inputting standardized fused coordinates into the retrieval stage as the core computation coordinates; negatively, it receives the grid index output by the retrieval algorithm and updates the data source type coefficients in real time through a correction formula, improving the accuracy of data quality calibration. The interaction with the adaptive fractional-order Kalman filter algorithm is positive, inputting data quality weight values ​​into the prediction stage as the core parameter for noise calculation; negatively, it receives the prediction deviation output by the prediction algorithm and periodically updates the filter projection weight coefficients through a correction formula, optimizing the data fusion effect. After the algorithm completes its calculations, it generates standardized fused track data containing the aircraft's unique identifier, longitude, latitude, altitude, speed, heading, generation timestamp, quality weight value, data source identifier, and local rectangular coordinates.

[0079] S3. 3D Spatial Mesh - Box Layered Modeling and Box Topology Preconstruction:

[0080] refer to Figure 3 This step, based on the spatiotemporal reference and global core parameters generated by S1, and combined with the dynamic coverage range of the monitoring airspace extracted from the standardized fused flight track data output by S2, is the core of the spatial discretization of the entire scheme. It constructs a three-layer discrete computational model consisting of a basic spatial grid, box units, and a box matrix, and pre-calculates the adjacent topological relationships of the box units in advance. This provides stable topological support for subsequent spatiotemporal retrieval algorithms, completely avoiding the problems of computational resource consumption, response latency, and efficiency reduction caused by real-time calculation of neighborhood relationships, thus achieving efficient management and rapid retrieval of large-scale three-dimensional airspace grids. First, the three-dimensional monitoring airspace is discretized into uniform cubic basic spatial grids according to the side length of the basic spatial grid, using the formula... , , Complete the conversion from latitude and longitude to metric coordinates, and assign a unique three-dimensional index to each basic spatial grid. This ensures that the spatial location of each basic spatial grid can be accurately identified, achieving a digital discrete representation of the three-dimensional spatial domain. Then, based on the number of basic spatial grids on one side of each box-shaped unit, continuous basic spatial grids are grouped into box-shaped units. Each box-shaped unit contains a fixed number of basic spatial grids, and a unique box index is assigned to each box-shaped unit. A hierarchical mapping relationship is established between the box index and the three-dimensional index of the basic spatial grid, achieving precise association and binding between the basic spatial grids and box-shaped units, forming a hierarchical grid management structure.

[0081] After layered modeling is completed, the set of three-dimensional adjacent box units for each box unit is pre-calculated. This is done by determining the adjacent boxes according to the three-dimensional spatial orientation (up / down, left / right, front / back, and diagonal), resulting in twenty-six adjacent boxes for each box unit. Box unit metadata is generated, containing the box center coordinates, the basic spatial grid index range, and a list of adjacent box unit indices. This metadata and adjacent topological relationships are persistently stored in an in-memory hash table as key-value pairs. The key is the box index, and the values ​​are the box metadata, adjacent topological relationships, activation status, and deactivation flag. Real-time read / write and status updates are supported, enabling millisecond-level read / write operations without recalculating topological relationships, significantly improving the efficiency of subsequent neighborhood retrieval and resource scheduling. The basic spatial grid index set, box unit index set, hierarchical mapping relationship, adjacent topological relationship table, and box unit metadata set constructed in this step will directly support the stable operation of subsequent core processes such as spatiotemporal coupling index construction, neighborhood retrieval, box scheduling, and layer binding.

[0082] S4. Spatiotemporal Coupled Spatial Hash Index Construction and Grid Mapping Writing:

[0083] refer to Figure 3 This step relies on the global core parameters of S1, the basic spatial grid index set generated by S3, and the timestamp information of the standardized fused track data output by S2. It is the core innovative link in improving the retrieval efficiency of the entire scheme. By constructing a spatiotemporally coupled spatial hash index, spatial retrieval is deeply bound to the time dimension, breaking through the limitation of traditional pure spatial indexes that can only achieve spatial filtering. This provides efficient index support for subsequent topology retrieval algorithms, significantly reducing the computational load of grid retrieval in large-scale three-dimensional spatial domains and achieving millisecond-level candidate grid filtering. First, based on the local rectangular coordinates of the center of the basic spatial grid, the formula... , , Calculate the spatial hash key This process converts continuous three-dimensional spatial coordinates into discrete hash indices, enabling rapid mapping of spatial locations. Combined with the generated timestamps of standardized fused track data, a four-bit time entropy value is generated through modulo operations. The temporal information is encoded as discrete numerical values, which are then combined with spatial hash keys to form unique spatiotemporal coupled hash keys. This achieves integrated encoding of spatial location and temporal information, giving each basic spatial grid a unique spatiotemporal identifier. Subsequently, all basic spatial grids are written into a hash index structure according to their spatiotemporal coupled hash keys. Each hash key corresponds to a list of basic spatial grids, completing the construction of the spatiotemporal coupled spatial hash index table. This index table enables rapid candidate grid selection with spatiotemporal integration, effectively reducing the computational load of retrieval in large-scale spatial domains. The time entropy value encoding will participate in the time dimension matching calculation in subsequent spatiotemporal dual-threshold retrieval, further improving the accuracy of the retrieval and preventing outdated data and invalid grids from participating in the retrieval calculation.

[0084] S5. Spatial hash nested octree topology search and accurate determination of the aircraft's grid location:

[0085] refer to Figure 3 This step, based on the spatiotemporal coupled spatial hash index table generated in S4, the standardized fused track data output in S2, the global core parameters in S1, and the grid index and adjacent topology relationships generated in S3, is the core link of the entire scheme for aircraft positioning and neighborhood acquisition. It uses a spatial hash nested octree topology retrieval algorithm to accurately determine the grid where the aircraft is located and acquire neighboring grids. This algorithm deeply integrates the rapid coarse screening capability of spatial hashing, the hierarchical fine division capability of octree, and the neighborhood constraint capability of box topology, effectively solving the technical problems of low efficiency, long computation time, and high resource consumption in large-scale three-dimensional airspace grid retrieval. At the same time, through a two-way interaction mechanism and linkage with data fusion and trajectory prediction algorithms, it realizes dynamic optimization of retrieval parameters, ensuring the real-time performance and accuracy of retrieval results.

[0086] The core formula for building the algorithm model is:

[0087] ;

[0088] ;

[0089] in This is the base space grid index of the aircraft output by the algorithm. This is a spatiotemporally coupled spatial hash retrieval function that performs coarse screening based on spatiotemporally coupled hash keys. This refers to the size of a spatiotemporally coupled hash unit; The function is a nested octree hierarchical retrieval function. The root node of the octree covers the entire monitored airspace. The node splitting condition is that a single node contains more than 64 grids. It is divided into eight fixed levels. During retrieval, it traverses from the root node to the leaf node layer by layer to accurately match the grid where the aircraft coordinates are located. In the single grid scenario, the target grid index is directly output. The initial value for the algorithm's search radius is fifty meters. This is the set of basic spatial grids for the twenty-six neighborhoods output by the algorithm; For the box topological neighborhood mapping function; The neighborhood grid level parameter is fixed at 26 neighborhoods.

[0090] The spatiotemporal coupled spatial hash retrieval function operates as follows: Using standardized and fused local geocentric rectangular coordinates as the calculation basis, and combining this with the spatiotemporal coupled hash cell size, a spatiotemporal coupled hash key is calculated. A coarse screening of candidate grids is performed using a spatiotemporal coupled spatial hash index table. Finally, invalid grids exceeding the threshold are removed using a spatiotemporal retrieval double threshold, outputting the coarsely screened set of candidate basic spatial grids. The expression is as follows:

[0091] ;

[0092] Standardized fusion of local geocentric rectangular coordinates; Round down; Spatiotemporal coupled hash key; Time entropy value encoding; : Matched candidate base space grid index; Modulo operation.

[0093] The nested octree hierarchical retrieval function operates as follows: Using the entire monitored airspace as the root node of the octree, it traverses and splits layer by layer according to 8 fixed levels. Within the coarsely screened candidate grid range, it precisely locates the aircraft using the retrieval radius, outputting the unique basic spatial grid index corresponding to the aircraft. The expression is as follows:

[0094] ;

[0095] : Search radius; Octrees have fixed hierarchical levels. : No. The coordinates of the grid center within the layer node; : Three-dimensional Euclidean distance operator.

[0096] The logic of the box topology neighborhood mapping function is as follows: Based on the basic spatial grid index of the aircraft, it matches the belonging box element, calls the pre-built box element adjacent topology relationships in S3, extracts adjacent grids according to the 26-neighbor standard, and outputs a set of basic spatial grids for the 26-neighbor standard. The expression is as follows:

[0097] ;

[0098] : The underlying spatial grid index of the aircraft; : 3D neighborhood offset; output is a set of 26 neighborhood basic spatial grids.

[0099] The algorithm model training focuses on optimizing the retrieval radius fit. Using the grid and topology data constructed by S3 as the training set, the baseline value of the retrieval radius is iteratively optimized. After training, an eight-layer octree structure is fixed, and the retrieval radius is dynamically adjusted only by receiving feedback parameters from subsequent prediction algorithms. When applying the algorithm model, it first receives the standardized fused coordinates output by S2. A coarse screening is performed using a spatiotemporally coupled spatial hash function, combined with the spatiotemporal retrieval dual thresholds of S1 to complete the judgment, eliminating invalid candidate grids that exceed the spatial distance threshold and time deviation threshold. Then, a fine search is performed using a nested octree function, narrowing the search range layer by layer to lock onto the unique basic spatial grid index of the aircraft. Finally, the box topology neighborhood mapping function calls the adjacent topology relationship of S3 to obtain a set of twenty-six neighboring basic spatial grids, and completes the index association between the grids and the corresponding box units.

[0100] This algorithm forms a real-time bidirectional interaction with S2 and S6. Its interaction with the local weighted projection homomorphic filter algorithm involves a forward output of the grid index to correct the data source type coefficients, and a reverse input of standardized fused coordinates as the core input for retrieval. Its interaction with the adaptive fractional Kalman filter algorithm involves a forward output of a 26-neighborhood basic spatial grid set as the spatial constraint for trajectory prediction, and a reverse input of the predicted coordinates, which are then processed using the formula... Real-time updates of the search radius To predict local geocentric rectangular coordinates, The initial value for the predicted coordinate correction coefficient is 0.02. This is a 3D Euclidean distance calculation operator, and the calculation formula is as follows: After the algorithm completes its calculations, the resulting basic spatial grid where the aircraft is located, the set of twenty-six neighboring basic spatial grids, and the index of associated container units will be directly applied to trajectory prediction, layer binding, container scheduling, and alarm calculation, becoming the core positioning data for subsequent situational awareness calculations.

[0101] S6. Adaptive Fractional Kalman Filter Trajectory Prediction and Grid Sequence Mapping:

[0102] refer to Figure 4 This step relies on the basic spatial grid of the aircraft output by S5, the standardized fused trajectory data output by S2, and the basic spatial grid index set generated by S3. It is the core link of the entire scheme for trajectory prediction and future situation inference. The adaptive fractional Kalman filter algorithm is used to complete the aircraft trajectory prediction and grid sequence mapping. This algorithm combines the dynamic filtering capability of fractional Kalman with the depth of grid spatial constraints, effectively solving the problem of insufficient prediction accuracy caused by the nonlinearity, strong disturbance and strong abruptness of low-altitude aircraft trajectories. At the same time, through the two-way interaction mechanism and the linkage with data fusion and spatiotemporal retrieval algorithms, the prediction parameters and retrieval parameters are optimized in a coordinated manner, thereby improving the accuracy and reliability of trajectory prediction.

[0103] The core formula for building the algorithm model is:

[0104] ;

[0105] ;

[0106] in For the algorithm to predict local geocentric rectangular coordinates at time k, The fractional state transition matrix is ​​constructed based on the Grünwald-Letnikov fractional difference operator, and the formula is as follows: , It is a fractional combination number. It is a conventional integer-order state transition matrix. It is the fractional order, with an initial value of 0.9 and a value range of 0.8 to 0.95; The fractional control input matrix is ​​constructed according to the same rules as the fractional state transition matrix, and is obtained by fractionalizing the conventional control input matrix. Let k be the aircraft control input vector, which includes speed and heading information; Let k be the process noise vector at time k, and the calculation formula is: , The basic process noise covariance matrix; This represents the prediction bias value output by the algorithm. Let be the standardized fused local geocentric rectangular coordinates at the k-th discrete time step.

[0107] The algorithm model training focuses on minimizing prediction bias. A training set is constructed using real aircraft flight path data. Iterative optimization of the fractional-order Kalman oscillator is performed until convergence is achieved when the prediction bias is less than five meters. Process noise weights are calibrated, and the neighborhood grid output by S5 is introduced as a spatial constraint. After training, the basic model parameters are fixed. When applying the algorithm model, the mass weight values ​​output by S2 are first received to calculate process noise. Then, the neighborhood grid output by S5 is received to limit the spatial range of the predicted coordinates, preventing the prediction results from exceeding a reasonable airspace range. The fractional-order Kalman oscillator is used to calculate the predicted coordinates for the next ten time points. The prediction bias is calculated and fed back to the S2 and S5 algorithms in real time to correct the corresponding parameters. The predicted local geocentric rectangular coordinates are then transformed using an inverse transformation formula. , , The data is converted to latitude, longitude, and altitude, and then mapped to the basic spatial grid using the S3 grid index formula. A predictive trajectory basic spatial grid sequence is generated in chronological order and stored in the memory cache pool using the aircraft's unique identifier as the key. When the track data is updated, only the differences between the old and new sequences are compared to perform incremental updates. If the cache has not been updated for more than 30 seconds, it is automatically cleaned up to balance cache efficiency and memory usage.

[0108] This algorithm forms a real-time bidirectional interaction with S2 and S5. Its interaction with the local weighted projection homomorphic filtering algorithm outputs the prediction deviation in the forward direction to correct the filtering projection weight coefficients, and receives the quality weight value in the reverse direction as the core input for process noise calculation. Its interaction with the spatial hash nested octree topology search algorithm outputs the predicted coordinates in the forward direction to correct the search radius, and receives the neighborhood grid in the reverse direction as the prediction spatial constraint. The generated basic spatial grid sequence of the predicted trajectory will be directly used for subsequent trajectory deviation alarm calculations, providing core data support for airspace conflict early warning and situation prediction.

[0109] S7. Import and parsing of multi-format spatial layer and feature-mesh-box binding mapping:

[0110] refer to Figure 4 This step, receiving layer import instructions from the system, is the core of the entire solution's digitization of airspace control rules. It is responsible for importing airspace layer files in all formats for the low-altitude region, parsing the layer data using a dedicated parser, and binding and mapping airspace control elements with the basic spatial grid and box units constructed by S3. This provides accurate spatial constraints for subsequent alarm calculations, achieving deep integration of airspace control rules and the grid-based monitoring system. The imported file formats fully cover common low-altitude monitoring standards, including CZML, GeoJSON, and custom JSON formats. Dedicated parsing logic is used for different formats: CZML formats parse dynamic airspace control elements according to time series, GeoJSON formats parse static control elements according to geometric coordinates, and custom JSON formats parse specific control elements according to custom fields. All airspace control elements are comprehensively extracted, with complete element types including no-fly zones, restricted flight zones, flight routes, airspace boundaries, and takeoff and landing fields, covering all low-altitude airspace control types. For each airspace element, the range of its underlying spatial grid index and container unit index is calculated using a spatial inclusion determination formula. This determines whether the grid center coordinates fall within the element's geometric range, establishing a unique binding relationship between the layer element identifier, the underlying spatial grid index, and the container unit index. Simultaneously, the alarm level of the bound grid is marked according to airspace control rules: Level 3 for no-fly zones, Level 2 for restricted-fly zones, and Level 1 for flight paths. Levels 3 and 2 are high alarm levels, corresponding to the highest priority control and scheduling strategies. The standardized airspace layer element set and binding relationship table, after binding and mapping, will be directly applied to container scheduling and airspace situation alarm calculation, ensuring that the grid-based monitoring system strictly adheres to airspace control rules in its calculations and scheduling.

[0111] S8. Container unit computational level three-level threshold dynamic hysteresis activation / deactivation scheduling:

[0112] refer to Figure 4This step is based on the core data of the container unit output by S3 and its memory hash table storage structure, the associated container unit index output by S5, the binding relationship between layer elements and grid containers and alarm level numerical identifiers output by S7, and the container computing-level scheduling three-level threshold initialized by S1. It is the core link in the entire solution's computing resource optimization and management. Dynamic hysteresis scheduling at the computing resource level with three-level thresholds is performed on all container units. Through precise allocation of computing resources, ineffective computing power consumption is avoided, ensuring the efficient and stable operation of critical airspace and core computing processes. Simultaneously, real-time recycling and status synchronization of computing resources are achieved, realizing refined and intelligent management of computing resources. The scheduling operation is triggered by a single standardized fused flight track data. Each time a set of standardized fused flight track data for an aircraft is received, a complete container scheduling process is immediately executed, ensuring that the scheduling logic is synchronized with the real-time position of the aircraft. First, using the associated container unit index output by S5 as the query key, the corresponding container unit metadata is accurately matched from the memory hash table constructed by S3. The local geocentric rectangular coordinates of the container unit center are extracted, and then calculated using a formula... Calculate the three-dimensional distance between the aircraft's current standardized fused coordinates and the center of the container unit. , , The coordinates are the local geocentric rectangular coordinates of the center of the box unit.

[0113] After the distance calculation is completed, hierarchical scheduling is performed according to the three-level threshold initialized in S1. When the three-dimensional distance is less than or equal to the activation threshold, the activation state of the box unit is set to valid, and all basic spatial grid calculation data corresponding to the box unit is loaded into memory to ensure real-time computing needs. When the three-dimensional distance is greater than the activation threshold but less than or equal to the maintenance threshold, the current activation state of the box unit is maintained, and neither new data is loaded nor existing data is unloaded, achieving hysteresis stability control and avoiding computational fluctuations caused by frequent state switching. When the three-dimensional distance is greater than the deactivation threshold, the activation state of the box unit is set to invalid, all basic spatial grid calculation data corresponding to the box unit is unloaded, and the deactivation flag of the box unit in the memory hash table is set to valid, completing the real-time reclamation of computing resources and releasing invalid computing power. During scheduling, the binding relationships and alarm level identifiers output by S7 are invoked synchronously. When the airspace element bound to a container unit includes a Level 3 no-fly zone or a Level 2 restricted flight zone, the Level 3 threshold judgment rule is ignored regardless of the distance value, and a forced activation strategy is executed to maintain the active state and loaded computational data of the container unit. If a container is simultaneously bound to both high alarm level and ordinary flight path elements, forced activation is prioritized, and the validity of the internal grid data is verified to ensure that the computational resources of critical controlled airspace are not released, thus guaranteeing the monitoring continuity of the core controlled area. After scheduling is completed, an active container unit computational dataset is generated. This dataset uses the container index as the association key to store the basic spatial grid computational data of all active containers, which can be directly called by subsequent alarm calculation stages. At the same time, a deactivated container unit identifier is generated, which is synchronously updated to the in-memory hash table of S3 to mark container units whose computational resources have been released, avoiding duplicate scheduling and resource waste.

[0114] S9. Airspace situational alarm calculation, closed-loop parameter adaptive optimization, and monitoring data archiving:

[0115] refer to Figure 4This step integrates the aircraft's grid and neighborhood set output by S5, the predicted trajectory grid sequence output by S6, the activated box cell calculation data output by S8, the layer element binding relationship output by S7, the core parameters of S1, and the standardized fused trajectory data output by S2. It is the final output stage of the entire solution, completing airspace situational awareness calculation, system parameter closed-loop adaptive optimization, and monitoring data archiving. This achieves a complete logical loop and data traceability throughout the entire process, fully outputting all core results of 3D airspace grid-based situational awareness monitoring and management. The alarm calculation stage accurately determines the aircraft's location based on layer binding relationships and real-time grid position. If the aircraft's grid belongs to a no-fly zone, a no-fly zone intrusion alarm is generated; if it belongs to a restricted flight zone, a restricted flight zone intrusion alarm is generated. If other aircraft exist in the aircraft's grid and neighboring grids, a neighborhood conflict alarm is generated. If the real-time grid and the predicted trajectory grid are inconsistent, a trajectory deviation alarm is generated. All alarm information is associated with the aircraft's unique identifier, time, spatial location, and grid information, ensuring the traceability and accuracy of the alarm information.

[0116] The parameter adaptive optimization process tracks the entire process, including retrieval time, fusion time, prediction deviation rate, and computational resource utilization. It then performs dynamic optimization based on the interaction results of the three core algorithms. When the retrieval time exceeds 100 milliseconds, the size of the spatiotemporal coupled hash unit is reduced by 50 meters to improve retrieval speed. When the computational resource utilization exceeds 80%, the number of basic spatial grids on one side of the container unit is adjusted to four to reduce computational load. When the prediction deviation rate exceeds 5%, the prediction sampling interval is shortened by 0.5 seconds to improve prediction accuracy. These optimized core parameters are fed back in real-time to the grid modeling, index construction, and spatiotemporal retrieval stages, achieving iterative optimization throughout the entire process and ensuring the system maintains optimal operating conditions. Simultaneously, it reads the deactivated container unit identifiers from the S3 memory hash table, counts the number of container activation and deactivation transitions, generates a container scheduling efficiency evaluation report, completes the closed-loop use of suspended data, and improves the system performance evaluation system.

[0117] The data archiving stage archives standardized, integrated flight track data, alarm information, optimization parameter records, and scheduling efficiency reports in JSON format, arranged chronologically. An index is created using the aircraft's unique identifier and timestamp to generate an anomaly traceability database and a monitoring data archive database, meeting the data retention and traceability requirements for low-altitude airspace supervision. The airspace situation alarm information set, the adaptively optimized full-domain core parameter package, the monitoring data archive database, the anomaly traceability database, and the container scheduling efficiency evaluation report output in this step are the final outputs of this method. This fully realizes all the core functions of three-dimensional airspace grid-based situation monitoring and management, providing comprehensive, high-precision, and high-efficiency technical support for low-altitude airspace safety control, situational awareness, and conflict early warning.

[0118] In one specific embodiment: This embodiment is used to specifically illustrate the complete implementation process of the three-dimensional airspace grid-based situational monitoring and management method based on spatial hashing of the present invention, which is applicable to the scenario of refined monitoring of urban low-altitude airspace and compliance management of UAVs.

[0119] This embodiment selects a typical low-altitude airspace in a certain district of a city as the monitoring scope, the monitoring object is a single light civilian drone, the data source is airborne Beidou high-precision positioning data, and the airspace control elements include a static layer of restricted flight zones.

[0120] Before performing situational monitoring, the underlying system parameters are first solidified, unifying spatial granularity, calculation thresholds, and algorithm hyperparameters to ensure consistent calculation methods and no spatiotemporal deviations throughout the entire process. This embodiment adopts the industry-standard low-altitude monitoring granularity, balancing positioning accuracy and computing power; a three-level hysteresis threshold is used to avoid frequent jitter in the container's status; and a spatiotemporal retrieval threshold is used to filter invalid data and improve retrieval efficiency.

[0121] The globally fixed parameters are as follows: Basic spatial grid side length =50m, number of grid cells on one side of the box unit =8, size of the spatiotemporally coupled hash unit =200m; Container scheduling level three threshold: activation threshold =500m, maintain threshold =800m, Discontinued threshold =1200m; Spatiotemporal retrieval dual thresholds: spatial distance threshold 100m, timestamp deviation threshold 2s; Algorithm hyperparameters: fractional order q=0.9, local weighted projection homomorphic filter weight coefficients =0.6, minimum constant = .

[0122] S1 provides a unified spatiotemporal reference for the entire process, eliminating system errors caused by inconsistencies in coordinates of multiple source devices and time drift. First, the geometric center of the monitored airspace is calculated as the local geocentric rectangular coordinate origin, unifying the conversion reference between latitude and longitude and plane metric coordinates; second, the UTC standard timestamp is calibrated to perform time alignment on all access data; finally, the core parameters of the entire domain are written into memory for global access in subsequent steps.

[0123] Calculation of latitude, longitude, and altitude of the geometric center of the airspace: ;

[0124] , These are the maximum and minimum longitudes, , These are the maximum and minimum latitudes, respectively. , These represent the maximum and minimum heights, respectively; UTC unified time base: =1713340800s;

[0125] The monitored airspace has the following horizontal range: longitude 105.80°~105.90°, latitude 29.30°~29.40°; and vertical range: altitude 0~1000m.

[0126] Substitute into the formula to calculate the center coordinates: ; ; ;

[0127] Time synchronization rules: all flight data timestamps and If the difference is greater than 1 second, a forced time calibration is performed.

[0128] Spatiotemporal reference parameters: Local coordinate origin =(105.85°, 29.35°, 500m), UTC time base =1713340800s; Global core parameter initialization complete.

[0129] S2 is the core of the system's data preprocessing, used to solve problems such as heterogeneous multi-source data, noise interference, and inconsistent coordinates. First, it receives the raw BeiDou positioning data from the UAV, filtering out outdated and disordered data; then, it converts latitude and longitude coordinates into local geocentric rectangular coordinates; it suppresses multiplicative noise through homomorphic filtering and fits coordinate offsets through local weighted projection; it calculates data quality weights based on data source quality, carrier-to-noise ratio, update frequency, and positioning error; finally, it outputs standardized, high-confidence fused trajectory data, simultaneously providing weight inputs for subsequent algorithms.

[0130] Raw BeiDou data for the UAV: ​​Longitude 105.8505°, Latitude 29.3503°, Altitude 520m, Timestamp T=1713340801s;

[0131] The signal carrier-to-noise ratio (SNR) is 45 dB / Hz, the data update frequency is 1 Hz, and the circular probability positioning error (CEP) is 0.8 m.

[0132] Convert latitude and longitude to local geocentric rectangular coordinates: ;

[0133] Locally weighted projection homomorphic filtering fusion model: ;

[0134] Homomorphic filtering noise reduction function: ;

[0135] Data quality weight calculation: ;

[0136] 1. Coordinate transformation calculation: cos(29.35°)≈0.871, substituting, we get: x=(105.8505-105.85)×111000×0.871=48.405m; y=(29.3503-29.35)×111000=33.3m; z=520-500=20m; Original local coordinates: =(48.405,33.3,20)m.

[0137] 2. Filtering and Fitting Calculation: Gaussian Low-Pass Filter Operator =0.95, coordinates after noise reduction: =(46.0,31.6,19.0)m; Locally weighted projection fitting coordinates: =(47.8,32.9,19.8)m; Merged coordinates: =0.6×(46.0,31.6,19.0)+0.4×(47.8,32.9,19.8)=(46.72,32.12,19.32)m;

[0138] 3. Quality weight calculation: Beidou data source coefficient =0.95, carrier-to-noise ratio =0.92, frequency coefficient =1, error coefficient =0.95; =0.95×0.92×1×0.95=0.831.

[0139] Standardized fusion of track coordinates =(46.72,32.12,19.32)m, data quality weights =0.831, timestamp is compliant and out of order, the difference between the data timestamp and the base time is 1 second, less than 3 seconds out of order tolerance window, no outdated data, no filtering required.

[0140] S3 implements 3D spatial discretization modeling, dividing the continuous spatial domain into a two-level structure of standardized grids and boxes, reducing the granularity of retrieval and computation. First, the spatial domain is uniformly cubically partitioned according to the grid side length to generate a unique 3D grid index; then, multiple grids are aggregated into box elements to establish a grid-box hierarchical mapping; the topological relationships of the 26 neighborhoods of the boxes are pre-calculated and cached in an in-memory hash table to avoid real-time redundant calculations, significantly improving neighborhood retrieval efficiency.

[0141] Basic spatial grid 3D index encoding: ;

[0142] Box unit index code: ;

[0143] 1. Grid index calculation: =50m, substitute into the fusion coordinates: =0, =0, =0, the grid index where the aircraft is located: =(0,0,0).

[0144] 2. Box index calculation: =8, a single box contains 8×8×8=512 basic grids, box index: =(0,0,0).

[0145] 3. Topology pre-construction: Generate twenty-six 3D neighboring box indexes for the box, encapsulate the box center coordinates, grid range, and neighbor list as metadata, and store them in an in-memory hash table with the box index as the key, supporting millisecond-level reads.

[0146] Finally, we obtain the basic grid index (0,0,0), the box index (0,0,0), the box 26 neighborhood topology table, and the grid-box mapping table.

[0147] S4 constructs a spatiotemporal integrated index to address the limitation of pure spatial indexes in filtering outdated data. It maps three-dimensional spatial coordinates to discrete hash keys, combines them with timestamps to generate time entropy values, and combines these into a spatiotemporally coupled hash primary key. Grid data is written to the hash index table according to the primary key, achieving rapid coarse screening in both spatial and temporal dimensions, significantly narrowing the search scope and reducing the computational cost of subsequent topology searches.

[0148] Spatial hash key and temporal entropy value encoding:

[0149] ;

[0150] Spatiotemporal coupled hash primary key ;

[0151] =200m, substitute into the fusion coordinates: =0, =0, =0;

[0152] Time entropy value: =1713340801 mod 10000=801;

[0153] Hash primary key: =(0,0,0,801);

[0154] Bind the grid (0,0,0) to the hash primary key and write it into the spatiotemporal coupled hash index table.

[0155] Finally, a spatiotemporal coupled spatial hash index table is obtained, completing the one-to-one mapping between the grid and the spatiotemporal primary key.

[0156] The S5 is the core of aircraft positioning, employing a three-level retrieval mechanism: hash coarse screening, octree fine screening, and topological neighborhood mapping. First, it quickly filters candidate grids using a hash index, filtering out spatiotemporal out-of-limit data. Then, it uses an 8-layer nested octree to converge layer by layer, accurately locating the grid where the aircraft is located. It then calls pre-built topological relationships to generate a set of 26 neighboring grids. Simultaneously, it receives feedback from the prediction algorithm, adaptively adjusting the retrieval radius to ensure real-time positioning accuracy.

[0157] Grid-based precise retrieval model: ;

[0158] Twenty-six neighborhood grid generation model:

[0159] ;

[0160] 3D Euclidean distance operator: ;

[0161] 1. Hash coarse screening: based on Candidate grids were filtered, and the spatiotemporal threshold verification passed, resulting in a candidate set {(0,0,0)}. The aircraft coordinate deviation was 46.72m < 100m spatial threshold, and the time deviation was 1s < 2s time threshold. The spatiotemporal dual threshold verification passed, and no invalid data was removed.

[0162] 2. Octree precise inspection: 8-level traversal, search radius =50m, lock target grid =(0,0,0); This embodiment is a minimalist single-mesh scenario with less than 64 grids in the entire domain. The octree node splitting was not triggered, and the target grid was directly locked.

[0163] 3. Neighborhood Extraction: Generates a set of 26 three-dimensional full neighborhood grids, covering the entire airspace around the aircraft;

[0164] 4. Parameter correction: There is currently no prediction bias, and the search radius remains unchanged.

[0165] Finally, we obtain the grid where the aircraft is located (0,0,0), the set of twenty-six neighboring grids, and the index of the associated box element (0,0,0).

[0166] S6 is used for aircraft future trajectory prediction, adapting to the nonlinear and highly disturbed characteristics of low-altitude flight. It constructs a state transition matrix based on fractional-order difference operators, adaptively adjusting process noise with data quality weights; it calculates multi-time-step predicted coordinates by combining neighborhood grid spatial constraints; it maps the predicted coordinates to a grid sequence, stores them in a memory cache pool, and uses an incremental update mechanism to reduce memory usage; it outputs the prediction bias and uses it to correct the data fusion and retrieval algorithm parameters, forming a closed-loop optimization.

[0167] Fractional Kalman state prediction equation: ;

[0168] Fractional state transition matrix: ;

[0169] Adaptive process noise: ;

[0170] Prediction deviation calculation: ;

[0171] 1. Input parameters: Control vector U=(2m / s,0°,0m / s), basic noise covariance =0.1;

[0172] 2. Noise Calculation: =0.1×(1-0.831)=0.0169;

[0173] 3. Single-step prediction: =(48.72,32.12,19.32)m;

[0174] 4. Deviation verification: =2.0m≤5m, meets the accuracy requirements;

[0175] 5. Mesh mapping: Predict the coordinates to correspond to the mesh (0,0,0), generate a time series mesh sequence and cache it.

[0176] The final predicted trajectory base grid sequence [G(0,0,0)] is obtained, with a prediction bias of 2.0m, and the cache incremental update is completed.

[0177] S7. Implement digital management rules by deeply binding static / dynamic airspace layers with 3D meshes. The system imports GeoJSON format restricted flight zone layers, parses geometric boundaries and management attributes; matches the meshes and boxes covered by the layer through spatial inclusion determination; and marks alarm levels according to management type (Level 3 for no-fly zones, Level 2 for restricted flight zones, and Level 1 for flight routes), generating a binding relationship table to provide a rule basis for subsequent alarm calculation and computing power scheduling.

[0178] Execution process: Import the restricted flight zone layer, and the geometric range covers the grid (0,0,0); the spatial inclusion determination result is that the center of the grid falls into the restricted flight zone; establish the mapping relationship: feature ID - grid (0,0,0) - box (0,0,0), mark the secondary alarm level, and define it as a high priority control grid.

[0179] Finally, the binding relationship table of airspace elements-grid-box is obtained, and the secondary alarm level marking is completed.

[0180] S8 serves as the core of computing power optimization, enabling on-demand allocation of computing resources. It uses the three-dimensional distance between the aircraft and the center of the container as the determining factor, combined with a three-level hysteresis threshold to switch between container activation / maintenance / deactivation states; it forces activation of containers with high-level alarms to ensure persistent computing power in the controlled area; and it synchronizes scheduling status to a memory hash table in real time, unloading invalid container data and significantly reducing system computing overhead.

[0181] Three-dimensional distance between the center of the aircraft and the enclosure:

[0182] ;

[0183] 1. Center coordinates of the box: =200m, =200m, =200m;

[0184] 2. Distance calculation: =268.3m≤500m, which meets the activation condition;

[0185] 3. Forced activation: The container is permanently activated by binding to a level 2 restricted flight zone, and the deactivation logic is not executed;

[0186] 4. State synchronization: Update the memory hash table and mark the box (0,0,0) as active.

[0187] The final result is the active container unit computing dataset, with high-priority container computing power always resident.

[0188] S9 is the final output and closed-loop optimization stage of the system. Based on grid location, neighborhood information, predicted trajectory and control rules, it generates multiple types of situational alarms; it calculates the system's retrieval time, computing power consumption, and prediction deviation, and adaptively adjusts the core parameters of the entire domain; all process data is archived in a standard format, and a unique traceability index is established to achieve a complete closed loop of situational monitoring, risk warning, performance optimization, and data retention.

[0189] Execution process:

[0190] 1. Alarm Calculation: When an aircraft is located in a Level 2 restricted flight zone grid, a restricted flight zone intrusion alarm is generated, which is associated with the aircraft ID, timestamp, and grid / box information;

[0191] 2. Closed-loop optimization: The retrieval time was 80ms, the computing power utilization was 65%, and the prediction error rate was 4.2%, all of which were better than the threshold, without any parameter adjustment;

[0192] 3. Data archiving: Archive track, alarm, scheduling, and optimization data in JSON format and establish dual-index retrieval.

[0193] Ultimately, we obtain intrusion alarm information in restricted flight zones, optimal system operating parameters, and traceable monitoring archive data throughout the entire process.

[0194] In summary, several specific embodiments of the present invention have been disclosed. Without contradiction, the various embodiments can be freely combined to form new embodiments. That is, embodiments that are alternative solutions can be freely substituted for each other, but cannot be combined with each other; embodiments that are not alternative solutions can be combined with each other. These new embodiments are also part of the substantive content of the present invention.

[0195] The above embodiments describe several specific implementations of the present invention. However, those skilled in the art should understand that various changes or modifications can be made to these implementations without departing from the principles and essence of the present invention, but all such changes and modifications fall within the protection scope of the present invention.

Claims

1. A three-dimensional spatial gridded situational awareness monitoring and management method based on spatial hashing, characterized in that, include: S1. Perform spatiotemporal reference calibration and global core parameter initialization on the three-dimensional monitoring airspace to obtain spatiotemporal reference parameters and global core parameters; S2. Acquire raw flight data of multi-source low-altitude aircraft, and process the data using a local weighted projection homomorphic filtering algorithm based on spatiotemporal reference parameters and global core parameters to obtain standardized fused flight track data; S3. Based on the core parameters of the entire domain and standardized fused track data, construct a three-dimensional airspace basic spatial grid and box unit hierarchical structure, pre-construct the adjacent topological relationship of box units, and obtain the grid index, box index and adjacent topological relationship of box units; S4. Based on the global core parameters, grid index and standardized fused track data, construct a spatiotemporal coupled spatial hash index to obtain the spatiotemporal coupled spatial hash index table. S5. Based on the spatiotemporal coupled spatial hash index table, standardized fused track data, global core parameters and adjacent topological relationships of container units, the spatial hash nested octree topological retrieval algorithm is used to process the data to obtain the basic spatial grid where the aircraft is located, the set of twenty-six neighboring basic spatial grids and the index of associated container units. S6. Based on the basic spatial grid where the aircraft is located, standardized fused track data and grid index, an adaptive fractional Kalman filter algorithm is used to process the data to obtain the basic spatial grid sequence of the predicted trajectory. S7. Import multi-format spatial layer files and complete the parsing. Based on the grid index and box index, complete the binding mapping between spatial features and basic spatial grids and box units to obtain the binding relationship table. S8. Based on the adjacent topology relationship of the container units, the associated container unit index, the binding relationship table and the global core parameters, perform dynamic hysteresis activation and deactivation scheduling of the container units at the three-level threshold to obtain the activated container unit calculation dataset. S9. Based on the aircraft's location in the basic spatial grid, the set of twenty-six neighboring basic spatial grids, the predicted trajectory basic spatial grid sequence, the activated box unit calculation dataset, the binding relationship table, and the standardized fused trajectory data, complete the calculation of airspace situational alarms, adaptive optimization of closed-loop parameters, and archiving of monitoring data.

2. The three-dimensional spatial gridded situation monitoring and management method based on spatial hashing according to claim 1, characterized in that, The local weighted projection homomorphic filtering algorithm first converts the original flight data of multi-source low-altitude aircraft into local geocentric rectangular coordinates, performs multiplicative noise reduction through homomorphic filtering, and then performs coordinate offset fitting through local weighted projection, finally outputting standardized fused flight track data.

3. The three-dimensional spatial gridded situation monitoring and management method based on spatial hashing according to claim 2, characterized in that, During the processing of the local weighted projection homomorphic filtering algorithm, outdated flight data is filtered through an out-of-order tolerance window. Corresponding coefficients are calibrated based on the data source type, signal-to-noise ratio, data update frequency, and data position error, and data quality weight values ​​are calculated based on the corresponding coefficients.

4. The three-dimensional spatial gridded situation monitoring and management method based on spatial hashing according to claim 3, characterized in that, The spatial hash nested octree topology retrieval algorithm first completes the coarse screening of candidate grids through a spatiotemporal coupled spatial hash index table, and then completes the precise positioning of the basic spatial grid where the aircraft is located by traversing the nested octree layer by layer. Combined with the adjacent topological relationship of the box unit, a set of twenty-six neighborhood basic spatial grids is obtained.

5. The three-dimensional spatial gridded situation monitoring and management method based on spatial hashing according to claim 4, characterized in that, The adaptive fractional Kalman filter algorithm constructs a fractional state transition matrix based on a fractional difference operator, calculates process noise using data quality weights, completes aircraft trajectory prediction, and maps it to a predictive trajectory base space grid sequence.

6. The three-dimensional spatial gridded situation monitoring and management method based on spatial hashing according to claim 5, characterized in that, After the adaptive fractional Kalman filter algorithm is completed, the predicted trajectory basic spatial grid sequence is stored in the memory cache pool with the aircraft's unique identifier as the key. When the trajectory data is updated, only the sequence difference part is updated incrementally.

7. The three-dimensional spatial gridded situation monitoring and management method based on spatial hashing according to claim 6, characterized in that, The local weighted projection homomorphic filtering algorithm, the spatial hash nested octree topology search algorithm, and the adaptive fractional Kalman filter algorithm form a real-time bidirectional interaction between each other; The local weighted projection homomorphic filtering algorithm outputs standardized fused track data in the forward direction to the spatial hash nested octree topology search algorithm, and receives the basic spatial grid index of the aircraft in the reverse direction to correct the data source type coefficient; The locally weighted projection homomorphic filtering algorithm outputs data quality weight values ​​in the positive direction to the adaptive fractional Kalman filtering algorithm, and receives prediction bias in the negative direction to correct the filtering projection weight coefficients. The spatial hash nested octree topology search algorithm outputs a set of 26 neighborhood basic spatial grids to the adaptive fractional Kalman filter algorithm in the forward direction, and receives the predicted coordinates in the reverse direction to correct the search radius.

8. The three-dimensional spatial gridded situation monitoring and management method based on spatial hashing according to claim 1, characterized in that, In the binding relationship table, the basic spatial grid corresponding to the airspace element is marked with an alarm level. The no-fly zone corresponds to the third-level alarm level, the restricted flight zone corresponds to the second-level alarm level, the flight path corresponds to the first-level alarm level, and the third-level and second-level alarm levels are high-level alarm grids. The container unit's three-level threshold dynamic hysteresis activation and deactivation scheduling switches the container state based on the three-dimensional distance between the aircraft and the center of the container unit. It executes a forced activation strategy on container units bound to high-level alarm grids and synchronously updates the scheduling state to the memory hash table.

9. The three-dimensional spatial gridded situation monitoring and management method based on spatial hashing according to claim 1, characterized in that, The airspace situation alarm calculation generates neighborhood conflict alarms, no-fly zone intrusion alarms, restricted-fly zone intrusion alarms, and trajectory deviation alarms based on the binding relationship table. All alarms are associated with the aircraft's unique identifier, time information, and grid information.

10. The three-dimensional spatial gridded situation monitoring and management method based on spatial hashing according to claim 1, characterized in that, The closed-loop parameter adaptive optimization dynamically adjusts the core parameters of the entire domain based on the system retrieval time, computing resource utilization, and prediction deviation rate, and feeds them back to the grid modeling, index construction, and spatiotemporal retrieval stages to form a closed-loop collaboration.

Images