A method for assessing the risk of drone interference in low-altitude airspace using big data processing

Abstract

This invention discloses a method for assessing the risk of unmanned aerial vehicle (UAV) interference in low-altitude airspace using big data processing, relating to the field of airspace safety technology. This method systematically models interference risks by progressively breaking down complex airspace safety issues into several sub-processes with clear physical meaning and computational logic. The front end of the process mainly completes the modeling and discretization of airspace structure and operational elements, providing a unified spatial and temporal scale for subsequent risk calculations. The intermediate steps characterize potential UAV attack paths and their spatial intersections with manned aircraft tracks. The final step identifies core risk areas and secondary risk areas through risk aggregation and threshold determination. This invention normalizes static collision risk values ​​to obtain standardized risk indicators, providing a more operational spatial basis for implementing layered, graded, and differentiated low-altitude integrated airspace management measures.

Description

Technical Field

[0001] This invention relates to the field of airspace safety technology, and more specifically, to a method for assessing the risk of unmanned aerial vehicle (UAV) interference in low-altitude airspace using big data processing. Background Technology

[0002] In recent years, with the popularization of drone technology, the deployment and construction of the low-altitude economy have been accelerating. However, the phenomenon of illegal drone flights is increasing, and drone flight accidents occur frequently, posing challenges to the healthy development of the low-altitude economy. In particular, the problem of unauthorized drone flights poses a safety risk to aviation safety. The abuse by some users is increasingly threatening the flight safety of manned aircraft. As an important area for flight activities, the low-altitude airspace where unauthorized and manned aircraft fly together faces particularly prominent risks of drone interference. When unauthorized drones fly near this area, they can affect the normal flight of manned aircraft and potentially lead to safety accidents. Summary of the Invention

[0003] To address the shortcomings of existing technologies, the present invention aims to provide a method for assessing the risk of drone interference in low-altitude airspace using big data processing.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] A method for assessing the risk of drone interference in low-altitude airspace using big data processing, the method comprising the following steps:

[0006] S1. Low-altitude unmanned and manned aircraft fusion airspace modeling; dividing the low-altitude unmanned and manned aircraft fusion airspace into first-class and second-class regions to achieve a hierarchical expression of the low-altitude fusion airspace structure; and uniformly mapping the low-altitude unmanned and manned aircraft fusion airspace to a three-dimensional grid space to provide a standardized spatial carrier for subsequent probability statistics and risk fusion.

[0007] The first type of region is a spatial region distributed along the flight path of manned aircraft and concentrated in the lateral and altitude directions; the second type of region is located outside the first type of region, extending in the flight path extension direction, lateral range, and altitude level, and is used to describe the extended space of the potential impact of UAV activities on manned aircraft operations within the low-altitude integrated airspace; the construction of the three-dimensional grid space is specifically as follows:

[0008] (1)

[0009] Where G represents the set of all 3D meshes in the 3D hierarchy; V represents the set of all 3D mesh nodes; E represents the set of connected edges in the 3D mesh; and adjacent 3D meshes are connected as follows:

[0010] (2)

[0011] Where l(e) is the center distance between the three-dimensional meshes, and c i c is the center coordinate of the i-th 3D grid. j It is the center coordinate of the j-th 3D grid.

[0012] S2. Evaluation of manned aircraft 3D grid distribution based on historical flight track data; At the spatial condition level, spatial probability distribution models are constructed for both manned and unmanned aircraft. The manned aircraft trajectory is obtained by cleaning, reconstructing the time series, and interpolating the spatial trajectory of historical flight track data. The take-off and landing flight trajectories are reproduced. By statistically analyzing the crossing frequency and flight exposure time of flight trajectories in each 3D grid, and calculating the probability of manned aircraft appearing in the 3D grid, the spatial distribution characteristics of manned aircraft in the 3D airspace are characterized.

[0013] The formula for calculating the probability of a human-machine interface appearing within a 3D mesh is:

[0014] (8)

[0015] Among them, A ijk Represents a three-dimensional mesh V ijk The probability of a space containing human or machine space. The number of times the 3D grid is traversed by all flights. This is the sum of the number of times all 3D meshes are traversed.

[0016] S3. Quantitative modeling of UAV intrusion probability based on shortest attack path in 3D mesh: Starting from potential disturbance behavior, with the second type of region and the boundary of the first type of region as the potential starting point set and the airspace of manned aircraft trajectory and its extension as the target set, a shortest attack path intrusion frequency model under the constraint of starting point-end point pair is constructed, and the intrusion probability of UAV in 3D mesh is quantitatively calculated.

[0017] For each attack source point s, Dijkstra's shortest path algorithm is used to calculate the shortest path from it to all target grid points, defined as:

[0018] (13)

[0019] Where t is any target grid point; u is any path; u( Let be a feasible path from source point s to target point t; dist(i,j) is the Euclidean distance between 3D meshes i and j; for attack source point s, select the shortest path among all target points:

[0020] (14)

[0021] Where T is the set of all target grid points; this path is called the optimal attack path of the attack source point s; the effective paths of all attack source points are accumulated and statistically analyzed, specifically as follows:

[0022] (15)

[0023] Where v is a single specific 3D mesh node; S is the set of all attack source points; s Let be the optimal path starting from s; the indicator function W is:

[0024] (16)

[0025] for When the three-dimensional mesh V ijk lie in The value is 1 if the source point s is on a valid attack path, otherwise it is 0.

[0026] Finally, the frequency of the shortest attack path in each 3D grid in the entire 3D spatial domain is obtained:

[0027] (17)

[0028] The intrusion intensity of all 3D meshes is uniformly processed using the Min–Max normalization method, and its expression is:

[0029] (18)

[0030] Among them, F ijk The shortest attack path for all attack sources through the 3D grid V ijk The number of times, F max With F min These represent the maximum and minimum values ​​of intrusion intensity across all 3D meshes within the study area, respectively. Finally, by probabilistically processing the normalized intrusion intensity, the intrusion probability of the UAV within the 3D mesh is obtained:

[0031] (19)

[0032] S4. Static collision risk assessment of UAVs and manned aircraft based on conditional probability decomposition; constructing the conditional probability C of UAVs and manned aircraft appearing simultaneously and forming spatial overlap. ijk The risk of static collisions between UAVs and manned aircraft is quantitatively characterized on a three-dimensional grid scale.

[0033] Conditional probability C ijk The calculation formula is:

[0034] (twenty one)

[0035] in, This indicates that the drone has entered the 3D grid V. ijk The probability is calculated using the drone's shortest attack path frequency model; A ijk This indicates that manned aircraft are in a 3D mesh V ijk The probability of it appearing within the range is obtained from historical flight track data.

[0036] S5. Generate a static collision risk map; conduct static risk simulation experiments for low-altitude airspace, quantitatively calculate the collision risk within each three-dimensional grid, and classify the three-dimensional grid units accordingly, dividing the airspace into high-risk, medium-risk, and low-risk zones to form a corresponding static collision risk distribution map, providing a basis for prevention and control zoning and risk management.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] (1) The new first-class risk area identified by the present invention is highly consistent with the area where UAVs and manned aircraft frequently encounter each other in spatial location. It can effectively cover the actual collision events and their corresponding axisymmetric locations, while avoiding the inclusion of a large area of ​​low-risk airspace into the high-level control scope.

[0039] (2) The zoning method based on conditional probability decomposition and relative risk classification improves the spatial precision and targeting of risk zoning while maintaining safety and conservatism.

[0040] (3) From the perspective of spatial structure, the zoning results generated based on the static collision risk field can naturally reflect the continuity and aggregation of risks as they change with spatial location, providing a more operational spatial basis for the subsequent implementation of layered, graded and differentiated low-altitude integrated airspace management measures. Attached Figure Description

[0041] Figure 1 This is a technical roadmap of an embodiment of the present invention;

[0042] Figure 2 This is a flowchart illustrating an embodiment of the present invention;

[0043] Figure 3 This is a schematic model diagram of the fused spatial domain according to an embodiment of the present invention;

[0044] Figure 4 This is a schematic diagram of the gridded splitting result in the first type of airspace region according to an embodiment of the present invention;

[0045] Figure 5 This is a schematic diagram of the spatial region gridding result in the fused spatial domain according to an embodiment of the present invention;

[0046] Figure 6This is a schematic diagram of the manned aircraft historical flight track data reproduction and grid probability distribution results according to an embodiment of the present invention;

[0047] Figure 7 This is a three-dimensional spatial distribution diagram of the drone mesh intrusion probability according to an embodiment of the present invention;

[0048] Figure 8 A histogram showing the distribution of the number of times a drone's flight path crosses a grid.

[0049] Figure 9 Airspace collision risk result diagram of this invention embodiment;

[0050] Figure 10 Comparison diagram of area delineation in embodiments of the present invention. Detailed Implementation

[0051] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to specific examples, such as... Figure 1 and Figure 2 As shown:

[0052] S1, Low-altitude unmanned and manned aircraft integrated airspace modeling

[0053] Under a unified three-dimensional mesh framework, the multi-layered airspace surrounding the fusion airspace of low-altitude unmanned and manned aircraft is discretized, and the fusion airspace of low-altitude unmanned and manned aircraft is uniformly mapped to a regular three-dimensional mesh space, providing a standardized spatial carrier for subsequent probability statistics and risk fusion.

[0054] Based on the survey and analysis of typical low-altitude unmanned and manned aircraft integrated operation scenarios, the integrated airspace can be divided into two types of spatial regions with different functional attributes. The first type of region is distributed along the main flight path of manned aircraft, maintaining a relatively concentrated spatial range in the lateral and altitude directions, and is used to characterize the key airspace where the probability of close-range aerial encounters between UAVs and manned aircraft is high. The second type of region is located outside the first type of region, extending in the flight path extension direction, lateral range, and altitude, and is used to describe the extended space of the potential impact of UAV activities on manned aircraft operations within the integrated airspace. Through regional division, a hierarchical expression of the low-altitude integrated airspace structure is achieved, providing a spatial basis for subsequent collision risk zoning modeling and quantitative assessment. Among them, the first type of region is a near-field low-altitude composite space constructed along the main flight path of manned aircraft, with its longitudinal range extending X1km to both ends of the flight path, its lateral range extending Y1km to both sides of the central axis, and its altitude range not exceeding Z1m. The second type of region extends outward from the first type of region, with its longitudinal extension range increasing to X2km, its lateral range expanding to Y2km on both sides of the central axis, and its upper limit of height increasing to Z2m. It also subtracts the area covered by the first type of region in space, and is used to characterize the extended influence area of ​​the fused airspace.

[0055] like Figure 3 Taking the example shown, the first type of area extends 6km longitudinally to both ends of the direction of operation, and 1.5km laterally to both sides of the central axis, with a height not exceeding 300m. The second type of area increases the longitudinal extension to 9km, expands the lateral extension to 4.5km to both sides of the central axis, raises the upper limit of height to 500m, and spatially deducts the area covered by the first type of area.

[0056] The fused airspace is discretized into a regular 3D grid. The resolution of the 3D grid is set considering the feasibility of risk assessment and the characteristics of typical low-altitude operating environments. The aim is to uniformly map and quantify multiple risk elements from different sources and with inconsistent dimensions onto the 3D grid, thereby forming a computable and superimposed standardized risk field. As the smallest risk calculation unit in space, the 3D grid effectively reduces the computational complexity of continuous spatial risk modeling while maintaining the geometric structure characteristics of the fused airspace, providing a data foundation for the fusion of multi-source risk elements. Numerical simulations are performed on different spatial levels within the fused airspace, dividing it into regular 3D grid units. The geometric scale is parametrically described using the form of "length × width × height." Each grid unit is considered the basic calculation unit for risk assessment. A bidirectional mapping relationship is established between geographic coordinates and the 3D grid index, and a corresponding spatial index structure is constructed to improve the efficiency of subsequent spatial queries, risk superimposition, and statistical calculations. The 3D grid model is constructed using a regular grid form with fixed length, width, and height, set according to the application scenario and actual operating conditions. The specific construction of the fused airspace element grid map is as follows:

[0057] (1)

[0058] Where G represents the set of all 3D meshes in the 3D layer; V represents the set of 3D mesh nodes; and E represents the set of connected edges in the 3D mesh. Connecting adjacent 3D meshes:

[0059] (2)

[0060] Where l(e) is the center distance between the three-dimensional meshes, and c i c is the center coordinate of the i-th 3D grid. j It is the center coordinate of the j-th 3D grid.

[0061] Taking a length * width * height of 800m * 800m * 30m as an example, its first type of region grid splitting construction is as follows: Figure 4 As shown, this illustrates the effect of discretizing the first type of region into a regular three-dimensional mesh; while the overall spatial region mesh is constructed as follows: Figure 5 As shown.

[0062] In the process of 3D spatial grid modeling and trajectory mapping, the spatial displacement of the local rectangular coordinate system is converted into the latitude and longitude changes of the geographic coordinate system, with the target spatial domain as a reference. Considering the relatively limited regional scale, an approximate Earth sphere model is adopted in the coordinate transformation, mapping the north-south and east-west displacements of the planar coordinate system into latitude and longitude increments, respectively. The calculation relationship can be expressed as:

[0063] (3)

[0064] (4)

[0065] in, This represents the amount of planar displacement along the north-south direction. It represents the amount of planar displacement along the east-west direction; This represents the Earth's average radius; its values ​​are 6,378,137 m. Indicates the initial latitude of the reference point, used to correct for changes in the spacing between meridians at different latitudes; and These represent the corresponding changes in latitude and longitude, respectively.

[0066] S2. Assessment of manned aircraft 3D grid distribution based on historical flight track data

[0067] At the spatial condition level, spatial probability distribution models are constructed for both manned and unmanned aircraft. The manned aircraft trajectory is obtained by cleaning, reconstructing the time series, and interpolating the spatial trajectory of historical flight track data. The take-off and landing trajectories are reproduced at the grid level. By statistically analyzing the crossing frequency and flight exposure time of the trajectory in each grid, the spatial distribution characteristics of manned aircraft in the three-dimensional airspace are characterized.

[0068] The civil aviation flight grid probability estimation based on historical flight track big data adopts a multi-source spatiotemporal data gridded statistical algorithm and a flight track probability field generation method. This method references gridded spatiotemporal data correlation analysis methods, providing a unified spatiotemporal grid framework for correlation analysis of multi-source heterogeneous data. Furthermore, the applicability of this method has been verified in low-altitude airspace planning for UAV swarms, and the constructed 3D grid model provides an effective reference for spatial index data structures and flight track representation methods.

[0069] S21, Manned machine data collection and processing

[0070] The manned aircraft operation data used in this method comes from the Quick Access Recorder (QAR). To highlight the actual impact of UAV interference risks on civil aviation safety and in line with the key concerns of UAV control in airspace, this method only selects low-altitude operation data during takeoff and approach / landing phases for analysis. Specifically, it uniformly extracts flight segments below the HMax altitude from the QAR data. In the data processing, the original QAR data is first synchronized in time and outlier removed to ensure data continuity and reliability. Then, based on the flight altitude information, data segments below the HMax are filtered and mapped to the aforementioned constructed three-dimensional airspace grid to statistically analyze the frequency of manned aircraft occurrence in each grid. Through the above processing, a spatial probability distribution of manned aircraft activity in low-altitude airspace can be formed, providing a basic input for subsequent fusion calculations with UAV occurrence probabilities.

[0071] S22, Manned Aircraft Flight Trajectory Reconstruction

[0072] The method for reconstructing manned aircraft flight trajectories involves projecting pre-processed historical manned aircraft trajectories one by one onto a 3D grid. The quantification method involves counting the frequency with which each 3D grid cell is traversed by a flight trajectory. A higher traversal frequency indicates a greater probability of manned aircraft activity in that airspace and a higher risk of conflict. The specific technical implementation is as follows:

[0073] Suppose the human-machine trajectory is a continuous spatial and temporal sequence. n =(X n ,Y n Z n ,t n ), where S n =(X n ,Y n Z n ,t n X represents the state information of the nth trajectory point. n Y n Z n Let t represent the position coordinates of the trajectory point in three-dimensional space. n This represents the corresponding timestamp, used to determine whether the track has passed through a certain 3D grid V. ijk We employ a "line segment-axis aligned mesh correlation determination method." By solving for the parameterized intersections of the line segment with the mesh boundary in three coordinate dimensions, we can obtain the parameter range for the line segment entering and leaving the mesh. Wherein, α... enter α represents the heading angle of a flying target as it enters the target area. exit This represents the heading angle of a flying target as it leaves the target area. All heading angles are defined in a unified angular reference frame. If α... enter ≤α exit α enter ≤1、α exit If the value is ≥0, then the flight path segment is considered to have spatially intersected with the 3D grid, meaning the flight actually passed through the 3D grid. Based on this, a 3D binary matrix O can be constructed. ijk This is used to describe whether a flight passes through a certain 3D grid:

[0074] (5)

[0075] The human-machine trajectory is set between two adjacent time points S n =(X n ,Y n Z n ,t n ) and S n+1 = (X n+1 ,Y n+1 Z n+1,t n+1 A line segment is formed. By determining the intersection of the line segment and the axis-aligned mesh, the position of the line segment within the 3D mesh V can be obtained. ijk The parameter enters the value α. enter With departure value α exit After confirming the alignment of the flight path segments with the 3D mesh V... ijk Given spatial intersections, the dwell time of a trajectory within this three-dimensional grid can be expressed as:

[0076] (6)

[0077] in, Indicates the first The flight path segment in the three-dimensional mesh Flight exposure time within the area, This represents the time span corresponding to the line segment. Furthermore, by summing all the flight path segments for the same flight, the time span of that flight within the three-dimensional grid V can be obtained. ijk Total stay time inside:

[0078] (7)

[0079] Where, N seg This indicates the relationship between the flight trajectory and the 3D grid V. ijk The number of intersecting line segments. Using the above method, accurate statistics of UAV flight exposure time can be achieved at the three-dimensional airspace grid level, providing a time-weighted basis for subsequent UAV-UAV static collision risk calculation. An embedded algorithm program is designed to reproduce the flight path of each flight, while the backend automatically counts all grids traversed by the flight and the time spent traversing each grid, such as... Figure 7 and Figure 8 As shown, the reproduction result is then output.

[0080] S23, Statistics on the Distribution of Manned and Machine-Manned 3D Grids

[0081] Each flight path in the historical flight track may traverse multiple 3D grids. The probability of a manned or aircraft appearing within a given 3D grid is determined by accumulating the number of traversals and the ratio of the number of times each 3D grid is traversed to the total number of times all 3D grids in the entire area are traversed. A higher probability of traversal indicates denser manned or aircraft activity at that location, and a greater risk. Specifically:

[0082] (8)

[0083] Among them, A ijk Represents a three-dimensional mesh V ijk The probability of a space containing human or machine space. The number of times the 3D grid is traversed by all flights. This is the sum of the number of times all 3D meshes are traversed.

[0084] like Figure 6 As shown, a total of 28 three-dimensional grids were recorded in the existing flight statistics, all within the first-class area and the space below 500m directly above the first-class area. Among them, the three-dimensional grid was passed through the fewest times (5 times) and the three-dimensional grid was passed through the most times (10 times). The darker the color, the greater the probability of a manned aircraft appearing on the three-dimensional grid; the maximum probability of a manned aircraft appearing on the three-dimensional grid is 0.05128, and the minimum probability of a manned aircraft appearing on the three-dimensional grid is 0.02564.

[0085] S3. Quantitative Modeling of Drone Intrusion Probability Based on Shortest Attack Path in 3D Mesh

[0086] Starting from potential disruptive behavior, a shortest attack path intrusion frequency model is constructed under the constraint of "start-end pair" with the boundary of the second type of region and the first type of region as the potential starting point set and the airspace of manned aircraft trajectory and its extension as the target set. A distance attenuation mechanism is introduced to describe the inhibitory effect of prevention and control capabilities on UAV intrusion behavior, thereby forming the spatial probability distribution of UAVs at the three-dimensional grid level.

[0087] The UAV mesh intrusion probability model comprehensively quantifies the threat level of UAVs to key targets at different airspace scales by considering both the frequency of the shortest attack path and the probability of the behavior occurring. This method aims to characterize the relative risk contribution of UAVs to various locations in space when they approach key targets along potentially optimal intrusion paths within a low-altitude fusion airspace.

[0088] During model construction, a regular 3D mesh was used to discretize the fused airspace, and analyses were conducted at different spatial levels. For each level, the mesh containing the key target was selected as the endpoint, and the shortest attack path from all potential intrusion sources to the target mesh was calculated. Intrusion sources were set in a multi-source distribution to reflect the situation where UAVs might enter the fused airspace from different spatial locations.

[0089] In the shortest path search process, a distance-based drone intrusion probability decay model is introduced and embedded layer by layer into the path calculation results to characterize the characteristic that the probability of drones committing intrusion behavior gradually decreases with increasing distance. By traversing all source points and counting the cumulative frequency of each grid cell appearing in all shortest attack paths, the path occurrence frequency can be transformed into a spatial probability distribution, thereby forming a drone intrusion risk heat field covering the entire domain.

[0090] Ultimately, the constructed UAV grid intrusion probability can intuitively reflect the relative importance of different spatial locations within the fused airspace in the potential UAV intrusion path system, providing a quantitative basis for subsequent risk superposition analysis and prevention and control strategy formulation.

[0091] The distance decay model assumes that the closer a drone is to the center of the airspace, the stronger the defenses become, and the probability of its successful intrusion decreases exponentially. Let d be the horizontal distance of a grid cell from the boundary of either the second or first type of region. The larger d is, the lower the probability of intrusion. The intrusion probability is defined as:

[0092] (9)

[0093] in, The attenuation coefficient determines the rate at which risk decreases with distance. Comprehensive research reveals that the attenuation coefficient is commonly used in UAV detection range performance and risk models for low-altitude, slow-moving, and small aircraft. This study adopts... This value reflects the reality of weaker monitoring capabilities in the outer perimeter and more stringent control in the inner perimeter. The probability of intrusion into the grid within the second type of area is:

[0094] (10)

[0095] As the distance from the grid to the target area increases, the intrusion probability decreases rapidly. Meanwhile, to maintain probability continuity at the boundary between the second and first type of regions, the probability of the first type of region is defined as:

[0096] (11)

[0097] in: (12)

[0098] d max The value of the second-class region model at its innermost boundary is used as the initial coefficient of the first-class region model to ensure a continuous transition of probabilities.

[0099] S31. Construction of Drone Intrusion Grid Model

[0100] Based on the aforementioned 3D mesh model, the first and second types of regions are uniformly represented as a weighted undirected graph G=(V,E) consisting of a set of nodes V and a set of connected edges E. Each mesh center serves as a graph node, and adjacent meshes are spatially connected through connected edges. The edge weight is determined by the Euclidean distance between adjacent mesh centers, representing the flight cost of the UAV in space. Using the designated UAV intrusion source points at each level as starting points, and the set of meshes containing the manned UAV trajectory and its extension as the target region, a decay model is embedded. A shortest path search algorithm is used to calculate the shortest attack path from each source point to the target region. By traversing all source points and recording the corresponding shortest path sets, the cumulative number of times each mesh is traversed in all shortest attack paths is counted, thus obtaining the frequency distribution of the UAV's shortest attack path. The specific implementation method is as follows:

[0101] Assuming the drone travels at V u The drone flies at a fixed speed, but this speed is only used to calculate the path length and does not participate in path optimization. This speed value is primarily based on typical performance parameters of current civilian-use lightweight drones. On one hand, according to the technical specifications of mainstream multi-rotor and small fixed-wing drones, their cruising speeds are typically distributed in the range of 10–30 m / s, often approaching their economic cruising speed limit when performing long-distance or rapid-crossing missions. On the other hand, in the highly sensitive environment of the surrounding airspace, if a drone engages in unauthorized flight or uncontrolled intrusion, it is more likely to cross critical airspace at a higher cruising speed, thus creating a more unfavorable situation for manned aircraft operations. For each attack source point s, the Dijkstra shortest path algorithm is used to calculate the shortest path from it to all target grid points, defined as:

[0102] (13)

[0103] Where t is any target grid point; u is any path; u( Let be a feasible path from source point s to target point t; dist(i,j) is the Euclidean distance between 3D meshes i and j; for attack source point s, select the shortest path among all target points:

[0104] (14)

[0105] Where T is the set of all target grid points; this path is called the effective attack path of the attack source point s; the grid attack traversal frequency is defined as the cumulative statistics of the effective paths of all source points, specifically:

[0106] (15)

[0107] Where v is a single specific 3D mesh node; S is the set of all attack source points; s Let be the optimal path starting from s; the indicator function W is:

[0108] (16)

[0109] for When the three-dimensional mesh V ijk The value is 1 if the attack source point s is on a valid attack path, otherwise it is 0.

[0110] Finally, the frequency of the shortest attack path in each 3D grid in the entire 3D spatial domain is obtained:

[0111] (17)

[0112] This frequency characterizes the attack channels that drones may use, forming a spatial distribution of "attack path heat".

[0113] S32, Probability Quantization of Drone Intrusion into Grid

[0114] Building upon the completed UAV mesh intrusion model, this study further quantifies the intrusion probability of UAVs within a 3D airspace mesh. This process transforms discrete path statistics into a spatially comparable meshed intrusion probability distribution. To eliminate numerical magnitude differences between different meshes and facilitate spatial comparative analysis, the study employs the Min–Max normalization method to standardize the intrusion intensity across all meshes, expressed as:

[0115] (18)

[0116] Among them, F ijk The shortest attack path through grid V for all attack sources ijk The number of times, F max With F min Let represent the maximum and minimum values ​​of the intrusion intensity for all grid cells within the study area, respectively. Finally, by probabilistically processing the normalized intrusion intensity, the probability of a UAV intruding into a grid cell can be obtained:

[0117] (19)

[0118] in For drones to appear in a 3D mesh The probability distribution, spatially covering the entirety of both the first and second category regions, reflects the relative threat intensity of UAVs to key target areas in the airspace, serving as a crucial input parameter for subsequent UAV-managed aircraft static collision risk assessment. Furthermore, all three-dimensional meshes satisfy:

[0119] (20)

[0120] Figure 7 This is a 3D spatial distribution map of UAV grid intrusion probability, quantifying the intrusion probability of each 3D grid within the fused airspace. Different colors are used to distinguish different levels of intrusion probability, with red representing the grids with the highest intrusion probability, followed by yellow, and white representing the lowest. Grids closer to Category I areas, runways, and flight path extensions have significantly higher intrusion probabilities, forming continuous high-probability intrusion channels along the manned aircraft flight path extension. These high-probability areas are mainly concentrated at low altitudes, with the intrusion probability gradually decreasing with increasing altitude, reflecting the trend of UAVs concentrating on the shortest path from the boundary of Category II areas to the core airspace. Figure 8 This is a histogram showing the distribution of the number of times the drone's flight path crossed the grid.

[0121] S4. Static Collision Risk Assessment of Unmanned and Manned Aircraft Based on Conditional Probability Decomposition

[0122] The conditional probability C of the simultaneous appearance of drones and manned aircraft forming spatial overlap. ijk This allows for the quantitative characterization of static collision risks between UAVs and manned aircraft on a three-dimensional spatial grid scale. To this end, the conditional probability C... ijk Further decomposed into the product of the probabilities of UAVs and manned aircraft appearing within the 3D mesh:

[0123] (twenty one)

[0124] in, This indicates that the drone has entered the 3D grid V. ijk The probability is calculated using the drone's shortest attack path frequency model; A ijk This indicates that manned aircraft are in a 3D mesh V ijk The probability of occurrence within the space is obtained from historical flight track data statistics; the two respectively characterize the possibility of overlap between UAVs and manned aircraft in space from the two dimensions of "potential threat path" and "actual operating trajectory".

[0125] S5. Generate a static collision risk map

[0126] Based on the S4 collision risk assessment method, a simulation calculation architecture for UAV-maned aircraft encounters in the air was constructed. On this basis, a static risk simulation experiment was carried out for low-altitude airspace. The collision risk of each spatial three-dimensional grid was quantitatively calculated, thereby generating a static collision risk map reflecting the surrounding area of ​​the target area, providing a basis for prevention and control zoning and risk management.

[0127] like Figure 9 The simulation results show that the collision risk of UAVs and manned aircraft in the low-altitude airspace of the target area exhibits a non-uniform spatial distribution. Darker colors indicate a higher collision risk. The figure shows that a large number of light-colored 3D grids have a lower collision risk, while a small number of dark-colored 3D grids have a significantly higher risk than the surrounding areas, forming a high-risk concentration zone. This phenomenon is closely related to the highly concentrated approach and departure trajectories of manned aircraft and the high consistency of their flight paths. In terms of altitude distribution, the collision risk is mainly concentrated in the range from the ground to 300m, forming a continuous distribution along the extension line of the manned aircraft's trajectory, thus exhibiting a significant path convergence effect.

[0128] To achieve interpretable evaluation and spatial differentiation of static collision risk results for UAVs and manned aircraft, conditional probability C at the grid scale is used. ijk Based on this, a unified risk assessment standard should be established. Because C ijkThis reflects the relative probability of spatial overlap between UAVs and manned aircraft within the same spatial grid. While its numerical value is spatially comparable, it does not directly correspond to the frequency of specific accidents. Therefore, a tiered evaluation method based on relative risk levels is used to determine static collision risk.

[0129] Specifically, the risks of all grids within the study area are used to construct a static collision risk set. Normalization is then applied to eliminate the impact of different magnitudes, resulting in a standardized risk index R. ijk This risk value reflects the relative probability of spatial conflict between UAVs and manned aircraft within corresponding spatial units, and is a continuous probability indicator. Based on the statistical distribution characteristics of the risk value across the entire spatial domain, the quantile threshold method is used to classify the risk level, dividing the merged airspace into regions with different risk levels. The higher the risk level, the greater the relative probability of spatial overlap between UAVs and manned aircraft within that grid, and the more significant the potential threat to flight safety. In spatial management and control applications, risk assessment is not based on the absolute value of a single grid, but rather focuses on the spatial clustering characteristics of high-risk grids and their spatial relationship with critical operational channels. When multiple adjacent grids continuously exhibit high risk levels, the area is considered to have formed a stable high-risk spatial structure and should be a key focus and priority control target. Through risk assessment standards, the static collision risk in continuous space is transformed into clear and identifiable risk zoning results, providing a quantitative basis for risk map generation and control strategy formulation. Figure 10 As shown, the spatial location of the collision event and its corresponding spatial axisymmetric grid are designated as the new first-class risk zone. It is believed that if an actual collision occurs, the spatial unit in which it is located poses a substantial threat to the safety target and should be given priority in the highest-level risk control scope.

[0130] and Figure 3 and Figure 4 In comparison, the new Category I risk zone spatially aligns closely with areas prone to frequent encounters between UAVs and manned aircraft, effectively covering actual collision events and their axially symmetrical locations, while avoiding the inclusion of large areas of low-risk airspace within high-level control zones. This indicates that the zoning method based on conditional probability decomposition and relative risk classification improves the spatial precision and targeting of risk zoning while maintaining safety conservatism. Furthermore, from a spatial structure perspective, the risk zone boundaries in the comparative scheme are relatively rigid, making it difficult to reflect the gradient changes in risk at different altitudes and lateral positions; whereas the zoning results generated based on the static collision risk field naturally reflect the continuity and clustering of risk with spatial location, providing a more operational spatial basis for subsequent implementation of layered, graded, and differentiated low-altitude integrated airspace management measures.

[0131] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for assessing the risk of drone interference in low-altitude airspace using big data processing, characterized in that, The method includes the following steps: S1. Low-altitude unmanned and manned aircraft integrated airspace modeling: The low-altitude unmanned and manned aircraft integrated airspace is divided into a first type of region and a second type of region to achieve a hierarchical expression of the low-altitude integrated airspace structure; and the low-altitude unmanned and manned aircraft integrated airspace is uniformly mapped to a three-dimensional mesh space. The first type of region is a spatial region distributed along the manned aircraft's flight path and concentrated in the lateral and altitude directions; the second type of region is located outside the first type of region and is extended in the flight path extension direction, lateral range, and altitude level to describe the extended space of the potential impact of UAV activities on manned aircraft operations within the low-altitude integrated airspace. S2. Evaluation of manned aircraft 3D grid distribution based on historical flight track data; At the spatial condition level, spatial probability distribution models are constructed for both manned and unmanned aircraft. The manned aircraft trajectory is obtained by cleaning, reconstructing the time series and interpolating the spatial trajectory of historical flight track data. The take-off and landing flight trajectories are reproduced. By statistically analyzing the crossing frequency and flight exposure time of flight trajectories in each 3D grid, and calculating the probability of manned aircraft appearing in the 3D grid, the spatial distribution characteristics of manned aircraft in the 3D airspace are characterized. S3. Quantitative modeling of UAV intrusion probability based on shortest attack path in 3D mesh: Starting from potential disturbance behavior, with the second type of region and the boundary of the first type of region as the potential starting point set and the airspace of manned aircraft trajectory and its extension as the target set, construct the shortest attack path intrusion frequency model under the constraint of starting point-end point pair, and quantitatively calculate the intrusion probability of UAV in 3D mesh. S4. Static collision risk assessment of UAVs and manned aircraft based on conditional probability decomposition; constructing the conditional probability C of UAVs and manned aircraft appearing simultaneously and forming spatial overlap. ijk The risk of static collision between UAVs and manned aircraft is quantitatively characterized on a three-dimensional grid scale. S5. Generate a static collision risk map; conduct static risk simulation experiments for low-altitude airspace, quantitatively calculate the collision risk within each three-dimensional grid, and form a corresponding static collision risk distribution map to provide a basis for prevention and control zoning and risk management.

2. The method for assessing the risk of drone interference in low-altitude airspace using big data processing according to claim 1, characterized in that, The process of constructing the three-dimensional mesh space is as follows: ； Where G represents the set of all 3D meshes in the 3D hierarchy; V represents the set of all 3D mesh nodes; E represents the set of connected edges in the 3D mesh; and adjacent 3D meshes are connected as follows: ； Where l(e) is the center distance between the three-dimensional meshes; c i c represents the center coordinates of the i-th 3D grid. j Let be the center coordinates of the j-th 3D grid.

3. The method for assessing the risk of drone interference in low-altitude airspace using big data processing according to claim 1, characterized in that, In step S2, the formula for calculating the probability of a human-machine interface appearing within the three-dimensional mesh is: ； Among them, A ijk Represents a three-dimensional mesh V ijk The probability of a space containing human or machine space. The number of times the 3D grid is traversed by all flights. This is the sum of the number of times all 3D meshes are traversed.

4. The method for assessing the risk of drone interference in low-altitude airspace using big data processing according to claim 3, characterized in that, In step S3, the calculation process of the shortest attack path intrusion frequency model is as follows: For each attack source point s, Dijkstra's shortest path algorithm is used to calculate the shortest path from it to all target grid points, defined as: ； Where t is any target grid point; u is any path; Let be a feasible path from source point s to target point t; dist(i,j) is the Euclidean distance between 3D meshes i and j; for attack source point s, select the shortest path among all target points: ； Where T is the set of all target grid points; Accumulate and statistically analyze the valid paths from all attack sources: ； Where v is a single specific 3D mesh node; S is the set of all attack source points; Let be the optimal path starting from s; the indicator function W is: ； for When the three-dimensional mesh V ijk A value of 1 is assigned if the attack path is on a valid attack path from the attack source point s, and 0 otherwise. This ultimately yields the frequency of the shortest attack path occurrence for each 3D grid in the entire 3D spatial domain. ； The intrusion intensity of all 3D meshes is uniformly processed using the Min–Max normalization method, and its expression is: ； Among them, F ijk The shortest attack path for all attack sources through the 3D grid V ijk The number of times, F max With F min These represent the maximum and minimum values ​​of intrusion intensity for all three-dimensional meshes within the study area, respectively. Finally, by probabilistically processing the normalized intrusion intensity, the intrusion probability of the drone in the 3D mesh is obtained: 。 5. The method for assessing the risk of drone interference in low-altitude airspace using big data processing according to claim 4, characterized in that, The conditional probability C described in S4 ijk The calculation formula is: 。

Images