Airspace delineation method and system based on historical flight data

Abstract

This invention discloses a method and system for airspace partitioning based on historical flight data. The method includes: historical flight data acquisition, flight parameter analysis, climate parameter analysis, generation of airspace group sets, and iterative updates. This invention belongs to the field of data processing technology, specifically referring to a method and system for airspace partitioning based on historical flight data. This solution employs a risk-aware kernel density model, comparing the risk-aware kernel density value with a local density threshold, comparing the bi-branch attention fusion similarity with a climate parameter adaptive clustering threshold, and combining spatial distance constraints to address the problem of inaccurate airspace partitioning. A risk deviation calibration factor is used to achieve dynamic iteration of grouping, a Bessel fitting algorithm is used to correct sawtooth boundaries, and a topology scaling operator is used to dynamically adjust the sector range to address the problem of excessively short crossing times. Real-time sector coordination data is combined to inversely optimize weight coefficients, reducing coordination frequency and improving airspace resource utilization.

Description

Technical Field

[0001] This invention belongs to the field of data processing technology, specifically an airspace partitioning method and system based on historical flight data. Background Technology

[0002] The airspace partitioning method based on historical flight data is a method that uses massive amounts of historical flight trajectory, flight traffic, flight speed, takeoff and landing frequency data, combined with algorithms to mine airspace traffic operation patterns, and then divides airspace units such as airspace sectors and airways.

[0003] However, existing airspace delineation methods based on historical flight data suffer from several technical problems. They lack comprehensive analysis of flight and climate parameters, as well as dynamic partitioning modeling of parameter similarity. They cannot build adaptive models based on the real-time correlation characteristics of parameters within the airspace, leading to inaccurate airspace delineation. Furthermore, the airspace delineation algorithms do not consider dynamic airspace characteristics and are based solely on static information, resulting in jagged boundaries and excessively short route crossing times between sectors. This increases the frequency of inter-sector coordination, making flight conflicts more likely and reducing the overall utilization rate of airspace resources. Summary of the Invention

[0004] To address the above issues and overcome the shortcomings of existing technologies, this invention provides an airspace partitioning method and system based on historical flight data. It addresses the technical problems of inaccurate airspace partitioning due to a lack of comprehensive analysis of flight and climate parameters, a lack of dynamic partitioning modeling for parameter similarity, and an inability to construct an adaptive model based on real-time correlation characteristics of parameters within the airspace. The invention employs a risk-aware kernel density model, comparing the risk-aware kernel density value with a local density threshold to generate a first airspace set. It then compares the similarity obtained by bi-branch attention fusion with an adaptive clustering threshold for climate parameters, incorporating spatial distance constraints to generate a second airspace set. Finally, it constructs a dual-objective optimization function for overlapping locations within the sets, resolving the insufficient dynamic partitioning modeling of parameter similarity. This ensures that the final generated airspace grouping set accurately adapts to the real-time correlation characteristics of airspace. This paper addresses the problem of inaccurate airspace delineation. Specifically, it identifies issues where airspace delineation algorithms fail to consider dynamic airspace characteristics, relying solely on static information for modeling. This leads to jagged boundaries and excessively short transit times between sectors, increasing inter-sector coordination frequency, increasing flight conflicts, and reducing overall airspace resource utilization. The paper employs a dual-objective optimization function, uses a risk deviation calibration factor for dynamic grouping iteration, uses a Bessel fitting algorithm to correct jagged boundaries, and dynamically adjusts sector ranges based on real-time transit times using a topology scaling operator to address excessively short transit times. Finally, it combines real-time sector coordination data with reverse optimization of weight coefficients to reduce coordination frequency. The resulting airspace grouping set adapts to dynamic airspace characteristics while avoiding boundary and transit time issues, thereby reducing flight conflicts and improving airspace resource utilization.

[0005] The technical solution adopted by this invention is as follows: The airspace delineation method based on historical flight data provided by this invention includes the following steps:

[0006] Step S1: Historical flight data collection;

[0007] Step S2: Flight parameter analysis;

[0008] Step S3: Climate parameter analysis;

[0009] Step S4: Generate a spatial grouping set;

[0010] Step S5: Iterative update.

[0011] Furthermore, in step S1, the historical flight data acquisition, used to obtain complete data of the target airspace and ensure data quality, includes the following steps:

[0012] Step S11: Data acquisition, collecting historical flight records and historical sensor data;

[0013] Step S12: Data preprocessing. The collected historical flight records and historical sensor data are preprocessed to obtain standard historical flight data.

[0014] Further, in step S2, the flight parameter analysis includes the following steps:

[0015] Step S21: Calculate the risk perception kernel density value, extract flight parameters from standard historical flight data, embed risk indicators from the flight parameters into the kernel function, the risk indicators include accident correlation index and risk avoidance index, and construct a risk perception kernel density model;

[0016] Step S22: Calculate the dynamic kernel bandwidth, which is dynamically adjusted according to the local flight density and risk level in the airspace;

[0017] Step S23: Design a local density threshold instead of a global threshold;

[0018] Step S24: Clustering is performed, calculating the location of each spatial domain. Risk perception kernel density With local density threshold ,like These are marked as core risk points. Clustering is then expanded around these core risk points to include those that satisfy the kernel density difference. And the weighted cosine similarity of flight parameters of Included in the same sub-cluster, the Indicates airspace location The risk perception kernel density is lower than the risk perception kernel density. Isolated points are removed from the locations, and subclusters are merged according to geographical continuity to generate the first spatial domain set. Each sub-cluster corresponds to an element in the set, and the flight behavior and risk characteristics of the location within the set are highly consistent.

[0019] Further, in step S3, the climate parameter analysis includes the following steps:

[0020] Step S31: Design a dual-branch attention module, extract climate parameters from standard historical flight data, and calculate the dual-branch attention fusion similarity;

[0021] Step S32: Design adaptive clustering thresholds for climate parameters ;

[0022] Step S33: Clustering is performed, and the similarity is fused using the dual-branch attention. As a condition, combined with spatial distance The constraints aggregate spatial locations that meet the conditions into subclusters, ultimately generating a second spatial set. The climate-flight coupling characteristics of locations within the set are highly consistent.

[0023] Further, in step S4, generating the spatial domain group set includes the following steps:

[0024] Step S41: For airspace locations that belong to both the first and second airspace sets, construct a bi-objective optimization function to select initial candidate airspace locations that meet the requirements of safety and efficiency.

[0025] For the selected airspace to be decided According to the overall matching degree from high to low, and combined with the affiliation relationship of the spatial domain sub-cluster centers, the spatial domains with similar matching degrees are divided into the same sub-cluster, and the set of all sub-clusters is defined as the initial spatial domain grouping set.

[0026] Step S42: Access real-time airspace operation data, design risk deviation calibration factors, introduce boundary smoothness deviation weights, dynamically iterate the initial airspace grouping results, so that the grouping results initially adapt to the actual airspace operation status, and generate a calibrated pre-optimized airspace grouping set.

[0027] Step S43: Sawtooth boundary smoothing correction. A Bézier fitting algorithm with curvature constraints is introduced to normalize and correct the calibrated spatial boundary, integrate all smoothed spatial domains, and finally output a set of boundary smoothed spatial domains.

[0028] Step S44: Dynamic adaptation of sector traversal time. Input the boundary smoothed airspace set, adjust the sector range according to the real-time traversal time data, and generate the final airspace grouping set.

[0029] Step S45: Back-optimize the objective function weights using real-time coordinated data.

[0030] Furthermore, in step S5, the iterative update includes the following steps: based on the actual flight trajectory of the UAV, airspace operation efficiency indicators and real-time risk data, reversely correct the group weight coefficient, boundary control points and risk thresholds, form a closed loop, and output an optimized group set that continuously adapts to the airspace state to realize airspace division.

[0031] The airspace partitioning system based on historical flight data provided by this invention includes a historical flight data acquisition module, a flight parameter analysis module, a climate parameter analysis module, an airspace grouping set generation module, and an iterative update module.

[0032] The historical flight data acquisition module specifically collects historical flight records and historical sensor data of the target airspace, generates standard historical flight data, and sends the data to the flight parameter analysis module and the climate parameter analysis module.

[0033] The flight parameter analysis module specifically constructs a risk perception kernel density model, calculates the risk perception kernel density value and dynamic kernel bandwidth, designs a local density threshold, compares the risk perception kernel density value with the local density threshold, and generates a first airspace set through clustering, outlier removal, and sub-cluster merging; and sends the data to the airspace grouping set generation module.

[0034] The climate parameter analysis module specifically calculates the bi-branch attention fusion similarity through the bi-branch attention module, designs an adaptive clustering threshold for climate parameters, compares the bi-branch attention fusion similarity with the adaptive clustering threshold for climate parameters, and generates a second spatial domain set by combining spatial distance constraints; and sends the data to the spatial domain grouping set generation module.

[0035] The module for generating the airspace group set specifically involves constructing a dual-objective optimization function for airspace locations that simultaneously belong to the first airspace set and the second airspace set, accessing real-time airspace operation data, performing risk deviation calibration, introducing a curvature-constrained Bezier fitting algorithm, and combining sector traversal time dynamic adaptation and weighted inverse optimization to generate the final airspace group set; the data is then sent to the iterative update module.

[0036] The iterative update module specifically corrects the core parameters of the grouping based on the actual flight trajectory of the UAV, airspace operation efficiency indicators and real-time risk data, and forms a closed loop, outputting an optimized grouping set that continuously adapts to the dynamic state of the airspace, thereby realizing airspace division.

[0037] The beneficial results achieved by the present invention using the above solution are as follows:

[0038] (1) To address the technical problem of inaccurate airspace division due to the lack of comprehensive analysis of flight parameters and climate parameters, as well as the lack of dynamic partitioning modeling of parameter similarity, and the inability to construct an adaptation model based on the real-time correlation characteristics of parameters within the airspace, a risk perception kernel density model is constructed. The risk perception kernel density value is compared with the local density threshold to generate the first airspace set. The similarity of the two-branch attention fusion is compared with the climate parameter adaptive clustering threshold. Combined with the spatial distance constraint, the second airspace set is generated. A dual-objective optimization function is constructed for the overlapping positions of the sets to solve the problem of insufficient dynamic partitioning modeling of parameter similarity. This makes the final generated airspace group set accurately adapt to the real-time correlation characteristics of the airspace and solves the problem of inaccurate airspace division.

[0039] (2) To address the technical problem that the airspace partitioning algorithm does not consider the dynamic characteristics of airspace and only models based on static information, the partitioning results are prone to jagged boundaries and short transit times between sectors, which leads to increased coordination frequency between sectors, flight conflicts, and reduced overall utilization of airspace resources, a dual-objective optimization function is constructed. A risk deviation calibration factor is used to achieve dynamic iteration of grouping. The Bessel fitting algorithm is used to correct jagged boundaries. The sector range is dynamically adjusted by the topology scaling operator based on the real-time transit time to solve the problem of short transit time. The weight coefficient is optimized in reverse by combining the real-time coordination data of the sectors to reduce the coordination frequency. The final generated airspace group set is both adapted to the dynamic characteristics of airspace and avoids boundary and transit time problems, thereby reducing flight conflicts and improving the utilization of airspace resources. Attached Figure Description

[0040] Figure 1 A flowchart illustrating the airspace partitioning method based on historical flight data provided by this invention;

[0041] Figure 2 This is a schematic diagram of the airspace partitioning system based on historical flight data provided by the present invention.

[0042] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0043] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0044] Example 1, see Figure 1 The present invention provides an airspace partitioning method based on historical flight data, which includes the following steps:

[0045] Step S1: Historical flight data collection, specifically, collecting historical flight records and historical sensor data of the target airspace to generate standard historical flight data;

[0046] Step S2: Flight parameter analysis, specifically, constructing a risk perception kernel density model, calculating the risk perception kernel density value and dynamic kernel bandwidth, designing a local density threshold, comparing the risk perception kernel density value with the local density threshold, and generating the first airspace set through clustering, outlier removal and sub-cluster merging.

[0047] Step S3: Climate parameter analysis, specifically, calculates the bi-branch attention fusion similarity through the bi-branch attention module, designs an adaptive clustering threshold for climate parameters, compares the bi-branch attention fusion similarity with the adaptive clustering threshold for climate parameters, and generates a second spatial domain set by combining spatial distance constraints;

[0048] Step S4: Generate a set of airspace groups. Specifically, for airspace locations that belong to both the first and second airspace sets, construct a dual-objective optimization function, access real-time airspace operation data, perform risk deviation calibration, introduce a Bézier fitting algorithm with curvature constraints, and combine sector traversal time dynamic adaptation and weighted inverse optimization to generate the final set of airspace groups.

[0049] Step S5: Iterative update, specifically based on the actual flight trajectory of the UAV, airspace operation efficiency indicators and real-time risk data, reversely correct the core parameters of the grouping and form a closed loop, output an optimized grouping set that continuously adapts to the dynamic state of the airspace, and realize airspace division.

[0050] Example 2, see Figure 1 This embodiment is based on the above embodiment. In step S1, the historical flight data acquisition is used to obtain complete data of the target airspace and ensure data quality, including the following steps:

[0051] Step S11: Data acquisition. Retrieve historical flight records of the target airspace from the airspace database and flight monitoring system. The historical flight records include flight routes, flight operation data, accident records, etc. of flight equipment within multiple historical time periods. Collect historical sensor data through meteorological sensors, remote sensing satellites, and environmental monitoring equipment. The historical sensor data includes multi-dimensional environmental data such as temperature, humidity, light intensity, light reflection, and sound.

[0052] Step S12: Data preprocessing. The collected historical flight records and historical sensor data are cleaned, missing trajectory values ​​in the flight records and abnormal noise in the sensor data are removed, the data format and spatiotemporal coordinate system are unified, and missing data is filled in to obtain standard historical flight data.

[0053] Example 3, see Figure 1 This embodiment is based on the above embodiment. In step S2, the flight parameter analysis includes the following steps:

[0054] Step S21: Calculate the risk perception kernel density value, extract flight parameters from standard historical flight data, embed risk indicators from the flight parameters into the kernel function, the risk indicators include accident correlation index and risk avoidance index, construct a risk perception kernel density model, which is used to assign a kernel density value with risk attributes to each airspace location, so that the risk perception kernel density not only reflects the flight trajectory clustering degree of the airspace location, but also highlights the characteristics of risk clustering. The formula used is as follows:

[0055] ;

[0056] ;

[0057] In the formula, Indicates airspace location The risk perception kernel density value, where n represents the total number of airspace locations involved in the kernel density calculation, h represents the kernel bandwidth, d represents the number of dimensions of flight parameters (including the total number of feature dimensions such as frequency of passage, risk avoidance index, accident correlation index, etc.), and j represents the traversal index, representing the area within the target airspace excluding... Other airspace locations K() represents the Gaussian kernel function. Indicates airspace location and European spatial distance between them Indicates airspace location The risk weights have a range of [0.4, 1.0]. Indicates airspace location Accident correlation index, Indicates airspace location Avoid the risks associated with the index;

[0058] Step S22: Calculate the dynamic kernel bandwidth, which is dynamically adjusted according to the local flight density and risk level in the airspace to improve the accuracy of clustering. The formula used is as follows:

[0059] ;

[0060] In the formula, Indicates airspace location Dynamic kernel bandwidth, This represents the base value of the kernel bandwidth. Indicates airspace location The density of flight paths This represents the maximum density of flight trajectories clustered across all airspace locations within the target airspace.

[0061] Step S23: Design a local density threshold instead of a global threshold to ensure the purity of core risk points and avoid missing potential risk points. The formula used is as follows:

[0062] ;

[0063] In the formula, Indicates airspace location The local density threshold, This represents the base value for the density threshold, which is 0.05. Indicates airspace location Standard deviation of flight parameters within the neighborhood This represents the global standard deviation of flight parameters for all locations within the target airspace;

[0064] Step S24: Clustering is performed, calculating the location of each spatial domain. Risk perception kernel density value With local density threshold ,like These are marked as core risk points. Clustering is then expanded around these core risk points to include those that satisfy the kernel density difference. And the weighted cosine similarity of flight parameters of Incorporated into the same sub-cluster ,in, Indicates airspace location The normalized value of the k1th flight parameter, Indicates airspace location The normalized value of the k1th flight parameter, This represents the weighting coefficient of the k1th flight parameter, where the risk perception kernel density value is lower than... Isolated points are removed from the locations, and subclusters are merged according to geographical continuity to generate the first spatial domain set. Each sub-cluster corresponds to an element in the set, and the flight behavior and risk characteristics of the location within the set are highly consistent.

[0065] Example 4, see Figure 1 This embodiment is based on the above embodiment. In step S3, the climate parameter analysis includes the following steps:

[0066] Step S31: Design a dual-branch attention module, extract climate parameters from standard historical flight data, and calculate the dual-branch attention fusion similarity using the following formula:

[0067] ;

[0068] ;

[0069] ;

[0070] In the formula, Indicates airspace location and The similarity of climate parameters between the two parameters is determined by a two-branch attention fusion method; a larger value indicates that the climate characteristics of the two parameters are more similar. Indicates the similarity of feature attention branches. and Representing airspace location and Climate parameter feature vectors The similarity of spatiotemporal attention branches is represented by k, which represents the dimensional index of the climate parameters, including climate type, climate intensity, climate-flight risk coupling coefficient, and risk duration. The attention weights representing the k-th dimension of the climate parameter are automatically learned and generated by a multilayer perceptron. Indicates airspace location The value of the k-th dimension climate parameter, Indicates airspace location The value of the k-th dimension climate parameter, This represents the spatial attenuation coefficient, with a value of 0.2. Indicates airspace location and The straight-line distance between them in geospatial space Indicates airspace location The timestamps corresponding to the climate parameters Indicates airspace location The timestamps corresponding to the climate parameters This represents the meteorological cycle, with a value of 24 hours.

[0071] Step S32: Design the adaptive clustering threshold for climate parameters, using the following formula:

[0072] ;

[0073] In the formula, Indicates airspace location The corresponding climate parameter adaptive clustering threshold, where 0.75 represents the base value of the clustering threshold and 0.15 represents the dynamic adjustment coefficient of the threshold. Indicates airspace location The standard deviation of climate parameters within a neighborhood; a larger value indicates greater climate instability. This represents the maximum standard deviation of climate parameters in the neighborhood of all locations within the target airspace;

[0074] Step S33: Clustering is performed, and the similarity is fused using the dual-branch attention. As a condition, combined with spatial distance The constraints aggregate spatial locations that meet the conditions into subclusters, ultimately generating a second spatial set. The climate-flight coupling characteristics of locations within the set are highly consistent.

[0075] Example 5, see Figure 1 This embodiment is based on the above embodiment. In step S4, generating the spatial domain group set includes the following steps:

[0076] Step S41: For airspace locations that simultaneously belong to the first and second airspace sets, construct a bi-objective optimization function. Using constraints such as distance, risk coupling, and traversal time, select initial candidate airspace locations that meet both safety and efficiency requirements. The formula used is as follows:

[0077] ;

[0078] ;

[0079] In the formula, Represents maximizing the spatial domain The overall matching degree objective function, This represents the location of the spatial domain to be decided, which simultaneously belongs to both the first spatial domain set and the second spatial domain set. This represents the weighting coefficient used to balance the similarity between flight parameters and climate parameters; it is 0.7 for safety-first scenarios and 0.4 for efficiency-first scenarios. This represents the similarity of average flight parameters between the airspace location to be decided and the airspace locations belonging to the first airspace set. This represents the similarity of average climate parameters between the airspace location to be decided and the airspace location belonging to the second airspace set. This represents the location of the airspace to be decided and the center location of the set of airspace subclusters to which it belongs. European spatial distance between them This represents the maximum permissible spatial distance threshold within the set, constraining the spatial location to be decided. The climate-flight risk coupling coefficient represents the location of the airspace to be decided. This represents the maximum allowable climate-flight risk coupling coefficient threshold for the set of airspace locations to be decided, preventing excessive clustering of high-risk locations. Indicates airspace location Real-time traffic coefficient, express Theoretical travel time for the area's air routes This indicates the minimum allowed traversal time for a sector. express Traffic-adaptive travel duration coefficient;

[0080] For the selected airspace to be decided According to the overall matching degree from high to low, and combined with the affiliation relationship of the spatial domain sub-cluster centers, the spatial domains with similar matching degrees are divided into the same sub-cluster, and the set of all sub-clusters is defined as the initial spatial domain grouping set.

[0081] Step S42: Integrate real-time airspace operation data, design a risk deviation calibration factor, introduce boundary smoothness deviation weights, dynamically iterate the initial airspace grouping results, correct grouping errors caused by real-time risk changes and boundary irregularities, avoid jagged boundaries after calibration, and allow the grouping results to initially adapt to the actual airspace operation state, generating a pre-optimized airspace grouping set after calibration. The formula used is as follows:

[0082] ;

[0083] In the formula, This represents the pre-optimized spatial grouping set after calibration. This represents the initial spatial grouping set at time t. Indicates the data update time interval. This represents the risk deviation calibration sensitivity coefficient, with a value of 0.12. This represents the deviation between the real-time climate-flight risk coupling coefficient and the historical coupling coefficient. This represents the boundary smoothness deviation weight, with a value of 0.3. This represents the deviation between the real-time boundary smoothness and the standard smoothness.

[0084] Step S43: Sawtooth boundary smoothing correction. A Bezier fitting algorithm with curvature constraints is introduced, combined with key control points such as route inflection points and navigation beacons, to regularize and correct the calibrated airspace boundaries. All smoothed airspaces are integrated, and the final output is a set of smoothed airspaces, resolving the sawtooth boundary problem and improving the rationality of airspace grouping. The formula used is as follows:

[0085] ;

[0086] ;

[0087] In the formula, Let n represent the boundary smoothness of the spatial grouping set G, and n1 represent the number of segments on the boundary of set G. This represents the curvature value of the i-th segment of the boundary. This represents the average curvature of all segments of the boundary. This represents the maximum permissible curvature threshold of the airspace boundary. This represents the smooth boundary curve generated after fitting, where m represents the order of the Bézier curve, determined by the number of boundary control points; 4 control points correspond to a 3rd-order curve. represents the binomial coefficient, used to allocate the contribution weight of each control point to the curve, and t represents the parameter variable of the curve, with a value range of [0, 1]. Key control points indicating boundaries are typically selected from route inflection points, navigation beacons, airport runway endpoints, etc.

[0088] All spatial subclusters that have undergone boundary smoothing correction are integrated, and spatial overlaps or gaps generated during the correction process are removed to obtain the final set of boundary-smoothed spatial domains. ;

[0089] Step S44: Dynamic adaptation of sector crossing time. Addressing the issue of route crossing sector times not meeting thresholds, a boundary-smoothed airspace set is input. The sector range is adjusted based on real-time crossing time data to generate a final airspace grouping set that balances boundary smoothness and sector crossing time compliance. The formula used is as follows:

[0090] ;

[0091] ;

[0092] In the formula, This indicates that after the time travel duration has been adapted, The spatial grouping set at time, This represents a sector topology scaling operator that triggers different topology adjustment actions based on the real-time duration of sector traversal. Indicates the duration of real-time sector traversal. Indicates the sector scaling factor. Indicates the maximum allowed traversal time for a sector. Indicates the minimum allowed traversal time for a sector;

[0093] Step S45: Use the real-time coordinated data to inversely optimize the objective function weights in S41, reducing the inter-sector coordination frequency. The formula used is as follows:

[0094] ;

[0095] In the formula, This represents the optimized weighting coefficients used to balance the similarity between flight parameters and climate parameters. Indicates the coordination frequency feedback weight, This indicates the real-time coordination frequency between sectors under the current airspace grouping scheme. This indicates the average coordination frequency of airspace grouping schemes during the same historical period.

[0096] By performing the above operations, a risk perception kernel density model is constructed. The risk perception kernel density value is compared with the local density threshold to generate a first airspace set. The similarity of the two-branch attention fusion is compared with the climate parameter adaptive clustering threshold. Combined with the spatial distance constraint, a second airspace set is generated. A dual-objective optimization function is constructed for the overlapping positions of the sets to solve the problem of insufficient dynamic partitioning modeling of parameter similarity. This makes the final generated airspace grouping set accurately adapt to the real-time correlation features of the airspace, solving the problem of inaccurate airspace division. It also solves the technical problem of inaccurate airspace division caused by the lack of comprehensive analysis of flight parameters and climate parameters, as well as the lack of dynamic partitioning modeling of parameter similarity, which prevents the construction of an adaptation model based on the real-time correlation features of parameters within the airspace.

[0097] Example 6, see Figure 1 This embodiment is based on the above embodiment. In step S5, the iterative update includes the following steps: based on the actual flight trajectory of the UAV, airspace operation efficiency indicators and real-time risk data, the group weight coefficient, boundary control points and risk thresholds are reversed to form a closed loop, and an optimized group set that continuously adapts to the airspace state is output to realize airspace division.

[0098] By performing the above operations, a dual-objective optimization function is constructed, a risk deviation calibration factor is used to achieve dynamic iteration of grouping, a Bessel fitting algorithm is used to correct sawtooth boundaries, and the sector range is dynamically adjusted based on real-time crossing time using a topology scaling operator to solve the problem of excessively short crossing time. Combined with real-time sector coordination data, the weight coefficients are optimized in reverse to reduce the coordination frequency. The final generated airspace group set not only adapts to the dynamic characteristics of airspace but also avoids boundary and crossing time issues, thereby reducing flight conflicts and improving airspace resource utilization. This solves the technical problem that airspace partitioning algorithms do not consider the dynamic characteristics of airspace and are based only on static information modeling, which easily leads to sawtooth boundaries and excessively short crossing times between sectors, resulting in increased coordination frequency between sectors, easy flight conflicts, and weakened overall airspace resource utilization.

[0099] Example 7, see Figure 2This embodiment is based on the above embodiment. The airspace division system based on historical flight data provided by the present invention includes a historical flight data acquisition module, a flight parameter analysis module, a climate parameter analysis module, an airspace grouping set generation module, and an iterative update module.

[0100] The historical flight data acquisition module specifically collects historical flight records and historical sensor data of the target airspace, generates standard historical flight data, and sends the data to the flight parameter analysis module and the climate parameter analysis module.

[0101] The flight parameter analysis module specifically constructs a risk perception kernel density model, calculates the risk perception kernel density value and dynamic kernel bandwidth, designs a local density threshold, compares the risk perception kernel density value with the local density threshold, and generates a first airspace set through clustering, outlier removal, and sub-cluster merging; and sends the data to the airspace grouping set generation module.

[0102] The climate parameter analysis module specifically calculates the bi-branch attention fusion similarity through the bi-branch attention module, designs an adaptive clustering threshold for climate parameters, compares the bi-branch attention fusion similarity with the adaptive clustering threshold for climate parameters, and generates a second spatial domain set by combining spatial distance constraints; and sends the data to the spatial domain grouping set generation module.

[0103] The module for generating the airspace group set specifically involves constructing a dual-objective optimization function for airspace locations that simultaneously belong to the first airspace set and the second airspace set, accessing real-time airspace operation data, performing risk deviation calibration, introducing a curvature-constrained Bezier fitting algorithm, and combining sector traversal time dynamic adaptation and weighted inverse optimization to generate the final airspace group set; the data is then sent to the iterative update module.

[0104] The iterative update module specifically corrects the core parameters of the grouping based on the actual flight trajectory of the UAV, airspace operation efficiency indicators and real-time risk data, and forms a closed loop, outputting an optimized grouping set that continuously adapts to the dynamic state of the airspace, thereby realizing airspace division.

[0105] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0106] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

[0107] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. An airspace delineation method based on historical flight data, characterized in that: The method includes the following steps: Step S1: Historical flight data collection, specifically, collecting historical flight records and historical sensor data of the target airspace, performing data preprocessing, removing missing trajectory values ​​and abnormal noise, unifying the data format and spatiotemporal coordinate system, completing missing data, and generating standard historical flight data. Step S2: Flight parameter analysis, specifically, constructing a risk perception kernel density model, calculating the risk perception kernel density value and dynamic kernel bandwidth, designing a local density threshold, comparing the risk perception kernel density value with the local density threshold, and generating the first airspace set through clustering, outlier removal and sub-cluster merging. Step S3: Climate parameter analysis, specifically, calculates the bi-branch attention fusion similarity through the bi-branch attention module, designs an adaptive clustering threshold for climate parameters, compares the bi-branch attention fusion similarity with the adaptive clustering threshold for climate parameters, and generates a second spatial domain set by combining spatial distance constraints; Step S4: Generate a set of airspace groups. Specifically, for airspace locations that belong to both the first and second airspace sets, construct a dual-objective optimization function, access real-time airspace operation data, perform risk deviation calibration, introduce a Bézier fitting algorithm with curvature constraints, and combine sector traversal time dynamic adaptation and weighted inverse optimization to generate the final set of airspace groups. Step S5: Iterative update, specifically based on the actual flight trajectory of the UAV, airspace operation efficiency indicators and real-time risk data, reversely correct the core parameters of the grouping and form a closed loop, output an optimized grouping set that continuously adapts to the dynamic state of the airspace, and realize airspace division.

2. The airspace partitioning method based on historical flight data according to claim 1, characterized in that: In step S2, the flight parameter analysis includes the following steps: Step S21: Calculate the risk perception kernel density value, extract flight parameters from standard historical flight data, embed risk indicators from the flight parameters into the kernel function, the risk indicators include accident correlation index and risk avoidance index, and construct a risk perception kernel density model; Step S22: Calculate the dynamic kernel bandwidth, which is dynamically adjusted according to the local flight density and risk level in the airspace; Step S23: Design a local density threshold instead of a global threshold; Step S24: Clustering is performed, calculating the location of each spatial domain. Risk perception kernel density With local density threshold ,like These are marked as core risk points. Clustering is then expanded around these core risk points to include those that satisfy the kernel density difference. And the weighted cosine similarity of flight parameters of Included in the same sub-cluster, the Indicates airspace location The risk perception kernel density is lower than the risk perception kernel density. Isolated points are removed from the locations, and subclusters are merged according to geographical continuity to generate the first spatial domain set. Each sub-cluster corresponds to an element in the set, and the flight behavior and risk characteristics of the location within the set are highly consistent.

3. The airspace partitioning method based on historical flight data according to claim 1, characterized in that: In step S3, the climate parameter analysis includes the following steps: Step S31: Design a dual-branch attention module, extract climate parameters from standard historical flight data, and calculate the dual-branch attention fusion similarity; Step S32: Design adaptive clustering thresholds for climate parameters ; Step S33: Clustering is performed, and the similarity is fused using the dual-branch attention. As a condition, combined with spatial distance The constraints aggregate spatial locations that meet the conditions into subclusters, ultimately generating a second spatial set. The climate-flight coupling characteristics of locations within the set are highly consistent.

4. The airspace partitioning method based on historical flight data according to claim 1, characterized in that: In step S4, generating the spatial domain group set includes the following steps: Step S41: For airspace locations that belong to both the first and second airspace sets, construct a bi-objective optimization function to select initial candidate airspace locations that meet the requirements of safety and efficiency. For the selected airspace to be decided According to the overall matching degree from high to low, and combined with the affiliation relationship of the spatial domain sub-cluster centers, the spatial domains with similar matching degrees are divided into the same sub-cluster, and the set of all sub-clusters is defined as the initial spatial domain grouping set. Step S42: Access real-time airspace operation data, design risk deviation calibration factors, introduce boundary smoothness deviation weights, dynamically iterate the initial airspace grouping results, so that the grouping results initially adapt to the actual airspace operation status, and generate a calibrated pre-optimized airspace grouping set. Step S43: Sawtooth boundary smoothing correction. A Bézier fitting algorithm with curvature constraints is introduced to normalize and correct the calibrated spatial boundary, integrate all smoothed spatial domains, and finally output a set of boundary smoothed spatial domains. Step S44: Dynamic adaptation of sector traversal time. Input the boundary smoothed airspace set, adjust the sector range according to the real-time traversal time data, and generate the final airspace grouping set. Step S45: Back-optimize the objective function weights using real-time coordinated data.

5. The airspace delineation method based on historical flight data according to claim 1, characterized in that: In step S1, the historical flight data collection includes the following steps: Step S11: Data acquisition, collecting historical flight records and historical sensor data; Step S12: Data preprocessing. The collected historical flight records and historical sensor data are preprocessed to obtain standard historical flight data.

6. The airspace partitioning method based on historical flight data according to claim 1, characterized in that: In step S5, the iterative update includes the following steps: based on the actual flight trajectory of the UAV, airspace operation efficiency indicators and real-time risk data, reversely correct the group weight coefficient, boundary control points and risk thresholds, form a closed loop, and output an optimized group set that continuously adapts to the airspace state to realize airspace division.

7. An airspace partitioning system based on historical flight data, used to implement the airspace partitioning method based on historical flight data as described in any one of claims 1-6, characterized in that: It includes a historical flight data acquisition module, a flight parameter analysis module, a climate parameter analysis module, a space grouping set generation module, and an iterative update module.

8. The airspace partitioning system based on historical flight data according to claim 7, characterized in that: The historical flight data acquisition module specifically collects historical flight records and historical sensor data of the target airspace, generates standard historical flight data, and sends the data to the flight parameter analysis module and the climate parameter analysis module. The flight parameter analysis module specifically constructs a risk perception kernel density model, calculates the risk perception kernel density value and dynamic kernel bandwidth, designs a local density threshold, compares the risk perception kernel density value with the local density threshold, and generates a first airspace set through clustering, outlier removal, and sub-cluster merging; and sends the data to the airspace grouping set generation module. The climate parameter analysis module specifically calculates the bi-branch attention fusion similarity through the bi-branch attention module, designs an adaptive clustering threshold for climate parameters, compares the bi-branch attention fusion similarity with the adaptive clustering threshold for climate parameters, and generates a second spatial domain set by combining spatial distance constraints; and sends the data to the spatial domain grouping set generation module. The module for generating the airspace group set specifically involves constructing a dual-objective optimization function for airspace locations that simultaneously belong to the first airspace set and the second airspace set, accessing real-time airspace operation data, performing risk deviation calibration, introducing a curvature-constrained Bezier fitting algorithm, and combining sector traversal time dynamic adaptation and weighted inverse optimization to generate the final airspace group set; the data is then sent to the iterative update module. The iterative update module specifically corrects the core parameters of the grouping based on the actual flight trajectory of the UAV, airspace operation efficiency indicators and real-time risk data, and forms a closed loop, outputting an optimized grouping set that continuously adapts to the dynamic state of the airspace, thereby realizing airspace division.

Images