Intelligent flight route planning system based on low-altitude flight service management platform

Abstract

The invention relates to the technical field of low-altitude flight route planning, in particular to a flight route intelligent planning system based on a low-altitude flight service management platform, which comprises an intention analysis module, an environment modeling module, a network generation module, an intelligent route planning module and a result processing module which are connected in sequence. The environment modeling module constructs a three-dimensional environment model containing a composite restriction layer through asynchronous fusion of multi-source dynamic information, and realizes dynamic accurate perception of airspace situations. The network generation module constructs an air route cellular network in the model and configures dynamic passing cost for each cell. The intelligent route planning module generates a preliminary planning route through a bidirectional optimization growth algorithm based on the network. According to the method, the adaptability of route planning to the dynamic change of the environment is improved, and a personalized optimized route meeting the requirements of a specific aircraft and a task can be generated.

Description

Flight route intelligent planning system based on low-altitude flight service management platform Technical Field

[0001] This invention relates to the field of low-altitude flight route planning technology, and in particular to an intelligent flight route planning system based on a low-altitude flight service management platform. Background Technology

[0002] In existing general aviation flight route planning technologies, the system's environmental modeling typically relies on static models constructed from basic geographic information data. These models mainly reflect unchanging elements such as terrain and fixed obstacles, while the actual airspace environment faced by flight activities is dynamically changing. Existing technical solutions are insufficient in integrating real-time meteorological information, ad-hoc airspace restriction notices, and real-time traffic conditions of other aircraft. Dynamic information is often obtained through independent channels and fails to be effectively integrated and uniformly expressed with the static environmental model, resulting in incomplete and outdated environmental conditions on which the planning is based.

[0003] At the level of route planning methodologies, existing technologies mostly employ path search algorithms based on fixed waypoint networks or pre-defined route structures. These methods have relatively fixed search network structures, making it difficult to flexibly adapt to changes in the origin and destination points of specific flight missions. Simultaneously, the cost evaluation criteria used in the planning process are typically singular or static, failing to comprehensively and dynamically incorporate the performance constraints of specific aircraft, the differentiated time or economic requirements of different missions, and the real-time changes in transit costs brought about by the aforementioned dynamic environmental information. This results in generated routes often lacking a refined balance of multi-dimensional and personalized constraints, limiting their practicality and optimization capabilities in complex low-altitude airspace.

[0004] In existing low-altitude flight route planning practices, airspace resource management often lacks a refined, layered design. Different altitude layers may correspond to different airspace usage rules, access rights, or operational standards, such as low-altitude logistics flight layers, general aviation cruise layers, and temporary isolated airspace layers. Existing environmental modeling methods typically fail to effectively reflect this airspace layering structure, making it difficult to integrate layered airspace usage strategies into the route generation process, thus limiting the efficiency of refined airspace resource utilization and the compliance of route planning. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and propose a flight route intelligent planning system based on a low-altitude flight service management platform.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a flight route intelligent planning system based on a low-altitude flight service management platform, comprising: an intention parsing module, wherein the low-altitude flight service management platform receives the original flight intention and parses out the flight airspace and time window; an environment modeling module, which establishes a three-dimensional spatial grid model corresponding to the flight airspace, wherein the three-dimensional spatial grid model includes a static geographic information layer, and asynchronously imports meteorological information, temporary airspace restriction information, and dynamic traffic situation information from multiple external data sources to form a composite restriction layer superimposed on the static geographic information layer; and a network generation module, which uses preset route cell generation rules. Within the passable space identified by the composite restriction layer, an iterative network of route cells is constructed. Responding to aircraft performance parameters and mission attribute parameters, a dynamic passage cost is configured for each route cell in the network. The intelligent route planning module invokes a route growth algorithm, using the origin and destination points of the flight intention as the growth origin and target points. It performs bidirectional optimization growth within the route cell network based on the dynamic passage cost until a connected path is generated. The connected path is then smoothed and conflict detected, and a preliminary planned route is output. The result processing module returns the preliminary planned route to the low-altitude flight service management platform for verification and confirmation.

[0007] As a further aspect of the present invention, the establishment of a three-dimensional spatial grid model corresponding to the flight airspace includes: defining a three-dimensional envelope space based on the latitude and longitude boundaries of the flight airspace and a preset vertical height range; dividing the three-dimensional envelope space into equal intervals along the longitude, latitude, and height directions to generate a basic grid array composed of a large number of cubic units; extracting surface elevation data within the three-dimensional envelope space from a terrain database, mapping the surface elevation data to the basic grid array, marking all grid units with elevation values ​​lower than the surface elevation as unusable, and forming the static geographic information layer.

[0008] As a further aspect of the present invention, the formation of the composite restriction layer superimposed on the static geographic information layer includes: establishing a real-time connection with a meteorological data server to obtain wind field, visibility, and precipitation intensity data of the flight airspace within the time window, and marking grid cells with wind field data exceeding a threshold or visibility below a threshold as meteorological restricted areas; accessing the notification system issued by the airspace management agency to parse the temporary airspace restriction information in effect within the time window and marking the corresponding grid cells as controlled restricted areas; receiving real-time flight track data from the low-altitude flight service management platform to predict the future positions of other aircraft and marking the predicted occupied grid cells as dynamic avoidance areas; and logically superimposing the meteorological restricted areas, the controlled restricted areas, and the dynamic avoidance areas to form the composite restriction layer.

[0009] As a further aspect of the present invention, the step of iteratively constructing a route cell network within the passable space identified by the composite restriction layer using a preset route cell generation rule includes: scanning the grid cells in the three-dimensional spatial grid model that are not marked by the composite restriction layer and treating them as potential passable cells; examining the twenty-six adjacent spatial neighboring cells of each potential passable cell as the center, and if all neighboring cells are unmarked, upgrading the center cell to a candidate route node; randomly sampling the candidate route node set according to a preset node density threshold and removing some candidate route nodes to prevent over-density; and calculating connection weights between the remaining candidate route nodes based on three-dimensional spatial distance and inter-node visibility to establish an undirected weighted graph structure as the route cell network.

[0010] As a further aspect of the present invention, configuring dynamic passage costs for each route cell in the route cell network in response to aircraft performance parameters and mission attribute parameters includes: obtaining the maximum rate of climb, minimum turning radius, and cruise speed of the aircraft as aircraft performance parameters; obtaining mission attribute parameters of the flight mission, including mission priority, economy coefficient, and safety level; calculating the three-dimensional length, turning angle change, and average altitude of each connecting edge in the route cell network; calculating the basic physical cost based on the three-dimensional length, the turning angle change, the average altitude, combined with the maximum rate of climb and the minimum turning radius; multiplying the basic physical cost by the economy coefficient, and then adding an additional safety cost according to the safety level to finally obtain the dynamic passage cost.

[0011] As a further aspect of the present invention, the invocation of the route growth algorithm, using the start and end points of the flight intention as the growth origin and target point, and performing bidirectional optimization growth in the route cell network based on the dynamic passage cost until a connected path is generated, includes: mapping the start and end points to the two nearest candidate route nodes in the route cell network, respectively serving as the root node of growth tree A and the root node of growth tree B; initializing two empty priority queues, respectively used to store the edge nodes to be expanded in growth tree A and growth tree B; in each iteration, selecting the node with the minimum cumulative dynamic passage cost from the priority queues for expansion, and exploring all its adjacent candidate route nodes; when the newly expanded node in growth tree A is already a node in growth tree B, or the newly expanded node in growth tree B is already a node in growth tree A, it is determined to be a bidirectional growth intersection; backtracking the path from the root node to the intersection node, and splicing the two paths to form the initial connected path.

[0012] As a further aspect of the present invention, the smoothing and conflict detection of the connected path to output a preliminary planned route includes: fitting a spline curve to the connected path to eliminate sharp angles in the path and generate a smooth three-dimensional spatial curve; performing discrete sampling along the smooth three-dimensional spatial curve at fixed time intervals to generate a series of time-space four-dimensional waypoints; checking whether each of the four-dimensional waypoints intrudes into the dynamic avoidance zone in the composite constraint layer; if an intrusion exists, locally adjusting the smooth three-dimensional spatial curve near the intrusion point and resampling and detecting until there is no conflict; and using the final conflict-free trajectory defined by the series of four-dimensional waypoints as the preliminary planned route.

[0013] As a further aspect of the present invention, the step of returning the preliminary planned route to the low-altitude flight service management platform for verification and confirmation includes: encapsulating the preliminary planned route into a formatted message conforming to aviation data exchange standards, and attaching a summary of key constraints generated during the process; sending the formatted message to the low-altitude flight service management platform to request airspace structure compliance verification and traffic feasibility verification; receiving verification feedback information from the low-altitude flight service management platform, the verification feedback information including approval, pending approval with modification suggestions, or rejection instructions; if it is a pending approval instruction with modification suggestions, automatically adjusting the preliminary planned route according to the modification suggestions, and resending the formatted message for confirmation.

[0014] As a further aspect of the present invention, the step of performing spline curve fitting on the connected path to eliminate sharp angles in the path and generate a smooth three-dimensional spatial curve includes: extracting the three-dimensional spatial coordinates of all sequentially arranged route nodes in the connected path; fitting the three-dimensional spatial coordinates using a non-uniform rational B-spline curve algorithm to generate an initial fitting curve; calculating the curvature value at each fitting point on the initial fitting curve, identifying fitting points whose curvature values ​​exceed the maximum allowable curvature of the aircraft as high curvature points; selecting multiple adjacent fitting points on the initial fitting curve as adjustment segments, with each high curvature point as the center; performing weighted average processing on the fitting points within each adjustment segment to generate a new control point sequence; using the new control point sequence to re-perform non-uniform rational B-spline curve fitting to generate an intermediate smooth curve; repeating the process of curvature value calculation, high curvature point identification, and adjustment segment processing until the curvature values ​​of the generated curve at all fitting points are less than or equal to the maximum allowable curvature of the aircraft, and using the final obtained curve as the smooth three-dimensional spatial curve.

[0015] As a further embodiment of the present invention, the system also includes a planning process traceability module, which records the input data, output data and key decision parameters of each logical step from receiving the original flight intention to outputting the preliminary planned route, forming a reproducible planning log, which is bound and stored with the final confirmed route planning result.

[0016] Compared with existing technologies, the advantages and positive effects of this invention are as follows: By asynchronously importing meteorological, temporary airspace restrictions, and dynamic traffic situation information from multiple external data sources, and superimposing these dynamic data layers on a static geographic information layer, a composite restriction layer is formed. This technology achieves a unified digital representation of multi-dimensional, time-varying constraints of the airspace environment. It enables the environmental model upon which route planning relies to synchronously reflect static geographic constraints and various dynamic restrictions, directly improving the completeness and timeliness of the system's perception of the real flight environment. The planning algorithm thus obtains a decision-making basis that more closely approximates actual operating conditions.

[0017] This method iteratively constructs a route cellular network within a space defined by a composite constraint layer using pre-defined rules, and assigns a dynamic passage cost to each cell that combines specific aircraft performance and mission attributes. Based on this, a route growth algorithm with the origin and destination points as bidirectional growth origins is employed to optimize the network. This technique discretizes a continuous space into a dynamically cost-definable cellular network structure, achieving flexibility in planning units and personalization of cost assessment. The bidirectional optimization growth mechanism improves the search convergence speed in complex networks. This method can comprehensively balance multiple objectives such as safety, efficiency, and economy in a single planning process based on real-time, variable cost assessments to generate optimized paths adapted to specific flight scenarios. Attached Figure Description

[0018] Figure 1 is a timing diagram of the intelligent flight route planning system based on the low-altitude flight service management platform described in this invention; Figure 2 is a flowchart of establishing a three-dimensional spatial grid model of the flight airspace; Figure 3 is a flowchart of configuring dynamic passage costs for route cells; Figure 4 is a radar chart of low-altitude flight route verification scores; Figure 5 is a bar chart comparing data complexity and parsing accuracy in the low-altitude flight intention parsing stage. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0021] Referring to Figure 1, the low-altitude flight service management platform receives raw flight intentions submitted by users, including takeoff and landing locations and desired times. The intention parsing module is responsible for parsing the raw flight intentions and extracting the target flight airspace and flight time window. The environment modeling module then starts, constructing a refined three-dimensional spatial mesh model for the parsed flight airspace and time window. The bottom layer of this model is a static geographic information layer reflecting the terrain, on which a composite constraint layer is dynamically superimposed by asynchronously calling meteorological, airspace notices, and real-time flight track data to jointly represent the spatial accessibility status at the planning time. Within the accessible space identified by the composite constraint layer, the network generation module generates a route cell network composed of nodes and edges according to preset rules, and assigns a dynamic access cost to each connecting edge (i.e., route cell) in the network, combining the aircraft performance parameters and mission attribute parameters of the specific flight mission. The intelligent route planning module uses the origin and destination points of the intended flight as growth targets. It runs a route growth algorithm on the aforementioned weighted network to perform bidirectional optimization search, generating an initial connected path. This path is then smoothed and subjected to four-dimensional spatiotemporal conflict detection, ultimately outputting a preliminary planned route. The result processing module encapsulates this preliminary planned route in a standard format and returns it to the low-altitude flight service management platform for final verification and confirmation, thus completing a full intelligent route planning process.

[0022] In one embodiment of the present invention, referring to Figure 2, an example scenario involves the analysis result of a flight intention. The intention analysis module outputs a flight airspace of 118.5°E to 119.5°E, 31.0°N to 32.0°N, with a vertical altitude range from the ground to 1500 meters. The time window is from 14:00 to 16:00 on October 10, 2025. The environment modeling module delineates a corresponding three-dimensional envelope space based on the latitude and longitude boundaries of the flight airspace and the preset vertical altitude range. The three-dimensional envelope space is a cuboid region that completely encloses the aforementioned geographical and altitude ranges. In a specific implementation, the three-dimensional envelope space is divided at equal intervals along the longitude, latitude, and altitude directions to generate a basic grid array composed of a large number of cubic units. For example, the intervals in the longitude direction are set to 0.01 degrees, the intervals in the latitude direction are set to 0.01 degrees, and the intervals in the altitude direction are set to 50 meters. The basic grid array contains more than hundreds of thousands of regularly arranged cubic units. In practice, surface elevation data within the three-dimensional envelope space is extracted from the terrain database and mapped to a basic grid array. The mapping process assigns an elevation value corresponding to the center point coordinates of each basic grid array cell. All grid cells with elevation values ​​lower than the surface elevation are marked as unusable, forming a static geographic information layer. For example, if the elevation of the center point of a basic grid cell is 300 meters, but the surface elevation data at that point is 350 meters, then this cell is marked as unusable.

[0023] In some embodiments, when constructing a 3D spatial mesh model, the environment modeling module can perform airspace layering design in the vertical direction based on airspace usage requirements or airspace structural characteristics. For example, a preset vertical height range can be divided into multiple height layers for different purposes, such as a low-altitude logistics flight layer, a general aviation cruise layer, and a temporary isolation airspace layer. Each height layer can be configured with its corresponding composite restriction layer independently or in combination. Airspace layering design allows the generation of route cell networks to be based on the passable space within a specific height layer, which helps to achieve refined management and use of airspace resources.

[0024] In some embodiments, forming a composite restriction layer superimposed on the static geographic information layer includes multiple asynchronous data import processes, establishing a real-time connection with a meteorological data server, and acquiring wind field, visibility, and precipitation intensity data of the flight airspace within a time window. For example, if the wind speed at a certain point in the gridded wind field data is 15 m / s and the visibility is 3000 m, grid cells with wind field data exceeding a threshold or visibility below a threshold are marked as meteorological restriction zones. The marking process uses a judgment formula: if the condition is met... or Then the grid cells Marked as a weather restriction zone, in the formula Represents grid cell Wind speed value at that location, This represents the preset wind field velocity threshold. Represents grid cell Visibility value at the location, This represents a preset visibility threshold. It can be understood that the system accesses the notification system issued by the airspace management agency, parses the temporary airspace restrictions effective within the time window, and marks the corresponding grid cells as controlled restricted areas. For example, if there is a circular temporary no-fly zone in the notification system, the environmental modeling module calculates and marks all grid cells whose center points fall within this circular area. In specific implementation, it receives real-time flight track data from the low-altitude flight service management platform, predicts the future positions of other aircraft, and marks the predicted occupied grid cells as dynamic avoidance zones. For example, if a flight track data indicates that an aircraft is flying at a specific speed and heading, the environmental modeling module uses a motion model to infer a series of grid cells it may occupy within a future time window. It can be understood that meteorological restricted areas, controlled restricted areas, and dynamic avoidance zones are logically superimposed to form a composite restriction layer. Logical superposition means that for the same grid cell, if it is marked as a restricted area by any source, that cell is marked as impassable in the composite restriction layer.

[0025] In some embodiments, data comparison reflects the differences in labeling of the same spatial area by different data sources. For example, at a certain altitude level in flight airspace, meteorological data indicates that the wind field is not exceeded, but airspace notices indicate that the area is a temporary control zone. In this case, the grid cell in this area is only marked as a control-restricted area, not a meteorological-restricted area. Similarly, if the dynamic avoidance zone of real-time flight path prediction may overlap with the terrain obstacle zone in the static geographic information layer, the overlapping cell has already been marked as unusable in the static geographic information layer, and remains unusable after the composite restriction layer is overlaid. Optionally, when mapping surface elevation data, the environmental modeling module uses digital elevation model (DEM) data with a resolution of 30 meters to ensure the matching accuracy between the basic grid array cells and the elevation data points. Optionally, when acquiring meteorological data, the temporal resolution of wind field data, visibility data, and precipitation intensity data is 1 hour, and the spatial resolution is 5 kilometers. The environmental modeling module uses an interpolation algorithm to match the meteorological data to each cell of the basic grid array. In practice, the prediction time step used when marking dynamic avoidance zones is 10 seconds, predicting the location within the next 5 minutes. Each predicted location corresponds to a grid cell. If multiple predicted locations occupy the same cell, that cell is marked only once. In the implementation, the logical overlay operation uses the grid cell as the basic unit, sequentially checking the status of each grid cell in the meteorological restriction zone, control restriction zone, and dynamic avoidance zone. Finally, it outputs a three-dimensional Boolean array with the same dimensions as the basic grid array, used to represent the composite restriction layer.

[0026] In one embodiment of the present invention, referring to Figure 3, when the network generation module constructs the route cell network, it can further constrain the passable space based on the airspace hierarchical structure. For example, if the airspace hierarchical design defines dedicated passage rules for a specific altitude layer, the network generation module can preferentially generate route cell nodes within that altitude layer and ignore spatial units at other altitude layers. This hierarchical spatial filtering mechanism can ensure that the generated route cell network is more in line with airspace management requirements and avoid conflicts between route planning results and hierarchical usage strategies. The network generation module scans the grid cells in the three-dimensional spatial grid model constructed by the environment modeling module that are not marked by the composite constraint layer, and regards these grid cells as potential passable units. In the example scenario, the three-dimensional spatial grid model contains a basic grid array with a longitude interval of 0.01 degrees, a latitude interval of 0.01 degrees, and an altitude interval of 50 meters. The total number of grid cells in the three-dimensional envelope space is approximately 1.2 million. After being marked by the composite constraint layer, approximately 450,000 grid cells are marked as meteorological restricted areas, control restricted areas, or dynamic avoidance areas, and the remaining approximately 750,000 grid cells become potential passable units. Centered on each potentially passable cell, the network generation module examines its 26 neighboring spatial cells, checking whether these cells are not marked by the composite confinement layer. If all 26 neighboring cells are passable, the central cell is upgraded to a candidate route node. After this step, approximately 200,000 grid cells in the example scenario meet the criteria to become candidate route nodes. From the candidate route node set, random sampling is performed according to a preset node density threshold to remove some candidate route nodes to prevent over-density. The node density threshold is set to retain a maximum of 8 nodes per cubic kilometer. After random sampling, the number of candidate route nodes is reduced to approximately 50,000. Between the remaining candidate route nodes, connection weights are calculated based on 3D spatial distance and inter-node visibility, establishing an undirected weighted graph structure as the route cell network. When calculating connection weights, for any two nodes, if their connection does not traverse any grid cells marked as impassable, they are considered to be visible. The initial value of the connection weight is the 3D Euclidean distance between the nodes.

[0027] In some embodiments, the node density threshold is adjusted based on the volume of the flight airspace and the planning accuracy requirements. For example, for high-precision planning, the node density threshold is set to 15 nodes per cubic kilometer, and for rapid planning, it is set to 5 nodes per cubic kilometer. It can be understood that visibility judgment uses a 3D ray tracing algorithm to check the state of the grid cells traversed by the node connection. If any grid cell traversed by the connection is marked in the composite constraint layer, it is determined to be non-visible. In a specific implementation, when establishing the undirected weighted graph structure, connection edges are only established for node pairs whose 3D Euclidean distance is less than the connection distance threshold and which pass the visibility judgment. The connection distance threshold is set to 3000 meters. In the example scenario, the final generated route cell network contains approximately 50,000 nodes and approximately 800,000 connection edges. Data comparison shows that when the connection distance threshold is reduced from 3000 meters to 1500 meters, the number of connection edges decreases to approximately 200,000, and the network sparsity increases.

[0028] In practical implementation, configuring dynamic passage costs for each connecting edge in the route cell network requires obtaining aircraft performance parameters and mission attribute parameters. Aircraft performance parameters include maximum climb rate, minimum turning radius, and cruise speed. In the example flight mission, the maximum climb rate is 8 m / s, the minimum turning radius is 300 m, and the cruise speed is 50 m / s. Mission attribute parameters include mission priority, economy coefficient, and safety level. In the example mission, the mission priority is ordinary, the economy coefficient is 0.9, and the safety level is level two. For each connecting edge in the route cell network, the three-dimensional length, turning angle change, and average altitude of the connecting edge are calculated. The turning angle change refers to the absolute value of the angle between the direction vectors of the connecting edge and the previous connecting edge in three-dimensional space. The average altitude refers to the arithmetic mean of the altitudes of the grid cells corresponding to the two endpoints of the connecting edge. Based on the three-dimensional length, turning angle change, and average altitude, combined with the maximum climb rate and minimum turning radius, the basic physical cost is calculated. From the formula:

[0029] in: Represents the three-dimensional length of the connecting edge. This represents the change in the turning angle of the connecting edge. This represents the average elevation of the connecting edges. Represents the minimum turning radius of the aircraft. Represents the aircraft's maximum rate of climb. , , This is the adjustment factor. The basic physical cost is multiplied by the economic coefficient, and then an additional security cost is added according to the security level to obtain the dynamic passage cost. In the example, when the security level is level two, the additional security cost is a fixed value of 100.

[0030] In some embodiments, when calculating the change in steering angle, the change in steering angle for the initial connecting edge in the route cell network is set to zero, and the average altitude is calculated by querying the elevation data of the corresponding grid cells at both ends of the connecting edge. Optionally, an adjustment coefficient... , , Based on historical flight data calibration, The value is usually greater than and The additional security cost is understood to be obtained from a preset mapping table based on the security level. Level 1 has an additional security cost of 0, Level 2 has an additional security cost of 100, and Level 3 has an additional security cost of 300. Optionally, an economic coefficient less than 1 indicates that the task is insensitive to economic costs, resulting in a reduced dynamic travel cost; an economic coefficient greater than 1 indicates that the task is sensitive to economic costs, resulting in a higher dynamic travel cost. Data comparison shows that for the same 3D length of 5000 meters, a turning angle change of 0.5 radians, and an average altitude of 1000 meters, when the economic coefficient changes from 0.9 to 1.1, the dynamic travel cost increases from 5500 to 6700; when the security level changes from Level 2 to Level 3, the dynamic travel cost increases from 5600 to 5900 due to the increased additional security cost.

[0031] In one embodiment of the present invention, the intelligent route planning module invokes a route growth algorithm to perform bidirectional optimization growth to generate a connected path. The route growth algorithm uses the origin and destination points of the flight intention as the growth origin and target points, respectively, and searches in the route cell network based on the dynamic passage cost. The takeoff point and landing point of the flight intention are mapped to the two nearest candidate route nodes in the route cell network, which are respectively used as the root nodes of growth tree A and growth tree B. In the example scenario, the takeoff point is mapped to candidate route node number N_4512, and the landing point is mapped to candidate route node number N_8734. Two empty priority queues are initialized to store the edge nodes to be expanded in growth tree A and growth tree B, respectively. Each element in the priority queue contains a node identifier and the cumulative dynamic passage cost from the root node to that node. In each iteration, the node with the smallest cumulative dynamic passage cost is selected from the two priority queues for expansion. All adjacent candidate route nodes of this node in the route cell network are explored. For each adjacent node, the cumulative dynamic passage cost from the root node to the adjacent node via the current node is calculated. If the adjacent node is not in the corresponding growth tree or the newly calculated cumulative dynamic passage cost is less than the recorded value, the node information is updated and inserted into the priority queue.

[0032] When executing the route growth algorithm in the intelligent route planning module, the growth strategy can be optimized by incorporating airspace hierarchical information. For example, if the flight mission specifies a preferred altitude layer, the growth algorithm can prioritize expanding nodes in the route cell network within that layer and limit the frequency of cross-layer searches. This hierarchical guidance mechanism enables the generated routes to better comply with airspace usage regulations, while reducing unnecessary three-dimensional spatial searches and improving planning efficiency.

[0033] When a newly expanded node in growth tree A is already a node in growth tree B, or a newly expanded node in growth tree B is already a node in growth tree A, it is determined to be a bidirectional growth intersection. In the example, when growth tree A expands to node N_6521, it is found that this node already exists in the node set of growth tree B. The paths from the root node of growth tree A to the intersection node and from the root node of growth tree B to the intersection node are backtracked, and the two paths are concatenated to form the initial connected path. In the example, the connected path contains the node sequence from N_4512 to N_6521 and then to N_8734.

[0034] In some embodiments, the distance from the origin and destination points to the nearest candidate route node is calculated using three-dimensional Euclidean distance, and the candidate route node with the smallest distance is selected. It can be understood that the priority queue implementation is based on a min-heap data structure to ensure efficient extraction of the node with the minimum cumulative dynamic passage cost each time. In specific implementations, when exploring adjacent candidate route nodes, only the connecting edges existing in the route cell network are considered, and the connecting edge information is pre-stored in an undirected weighted graph structure. Data comparison shows that when using a unidirectional growth algorithm, the number of searched nodes is approximately 8500, while when using a bidirectional growth algorithm, growth tree A and growth tree B each search approximately 3000 nodes before reaching intersection, reducing the total number of searched nodes. Optionally, the parent node pointer is recorded each time a node is expanded to facilitate path backtracking.

[0035] In practical implementation, spline curve fitting is performed on the connected path to eliminate sharp angles and generate a smooth 3D spatial curve. The 3D spatial coordinates of all sequentially arranged waypoint nodes in the connected path are extracted; in this example, the connected path contains a coordinate sequence of 45 waypoint nodes. A non-uniform rational B-spline curve algorithm is used to fit the 3D spatial coordinates to generate an initial fitted curve. This algorithm employs cubic B-spline basis functions and a control point inverse calculation method. The curvature value at each fitting point on the initial fitted curve is calculated, and fitting points with curvature values ​​exceeding the aircraft's maximum permissible curvature are identified as high curvature points. From the formula:

[0036] in: The representative curve is represented by parameters The position vector represents... yes For parameters The first derivative vector, It is a unit tangent vector. It is the unit tangent vector with respect to the parameters The derivative vector, This represents the Euclidean norm of the vector. Centered on each high curvature point, several adjacent fitting points on the initial fitted curve are selected as adjustment segments. Each adjustment segment contains the high curvature point and two fitting points before and after it, for a total of five points. A weighted average is performed on the fitting points within each adjustment segment to generate a new control point sequence. During weighted averaging, the weight of the central high curvature point is reduced, while the weight of adjacent points is increased. The new control point sequence is used to re-fit a non-uniform rational B-spline curve to generate an intermediate smooth curve. The process of curvature value calculation, high curvature point identification, and adjustment segment processing is repeated until the curvature values ​​of the generated curve at all fitting points are less than or equal to the maximum permissible curvature of the aircraft. The final curve is then used as the smoothed three-dimensional space curve.

[0037] In some embodiments, the node vectors are determined using the chord length parameterization method during the fitting of the non-uniform rational B-spline curve. It can be understood that the maximum permissible curvature of the aircraft is set based on the reciprocal of the aircraft's minimum turning radius; in the example, the minimum turning radius is 300 meters, so the maximum permissible curvature is 0.0033 per meter. Data comparison shows that 8 nodes in the initial connected path have turning angles exceeding 30 degrees. After spline curve fitting, the curvature values ​​at all points are lower than the maximum permissible curvature. Optionally, in the weighted average processing, the weight of the center point is set to 0.2, and the weights of adjacent points are set to 0.4 and 0.4, respectively. In a specific implementation, smoothing iterations are repeated until the number of high curvature points no longer decreases in two consecutive iterations or the maximum number of iterations (10) is reached.

[0038] In one embodiment of the present invention, the intelligent route planning module performs discrete sampling and conflict detection on a smooth three-dimensional spatial curve. The smooth three-dimensional spatial curve is generated by the spline curve fitting process of the embodiment. Discrete sampling is performed along the smooth three-dimensional spatial curve at fixed time intervals to generate a series of time-space four-dimensional waypoints. The fixed time interval is set to 30 seconds. In the example scenario, based on the aircraft's cruising speed of 50 meters per second, a total of 800 four-dimensional waypoints are generated by sampling along the smooth curve with a total length of 120 kilometers. Each four-dimensional waypoint is checked to see if it intrudes into the dynamic avoidance zone in the composite restriction layer. The intrusion judgment is based on whether the three-dimensional spatial coordinates and the expected time of the waypoint overlap with the spatiotemporal range defined by the dynamic avoidance zone. The dynamic avoidance zone is generated by the environment modeling module based on real-time trajectory prediction. If an intrusion exists, the smooth three-dimensional spatial curve is locally adjusted near the intrusion point and resampling and detection are performed. The local adjustment is achieved by applying a small displacement vector perpendicular to the tangent of the original curve to the smooth curve at the parameter position corresponding to the intrusion point. The direction of the displacement vector is from the intrusion point to the center of the nearest passable grid cell. The magnitude of the displacement is dynamically calculated according to the intrusion depth, and the formula is:

[0039] in: This represents the magnitude of the displacement in a local adjustment. The adjustment factor is set to 1.5. The preset safe distance threshold is set to 200 meters. This is the minimum spatial distance between the current intrusion point and the boundary of the dynamic avoidance zone. Data comparison shows that the initial sampling detection found 15 intrusions at four-dimensional waypoints. After three rounds of local adjustments and re-detection, all waypoints passed the detection. The final conflict-free trajectory defined by a series of four-dimensional waypoints is used as the preliminary planned route. The preliminary planned route contains a complete spatiotemporal description of 800 waypoints. See Table 1.

[0040] Table 1: Data Table of Conflict Detection and Adjustment Process

[0041] In some embodiments, the fixed time interval can be configured between 10 and 60 seconds, depending on the accuracy requirements of the flight mission. It can be understood that resampling detection is an iterative process; after each local adjustment, a new smooth curve is generated, and the entire process from sampling to detection is re-executed until there are no conflicts or the maximum number of iterations (5) is reached. In a specific implementation, the direction calculation of the displacement vector calls the spatial geometry library to find the direction vector from the intrusion point to the center of the nearest non-dynamic avoidance zone grid cell.

[0042] In practice, the results processing module returns the preliminary planned route to the Low Altitude Flight Service Management Platform for verification and confirmation. It encapsulates the preliminary planned route into a formatted message conforming to aviation data exchange standards, attaching a summary of key constraints generated during the process. The formatted message uses the AIXM5.1 data standard, and the key constraint summary includes the release time of the meteorological data used, the list of included temporary airspace restriction notice numbers, and the dynamic avoidance zone version identifier used for conflict detection. The module then sends a formatted message request to the Low Altitude Flight Service Management Platform for airspace structure compliance verification and traffic feasibility verification. Airspace structure compliance verification checks whether the preliminary planned route contradicts the permanent airspace structure. Traffic feasibility verification is based on the platform's traffic management module predicting the congestion level of the route within the application time window. The module receives verification feedback information from the Low Altitude Flight Service Management Platform, including approval, pending approval with modification suggestions, or rejection instructions. In the example, a pending approval instruction with modification suggestions is received: "It is recommended to raise the altitude of the waypoint segment from P_301 to P_320 by 150 meters to avoid planned air activity." If it is an instruction pending approval with modification suggestions, the preliminary planned route will be automatically adjusted according to the modification suggestions, and a new format message will be sent for confirmation. The text description in the modification suggestions will be automatically adjusted and parsed to extract key parameters. The altitude coordinates of the specified route points in the preliminary planned route will be uniformly increased by 150 meters, and the collision detection near the point points will be re-executed. After confirmation that there are no errors, a new format message will be encapsulated and submitted again.

[0043] Referring to Figure 4, this is a radar chart of low-altitude flight route verification scores. By comparing the score changes before adjustment (orange) and after adjustment (green), the optimization effect of route planning is evaluated from multiple dimensions. The radar chart visually shows the score improvement in the five core verification dimensions after route adjustment, with "safe distance compliance" and "altitude compliance" showing the largest improvements, reflecting the core effectiveness of the local adjustment strategy. The score changes directly correspond to the technical aspect of "automatically adjusting the preliminary planned route based on modification suggestions" in the patent, which can verify the effectiveness of the adjustment strategy and provide a quantitative basis for subsequent route optimization. By analyzing the score differences in each dimension, the focus of subsequent optimization can be identified. For example, there is still room for improvement in "flow feasibility," and the route can be further adjusted in conjunction with real-time traffic conditions.

[0044] In one embodiment of the present invention, the planning process traceability module is activated synchronously when the system initiates the route planning process. The planning process traceability module runs through the entire process from receiving the original flight intention to outputting the preliminary planned route. In the example scenario, a complete planning process involves the sequential execution of the intention parsing module, environment modeling module, network generation module, intelligent route planning module, and result processing module. The planning process traceability module immediately captures and records the input data snapshot, output data snapshot, and key decision parameters of each logical step upon completion. It records the input data snapshot, output data snapshot, and key decision parameters that play a decisive role in each core logical step. In the example, the intention parsing step records the original flight intention text string "from Airport A to Airport B, time window 2025-10-10 14:00 to 16:00" and the parsed structured flight airspace range and time window object. The key decision parameter is recorded as the version number of the regular expression pattern used for airspace parsing. In the environmental modeling step, the raw meteorological data reports, airspace notice texts, and real-time waypoint lists imported from external data sources are recorded as input data snapshots, and the generated 3D mesh Boolean array is recorded as output data snapshots. Key decision parameters include mesh resolution parameters, meteorological threshold parameters, and dynamic avoidance prediction duration parameters. In the network generation step, the 3D mesh Boolean array and aircraft performance parameters are recorded as inputs, and the generated route cell network graph structure is recorded as output. Key decision parameters include node density threshold, connection distance threshold, and visibility judgment algorithm identifier. In the intelligent route planning step, the route cell network graph and dynamic passage cost matrix are recorded as inputs, and the four-dimensional waypoint sequence of the preliminary planned route is recorded as output. Key decision parameters include route growth algorithm type, heuristic function weights for bidirectional search, and curvature tolerance parameters for spline curve fitting.

[0045] In some embodiments, input data snapshots store the original data or a hash digest of the data in a serialized format; for large datasets such as 3D mesh Boolean arrays, the hash digest is stored. It is understood that output data snapshots store key result data; for example, a route cellular network stores a list of nodes and edges rather than the complete graph structure. In specific implementations, key decision parameters are recorded in key-value pair form, such as "node density threshold: 8 nodes / cubic kilometer" and "connection distance threshold: 3000 meters". Data comparison shows that when the node density threshold is adjusted from 8 nodes / cubic kilometer to 15 nodes / cubic kilometer, the number of nodes in the network generation step output data snapshot recorded by the planning process tracing module increases from approximately 50,000 to approximately 90,000, and the key decision parameter records are updated accordingly.

[0046] In implementation, the planning process traceability module organizes all records chronologically and logically to form a structured planning log that fully reproduces the entire route planning process. The planning log uses timestamps as the main thread to link the records of each step and marks the dependencies of each step's output. The planning log is bound to the final route planning result confirmed by the low-altitude flight service management platform. This binding is achieved by assigning a globally unique task identifier to each planning task. Both the planning log and the final confirmed route planning result use the same task identifier as the index key. The bound planning log and route planning result are persistently stored in a database using a relational database table structure. The planning log is stored in a separate table field in JSON format. Data integrity is guaranteed through checksums. formula:

[0047] in: The binary content representing the planning log. The binary content representing the final confirmed route planning result. Represents the task identifier. This indicates a string concatenation operation; Hash represents the checksum calculated using the SHA-256 hash function. Store them together for subsequent verification.

[0048] In some embodiments, the planning log structure includes a header, body, and footer. The header records the task identifier, planning start time, and end time; the body records the tracking entries for each step in chronological order; and the footer records the integrity checksum. It is understood that the JSON content of the planning log is compressed when persistently stored in the database to reduce storage space usage. Optionally, the planning log supports querying and retrieval by task identifier, planning time range, or flight airspace conditions. Data comparison shows that the planning log size generated by a typical planning task is approximately 2MB, while the final confirmed route planning result is approximately 50KB.

[0049] Referring to Figure 5, this is a bar chart comparing data complexity and parsing accuracy during the low-altitude flight intention parsing stage, showing the changes in two core indicators: airspace range and time window, from the original flight intention to the parsed airspace range. The chart clearly presents the negative correlation between "decreased data complexity" and "improved parsing accuracy," verifying the technical effectiveness of the intention parsing module—through structured processing, the original non-standardized flight intention is transformed into highly accurate airspace and time window information. The stepwise decrease in data complexity reflects the ability to filter redundant information and extract core elements during the parsing process; while the continuous improvement in parsing accuracy proves the algorithm's accurate understanding of flight intentions, providing reliable input for subsequent route planning. This chart can be used to verify the performance baseline of the intention parsing module and provide quantitative basis for system iteration and upgrades. For example, when the data complexity drops to 45%, the parsing accuracy reaches 98%, indicating that the current algorithm is close to its optimal state in time window parsing.

[0050] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A flight route intelligent planning system based on a low-altitude flight service management platform, characterized in that: include: The intention parsing module is used by the low-altitude flight service management platform to receive the original flight intention and parse out the flight airspace and time window. The environmental modeling module establishes a three-dimensional spatial grid model corresponding to the flight airspace. The three-dimensional spatial grid model includes a static geographic information layer. Meteorological information, temporary airspace restriction information, and dynamic traffic situation information are asynchronously imported from multiple external data sources to form a composite restriction layer superimposed on the static geographic information layer. The network generation module uses preset route cell generation rules to iteratively construct a route cell network within the passable space identified by the composite restriction layer. In response to aircraft performance parameters and mission attribute parameters, it configures a dynamic passage cost for each route cell in the route cell network. The intelligent route planning module calls the route growth algorithm, using the start and end points of the flight intention as the growth origin and target point, and performs bidirectional optimization growth in the route cell network based on the dynamic passage cost until a connected path is generated. The connected path is smoothed and conflict detected, and a preliminary planned route is output. The results processing module returns the preliminary planned route to the low-altitude flight service management platform for verification and confirmation.

2. The intelligent flight route planning system based on a low-altitude flight service management platform as described in claim 1, characterized in that, The establishment of a three-dimensional spatial grid model corresponding to the flight airspace includes: defining a three-dimensional envelope space based on the latitude and longitude boundaries of the flight airspace and a preset vertical height range; dividing the three-dimensional envelope space into equal intervals along the longitude, latitude, and height directions to generate a basic grid array composed of a large number of cubic units; extracting surface elevation data within the three-dimensional envelope space from a terrain database, mapping the surface elevation data to the basic grid array, marking all grid units with elevation values ​​lower than the surface elevation as unusable, and forming the static geographic information layer.

3. The intelligent flight route planning system based on a low-altitude flight service management platform as described in claim 2, characterized in that, The process of forming a composite restriction layer superimposed on the static geographic information layer includes: establishing a real-time connection with a meteorological data server to acquire wind field, visibility, and precipitation intensity data of the flight airspace within the time window, and marking grid cells with wind field data exceeding a threshold or visibility below a threshold as meteorological restricted areas; accessing the notification system issued by the airspace management agency to parse the temporary airspace restriction information in effect within the time window and marking the corresponding grid cells as controlled restricted areas; receiving real-time flight track data from the low-altitude flight service management platform to predict the future positions of other aircraft and marking the predicted occupied grid cells as dynamic avoidance areas; and logically superimposing the meteorological restricted areas, the controlled restricted areas, and the dynamic avoidance areas to form the composite restriction layer.

4. The intelligent flight route planning system based on a low-altitude flight service management platform as described in claim 1, characterized in that, The step of iteratively constructing a route cell network within the passable space identified by the composite constraint layer using a preset route cell generation rule includes: scanning the grid cells in the three-dimensional spatial grid model that are not marked by the composite constraint layer and treating them as potential passable cells; examining the twenty-six adjacent spatial neighbor cells of each potential passable cell as the center, and upgrading the center cell to a candidate route node if all neighbor cells are unmarked; randomly sampling the candidate route node set according to a preset node density threshold and removing some candidate route nodes to prevent over-density; and calculating connection weights between the remaining candidate route nodes based on three-dimensional spatial distance and inter-node visibility to establish an undirected weighted graph structure as the route cell network.

5. The intelligent flight route planning system based on a low-altitude flight service management platform as described in claim 4, characterized in that, The process of configuring dynamic passage costs for each route cell in the route cell network in response to aircraft performance parameters and mission attribute parameters includes: obtaining the maximum rate of climb, minimum turning radius, and cruise speed of the aircraft as aircraft performance parameters; obtaining mission attribute parameters of the flight mission, including mission priority, economy coefficient, and safety level; calculating the three-dimensional length, turning angle change, and average altitude of each connecting edge in the route cell network; calculating the basic physical cost based on the three-dimensional length, the turning angle change, the average altitude, and the maximum rate of climb and the minimum turning radius; multiplying the basic physical cost by the economy coefficient, and then adding an additional safety cost according to the safety level to finally obtain the dynamic passage cost.

6. The intelligent flight route planning system based on a low-altitude flight service management platform as described in claim 5, characterized in that, The invocation of the route growth algorithm, using the origin and destination points of the flight intention as the origin and destination points, and performing bidirectional optimization growth in the route cell network based on the dynamic passage cost until a connected path is generated, includes: mapping the origin and destination points to the two nearest candidate route nodes in the route cell network, which are respectively the root nodes of growth tree A and growth tree B; initializing two empty priority queues to store the edge nodes to be expanded in growth tree A and growth tree B, respectively; in each iteration, selecting the node with the minimum cumulative dynamic passage cost from the priority queues for expansion, and exploring all its adjacent candidate route nodes; when the newly expanded node in growth tree A is already a node in growth tree B, or the newly expanded node in growth tree B is already a node in growth tree A, it is determined to be a bidirectional growth intersection; backtracking the path from the root node to the intersection node, and splicing the two paths to form the initial connected path.

7. The intelligent flight route planning system based on a low-altitude flight service management platform as described in claim 6, characterized in that, The process of smoothing and detecting conflicts in the connected path to output a preliminary planned route includes: fitting a spline curve to the connected path to eliminate sharp angles and generate a smooth three-dimensional spatial curve; performing discrete sampling along the smooth three-dimensional spatial curve at fixed time intervals to generate a series of time-space four-dimensional waypoints; checking whether each of the four-dimensional waypoints intrudes into the dynamic avoidance zone in the composite constraint layer; if an intrusion occurs, locally adjusting the smooth three-dimensional spatial curve near the intrusion point and resampling and detecting until there is no conflict; and using the final conflict-free trajectory defined by the series of four-dimensional waypoints as the preliminary planned route.

8. The intelligent flight route planning system based on a low-altitude flight service management platform as described in claim 1, characterized in that, The step of returning the preliminary planned route to the low-altitude flight service management platform for verification and confirmation includes: encapsulating the preliminary planned route into a formatted message conforming to aviation data exchange standards, and attaching a summary of key constraints generated during the process; sending the formatted message to the low-altitude flight service management platform to request airspace structure compliance verification and traffic feasibility verification; receiving verification feedback information from the low-altitude flight service management platform, the verification feedback information including approval, pending approval with modification suggestions, or rejection instructions; if it is a pending approval instruction with modification suggestions, automatically adjusting the preliminary planned route according to the modification suggestions, and resending the formatted message for confirmation.

9. The intelligent flight route planning system based on a low-altitude flight service management platform as described in claim 6, characterized in that, The process of fitting spline curves to the connected path to eliminate sharp angles and generate a smooth three-dimensional spatial curve includes: extracting the three-dimensional spatial coordinates of all sequentially arranged route nodes in the connected path; fitting the three-dimensional spatial coordinates using a non-uniform rational B-spline curve algorithm to generate an initial fitted curve; calculating the curvature value at each fitted point on the initial fitted curve, identifying fitted points whose curvature values ​​exceed the maximum allowable curvature of the aircraft as high curvature points; selecting multiple adjacent fitted points on the initial fitted curve as adjustment segments, centered on each high curvature point; performing weighted averaging on the fitted points within each adjustment segment to generate a new control point sequence; re-fitting the non-uniform rational B-spline curve using the new control point sequence to generate an intermediate smooth curve; repeating the process of curvature value calculation, high curvature point identification, and adjustment segment processing until the curvature values ​​of the generated curve at all fitted points are less than or equal to the maximum allowable curvature of the aircraft, and using the final obtained curve as the smooth three-dimensional spatial curve.

10. The intelligent flight route planning system based on a low-altitude flight service management platform as described in claim 8, characterized in that, The system also includes a planning process traceability module, which records the input data, output data, and key decision parameters of each logical step from receiving the original flight intention to outputting the preliminary planned route, forming a reproducible planning log. The planning log is bound and stored with the final confirmed route planning result.

Images