Low-altitude comprehensive management system based on grid digital twin model and intelligent algorithm

Abstract

The application discloses a low-altitude comprehensive management system based on a grid digital twinborn model and an intelligent algorithm, and relates to the technical field of low-altitude aircrafts.The application comprises the following steps: multi-source sensing data fusion and grid coding, constructing an airspace environment space-time database; constructing the airspace environment space-time database to perform three-dimensional solid sectioning on the airspace; generating a multi-scale grid twinborn body, and correlating the attributes of each grid according to management requirements; generating an airspace risk heat map for the next 5-30 minutes; establishing a grid state change triggering rule; and taking the grid navigation cost as a weight to generate a multi-target optimal path set.The application combines real-time operation data of low-altitude aircrafts, meteorological environment data and the like with an artificial intelligence algorithm and a navigation rule base, analyzes airspace traffic flow, optimizes a flight plan, intelligently allocates a flight route, realizes automatic setting of an arbitrary two-point route, and improves the response speed of a low-altitude flight service system and the processing capacity of various flight data.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of low-altitude aircraft, and integrates satellite navigation, Internet of Things, AI algorithm, Beidou grid code and real scene three-dimensional digital twin technologies to build a low-altitude comprehensive intelligent management platform, in particular to a low-altitude comprehensive management system based on a grid digital twin model and an intelligent algorithm. BACKGROUND

[0002] Low-altitude navigation refers to daily low-altitude flight activities in an environment where low-altitude airspace is open, relying on smooth guidance of ground support facilities and autonomous avoidance functions of low-altitude aircraft. In W-class airspace with a true height of 120 meters or below, mainly micro, light, small and medium-sized unmanned aerial vehicles are operated. In G-class airspace with a true height of 120 meters to 300 meters, light, small and medium-sized commercial unmanned aerial vehicles are allowed to fly. Above that, in E-class airspace with a true height of 300 meters to 1000 meters, medium and large-sized unmanned aerial vehicles and innovative aircraft such as eVTOL and flying cars are suitable for activities. Looking forward to the future, with the continuous development of the aviation industry, its activity airspace is expected to gradually expand to a true height of 3000 meters.

[0003] Low-altitude air traffic control needs to realize multiple functions such as air traffic control, flow management and airspace management included in traditional civil air traffic control. However, due to the integration of unmanned aerial vehicles, eVTOL, helicopters, traditional fixed-wing aircraft and other aircraft in low-altitude flight, it will promote the flight volume to double, and the mixed operation of various aircraft, high density of aircraft, large number of obstacles and complex weather environment will put higher requirements on the carrying capacity and technological innovation of the air traffic control system. Under this background, a low-altitude comprehensive intelligent system is urgently needed to have core capabilities such as airspace management, planning service, inspection and surveillance, operation management and safety supervision. SUMMARY

[0004] The purpose of the present application is to provide a low-altitude comprehensive management system based on a grid digital twin model and an intelligent algorithm, which integrates satellite navigation, Internet of Things, AI algorithm, Beidou grid code and real scene three-dimensional digital twin technologies to build a low-altitude comprehensive intelligent management platform, solving the problems of slow response speed and low efficiency of flight data processing in existing low-altitude management.

[0005] To solve the above technical problems, the present application is realized by the following technical scheme:

[0006] The present application is a low-altitude comprehensive management system based on a grid digital twin model and an intelligent algorithm, which includes a device terminal and an infrastructure layer, a data and service support layer and an application layer.

[0007] The device terminal and infrastructure layer includes device terminals and infrastructure; the device terminal is a core hard core component responsible for communication and data processing in the airborne system, and the main role is to realize the two-way information interaction between the device terminal and the ground control station, other devices or network through the data link, which specifically includes low-altitude aircraft, airborne communication equipment, multi-source navigation equipment, airborne perception and identification equipment and intelligent flight path equipment; the low-altitude aircraft includes unmanned aerial vehicles, eVTOL aircraft and flying cars; the airborne communication equipment includes private network terminal, satellite communication equipment, airborne ad hoc network and airborne digital / video transmission; the multi-source navigation equipment includes inertial navigation, visual navigation, satellite navigation and altimeter; the altimeter is a barometric altimeter or an ultrasonic altimeter; wherein the barometric altimeter is used to measure atmospheric pressure, thereby calculating the relative height; the ultrasonic altimeter is used to measure the height of the aircraft relative to the ground or sea level by ultrasonic waves; the airborne perception and identification equipment includes identity recognition equipment, visual perception equipment and airborne traffic radar; the intelligent flight path equipment includes intelligent flight tube module, autonomous avoidance module and airborne electronic fence; the intelligent flight tube module, i.e. flight control system, is used to be responsible for low-altitude aircraft flight attitude control, task execution and safety guarantee; wherein the flight attitude control monitors the roll, pitch and yaw angle of the unmanned aerial vehicle in real time through sensors such as accelerometers and gyroscopes, dynamically adjusts the motor speed to keep the flight stable, the task execution is used to analyze and execute operation instructions (such as take-off, landing and flight path), and coordinates the power components to accurately complete the action and safety guarantee is used to monitor parameters such as battery power and signal strength, and triggers automatic return or emergency landing to deal with abnormal situations; the autonomous avoidance module realizes dynamic obstacle avoidance through environmental perception and intelligent algorithm, including environmental perception, dynamic avoidance and cooperative avoidance; environmental perception combines sensors such as laser radar, binocular vision and millimeter wave radar to detect the distance, shape and motion state of obstacles in real time; dynamic avoidance generates real-time obstacle avoidance path based on AI algorithm, supports detouring or hovering in complex scenarios, and cooperative avoidance shares airspace information with surrounding unmanned aerial vehicles and unmanned vehicles through the low-altitude control platform, and optimizes the global obstacle avoidance strategy; the airborne electronic fence limits the activity range of the unmanned aerial vehicle through virtual boundary, and prevents illegal flight; mainly based on GPS or Wi-Fi positioning technology to demarcate no-fly zones (such as airports and sensitive areas), and when the unmanned aerial vehicle approaches the boundary, an alarm or forced hovering is triggered;

[0008] The infrastructure includes low-altitude communication facilities, low-altitude navigation facilities, low-altitude monitoring facilities and support facilities; the low-altitude communication facilities include mobile public networks, satellite communication networks and low-altitude communication private networks; the low-altitude navigation facilities include network / area RTK, ground-based / satellite-based facilities and navigation monitoring equipment; the low-altitude monitoring facilities include identity recognition receiving equipment, integrated sensing equipment, low-altitude radars, spectrum detection equipment and photoelectric / infrared detection equipment; the support facilities include low-altitude intelligence facilities, low-altitude meteorological facilities, take-off and landing facilities, computing power facilities and low-altitude countermeasure facilities;

[0009] The data and service support layer includes a communication access module, a navigation access module, a monitoring access module and a peripheral device access module, which are used to realize communication, navigation, monitoring and information services through a data exchange network; the communication access module is used to integrate C-band (wideband data transmission), UHF-band (remote control / telemetry) and Ka-band satellite communication links, support line-of-sight and beyond-line-of-sight communication, realize real-time data backhaul and remote control through a 4G / 5G cellular network; the transmitted data includes geographic information data, low-altitude thematic data and CIM data and the like; the navigation access module is compatible with satellite navigation systems such as Beidou, GPS and Galileo, and realizes high-precision positioning; the monitoring access module obtains real-time position and height information of the unmanned aerial vehicle through a ground radar station, and actively broadcasts its own state (position, speed and the like) for monitoring by an air traffic control system; the peripheral device access module is a standard interface such as a CAN bus, a USB and an Ethernet interface, which is used to connect task loads (cameras, infrared sensors and the like) and external equipment (meteorological instruments, laser radars);

[0010] The application layer is deployed with a low-altitude service platform; the low-altitude service platform includes a registration and authentication unit, low-altitude traffic services, a low-altitude traffic flow management unit, a low-altitude airspace management unit, a low-altitude data management unit and an application system interfacing unit; the registration and authentication unit includes an unmanned aerial vehicle registration module, an operator management module, a take-off and landing point management and a human operator management; the low-altitude traffic services include control services, coordination services and information services; the low-altitude traffic flow management unit includes low-altitude traffic rule management, low-altitude aerial vehicle management, low-altitude navigation management and low-altitude air route management; the low-altitude airspace management includes airspace division management, airspace flow management, electronic fence management, route planning management, three-dimensional grid management of airspace and terrain and ground obstacle management; the low-altitude data management includes ADS-B management data management and flight information data; the application system interfacing unit includes a system management module, a system configuration module, a unified authentication module and an application data interface module.

[0011] As a preferred technical solution, the low-altitude airspace management unit workflow is as follows:

[0012] Step S1: multi-source perception data fusion and grid coding, constructing a spatio-temporal database of airspace environment;

[0013] Step S2: three-dimensional solid subdivision of airspace, giving each grid a unique code, and establishing a dynamic mapping relationship between airspace resources and grid codes;

[0014] Step S3: generating multi-scale grid twin, associating corresponding attributes for each grid according to management needs;

[0015] Step S4: using time series prediction model to simulate the coupling effect of aircraft trajectory, weather changes and signal interference in the grid, and generating airspace risk heat map for the next 5-30 minutes;

[0016] Step S5: establishing grid state change trigger rules, when the aircraft crosses the grid boundary or the environmental parameters in the grid exceed the threshold, automatically triggering local update of the twin model;

[0017] Step S6: real-time calculation of the spatio-temporal intersection probability of aircraft trajectory in the grid, combined with Monte Carlo simulation to generate conflict warning levels;

[0018] Step S7: generating a set of multi-objective optimal paths with grid navigation cost as the weight.

[0019] As a preferred technical solution, in step S1, the spatio-temporal database of airspace environment is constructed by collecting aircraft position, speed, weather parameters and electromagnetic spectrum data from device terminals and infrastructure layer, and the specific construction method is as follows:

[0020] Step S11: obtaining the ID, speed, heading, grid code and three-dimensional coordinates of the aircraft through millimeter wave radar, ADSB broadcast receiving station and Beidou grid code positioning terminal;

[0021] Step S12: collecting wind speed, temperature and humidity, and visibility through multi-spectral weather sensors;

[0022] Step S13: using federated Kalman filter algorithm for time synchronization and spatial registration of radar point cloud, ADSB data and weather data;

[0023] Step S14: for abnormal data, combined with historical data trend prediction for difference completion;

[0024] Step S15: completing the construction of the spatio-temporal database of airspace environment.

[0025] As a preferred technical solution, in step S2, when the airspace is three-dimensionally subdivided, the Beidou grid position code standard is used to divide the airspace into a three-dimensional network structure with adjustable edge length, each grid including static and dynamic attributes;

[0026] The static attributes include a grid unique code, airspace regulation rules of a grid area, and elevation data of building obstacles in the grid area;

[0027] The dynamic attributes include a real-time number of aircraft in the grid, a meteorological risk index of the grid, and an electromagnetic interference intensity of the grid.

[0028] As a preferred technical solution, in the step S3, the multi-scale grid twin generation process is as follows:

[0029] Step S31: uniform spatiotemporal coding of multi-source data, establishing a grid coordinate system based on a Beidou grid position code, and aligning timestamps of different devices to UTC standard time;

[0030] Step S32: extracting grid-related features locally, and protecting sensitive data by differential privacy technology;

[0031] Step S33: fusing dynamic attributes and static attributes of different grid levels into a unified tensor, designing federal aggregation weights according to the importance of grid levels, and dynamically adjusting the data contribution of different levels of grids by using an attention mechanism;

[0032] Step S34: mapping multi-source data to grid levels according to a grid level mapping mechanism;

[0033] Step S35: coupling multi-level models to predict cross-regional flight flow distribution;

[0034] Step S36: performing lightweight processing on the model, and merging grids in low-risk areas;

[0035] Step S37: setting an incremental update strategy, and performing virtual-real synchronization verification.

[0036] As a preferred technical solution, in the step S3, the multi-scale grid twin generation process is as follows:

[0037] Step S31: uniform spatiotemporal coding of multi-source data, establishing a grid coordinate system based on a Beidou grid position code, and aligning timestamps of different devices to UTC standard time;

[0038] Step S32: extracting grid-related features locally, and protecting sensitive data by differential privacy technology;

[0039] Step S33: fusing dynamic attributes and static attributes of different grid levels into a unified tensor, designing federal aggregation weights according to the importance of grid levels, and dynamically adjusting the data contribution of different levels of grids by using an attention mechanism;

[0040] Step S34: mapping multi-source data to grid levels according to a grid level mapping mechanism;

[0041] Step S35: coupling the multi-level model to predict the cross-region flight flow distribution;

[0042] Step S36: lightening the model and merging the grids in low-risk areas;

[0043] Step S37: setting an incremental update strategy and performing virtual-real synchronization verification.

[0044] As a preferred technical solution, in the step S4, the time series prediction model is used to simulate the unified mapping of the aircraft trajectory, weather changes, and signal interference in the grid to the three-dimensional grid. The timestamp difference and coordinate system offset are eliminated through federated Kalman filtering. The features of the aircraft trajectory, weather changes, and signal interference are extracted respectively. A multi-modal time series prediction model is constructed. The cross-attention mechanism is introduced to dynamically adjust the weights of the aircraft trajectory, weather, and electromagnetic interference. The conflict probability generated by the superposition of the three is predicted. The risk index calculation formula is as follows:

[0045] ;

[0046] In the formula, is the risk index, is the weight coefficient, is the conflict risk based on the predicted trajectory overlap probability, is the weather mutation intensity, is the control failure probability caused by signal interference;

[0047] The airspace risk heat map is divided into four heat map layers according to the risk value. For the grids that are not directly monitored, the Kriging spatial interpolation algorithm is used to fill the risk value combined with historical similar scene data.

[0048] As a preferred technical solution, in the step S6, when calculating the airspace grid conflict warning, the aircraft ADSB / radar data is received. The trajectory points are sampled at 1 second intervals. The three-dimensional coordinates are mapped to the target grid through the Beidou grid location code. The speed vector, heading angle, and acceleration features are extracted. A space-time window is constructed. The trajectory sequence is divided according to the sliding window. The grid-level space-time trajectory matrix is constructed. According to the historical data and real-time state, the random variable distribution of Monte Carlo simulation is defined. For each Monte Carlo sample, the space-time overlap condition of the aircraft trajectory in the grid is calculated. The conflict event frequency is counted. The kernel density estimation is used to generate the grid-level conflict probability surface. The bias of small probability events is corrected through importance sampling. According to the probability value of the conflict, the conflict warning level is divided.

[0049] As a preferred technical solution, in the step S7, the specific formula for generating a multi-objective optimal path set with grid navigation cost as the weight is as follows:

[0050] ;

[0051] In the formula, is a flight path planning cost of the aircraft, is a flight range cost of the aircraft, is an aircraft avoidance cost, is a flight range height cost of the aircraft, is a flight turning cost of the aircraft, are weight proportions of the flight path planning cost, the flight range cost, the avoidance cost and the flight range height cost of the aircraft respectively.

[0052] As a preferred technical solution, the flight range cost of the aircraft is calculated according to the following formula:

[0053] ;

[0054] In the formula, is an electric quantity consumption of the aircraft per unit flight range, is a maximum rated battery capacity of the aircraft, is a length of a flight grid i of the aircraft to an adjacent grid i+1, is a set of all grids flown through by the aircraft, ;

[0055] The flight avoidance cost of the aircraft is calculated according to the following formula:

[0056] ;

[0057] In the formula, represents a cost of infinite collision with an obstacle, represents a radius of an obstacle in a jth grid area, represents a Euclidean distance between the current grid i of the aircraft and the obstacle;

[0058] The flight range height cost of the aircraft is calculated according to the following formula:

[0059] ;

[0060] In the formula, represents a height value of the aircraft in the current grid i, represents a maximum value of a flight height of the aircraft, represents a minimum value of the flight height of the aircraft;

[0061] The flight turning cost of the aircraft is calculated according to the following formula:

[0062] ;

[0063] wherein, denotes the maximum turning angle of the aircraft, denotes the turning angle of the current grid i of the aircraft, denotes the vector of the i-th grid, denotes the length of the vector .

[0064] The present application has the following beneficial effects:

[0065] (1) The present application combines real-time operation data and meteorological environment data of low-altitude aircraft with artificial intelligence algorithms and navigation rule library, analyzes airspace traffic flow, optimizes flight plan, intelligently allocates flight route, realizes automatic setting of any two-point route, and improves the response speed of low-altitude flight service system and the processing capacity of various flight data.

[0066] (2) The real scene three-dimensional of the present application provides high-precision three-dimensional space data basis for digital twin model by collecting real geographic information of the target area. Real scene three-dimensional data and low-altitude elements are deeply integrated to realize the visualization display of low-altitude road network, no-fly area and controlled airspace. Through real-time rendering technology, the accurate presentation of aircraft position, airspace conflict and other dynamic information is supported to provide intuitive basis for monitoring and early warning.

[0067] (3) The present application constructs a grid airspace model through real scene three-dimensional data, supports airspace safety analysis, landing site selection and other scenes, discretizes continuous airspace through grid coding technology, simplifies the complexity of massive data processing, and improves the efficiency of airspace resource allocation and conflict detection.

[0068] (4) The present application adopts multi-source heterogeneous data association, realizes the spatial strong association of aircraft trajectory, meteorological and electromagnetic data through Beidou grid code, and solves the data island problem in traditional airspace database.

[0069] (5) The present application can perform dynamic compression storage through the constructed airspace environment spatio-temporal database, improve the data compression efficiency, set local update mechanism, make the database write delay reach millisecond level, and meet the real-time demand of low-altitude dense flight scene.

[0070] Of course, any product implementing the present application does not necessarily need to achieve all the advantages described above at the same time. BRIEF DESCRIPTION OF DRAWINGS

[0071] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed for the description of the embodiments will be briefly introduced as follows. Obviously, the drawings described below are only some of the embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0072] Figure 1 A structure schematic diagram of a low-altitude comprehensive management system based on a grid digital twin model and an intelligent algorithm according to the present application;

[0073] Figure 2 A structure schematic diagram of a low-altitude service platform;

[0074] Figure 3 A data transmission timing diagram of the unmanned aerial vehicle in Embodiment 2. DETAILED DESCRIPTION

[0075] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0076] In addition, the technical features involved in each of the embodiments of the present application described below can be combined with each other as long as there is no conflict between them.

[0077] In order to make the purpose, technical solutions and advantages of the present application more clear, the following will be combined with the drawings of the present application to further specifically describe the embodiments of the present application. Figure 1 The embodiments of the present application will be further described in detail.

[0078] Before introducing the embodiments of the present application, first, the grid digital twin model will be described.

[0079] In data processing and spatial information representation, a grid digital twin model refers to a set of data associated with spatial positions, usually organized in the form of a spatial grid. Each grid cell can represent a spatial unit (such as a pixel or a square), and the data in these grid cells can contain different types of information.

[0080] A twin model is a technology based on virtual simulation, which connects the actual physical entity with its digital twin representation. It collects and integrates device operation data, sensor data and environmental data to simulate the behavior and state of the device in real time. The digital twin model can achieve accurate monitoring, analysis and optimization of the device by establishing a virtual copy of the device.

[0081] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application 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 only used to explain the present application and not used to limit the present application. Figure 1 The present application 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 only used to explain the present application and not used to limit the present application.

[0082] Embodiment one

[0083] Please refer to Figures 1-2 The present application is a low-altitude comprehensive management system based on a grid digital twin model and an intelligent algorithm, which includes a device terminal and an infrastructure layer, a data and service support layer, and an application layer.

[0084] The device terminal and infrastructure layer includes device terminals and infrastructure; the device terminal is a core hard core component responsible for communication and data processing in the airborne system, and the main role is to realize the two-way information interaction between the device terminal and the ground control station, other devices or network through the data link, specifically including low-altitude aircraft, airborne communication equipment, multi-source navigation equipment, airborne perception and identification equipment and intelligent flight path equipment; the low-altitude aircraft includes unmanned aerial vehicles, eVTOL aircraft and flying cars; the airborne communication equipment includes private network terminals, satellite communication equipment, airborne ad hoc networks and airborne data / video transmission; the multi-source navigation equipment includes inertial navigation, visual navigation, satellite navigation and altimeter; the altimeter is a barometric altimeter or an ultrasonic altimeter; the barometric altimeter is used to measure atmospheric pressure, thereby calculating the relative height; the ultrasonic altimeter is used to measure the height of the aircraft relative to the ground or sea level by ultrasonic waves; the airborne perception and identification equipment includes identity recognition equipment, visual perception equipment and airborne traffic radar; the intelligent flight path equipment includes intelligent flight tube modules, autonomous avoidance modules and airborne electronic fences; the intelligent flight tube module is a flight control system, which is responsible for low-altitude aircraft flight attitude control, task execution and safety guarantee; wherein, the flight attitude control monitors the roll, pitch and yaw angles of the unmanned aerial vehicle in real time through sensors such as accelerometers and gyroscopes, dynamically adjusts the motor speed to keep the flight stable, the task execution is used to analyze and execute operation instructions (such as take-off, landing and flight path), and coordinates the power components to accurately complete the action and safety guarantee is used to monitor parameters such as battery power and signal strength, and trigger automatic return or emergency landing to deal with abnormal situations; the autonomous avoidance module realizes dynamic obstacle avoidance through environmental perception and intelligent algorithm, including environmental perception, dynamic avoidance and cooperative avoidance; environmental perception combines sensors such as laser radar, binocular vision and millimeter wave radar to detect the distance, shape and motion state of obstacles in real time; dynamic avoidance generates real-time obstacle avoidance path based on AI algorithm, supports detouring or hovering in complex scenarios, and cooperative avoidance shares airspace information with surrounding unmanned aerial vehicles and unmanned vehicles through the low-altitude control platform to optimize the global obstacle avoidance strategy; the airborne electronic fence limits the activity range of the unmanned aerial vehicle through virtual boundaries to prevent illegal flight; it is mainly based on GPS or Wi-Fi positioning technology to demarcate no-fly zones (such as airports and sensitive areas), and triggers an alarm or forces the unmanned aerial vehicle to hover when it approaches the boundary;

[0085] The infrastructure includes low-altitude communication facilities, low-altitude navigation facilities, low-altitude monitoring facilities and support facilities; the low-altitude communication facilities include mobile public networks, satellite communication networks and low-altitude communication private networks; the low-altitude navigation facilities include network / area RTK, ground / satellite-based facilities and navigation monitoring equipment; the low-altitude monitoring facilities include identity recognition receiving equipment, integrated sensing equipment, low-altitude radar, spectrum detection equipment and photoelectric / infrared detection equipment; the support facilities include low-altitude intelligence facilities, low-altitude meteorological facilities, take-off and landing facilities, computing power facilities and low-altitude countermeasure facilities;

[0086] The data and service support layer includes a communication access module, a navigation access module, a monitoring access module and a peripheral device access module, which are used to realize communication, navigation, monitoring and information service through a data exchange network; the communication access module is used to integrate C-band (wideband data transmission), UHF-band (remote control / telemetry) and Ka-band satellite communication links, support line-of-sight and beyond line-of-sight communication, realize real-time data backhaul and remote control through a 4G / 5G cellular network; the transmitted data includes geographic information data, low-altitude thematic data and CIM data, etc.; the navigation access module is compatible with satellite navigation systems such as Beidou, GPS and Galileo, and realizes high-precision positioning; the monitoring access module obtains real-time position and height information of the unmanned aerial vehicle through a ground radar station, and actively broadcasts its own state (position, speed, etc.) for monitoring by the air traffic control system; the peripheral device access module is a standard interface, such as a CAN bus, a USB and an Ethernet interface, which is used to connect task loads (cameras, infrared sensors, etc.) and external devices (weather meters, laser radars);

[0087] The application layer includes an application layer in which a low-altitude service platform is deployed; the low-altitude service platform includes a registration and authentication unit, a low-altitude traffic service, a low-altitude traffic flow management unit, a low-altitude airspace management unit, a low-altitude data management unit and an application system interfacing unit; the registration and authentication unit includes an unmanned aerial vehicle registration module, an operator management module, a take-off and landing point management and a human operation management; the low-altitude traffic service includes a control service, a coordination service and an information service; the low-altitude traffic flow management unit includes low-altitude traffic rule management, low-altitude aerial vehicle management, low-altitude navigation management and low-altitude air route management; the low-altitude airspace management includes airspace division management, airspace flow management, electronic fence management, air route planning management, three-dimensional grid management of airspace and terrain and ground obstacle management; the low-altitude data management includes ADS-B management data management and flight information data; the application system interfacing unit includes a system management module, a system configuration module, a unified authentication module and an application data interface module.

[0088] The registration and authentication unit is used to receive registration information submitted by a user of an aerial vehicle or an operator of an aerial vehicle, and the registration information at least contains the following contents:

[0089] a) name of the user, valid certificate number (such as ID number, passport number, etc.), mobile phone, email, purpose of use, flight license information, etc.;

[0090] b) the enterprise and institution user needs to additionally fill in the name of the unit, the unified social credit code or the organization code, etc.;

[0091] c) for special users (such as rural cooperatives, etc.), a separate application needs to be submitted to the civil aviation administration to obtain a special code.

[0092] The aircraft cloud system shall have the ability to verify the identity of the user or operator when the aircraft is registered for use under a person operation license, which shall include identity verification of name, valid certificate number, mobile phone, flight license information, etc.

[0093] The airborne electronic fence configuration can be divided into the following three types according to its horizontal projection geometry:

[0094] The longitude and latitude coordinate points used by the UAV fence are WGS-84 coordinates, and the units of longitude and latitude are degrees; the northern latitude is positive, the southern latitude is negative, the eastern longitude is positive, and the western longitude is negative.

[0095] The workflow of the low-altitude airspace management unit is as follows:

[0096] Step S1: Multi-source perception data fusion and grid coding, constructing airspace environment spatio-temporal database;

[0097] Step S2: Three-dimensional solid subdivision of airspace, giving each grid a unique code, and establishing a dynamic mapping relationship between airspace resources and grid codes;

[0098] Step S3: Generate multi-scale grid twin, associate corresponding attributes to each grid according to management needs;

[0099] Step S4: Use time series prediction model to simulate the coupling effect of aircraft trajectory, weather changes and signal interference in the grid, and generate airspace risk heat map for the next 5-30 minutes;

[0100] Step S5: Establish grid state change trigger rules, when the aircraft crosses the grid boundary or the environmental parameters in the grid exceed the threshold, automatically trigger local update of the twin model;

[0101] Step S6: Real-time calculation of the spatio-temporal intersection probability of the aircraft trajectory in the grid, combined with Monte Carlo simulation to generate conflict warning levels;

[0102] Step S7: Generate a set of multi-objective optimal paths with grid navigation cost as the weight.

[0103] In step S1, the airspace environment spatio-temporal database is constructed by collecting aircraft position, speed, weather parameters and electromagnetic spectrum data from device terminals and infrastructure layers, and the specific construction method is as follows:

[0104] Step S11: Obtain the ID, speed, heading, grid code, and three-dimensional coordinates of the aircraft through millimeter wave radar, ADSB broadcast receiving station, and Beidou grid code positioning terminal; wherein the detection accuracy of the millimeter wave radar reaches 0.1 meter level; the ADSB broadcast receiving station is used to capture the ID, speed, and heading of the aircraft; the Beidou grid code positioning terminal outputs the grid code and three-dimensional coordinates in real time;

[0105] Step S12: Collect wind speed, temperature and humidity, and visibility through a multispectral weather sensor; a laser radar LiDAR can also be used to construct a three-dimensional atmospheric turbulence model;

[0106] The multi-source data can also include electromagnetic sensing, such as monitoring communication frequency band occupancy and interference signals using a wide frequency spectrum analyzer;

[0107] The obtained multi-source data is timestamped, coordinated (WGS84+Beidou grid code), and data formatted (JSON-LD semantic description) through the IEEE 1851-2025 airspace data interface protocol, facilitating data processing;

[0108] Step S13: Adopt federated Kalman filtering algorithm to time synchronize and spatially register radar point cloud, ADSB data, and weather data, eliminate clock bias and coordinate system difference between devices, achieve spatial strong correlation of aircraft trajectory, weather, and electromagnetic data through Beidou grid code, and solve the data island problem in traditional airspace database;

[0109] Step S14: For data detected as abnormal, difference completion is performed in combination with historical data trend prediction to ensure data integrity; abnormal data can be detected using the isolation forest algorithm in many cases, such as radar false alarm and weather sensor failure;

[0110] Step S15: Complete the construction of the airspace environment spatio-temporal database; after the construction of the airspace environment spatio-temporal database is completed, a spatio-temporal index can be created to improve query efficiency, such as creating a hybrid index of quad-tree (Quad-Tree) and R-tree (R-Tree) to support fast retrieval of grid data within a specific spatio-temporal range; the query conditions can be one or more of "2025-03-06, 14:00 to 14:30, altitude 0-500 meters, and conflict risk > 70% grid".

[0111] The airspace environment spatio-temporal database adopts a hierarchical storage architecture, which is divided into raw data layer, fused data layer, and business data layer, and the specific storage is as follows:

[0112] ‌Raw data layer: store raw sensor data (retention period ≤24 hours) based on Apache Kafka real-time stream processing platform;

[0113] ‌Fusion data layer: use the time series database InfluxDB to store the cleaned spatio-temporal correlation data (time series compression rate ≥80%);

[0114] ‌Business data layer: use PostgreSQL+PostGIS to store the spatial management rules and dynamic state of the grid encoding.

[0115] The spatio-temporal data of the airspace environment is also configured with an update strategy: it can be updated according to events, such as when an aircraft enters / leaves a grid, a meteorological parameter mutates (such as wind speed changes >5m / s), only the affected grid data is updated; or set to periodic trigger update, such as checking the data consistency every 5 minutes, repairing the local state deviation caused by network delay, this local update mechanism reduces the database write delay from minutes to milliseconds, meeting the real-time requirements of low-altitude dense flight scenarios‌.‌‌‌

[0116] In step S2, when the airspace is three-dimensionally dissected, the Beidou grid position code standard (BGC standard) is used to divide the airspace into a three-dimensional network structure with adjustable edge length (national level 1km³, city level 100m³, facility level 1m³), each grid includes static attributes and dynamic attributes;

[0117] The static attributes include the grid unique code, the coding rule such as BGCN-31-012-05-001, the airspace regulation rules of the grid area, such as whether to fly, whether to limit height and communication frequency band limitation, and the elevation data of the building obstacles in the grid area;

[0118] The dynamic attributes include the number of real-time aircraft in the grid, the meteorological risk index of the grid (0-100), and the electromagnetic interference intensity of the grid (dBm).

[0119] In step S3, the multi-scale grid twin generation process is as follows:

[0120] Step S31, spatio-temporal standardization: unify the time and space coding of multi-source data, align the time stamps of different devices to UTC standard time based on Beidou grid position code to establish a grid coordinate system‌;

[0121] Step S32: extract grid-related features locally, and protect sensitive data through differential privacy technology‌;

[0122] In constructing the federated learning model, the horizontal federated learning and the vertical federated learning are implemented; wherein, the horizontal federated learning is directed to cross-regional data of the same grid level (such as different city-level hundred-meter grids), each node trains a local model (such as an LSTM trajectory prediction model), and a central server aggregates model parameters to generate a global grid state inference model; the vertical federated learning is directed to multi-level grid data (such as national and city-level grids), and through homomorphic encryption, cross-level data samples are aligned, for example, the traffic statistical features of the national grid are trained jointly with the real-time dynamic features of the city-level grid;

[0123] Step S33: The dynamic attributes of different grid levels (such as the electromagnetic interference intensity of the facility-level sub-meter grid) and the static attributes (such as the airspace rules of the national grid) are fused into a unified tensor, the federated aggregation weight is designed according to the importance of the grid level, and the attention mechanism is used to dynamically adjust the data contribution of different levels of grids;

[0124] In order to improve the safety of the unmanned aerial vehicle calculation transmission process, an encryption transmission protocol can also be implemented in the transmission process, such as using the Paillier homomorphic encryption algorithm to transmit gradient parameters, to ensure that sensitive information such as grid encoding and aircraft ID cannot be reversely analyzed in the transmission process; at the same time, an edge node is deployed at the regional grid to preprocess high-frequency data (such as 10Hz radar point cloud), and only a lightweight feature vector (such as the number of targets in the grid) is uploaded, thereby reducing the communication overhead;

[0125] Step S34: According to the grid level mapping mechanism, the multi-source data is mapped to the grid level;

[0126] The grid levels are divided into:

[0127] Spatial subdivision level: the Beidou grid code adopts a multi-scale earth subdivision system, which is divided into 32 levels of grids, and the smallest unit can reach a cube precision of 1.5 centimeters. The specific division standards include:

[0128] First-level grid: 4° latitude x 6° longitude;

[0129] Five-level grid: 4 seconds (about 120 meters);

[0130] Six-level grid: 2 seconds (about 60 meters);

[0131] Ten-level grid: 1 / 2048 seconds (about 1.5 centimeters).

[0132] Height dimension division: the height direction is divided into multiple levels according to the logarithmic law, from the ground to 500,000 kilometers high, supporting accurate identification and management of the flight height of unmanned aerial vehicles.

[0133] Step S35: coupling the multi-level model to predict the cross-regional flight flow distribution; predicting the cross-regional flight flow distribution by predicting the cross-regional flight flow distribution based on a graph neural network (GNN), or using a physical engine (such as NVIDIA PhysX) to simulate the interaction effect of the unmanned aerial vehicle and the building turbulence to predict the collision probability;

[0134] Step S36: lightening the model and merging the grids in the low-risk area; in order to merge the grids in the low-risk area (such as the suburban airspace) and reduce the model rendering load, 100 1km³ grids are merged into 1 10km³ unit;

[0135] Step S37: setting an incremental update strategy and performing virtual-real synchronization verification; when setting the incremental update strategy, when the aircraft enters the grid or the weather changes suddenly, only the twin of the affected grid is recalculated or the model is checked with the real environment every 30 minutes, and the differential evolution algorithm is used to correct the parameter deviation; when performing virtual-real synchronization verification, a virtual test environment is constructed, simulated data is injected, the response accuracy of the twin is verified, key grid state change events are recorded, and data traceability is ensured.

[0136] In step S4, a time series prediction model is used to simulate the mapping of the aircraft trajectory, weather changes, and signal interference in the grid to the three-dimensional grid, the timestamp difference and coordinate system offset are eliminated by federated Kalman filtering, and the features of the aircraft trajectory, weather changes, and signal interference are extracted respectively; the aircraft trajectory extracts the speed change rate and heading angle deviation of the aircraft in the grid as dynamic features; the weather parameters calculate the wind speed gradient and visibility attenuation rate as weather mutation indicators; the signal interference calculates the communication frequency band occupancy rate and signal-to-noise ratio degradation trend in the grid;

[0137] A multi-modal time series prediction model is constructed, a cross-attention mechanism is introduced, the weights of the aircraft trajectory, weather, and electromagnetic interference are dynamically adjusted, the conflict probability generated by the superposition of the three is predicted, and the risk index calculation formula is as follows:

[0138] ;

[0139] In the formula, is the risk index, is the weight coefficient, is the conflict risk based on the predicted trajectory overlap probability, is the weather mutation intensity, is the control failure probability caused by signal interference;

[0140] The airspace risk heat map is divided into four levels of heat map layers according to the risk value, and the four-level distribution indexes are as follows: green: R<30; yellow: 30≤R<60; orange: 60≤R<90; red: R≥90, and the Kriging spatial interpolation algorithm is used to fill in the risk value in combination with the historical similar scene data for the grid not directly monitored; an incremental prediction mechanism can also be used: the model parameters are updated in full every 5 minutes, and local re-prediction is carried out based on event triggering (such as a new aircraft entering the grid) with a delay control within 200 ms; when performing virtual closed-loop verification, simulated emergencies (such as thunderstorm generation and GPS spoofing attack) are injected in the digital twin sandbox, the error rate of the predicted heat map compared with the actual evolution result is compared, and the model weight is continuously optimized (target error rate <5%).

[0141] In step S6, when the airspace grid conflict warning is calculated, the ADSB / radar data of the receiving aircraft is sampled at an interval of 1 second, the three-dimensional coordinates are mapped to the target grid through the Beidou grid location code, the speed vector, heading angle and acceleration characteristics are extracted; a space-time window is constructed, a prediction time window of 5-30 minutes in the future is defined, the trajectory sequence is segmented according to the sliding window, a grid-level space-time trajectory matrix is constructed, and the dimension of the matrix adopts time×space×motion parameter; according to the historical data and real-time state, the random variable distribution of Monte Carlo simulation is defined; the random variable distribution includes:

[0142] ‌The Gaussian distribution of the aircraft motion model describes the speed fluctuation and the wind speed of the environmental disturbance factor obeys the Weibull distribution, and the electromagnetic interference intensity is modeled according to the Poisson process;‌

[0143] For each Monte Carlo sample, the space-time overlap condition of the aircraft trajectory in the grid is calculated, the space-time overlap condition takes the meeting time of not more than 30s as the time threshold and 50m as the safety distance; and the conflict event frequency is counted, the kernel density estimation is used to generate the grid-level conflict probability surface, and the bias of small probability events is corrected through importance sampling, and the conflict warning level is divided according to the probability value of the conflict; the conflict warning levels are as follows:

[0144] Low risk (<0.1%) Green, log only Medium risk (0.1-1%) Yellow alert, trigger manual review High risk (>0.1%) Red alert, automatically send reroute instructions to affected aircraft

[0145] In step S7, the specific formula of the multi-objective optimal path set is as follows:

[0146] ;

[0147] In the formula, is the flight path planning cost of the aircraft, is the flight range cost of the aircraft, is the flight avoidance cost of the aircraft, is the flight range height cost of the aircraft, a turning cost of the aerial vehicle, respectively, are weight proportions of a route path planning cost, a flight distance cost, an avoidance cost and a flight height cost of the aerial vehicle.

[0148] a flight distance cost of the aerial vehicle is calculated by the following formula:

[0149] ;

[0150] wherein, is an electric quantity consumption of the aerial vehicle per unit flight distance, is a maximum rated battery capacity of the aerial vehicle, is a length of a grid i to an adjacent grid i+1 of the aerial vehicle, is a set of all grids through which the aerial vehicle flies, ;

[0151] an avoidance cost of the aerial vehicle is calculated by the following formula:

[0152] ;

[0153] wherein, represents a cost of collision with an obstacle being infinite, represents a radius of an obstacle in a jth grid region, represents a Euclidean distance between a current grid i of the aerial vehicle and the obstacle;

[0154] a flight height cost of the aerial vehicle is calculated by the following formula:

[0155] ;

[0156] wherein, represents a height value of the aerial vehicle in a current grid i, represents a maximum value of a flight height of the aerial vehicle, represents a minimum value of the flight height of the aerial vehicle;

[0157] a turning cost of the aerial vehicle is calculated by the following formula:

[0158] ;

[0159] wherein, represents a maximum turning angle of the aerial vehicle, represents a turning angle of the aerial vehicle in a current grid i, represents a vector of an i th grid, represents a length of the vector .

[0160] Example two

[0161] Please refer to Figure 3 The embodiment takes a drone as an example:

[0162] The registration information is the information of the drone related identity transmitted by the drone system to the drone cloud system and the code generated by the drone cloud system for the drone. It should at least include:

[0163] Product serial number (MSN): The factory number of the drone system defined by each drone manufacturer, composed of multiple characters. The product serial number is required to be consistent with the real-name authentication system registration;

[0164] Flight control system serial number (FCSN): The flight control system number defined by each manufacturer, composed of multiple characters;

[0165] Nationality registration mark or real-name registration code (REG): The registration number is determined by the airworthiness department;

[0166] The number generated by the drone cloud operator (hereinafter referred to as the operator) in the cloud system for the drone (CPN): composed of multiple characters, only the first six characters are required, the first three characters in the six characters are the operator code, and the fourth to sixth characters represent the category of the drone. The fourth character represents the category in the drone operation and management category, that is, I~Ⅶ.

[0167] In order to improve the efficiency of data transmission, when the amount of data to be transmitted is large, the data items that are different from the previous data packet are provided according to the time unit of update under the condition of meeting the data update rate requirement. According to the characteristics of dynamic information and static information, the difference data only includes but not completely includes the content of dynamic information in the interface transmission data.

[0168] The drone cloud system should have wired or wireless communication function, and the data transmitted between the drone cloud system and the drone system should include the following contents:

[0169] Instruction one, code: MAYDAY MAYDAY MAYDAY, indicating that the drone receiving the instruction needs to land immediately in the specified area;

[0170] Instruction two, code: PAN PAN PAN PAN PAN PAN, indicating that the drone receiving the instruction should leave the specified area within ten minutes, and those who cannot leave should complete the return and prepare for landing;

[0171] Instruction three, code: CLEAN CLEAN CLEAN, indicating that the drone receiving the instruction should leave the specified area within half an hour, and those who cannot leave should complete the return and prepare for landing;

[0172] Instructions four and five are backup instructions.

[0173] ‌It is worth noting that in the above system embodiments, each unit included is only divided according to functional logic, but is not limited to the above division, as long as the corresponding function can be realized; in addition, the specific name of each functional unit is only for the convenience of mutual differentiation, and is not used to limit the protection scope of the present application. In addition, those skilled in the art can understand that all or part of the steps in the above-mentioned embodiment methods can be completed by programs instructing relevant hardware, and the corresponding programs can be stored in a computer readable storage medium. The preferred embodiments of the application disclosed above are only used to help explain the application. The preferred embodiments do not describe all the details and do not limit the application to the specific embodiments described. Obviously, many modifications and changes can be made according to the content of the present application. The present application selects and describes these embodiments in order to better explain the principles and practical applications of the present application, so that those skilled in the art can well understand and utilize the present application. The present application is limited only by the claims and their full scope and equivalents.

Claims

1. A low-altitude comprehensive management system based on a grid digital twin model and an intelligent algorithm, characterized in that, This includes the device terminal and infrastructure layer, the data and service support layer, and the application layer; The device terminal and infrastructure layer includes device terminals and infrastructure; the device terminals include low-altitude aircraft, airborne communication equipment, multi-source navigation equipment, airborne sensing and identification equipment, and intelligent route equipment; the low-altitude aircraft include unmanned aerial vehicles (UAVs), eVTOL aircraft, and flying cars; the airborne communication equipment includes private network terminals, satellite communication equipment, airborne ad hoc networks, and airborne data / image transmission; the multi-source navigation equipment includes inertial navigation, visual navigation, satellite navigation, and altimeters; the airborne sensing and identification equipment includes identity recognition equipment, visual sensing equipment, and airborne traffic radar; the intelligent route equipment includes intelligent... The system includes a flight control module, an autonomous obstacle avoidance module, and an airborne electronic fence; the infrastructure includes low-altitude communication facilities, low-altitude navigation facilities, low-altitude surveillance facilities, and support facilities; the low-altitude communication facilities include public mobile networks, satellite communication networks, and private low-altitude communication networks; the low-altitude navigation facilities include network / regional RTK, ground-based / satellite-based facilities, and navigation and surveillance equipment; the low-altitude surveillance facilities include identification receiving equipment, integrated sensing equipment, low-altitude radar, spectrum detection equipment, and photoelectric / infrared detection equipment; the support facilities include low-altitude intelligence facilities, low-altitude meteorological facilities, take-off and landing facilities, computing power facilities, and low-altitude countermeasure facilities. The data and service support layer includes a communication access module, a navigation access module, a monitoring access module, and a peripheral device access module, which are used to realize communication, navigation, monitoring, and information services through a data exchange network; The application layer is equipped with a low-altitude service platform; the low-altitude service platform includes a registration and authentication unit, a low-altitude traffic service, a low-altitude traffic flow management unit, a low-altitude airspace management unit, a low-altitude data management unit, and an application system interface unit; The registration and authentication unit includes a drone registration module, an operator management module, a take-off and landing point management module, and an operator management module; the low-altitude traffic service includes control services, coordination services, and information services; the low-altitude traffic flow management unit includes low-altitude traffic rule management, low-altitude aircraft management, low-altitude navigation management, and low-altitude route management; the low-altitude airspace management includes airspace allocation management, airspace flow management, electronic fence management, route planning management, airspace 3D grid management, and terrain and ground obstacle management; the low-altitude data management includes ADS-B management data management and flight information data; the application system interface unit includes a system management module, a system configuration module, a unified authentication module, and an application data interface module. The workflow of the low-altitude airspace management unit is as follows: Step S1: Multi-source sensing data fusion and grid coding to construct a spatiotemporal database of the airspace environment; Step S2: Perform three-dimensional partitioning of the airspace, assign a unique code to each grid, and establish a dynamic mapping relationship between airspace resources and grid codes; Step S3: Generate a multi-scale mesh twin, and associate the corresponding attributes of each mesh level according to management requirements; Step S4: Use a time series prediction model to simulate the coupling effect of aircraft trajectory, weather changes, and signal interference within the grid, and generate an airspace risk heat map for the next 5-30 minutes; Step S5: Establish grid state change triggering rules. When the aircraft crosses the grid boundary or the environmental parameters within the grid exceed the threshold, the twin model is automatically updated locally. Step S6: Calculate the spatiotemporal intersection probability of the aircraft trajectories within the grid in real time, and generate a conflict warning level by combining Monte Carlo simulation; Step S7: Generate a set of optimal paths for multiple objectives, using grid navigation costs as weights.

2. The low-altitude comprehensive management system based on the grid digital twin model and intelligent algorithm according to claim 1, characterized in that, In step S1, a spatiotemporal database of the airspace environment is constructed by collecting aircraft position, speed, meteorological parameters, and electromagnetic spectrum data from the device terminal and infrastructure layer. The specific construction method is as follows: Step S11: Obtain the aircraft's ID, speed, heading, grid code, and three-dimensional coordinates through millimeter-wave radar, ADSB broadcast receiving station, and Beidou grid code positioning terminal; Step S12: Collect wind speed, temperature, humidity, and visibility data using a multispectral meteorological sensor; Step S13: Use the federated Kalman filter algorithm to perform time synchronization and spatial registration of radar point cloud, ADSB data, and meteorological data; Step S14: For the data that detects anomalies, perform interpolation by combining historical data trend predictions; Step S15: Complete the construction of the spatial-temporal database of the airspace environment.

3. The low-altitude comprehensive management system based on the grid digital twin model and intelligent algorithm according to claim 2, characterized in that, In step S2, when performing three-dimensional segmentation of the airspace, the BeiDou grid position code standard is adopted to divide the airspace into a three-dimensional network structure with adjustable side lengths. Each grid includes static and dynamic attributes. The static attributes include the grid's unique code, the airspace control rules for the grid area, and the elevation data of building obstacles within the grid area; The dynamic attributes include the real-time number of aircraft in the grid, the weather risk index of the grid, and the electromagnetic interference intensity of the grid.

4. The low-altitude integrated management system based on a grid digital twin model and intelligent algorithm according to claim 3, characterized in that, In step S3, the multi-scale mesh twin generation process is as follows: Step S31: Perform unified spatiotemporal coding on multi-source data, establish a grid coordinate system based on the BeiDou grid location code, and align the timestamps of different devices to UTC standard time; Step S32: Extract grid-related features locally and protect sensitive data using differential privacy technology; Step S33: Merge the dynamic and static attributes of different grid levels into a unified tensor, design federated aggregation weights according to the importance of grid levels, and use an attention mechanism to dynamically adjust the data contribution of different grid levels; Step S34: Map the multi-source data to the grid level according to the grid level mapping mechanism; Step S35: Couple the multi-level model to predict cross-regional flight traffic distribution; Step S36: Perform lightweight processing on the model and merge the meshes in low-risk areas; Step S37: Set the incremental update strategy and perform virtual-real synchronization verification.

5. The low-altitude integrated management system based on a grid digital twin model and intelligent algorithm according to claim 4, characterized in that, In step S4, a time series prediction model is used to simulate the unified mapping of aircraft trajectory, weather changes, and signal interference within the grid to a three-dimensional grid. Federated Kalman filtering is used to eliminate timestamp differences and coordinate system offsets. Feature extraction is performed on aircraft trajectory, weather changes, and signal interference respectively to construct a multimodal time series prediction model. A cross-attention mechanism is introduced to dynamically adjust the weights of aircraft trajectory, weather, and electromagnetic interference, predicting the probability of conflict caused by their superposition. The risk index calculation formula is as follows: ； In the formula, As a risk index, These are the weighting coefficients. To assess the conflict risk based on the predicted trajectory overlap probability, The intensity of meteorological changes, The probability of control failure caused by signal interference; The airspace risk heat map is divided into four levels of heat map layers according to the risk value, and for grids that are not directly monitored, the risk value is filled in by the Kriging spatial interpolation algorithm combined with historical similar scene data.

6. The low-altitude integrated management system based on a grid digital twin model and intelligent algorithm according to claim 5, characterized in that, The multimodal time series prediction model adopts an LSTM-Transformer hybrid network. In the LSTM-Transformer hybrid network, the bottom layer LSTM processes the aircraft trajectory sequence, the middle layer Transformer captures the long-term dependence of meteorological parameters, and the top layer graph convolutional network models the signal interference propagation effect between grids.

7. The low-altitude integrated management system based on a grid digital twin model and intelligent algorithm according to claim 6, characterized in that, In step S6, when calculating the airspace grid conflict early warning, the aircraft's ADSB / radar data is received, trajectory points are sampled at 1-second intervals, and the three-dimensional coordinates are mapped to the target grid through the BeiDou grid position code to extract velocity vector, heading angle and acceleration features. A spatiotemporal window is constructed, and the trajectory sequence is divided by a sliding window to construct a grid-level spatiotemporal trajectory matrix. Based on historical data and real-time status, the distribution of random variables in Monte Carlo simulation is defined. For each Monte Carlo sample, the spatiotemporal overlap condition of the aircraft trajectory within the grid is calculated, and the frequency of conflict events is counted. A grid-level conflict probability surface is generated using kernel density estimation, and the bias of low-probability events is corrected by importance sampling. The conflict warning level is divided according to the probability value of the conflict.

8. The low-altitude integrated management system based on a grid digital twin model and intelligent algorithm according to claim 7, characterized in that, In step S7, the specific formula for generating the multi-objective optimal path set using grid navigation cost as the weight is as follows: ； In the formula, Cost of aircraft route planning For the cost of aircraft range, To cover the cost of aircraft avoidance, Cost based on aircraft range and altitude, For the cost of aircraft turning, These are the weighting ratios of the aircraft's route planning cost, range cost, obstacle avoidance cost, and range-altitude cost, respectively.

9. The low-altitude integrated management system based on a grid digital twin model and intelligent algorithm according to claim 8, characterized in that, The flight cost of the aircraft The calculation formula is as follows: ； In the formula, This refers to the power consumption per unit distance of the aircraft. This is the maximum rated battery capacity of the aircraft. Let i be the length from the spacecraft's navigation grid i to its adjacent grid i+1. The set of all grids traversed by the aircraft during its flight. ; The aircraft avoidance cost The calculation formula is: ； In the formula, This indicates that the cost of colliding with an obstacle is infinite. This represents the radius of the obstacle within the j-th grid region. This represents the Euclidean distance between the current grid i of the aircraft and the obstacle; The cost of the aircraft's range and altitude The calculation formula is: ； In the formula, This represents the altitude value of the aircraft in the current grid i. This indicates the highest possible altitude at which the aircraft is flying. This indicates the minimum altitude at which the aircraft is flying. The aircraft turning cost The calculation formula is: ； In the formula, Indicates the aircraft's maximum turning angle. This represents the turning angle of the aircraft at the current grid i. The vector representing the i-th grid. Represents vector The length.

Images