Low-altitude airspace dynamic decision-making system based on multi-modal large model

Abstract

The invention provides a low-altitude airspace dynamic decision-making system based on a multi-modal large model, and relates to the technical field of low-altitude airspace traffic management.The low-altitude airspace dynamic decision-making system is characterized in that a first thermal disturbance coverage map, a second shear wind high-risk area map, a third turbulence risk drainage basin map and a fourth sudden obstacle prediction area map are generated by a map construction module; the aircraft can judge whether the aircraft enters the dangerous area in real time and assess the safety of the flight path according to the superposed risk value. When the superposition risk value # imgabs0 # of the flight path section passing through the jth area in the first original flight path of the ith aircraft is in a middle risk section, the system generates a first-level avoidance strategy, and the height, the speed and the course of the aircraft are adjusted; and when the superposed risk value # imgabs1 # is a high-risk section, the system generates a secondary avoidance strategy, and bypasses a high-risk area by remarkably increasing the flight height and reducing the speed.

Description

Technical Field

[0001] The present invention relates to the technical field of low-altitude airspace traffic management, and specifically to a low-altitude airspace dynamic decision-making system based on a multimodal large model. Background Art

[0002] With the increasingly mature operating environment of the low-altitude airspace, various flight platforms such as unmanned aerial vehicles and low-altitude manned aircraft are widely used in scenarios such as inspection, logistics transportation, emergency response, and environmental monitoring in the near-earth airspace of 120 to 1000 meters. Compared with high-altitude flight routes, the low-altitude flight environment is more complex. During the mission execution, the aircraft faces significant meteorological interference and dynamic environmental changes, posing higher requirements for its safety and the intelligence of path planning.

[0003] In traditional low-altitude flight missions, flight path planning mainly relies on static charts and limited meteorological data, making it difficult to accurately capture and evaluate non-linear meteorological factors such as instantaneous thermal disturbances, shear winds, and sudden turbulence changes in the low altitude. For example: Surface thermal disturbance effect: Due to factors such as the difference in the heat capacity of ground materials and the urban heat island effect, there are strong vertical heat flow disturbances in the low altitude, which are likely to cause fluctuations in the aircraft's altitude and unstable attitude, especially significant in summer or complex terrains.

[0004] Shear wind and sudden turbulence phenomena: In near-earth airspace areas such as mountains, building complexes, and viaducts, aircraft often encounter "shear winds" with rapid changes in wind speed and direction, or local turbulence formed due to terrain interference. These phenomena have a great impact on small aircraft and may lead to flight path deviation or even flight control failure.

[0005] Dynamic targets and sudden obstacles: There are a large number of dynamically changing targets (such as other aircraft, bird flocks) in the low-altitude airspace, and traditional rule-based path planning is difficult to identify and avoid them in a timely manner.

[0006] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure. Therefore, it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0007] The purpose of the present invention is to provide a low-altitude airspace dynamic decision-making system based on a multimodal large model to solve the problems raised in the above background art.

[0008] To achieve the above purpose, the present invention provides the following technical solutions: A low-altitude airspace dynamic decision-making system based on a multimodal large model, comprising: An airspace modeling module, which is used for the flight tasks of multiple aircraft in the low-altitude airspace of 120 to 1000 meters. Taking the i-th aircraft as an example, it establishes a dynamic three-dimensional floating coordinate system for the target operation airspace, synchronously collects meteorological data, remote sensing images, visual images, and navigation data related to the airspace where the i-th aircraft is located, and establishes a multi-source data set; A trajectory analysis module, which is used to collect the first original flight trajectory of the i-th aircraft in real time, establish its first original flight trajectory in the dynamic three-dimensional floating coordinate system, and identify the environmental data of the j-th area in the first original flight trajectory to construct the heat flux of the j-th area , the shear wind intensity , the turbulence intensity and the dynamic target density index and evaluate; A map construction module, which is used to identify whether each trajectory point of the i-th aircraft crosses a risk area in the first original flight trajectory, obtain the path recognition result of the first original flight trajectory, and construct a first heat disturbance coverage map, a second shear wind high-risk area map, a third turbulence risk watershed map, and a fourth sudden obstacle prediction area map; A flight risk superposition module, which constructs the superposition risk value of the j-th area based on the first heat disturbance coverage map, the second shear wind high-risk area map, the third turbulence risk watershed map, and the fourth sudden obstacle prediction area map , and evaluate, , normal passage, when It indicates that the superposition risk is in the medium-risk section, and generates a first-level avoidance strategy; when It indicates that the superposition risk is in the high-risk section, and generates a second-level avoidance strategy.

[0009] Furthermore, the airspace modeling module includes a three-dimensional floating coordinate system construction unit and a multi-source data synchronous acquisition unit; The three-dimensional floating coordinate system construction unit is used for the flight tasks of multiple aircraft in the low-altitude airspace of 120 to 1000 meters. Taking the i-th aircraft as an example, where i ∈ [1, n], and n is the total number of aircraft operating simultaneously; Taking the current position of the i-th aircraft as a reference point, it establishes a dynamic three-dimensional floating coordinate system in real time. The dynamic three-dimensional floating coordinate system is adaptively adjusted according to the heading and attitude of the i-th aircraft to ensure that the dynamic three-dimensional floating coordinate system is aligned with the aircraft task path; The multi-source data synchronous acquisition unit is used to synchronously collect meteorological data, remote sensing images, visual images, and navigation data related to the airspace where the i-th aircraft is located, establish a multi-source data set, and construct and train a three-dimensional dynamic environment model. It uses the time stamp alignment and spatial interpolation methods to perform multi-modal fusion on various types of data to ensure that different data sources have a consistent spatio-temporal reference system in the coordinate domain of the i-th aircraft.

[0010] Furthermore, the trajectory analysis module includes a first trajectory acquisition unit, a ground heat flux identification unit, and a local mutation shear wind field identification unit; The first trajectory acquisition unit is used to collect the position, attitude angle, velocity vector of the i-th aircraft in real time, and the first original flight trajectory; The ground heat flux identification unit is used to deploy heat flux sensors on the ground, collect heat flux data, define the spatial mapping of heat flux perturbation and ground temperature gradient according to the relationship between heat flux and ground temperature, and obtain the heat flux of the j-th region : ; where is the ground temperature of the j-th region, is the ambient temperature between 120 m and 1000 m above the ground in the j-th region, h is the heat transfer coefficient, and the value is taken according to the actual scenario, including: h = 5 - 25 W / m²·K for natural convection; h = 25 - 250 W / m²·K for forced convection under strong wind; h = 10 - 50 W / m²·K for indoor air flow; h = 50 - 500 W / m²·K near the heat source of industrial chimneys; According to the heat flux of the j-th region , judge the abnormal level of heat flux, including: When the heat flux of the j-th region <100 W / m 2 ; it is judged as a normal heat flux region; When 100 W / m 2 ≤ ≤ 300 W / m 2 ; it is judged that there is a heat interference anomaly and marked as a first-order heat perturbation region; When > 300 W / m 2 ; it is judged that there is a heat interference anomaly and marked as a second-order heat perturbation region.

[0011] Furthermore, the local mutation shear wind field identification unit is used to deploy a three-dimensional ultrasonic anemometer group and a laser wind measurement radar device on the ground and in the flight channel, respectively collect the wind speed vector data within 1000 m above the ground in the target area, and obtain the shear wind intensity of the j-th region based on the wind speed difference between each height layer : ; where and are the wind speed vectors of the j-th region at the second height and the first height respectively, and the second height is set to 100 m, and the first height Set to 50m; According to the shear wind intensity of the j-th area , judge the shear wind disturbance level, including: When the shear wind intensity of the j-th area <1.0m / s·100m -1 ; Judge as a normal shear wind intensity area; When 1.0m / s·100m -1 ≤ ≤2.5m / s·100m -1 ; Judge that there is shear wind anomaly and mark it as a first-order shear wind disturbance area; When >2.5m / s·100m -1 ; Judge that there is shear wind anomaly and mark it as a second-order shear wind disturbance area.

[0012] Furthermore, the trajectory analysis module further includes a turbulent area boundary recognition unit and a dynamic target obstacle distribution recognition unit; The turbulent area boundary recognition unit is used to deploy an anemometer device in the target area, collect the instantaneous change data of the wind speed, and obtain the turbulent intensity of the j-th area : ; Wherein, is the standard deviation of the wind speed in the j-th area, reflecting the degree of violent fluctuation, represents the average wind speed in the j-th area; According to the turbulent intensity of the j-th area , judge the turbulent risk level, including: When the turbulent intensity of the j-th area <0.1; Judge as a normal turbulent area; When 0.1≤ ≤0.2; Judge that there is turbulent anomaly and mark it as a first-order turbulent disturbance area; When >0.2; Judge that there is turbulent anomaly and mark it as a second-order turbulent disturbance area.

[0013] Furthermore, the dynamic target obstacle distribution recognition unit is used to collect the trajectory, speed, size and distribution density data of moving targets in the target area by deploying lidar, vision sensors, millimeter wave radar, ADS-B or V2X devices, and construct a real-time distribution map of dynamic obstacles to obtain the dynamic target density index of the j-th area : ; Wherein, is the number density of dynamic obstacles per unit area in the j-th area, with the unit of number / m2 , represents the number of currently detected dynamic obstacles in the j-th area, represents the horizontal projected area of the j-th area; Based on the dynamic target density index of the j-th area , judge the flight obstacle avoidance risk level, including: When the dynamic target density index of the j-th area < 0.1 per m 2 ; It is judged as a normal obstacle avoidance risk area; When 0.1 per m 2 ≤ ≤ 0.5 per m 2 ; Marked as a first-order obstacle avoidance prompt area; When > 0.5 per m 2 ; Marked as a second-order high-density obstacle avoidance warning area.

[0014] Furthermore, the map construction module includes a first original flight trajectory recognition unit and a map generation unit; The first original flight trajectory recognition unit is used to recognize the flight path segment passing through the j-th area in the first original flight trajectory, and obtain the heat flux of the j-th area corresponding to the flight path segment of the j-th area, the shear wind intensity of the j-th area, the turbulence intensity of the j-th area and the dynamic target density index of the j-th area; Judge whether each trajectory point in the flight trajectory crosses the following risk areas, obtain the first original flight trajectory path recognition result, and perform hierarchical marking, including: First-order or second-order thermal disturbance area, first-order or second-order shear wind disturbance area, first-order or second-order turbulence disturbance area, first-order obstacle avoidance prompt area or second-order high-density obstacle avoidance warning area; Construct a first thermal disturbance coverage map, a second shear wind high-risk area map, a third turbulence risk basin map and a fourth sudden obstacle prediction area map based on the first original flight trajectory path recognition result.

[0015] Furthermore, the flight risk superposition module includes a flight risk superposition recognition unit and a hierarchical response unit; The flight risk superposition recognition unit is used to calculate and obtain the superposition risk value of the j-th area according to the first thermal disturbance coverage map, the second shear wind high-risk area map, the third turbulence risk basin map and the fourth sudden obstacle prediction area map through the following formula : ; In the formula, is the heat flux score of the j-th area, the normal heat flux area is 0 points, the first-order thermal disturbance area is 1 point, and the second-order thermal disturbance area is 2 points; is the shear score for the j-th area. The normal shear wind intensity area is scored 0, the first-order shear wind disturbance area is scored 1, and the second-order shear wind disturbance area is scored 2; is the turbulence intensity score for the j-th area. The normal turbulence area is scored 0, the first-order turbulence disturbance area is scored 1, and the second-order turbulence disturbance area is scored 2; is the dynamic target density score for the j-th area. The normal obstacle avoidance risk area is scored 0, the first-order obstacle avoidance prompt area is scored 1, and the second-order high-density obstacle avoidance warning area is scored 2; When < 2, indicating the superimposed risk value of the flight path segment passing through the j-th area in the first original flight trajectory of the i-th aircraft is a low-risk segment, and the visualization of the j-th area is set to green in the dynamic three-dimensional floating coordinate system; When , indicating the superimposed risk value of the flight path segment passing through the j-th area in the first original flight trajectory of the i-th aircraft is a medium-risk segment, and the visualization of the j-th area is set to yellow in the dynamic three-dimensional floating coordinate system; When , indicating the superimposed risk value of the flight path segment passing through the j-th area in the first original flight trajectory of the i-th aircraft is a high-risk segment, and the visualization of the j-th area is set to red in the dynamic three-dimensional floating coordinate system.

[0016] Furthermore, the grading response unit is used to, when the i-th aircraft passes through the flight path segment of the j-th area according to the first original flight trajectory, if it is identified as a low-risk segment, maintain the current flight state; It is used to, when the i-th aircraft passes through the flight path segment of the j-th area according to the first original flight trajectory, if it is identified as a medium-risk segment, generate a first-level avoidance strategy, including: Increase the vertical flight height of the i-th aircraft by 20m, reduce the flight speed by 20 - 30% on the stable wall, including reducing from 15m / s to 8 - 10m / s, and adjust the heading angle by ±5 - 10° to avoid the second-order high-density obstacle avoidance warning area or the second-order turbulence disturbance area; It is used to, when the i-th aircraft passes through the flight path segment of the j-th area according to the first original flight trajectory, if it is identified as a high-risk segment, generate a second-level avoidance strategy, including: Increase the vertical flight altitude of the i-th aircraft by 50 - 80 m, reduce the flight speed by 40 - 50% on the stable wall, including reducing from 15 m / s to 6 - 7 m / s, and adjust the course angle by ±15 - 30° to bypass the second-order high-density obstacle avoidance warning area or the second-order turbulence disturbance area; determine whether there is a continuous band of risk voxels in the area where the current original path segment is located in the high-risk section; if so, call the path reconstruction algorithm to plan a new flight path segment with the minimum risk cost; splice the new path segment into the aircraft trajectory control system, issue path points in real time, and synchronously mark the current path segment as a high-risk passage record area.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: The low-altitude airspace dynamic decision-making system of the present invention collects multi-source data in real time (including meteorological data, remote sensing images, visual images, and navigation data), establishes a dynamic three-dimensional floating coordinate system, and combines the real-time trajectory of the aircraft to accurately identify meteorological factors such as thermal disturbances, shear winds, and turbulence mutations existing in the low-altitude airspace, as well as dynamic targets and sudden obstacles. This accurate identification ability greatly improves the adaptability of the aircraft to complex low-altitude environments, can effectively avoid risk factors that may cause the aircraft's flight path to deviate or the flight control to fail, thereby improving flight safety.

[0018] Through the first thermal disturbance coverage map, the second shear wind high-risk area map, the third turbulence risk basin map, and the fourth sudden obstacle prediction area map generated by the map construction module of the present invention, the aircraft can judge in real time whether it enters a dangerous area and evaluate the safety of the flight path according to the superimposed risk value. When the superimposed risk value of the flight path segment passing through the j-th area in the first original flight trajectory of the i-th aircraft is in the medium-risk section, the system generates a first-level avoidance strategy to adjust the aircraft's altitude, speed, and course; when the superimposed risk value is in the high-risk section, the system generates a second-level avoidance strategy to bypass the high-risk area by significantly increasing the flight altitude and reducing the speed. This intelligent path planning technology can dynamically adapt to the risk changes in the low-altitude environment and achieve the optimal execution of flight tasks.

[0019] The present invention enhances the aircraft's autonomous decision-making ability in complex environments by generating a superimposed risk value and through an intelligent response mechanism (such as the first-level avoidance and the second-level avoidance strategies). Especially in the dynamically changing low-altitude airspace, the aircraft can automatically adjust its flight parameters to avoid unstable airflows, dynamic targets, and obstacles without manual intervention. This technology improves the autonomous obstacle avoidance ability of low-altitude aircraft, which is particularly important in high-density and complex low-altitude flight environments such as cities and mountains. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1This is a schematic diagram of the low-altitude airspace dynamic decision-making system based on the multi-modal large model of the present invention. Specific embodiments

[0021] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the following further details the present invention in conjunction with specific embodiments.

[0022] It should be noted that unless otherwise defined, the technical terms or scientific terms used in the present invention should have the ordinary meanings understood by those of ordinary skill in the field to which the present invention belongs. The "first", "second", and similar terms used in the present invention do not indicate any order, quantity, or importance, but are only used to distinguish different components. The terms such as "including" or "comprising" mean that the elements or objects appearing before this word cover the elements or objects listed after this word and their equivalents, without excluding other elements or objects. The terms such as "connected" or "linked" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "upper", "lower", "left", and "right" are only used to represent relative position relationships, and when the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0023] Example 1: Please refer to Figure 1 , the present invention provides a technical solution: a low-altitude airspace dynamic decision-making system based on a multi-modal large model, including: An airspace modeling module, which is used for the flight tasks of multiple aircraft in the low-altitude airspace of 120 to 1000 meters. Taking the i-th aircraft as an example, it establishes a dynamic three-dimensional floating coordinate system for the target operation airspace, synchronously collects meteorological data, remote sensing images, visual images, and navigation data related to the airspace where the i-th aircraft is located, and establishes a multi-source data set; A trajectory analysis module, which is used to collect the first original flight trajectory of the i-th aircraft in real time, establish its first original flight trajectory in the dynamic three-dimensional floating coordinate system, and identify the environmental data of the j-th area in the first original flight trajectory to construct the heat flux of the j-th area , shear wind intensity , turbulence intensity and dynamic target density index and evaluate; A map construction module, which is used to obtain the first original flight trajectory path recognition result when identifying whether each trajectory point of the i-th aircraft in the first original flight trajectory crosses a risk area, and construct a first heat disturbance coverage map, a second shear wind high-risk area map, a third turbulence risk basin map, and a fourth sudden obstacle prediction area map; The flight risk superposition module constructs the superposition risk value of the jth region based on the first thermal disturbance coverage map, the second high-risk shear wind area map, the third turbulence risk basin map, and the fourth sudden obstacle prediction area map. , and evaluate, , normal passage, when indicates that the superposition risk is in the medium-risk section, and a first-level avoidance strategy is generated; when indicates that the superposition risk is in the high-risk section, and a second-level avoidance strategy is generated.

[0024] In this embodiment, the trajectory analysis module extracts meteorological factors such as heat flux, shear wind intensity, and turbulence intensity in real time to realize accurate modeling of the airspace environment where the aircraft is located, and can effectively identify and evaluate the thermal disturbance area, shear wind belt, and turbulence risk area, improving the risk perception ability of low-altitude flight. The system generates multiple types of risk maps (the first thermal disturbance coverage map, the second high-risk shear wind area map, the third turbulence risk basin map, and the fourth sudden obstacle prediction area map) through the map construction module, and realizes the comprehensive calculation and dynamic evaluation of multi-source risks in the flight risk superposition module, thus avoiding the problem that a single risk model cannot cover complex scenarios. This system can divide the risk level according to the real-time evaluation results and automatically generate hierarchical avoidance strategies, including first-level avoidance (medium risk) and second-level avoidance (high risk), enabling the flight path planning to have the ability of active avoidance, thereby significantly reducing the instability rate and collision risk of the aircraft during low-altitude mission execution. The system can sense and predict the dynamic target density and potential sudden obstacles in the low-altitude airspace, and is especially suitable for low-altitude application scenarios with high dynamics, complex terrain, or variable weather, enhancing the stability and mission completion rate during the flight mission execution.

[0025] Embodiment 2: This embodiment is an explanatory description based on Embodiment 1. Please refer to Figure 1 , specifically, the airspace modeling module includes a three-dimensional floating coordinate system construction unit and a multi-source data synchronous acquisition unit; The three-dimensional floating coordinate system construction unit is used for the flight missions of multiple aircraft in the low-altitude airspace of 120 to 1000 meters, and is used for the ith aircraft, where i ∈ [1, n], and n is the total number of aircraft operating simultaneously; Taking the current position of the ith aircraft as a reference point, a dynamic three-dimensional floating coordinate system is established in real time, and the dynamic three-dimensional floating coordinate system is adaptively adjusted according to the heading and attitude of the ith aircraft to ensure that the dynamic three-dimensional floating coordinate system is aligned with the aircraft mission path; The multi-source data synchronous acquisition unit is used to synchronously acquire meteorological data, remote sensing images, visual images, and navigation data related to the airspace where the i-th aircraft is located, establish a multi-source data set, construct a three-dimensional dynamic environment model and train it, and perform multi-modal fusion on various types of data using timestamp alignment and spatial interpolation methods to ensure that different data sources have a consistent spatio-temporal reference system within the coordinate domain of the i-th aircraft.

[0026] In this embodiment, the three-dimensional floating coordinate system construction unit takes the current position of the aircraft as the reference origin and establishes a three-dimensional coordinate system that adaptively changes with the heading and attitude of the aircraft in real time, so that the constructed environment model is always aligned with the mission direction of the aircraft, avoiding the problem of mismatch between the static geographic coordinate system and the motion state of the aircraft, thereby significantly improving the real-time performance and adaptability of spatial modeling. The multi-source data synchronous acquisition unit realizes the mapping of a unified spatio-temporal reference system for various types of heterogeneous data such as remote sensing images, visual images, meteorological data, and navigation data within the coordinate domain of the aircraft through timestamp alignment and spatial interpolation strategies, effectively solving the modeling deviation problem caused by time / space asynchronization of multi-modal data in traditional systems. This module is particularly suitable for coping with non-linear disturbances caused by factors such as urban heat island effect, complex terrain, or sudden airflow during low-altitude flight, providing more stable and reliable environmental modeling support for the aircraft to perform tasks such as inspection and emergency response.

[0027] Embodiment 3: This embodiment is an explanatory description based on Embodiment 1. Please refer to Figure 1 , specifically, the trajectory analysis module includes a first trajectory acquisition unit, a ground heat flux identification unit, and a local mutation shear wind field identification unit; The first trajectory acquisition unit is used to collect the position, attitude angle, velocity vector of the i-th aircraft in real time, and the first original flight trajectory; The ground heat flux identification unit is used to deploy heat flux sensors on the ground, collect heat flux data, define the spatial mapping of heat flux perturbation and ground temperature gradient according to the relationship between heat flux and ground temperature, and obtain the heat flux of the j-th region : ; where is the ground temperature of the j-th region; is the environmental temperature between 120 meters and 1000 meters above the ground in the j-th region, which is obtained by extracting from the meteorological data related to the airspace where the i-th aircraft is located; h is the heat transfer coefficient, which is determined according to the actual scenario and includes: h = 5 - 25 W / m²·K for natural convection; h = 25 - 250 W / m²·K for forced convection under strong wind; h = 10 - 50 W / m²·K for indoor air flow; h = 50 - 500 W / m²·K near the heat source of industrial chimneys; According to the heat flux of the j-th region , judge the heat flux anomaly level, including: When the heat flux of the j-th region <100W / m 2 ; It is judged as a normal heat flux region; When 100W / m 2 ≤ ≤300W / m 2 ; It is judged that there is a heat interference anomaly and marked as a first-order heat disturbance area; When >300W / m 2 ; It is judged that there is a heat interference anomaly and marked as a second-order heat disturbance area.

[0028] In this embodiment, the ground heat flux identification unit introduces the heat transfer coefficient h for the first time to classify the heat transfer mechanisms under different natural or artificial environments by deploying heat flux sensors in the flight area and establishing the spatial mapping relationship between the heat flux and the ground temperature gradient, effectively improving the identification ability of heat disturbance factors such as urban heat island effect and complex terrain radiation heat sources.

[0029] The module divides the heat flux anomaly level into three levels (normal area, first-order heat disturbance area, second-order heat disturbance area), uses the heat flux thresholds (<100W / m², 100 - 300W / m², >300W / m²) as the criteria, provides a quantitative index that can be directly docked with the path risk assessment and avoidance strategy generation, enables the flight system to have a hierarchical response mechanism, and enhances the flight safety and strategy intelligence. This module can be widely applied to typical low-altitude heat anomaly environments such as near-ground urban heat areas, industrial heat source peripheries, and complex terrain heat stagnation areas, and can significantly improve the stability and reliability of small flight platforms in summer, high-temperature areas, and multi-obstacle environments.

[0030] Example 4: This example is an explanatory description based on Example 1. Please refer to Figure 1 , specifically, the local mutation shear wind field identification unit is used to deploy a three-dimensional ultrasonic anemometer group and a laser wind measurement radar device on the ground and in the flight channel, respectively collect the wind speed vector data from the ground to a height of 1000 meters in the target area, and obtain the shear wind intensity of the j-th region based on the wind speed difference between each height layer : ; Wherein, and are the wind speed vectors of the j-th region at the second height and the first height respectively. The second height is set to 100m, and the first height is set to 50m; According to the shear wind intensity of the j-th area , determine the shear wind disturbance level, including: When the shear wind intensity of the j-th area <1.0m / s·100m -1 ; It is judged as a normal shear wind intensity area; When 1.0m / s·100m -1 ≤ ≤2.5m / s·100m -1 ; It is judged that there is shear wind anomaly and marked as a first-order shear wind disturbance area; When >2.5m / s·100m -1 ; It is judged that there is shear wind anomaly and marked as a second-order shear wind disturbance area.

[0031] In this embodiment, by deploying three-dimensional ultrasonic anemometers and lidar along the ground and flight path, continuous wind speed vector data acquisition in the range from the ground to 1000 meters in height is realized, effectively covering key airspaces such as building-dense areas, valley channels, and bridge-tunnel exits where shear winds are frequent, improving the vertical resolution and dynamic response ability of wind field identification. The local mutation shear wind field identification unit constructs the height difference shear wind intensity to adapt to the non-uniform wind field identification requirements: this unit introduces a wind speed difference calculation model, calculates the shear wind intensity based on the wind speed vector difference between 50 meters and 100 meters in height in the j-th area, accurately captures the local disturbance caused by wind speed mutation, and is especially suitable for identifying flight dangerous points with uneven distribution of sudden vertical airflows.

[0032] By setting the shear wind intensity threshold (<1.0m / s·100m -1 is normal, 1.0–2.5m / s·100m -1 is first-order, >2.5m / s·100m -1 is the second-order disturbance area), a multi-level determination mechanism for shear wind risk is realized, which facilitates the system to automatically generate corresponding flight control instructions such as avoidance routes, deceleration, or hovering according to the risk level, and improves the safety decision-making efficiency of flight tasks. This identification unit is especially suitable for flight scenarios in complex terrains or high wind change areas, such as ridge wind mouths, wind corridors between urban high-rise buildings, and coastal wind pressure areas, and can effectively improve the path keeping and flight control stability capabilities of unmanned aerial vehicles and small flight platforms in sudden wind disturbances, reducing safety risks such as attitude deviation and trajectory yaw.

[0033] Example 5: This example is an explanatory description based on Example 1. Please refer to Figure 1 , specifically, the trajectory analysis module further includes a turbulent area boundary identification unit and a dynamic target obstacle distribution identification unit; The turbulent region boundary recognition unit is used to deploy an anemometer device in the target area, collect the instantaneous change data of the wind speed, and obtain the turbulence intensity of the j-th region. : ; In the formula, is the standard deviation of the wind speed in the j-th region, reflecting the degree of fluctuation, represents the average wind speed of the j-th region; According to the turbulence intensity of the j-th region , judge the turbulence risk level, including: When the turbulence intensity of the j-th region < 0.1; It is judged as a normal turbulence region; When 0.1 ≤ ≤ 0.2; It is judged that there is a turbulence anomaly and marked as a first-order turbulence disturbance area; When > 0.2; It is judged that there is a turbulence anomaly and marked as a second-order turbulence disturbance area.

[0034] In this embodiment, by collecting the standard deviation and average wind speed of the wind speed in the j-th region, a dimensionless turbulence intensity index is constructed, which can accurately characterize the local air flow disturbance intensity and the degree of change, and provide key quantitative support for identifying the low-altitude unsteady wind field. According to the numerical range of the turbulence intensity, a risk classification rule is set (<0.1 is the normal area, 0.1 - 0.2 is the first-order disturbance area, >0.2 is the second-order disturbance area), which can realize the rapid classification and spatial boundary delineation of different levels of turbulence risk areas, and provide a decision-making basis for subsequent flight path correction and attitude adjustment. Based on the real-time wind speed data input and the sliding time window processing mechanism, this unit can dynamically update the distribution range of the turbulent region, and is applicable to short-term sudden turbulence caused by weather changes, terrain changes, heat island effects, etc., improving the overall response speed and prediction ability of the system to unstable air flow regions. By identifying and accurately positioning the boundary of the turbulent region, the system can implement response operations such as real-time avoidance of high-risk path segments, flight altitude adjustment, or delayed entry, effectively reducing the risks such as vibration, yaw, and attitude out-of-control caused by the influence of air flow disturbance on the aircraft, and improving the overall flight path smoothness and flight control stability.

[0035] Example 6: This example is an explanatory description based on Example 1. Please refer to Figure 1 , specifically, the dynamic target obstacle distribution recognition unit is used to collect the trajectory, speed, size, and distribution density data of moving targets in the target area by deploying lidar, vision sensors, millimeter-wave radars, ADS-B, or V2X devices, and construct a real-time distribution map of dynamic obstacles to obtain the dynamic target density index of the j-th region : ; In the formula, is the dynamic obstacle number density per unit area in the j-th region, with the unit of number per m² 2 , represents the number of currently detected dynamic obstacles in the j-th region, represents the horizontal projected area of the j-th region; Through the target detection algorithm, dynamic targets in the remote sensing images and visual images related to the airspace where the i-th aircraft is located can be identified and extracted; through image difference technology or optical flow method, which are relatively mature in the prior art, the motion trajectories of dynamic targets are extracted to judge the attributes such as the speed, direction, and size of the targets. For each detected dynamic target, its position in the remote sensing image (usually calibrated by coordinates) can be combined with the horizontal projected area of the target region to calculate the density of the target in a specific region. In image processing, the image is usually divided into several grids (such as units with a resolution of per square meter), and then the number density of dynamic targets (i.e., the number of dynamic targets per unit area) in each grid is calculated. The dynamic target information obtained through remote sensing images can be fused with other sensor data (such as lidar, millimeter-wave radar, etc.) to improve the accuracy of target recognition. The dynamic target density index (such as "number / m²" in the figure) is calculated based on this fused data and can be used to construct a real-time distribution map of dynamic obstacles to reflect the distribution of dynamic targets in the target region.

[0036] According to the dynamic target density index of the j-th region , judge the flight obstacle avoidance risk level, including: When the dynamic target density index of the j-th region < 0.1 number / m² 2 ; It is judged as a normal obstacle avoidance risk area; When 0.1 number / m² 2 ≤ ≤ 0.5 number / m² 2 ; It is marked as a first-order obstacle avoidance prompt area; When > 0.5 number / m² 2 ; It is marked as a second-order high-density obstacle avoidance warning area.

[0037] In this embodiment, the unit comprehensively utilizes the high-precision distance measurement ability of lidar, the morphological recognition characteristics of visual sensors, the all-weather penetration advantage of millimeter-wave radars, and the broadcast perception capabilities of ADS-B and V2X to achieve efficient detection and information complementation of various types of dynamic obstacles (such as drones, bird flocks, airborne objects, etc.) in complex airspaces, effectively alleviating the limitations of single sensors, such as being easily blocked and having a high false alarm rate. By counting the number of dynamic obstacles per unit area, a dynamic target density index (unit: number / m²) is established to quantitatively evaluate the spatial density of obstacles in the flight path. Combining with the set risk classification criteria (<0.1 is normal, 0.1 - 0.5 is the warning area, >0.5 is the alarm area), it provides a clear and controllable decision-making basis for obstacle avoidance strategies for the aircraft. Through the dynamic target distribution map and real-time risk level determination, the system can achieve automatic route adjustment, obstacle avoidance path reconstruction, or waiting mechanisms during flight path planning or navigation, which is particularly suitable for low-altitude multi-aircraft collaborative operations or urban air mobility (UAM) scenarios, effectively preventing airspace congestion, collision risks, or attitude disturbance problems.

[0038] Embodiment 7: This embodiment is an explanatory description based on Embodiment 1. Please refer to Figure 1 , specifically, the map construction module includes a first original flight trajectory recognition unit and a map generation unit; The first original flight trajectory recognition unit is used to identify the flight path segment passing through the j-th area in the first original flight trajectory, and obtain the heat flux of the j-th area, the shear wind intensity of the j-th area, the turbulence intensity of the j-th area, and the dynamic target density index of the j-th area corresponding to the flight path segment of the j-th area. Judge whether each trajectory point in the flight trajectory crosses the following risk areas to obtain the first original flight trajectory path recognition result and perform hierarchical marking, including: First-order or second-order thermal disturbance area, first-order or second-order shear wind disturbance area, first-order or second-order turbulence disturbance area, first-order obstacle avoidance warning area or second-order high-density obstacle avoidance alarm area; Construct a first thermal disturbance coverage map, a second shear wind high-risk area map, a third turbulence risk basin map, and a fourth sudden obstacle prediction area map based on the first original flight trajectory path recognition result.

[0039] In this embodiment, this module relies on the first original flight trajectory recognition unit to dynamically extract the heat flux, shear wind intensity, turbulence intensity, and dynamic target density indicators in the traversed area of the flight path, construct the spatial correlation relationship between the trajectory and the disturbance points, realize the recognition and level determination of multiple risk factors at the trajectory level, and significantly improve the refined analysis ability of the risk area. Respectively construct the heat disturbance coverage map, the high-risk shear wind area map, the turbulence risk basin map, and the sudden obstacle prediction area map, so that the disturbance factors in the airspace operation environment have the ability of visual expression, can provide an intuitive path avoidance basis for the flight path planning system, and realize the early detour and dynamic closure strategy for potential high-risk airspace. By marking the traversal of the flight path segment in various types of risk areas, the disturbance level classification of the trajectory segment (such as the first-order or second-order heat disturbance area, etc.) is realized, which can provide a data basis for subsequent path safety level assessment, flight performance assessment, and mission assignment and scheduling, and is particularly suitable for intelligent flight path planning and risk avoidance guidance in low-altitude complex scenarios.

[0040] Embodiment 8: This embodiment is an explanatory description carried out in Embodiment 1. Please refer to Figure 1 , specifically, the flight risk superposition module includes a flight risk superposition recognition unit and a hierarchical response unit; The flight risk superposition recognition unit is used to calculate and obtain the superposition risk value of the jth area according to the first heat disturbance coverage map, the second high-risk shear wind area map, the third turbulence risk basin map, and the fourth sudden obstacle prediction area map through the following formula : ; In the formula, is the heat flux score of the jth area. The normal heat flux area is 0 points, the first-order heat disturbance area is 1 point, and the second-order heat disturbance area is 2 points; is the shear score of the jth area. The normal shear wind intensity area is 0 points, the first-order shear wind disturbance area is 1 point, and the second-order shear wind disturbance area is 2 points; is the turbulence intensity score of the jth area. The normal turbulence area is 0 points, the first-order turbulence disturbance area is 1 point, and the second-order turbulence disturbance area is 2 points; is the dynamic target density score of the jth area. The normal obstacle avoidance risk area is 0 points, the first-order obstacle avoidance prompt area is 1 point, and the second-order high-density obstacle avoidance warning area is 2 points; When <2, it means that the superposition risk value of the flight path segment passing through the jth area in the first original flight trajectory of the ith aircraft is a low-risk segment, and the visualization of the jth area is set to green in the dynamic three-dimensional floating coordinate system; When , representing the superimposed risk value of the flight path segment passing through the j-th area in the first original flight trajectory of the i-th aircraft is a medium-risk segment, and the visualization of the j-th area is set to yellow in the dynamic three-dimensional floating coordinate system; When , representing the superimposed risk value of the flight path segment passing through the j-th area in the first original flight trajectory of the i-th aircraft is a high-risk segment, and the visualization of the j-th area is set to red in the dynamic three-dimensional floating coordinate system.

[0041] In this embodiment, by defining four types of risk scores for heat flux, shear wind, turbulence, and dynamic obstacles, the perturbation information in different physical dimensions and perception sources is normalized and uniformly weighted and scored, and the flight superimposed risk value of the j-th area is constructed, realizing the fusion expression of complex perturbation information and effectively improving the comprehensiveness and scientificity of airspace risk identification. By mapping the superimposed risk value to three types of dynamic color identifiers: green (low risk), yellow (medium risk), and red (high risk), the risk level of each area is marked in real time in the dynamic three-dimensional floating coordinate system, and a flight visualization risk map is constructed, providing an intuitive risk decision-making basis for aircraft path avoidance, route reconstruction, and emergency dispatch. Identifying and marking risk areas at the granularity of flight path segments can realize the differential assessment of the risk levels of different sections in the same path, avoid a one-size-fits-all approach to the entire path, and thus improve the accuracy and flexibility of flight path planning and dynamic correction. Through the setting of the hierarchical response unit, the system can automatically trigger different levels of risk avoidance response strategies (such as flight altitude adjustment, speed limit, route deviation suggestion, or abort command) according to the risk level of the flight path segment, constructing an active, multi-level, and scalable flight safety guarantee mechanism, and improving the system's real-time autonomous response ability in complex low-altitude scenarios.

[0042] Embodiment 9: This embodiment is an explanatory description based on Embodiment 8. Please refer to Figure 1 , specifically, the hierarchical response unit is used to maintain the current flight state when the i-th aircraft passes through the flight path segment of the j-th area according to the first original flight trajectory; this unit can automatically trigger appropriate avoidance measures according to the risk level of the flight path segment. In a low-risk segment, the aircraft maintains its current flight state to avoid ineffective operations; while in medium- and high-risk segments, by adjusting flight parameters such as flight altitude, speed, and heading angle, it ensures that the aircraft can safely avoid potential dangerous areas. Through this intelligent response, the safety hazards during flight are effectively reduced.

[0043] It is used to generate a first-level avoidance strategy when the i-th aircraft passes through the flight path segment of the j-th area according to the first original flight trajectory. The strategy includes: Increase the vertical flight altitude of the i-th aircraft by 20 m, and reduce the flight speed by 20 - 30% on the stable wall, including reducing it from 15 m / s to 8 - 10 m / s, and adjust the course angle by ±5 - 10° to avoid the second-order high-density obstacle avoidance warning area or the second-order turbulence disturbance area; When the i-th aircraft passes through the flight path segment of the j-th area according to the first original flight trajectory, if it is identified as a high-risk segment, a secondary avoidance strategy is generated, including: Increase the vertical flight altitude of the i-th aircraft by 50 - 80 m, and reduce the flight speed by 40 - 50% on the stable wall, including reducing it from 15 m / s to 6 - 7 m / s, and adjust the course angle by ±15 - 30° to avoid the second-order high-density obstacle avoidance warning area or the second-order turbulence disturbance area; Determine whether there is a continuous band of risk voxels in the area where the current original path segment is located in the high-risk paragraph; If so, call the path reconstruction algorithm to plan a new flight path segment with the minimum risk cost; splice the new path segment into the aircraft trajectory control system, send the path points in real time, and synchronously mark the current path segment as a high-risk passage record area.

[0044] In this embodiment, for different risk levels, the hierarchical response unit generates precise flight strategies. For example, for medium-risk segments, the generated primary avoidance strategy includes a moderate increase in altitude and a reduction in speed, reasonably reducing the impact range of the aircraft; while for high-risk segments, the secondary avoidance strategy is more radical, requiring a significant increase in flight altitude and a significant reduction in flight speed to further ensure that the aircraft stays away from the high-density obstacle area or the turbulence disturbance area. This detailed hierarchical response provides comprehensive safety protection for the aircraft. When the aircraft enters a high-risk segment and detects a continuous band of risk voxels, the system can call the path reconstruction algorithm in real time to plan a new path with the minimum risk cost. This dynamic path reconstruction ability enables the aircraft to flexibly adjust its flight path, avoid risk areas, and at the same time ensure the efficient completion of the flight mission. The system splices the new avoidance path segment into the aircraft's trajectory control system in real time and synchronously sends the path points to achieve real-time monitoring and adjustment of the aircraft's path. At the same time, the system marks the high-risk areas passed through and automatically references them in subsequent flights to form a complete safety monitoring record. The accumulation of this record will contribute to the long-term data accumulation and optimization of the flight mission, improving the intelligence level of the flight mission. Through the hierarchical response mechanism, the system can intelligently adjust the action strategy of the aircraft according to the actual situation of different airspace risks, providing support for large-scale, multi-aircraft collaborative operations. Promote the improvement of the intelligence level of airspace management and adapt to the complex requirements of future flight missions.

[0045] It should be noted that all calculation formulas in this application document adopt regression analysis including but not limited to machine learning algorithms to deeply analyze the relevant parameters collected, identify their natural trends and interrelationships. Using professional software such as the Scikit-learn library of Python or the R language, a mathematical model matching the data is automatically generated. Then, the performance of the model is objectively evaluated through methods such as cross-validation, and combined with continuous feedback and optimization to ensure that the created formula truly reflects the internal laws of the data, thereby ensuring its effectiveness and accuracy, and ensuring that the calculation process conforms to the constraints of natural laws rather than being based on artificially set rules.

[0046] Essentially, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk or optical disc of a computer, etc., including several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods of various embodiments of the present invention.

[0047] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus or device and execute the instructions), or used in combination with these instruction execution systems, apparatus or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by or in combination with an instruction execution system, apparatus or device.

[0048] It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and not to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solution of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solution of the present invention, and all of them should be covered by the scope of the claims of the present invention.

[0049] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A low-altitude airspace dynamic decision-making system based on a multimodal large model, characterized in that, Including: An airspace modeling module, which is used for the flight tasks of multiple aircraft in the low-altitude airspace of 120 to 1000 meters. Taking the i-th aircraft as an example, it establishes a dynamic three-dimensional floating coordinate system for the target operation airspace, synchronously collects meteorological data, remote sensing images, visual images and navigation data related to the airspace where the i-th aircraft is located, and establishes a multi-source data set; A trajectory analysis module, which is used to collect the first original flight trajectory of the i-th aircraft in real time, establish its first original flight trajectory in the dynamic three-dimensional floating coordinate system, and identify the environmental data of the j-th area in the first original flight trajectory, so as to construct the heat flux of the j-th area , shear wind intensity , turbulence intensity and dynamic target density index and evaluate; A spectrum map construction module, which is used to identify whether each trajectory point of the i-th aircraft in the first original flight trajectory crosses a risk area, obtain the path recognition result of the first original flight trajectory, and construct a first thermal disturbance coverage map, a second shear wind high-risk area map, a third turbulence risk basin map and a fourth sudden obstacle prediction area map; The flight risk superposition module constructs the superposition risk value of the j-th region based on the first thermal disturbance coverage map, the second shear wind high-risk area map, the third turbulence risk basin map, and the fourth sudden obstacle prediction area map , and evaluates , for normal passage when indicates that the superposition risk is in the medium-risk section, and a first-level avoidance strategy is generated; when indicates that the superposition risk is in the high-risk section, and a second-level avoidance strategy is generated.

2. The low-altitude airspace dynamic decision-making system based on a multi-modal large model according to claim 1, wherein: The airspace modeling module includes a three-dimensional floating coordinate system construction unit and a multi-source data synchronous collection unit; The three-dimensional floating coordinate system construction unit is used for the flight tasks of multiple aircraft in the low-altitude airspace of 120 to 1000 meters. Taking the i-th aircraft as an example, where i ∈ [1, n], and n is the total number of aircraft operating simultaneously; Taking the current position of the i-th aircraft as a reference point, it establishes a dynamic three-dimensional floating coordinate system in real time. The dynamic three-dimensional floating coordinate system is adaptively adjusted according to the heading and attitude of the i-th aircraft to ensure that the dynamic three-dimensional floating coordinate system is aligned with the aircraft mission path; The multi-source data synchronous collection unit is used to synchronously collect meteorological data, remote sensing images, visual images and navigation data related to the airspace where the i-th aircraft is located, establish a multi-source data set, construct and train a three-dimensional dynamic environment model, and perform multi-modal fusion on various types of data using timestamp alignment and spatial interpolation methods to ensure that different data sources have a consistent spatio-temporal reference system within the coordinate domain of the i-th aircraft.

3. The low-altitude airspace dynamic decision-making system based on a multimodal large model according to claim 2, wherein: The trajectory analysis module includes a first trajectory collection unit, a ground heat flux identification unit and a local mutation shear wind field identification unit; The first trajectory collection unit is used to collect the position, attitude angle, velocity vector of the i-th aircraft and the first original flight trajectory in real time; The ground heat flux identification unit is used to deploy heat flux sensors on the ground, collect heat flux data, define the spatial mapping of heat flux perturbation and ground temperature gradient according to the relationship between heat flux and ground temperature, and obtain the heat flux of the j-th region : ; where, is the ground temperature of the j-th area, is the ambient temperature between 120 m and 1000 m above the ground of the j-th area, h is the heat transfer coefficient, which is taken according to the actual scenario, including: h = 5 - 25 W / m²·K for natural convection; h = 25 - 250 W / m²·K for forced convection under strong wind; h = 10 - 50 W / m²·K for indoor air flow; h = 50 - 500 W / m²·K near the heat source of industrial chimneys; According to the heat flux of the j-th region , determine the heat flux anomaly level, including: When the heat flux of the j-th region <100 W / m 2 ; it is determined as a normal heat flux region; When 100 W / m 2 ≤ ≤ 300 W / m 2 ; It is judged that there is a thermal interference anomaly and marked as the first-order thermal disturbance area; When > 300 W / m 2 ; It is determined that there is a thermal interference anomaly and marked as the second-order thermal disturbance area.

4. The low-altitude airspace dynamic decision-making system based on the multimodal large model according to claim 3, wherein: The local mutation shear wind field identification unit is used to deploy a three-dimensional ultrasonic wind speed sensor group and a lidar device in the ground and flight channels, respectively collect wind speed vector data within the height range from the ground to 1000 meters in the target area, and obtain the shear wind intensity of the j-th area based on the wind speed difference at each height layer : ; where, and are the wind speed vectors at the second height and the first height in the j-th area, respectively. The second height is set to 100 m, and the first height is set to 50 m; According to the shear wind intensity of the j-th region , determine the shear wind disturbance level, including: When the shear wind intensity in the j-th area <1.0 m / s·100 m -1 ; it is judged as a normal shear wind intensity area; When 1.0 m / s·100 m -1 ≤ ≤2.5 m / s·100 m -1 ; It is judged that there is an abnormal shear wind and marked as a first-order shear wind disturbance area; When > 2.5 m / s·100 m -1 ; it is determined that there is an abnormal shear wind and it is marked as a second-order shear wind disturbance area.

5. The low-altitude airspace dynamic decision-making system based on the multi-modal large model according to claim 3, wherein: The trajectory analysis module further includes a turbulence area boundary identification unit and a dynamic target obstacle distribution identification unit; The turbulent region boundary recognition unit is used to deploy an anemometer device in the target area, collect the instantaneous wind speed change data, and obtain the turbulence intensity of the j-th region : ; where, is the standard deviation of the wind speed in the j-th region, reflecting the degree of fluctuation intensity, represents the average wind speed in the j-th region; According to the turbulence intensity of the j-th region , determining the turbulence risk level, including: When the turbulence intensity of the j-th region <0.1; it is judged as a normal turbulence region; When 0.1 ≤ ≤ 0.2; it is judged that there is a turbulent anomaly and it is marked as a first-order turbulent perturbation area; When > 0.2; it is determined that there is a turbulent anomaly and it is marked as a second-order turbulent perturbation zone.

6. The low-altitude airspace dynamic decision-making system based on a multimodal large model according to claim 5, characterized in that: The dynamic target obstacle distribution recognition unit is used to collect the trajectory, speed, size, and distribution density data of moving targets in the target area by deploying lidar, vision sensors, millimeter-wave radars, ADS-B, or V2X devices, construct a real-time distribution map of dynamic obstacles, and obtain the dynamic target density index of the jth area : ; where, is the dynamic obstacle number density per unit area in the j-th region, with the unit of number per m 2 , represents the number of currently detected dynamic obstacles in the j-th region, represents the horizontal projection area of the j-th region; Based on the dynamic target density index of the j-th region , determining the flight obstacle avoidance risk level, including: When the dynamic target density index of the j-th area <0.1 per m 2 ; it is judged as a normal obstacle avoidance risk area; When 0.1 per m 2 ≤ ≤ 0.5 per m 2 ; Marked as the first-order obstacle avoidance prompt area; When > 0.5 per m 2 ; Marked as a second-order high-density obstacle avoidance warning area.

7. The low-altitude airspace dynamic decision-making system based on a multi-modal large model according to claim 4, characterized in that: The spectrum map construction module includes a first original flight trajectory identification unit and a spectrum map generation unit; The first original flight trajectory identification unit is used to identify the flight path segment passing through the j-th area in the first original flight trajectory, and obtain the heat flux of the j-th area corresponding to the flight path segment of the j-th area, the shear wind intensity of the j-th area, the turbulence intensity of the j-th area and the dynamic target density index of the j-th area; Judge whether each trajectory point in the flight trajectory crosses the following risk areas, obtain the path recognition result of the first original flight trajectory, and perform hierarchical marking, including: First-order or second-order thermal disturbance area, first-order or second-order shear wind disturbance area, first-order or second-order turbulence disturbance area, first-order obstacle avoidance prompt area or second-order high-density obstacle avoidance warning area; Construct a first thermal disturbance coverage map, a second shear wind high-risk area map, a third turbulence risk basin map and a fourth sudden obstacle prediction area map based on the path recognition result of the first original flight trajectory.

8. The low-altitude airspace dynamic decision-making system based on a multimodal large model according to claim 7, characterized in that: The flight risk superposition module includes a flight risk superposition identification unit and a hierarchical response unit; The flight risk superposition recognition unit is used to calculate and obtain the superposition risk value of the jth region according to the first thermal disturbance coverage map, the second high-shear-wind risk area map, the third turbulence risk basin map, and the fourth sudden obstacle prediction area map through the following formula :[[]]END]] ; where is the heat flux score of the j-th region, with a normal heat flux region being 0 points, a first-order heat perturbation region being 1 point, and a second-order heat perturbation region being 2 points; is the shear score of the j-th region, with a normal shear wind intensity region being 0 points, a first-order shear wind perturbation region being 1 point, and a second-order shear wind perturbation region being 2 points; is the turbulence intensity score of the j-th region, with a normal turbulence region being 0 points, a first-order turbulence perturbation region being 1 point, and a second-order turbulence perturbation region being 2 points; is the dynamic target density score of the j-th region, with a normal obstacle avoidance risk region being 0 points, a first-order obstacle avoidance prompt region being 1 point, and a second-order high-density obstacle avoidance warning region being 2 points; When <2, it represents the superimposed risk value of the flight path segment passing through the j-th area in the first original flight trajectory of the i-th aircraft is a low-risk segment, and the visualization of the j-th area is set to green in the dynamic three-dimensional floating coordinate system; When , it represents the superimposed risk value of the flight path segment passing through the j-th area in the first original flight trajectory of the i-th aircraft is a medium-risk segment, and the visualization of the j-th area is set to yellow in the dynamic three-dimensional floating coordinate system; When , it represents the superimposed risk value of the flight path segment passing through the j-th area in the first original flight trajectory of the i-th aircraft is a high-risk segment, and the visualization setting of the j-th area is marked as red in the dynamic three-dimensional floating coordinate system.

9. The low-altitude airspace dynamic decision-making system based on a multimodal large model according to claim 8, characterized in that: The hierarchical response unit is configured to, when the i-th aircraft passes through the flight path segment of the j-th area according to the first original flight trajectory, if it is identified as a low-risk segment, maintain the current flight state; When the i-th aircraft passes through the flight path segment of the j-th area according to the first original flight trajectory, if it is identified as a medium-risk segment, generate a first-level avoidance strategy, including: Increase the vertical flight altitude of the i-th aircraft by 20 m, reduce the flight speed by 20 - 30% on the stable wall, including reducing from 15 m / s to 8 - 10 m / s, and adjust the heading angle by ±5 - 10° to bypass the second-order high-density obstacle avoidance warning area or the second-order turbulence disturbance area; When the i-th aircraft passes through the flight path segment of the j-th area according to the first original flight trajectory, if it is identified as a high-risk segment, generate a second-level avoidance strategy, including: Increase the vertical flight altitude of the i-th aircraft by 50 - 80 m, reduce the flight speed by 40 - 50% on the stable wall, including reducing from 15 m / s to 6 - 7 m / s, and adjust the heading angle by ±15 - 30° to bypass the second-order high-density obstacle avoidance warning area or the second-order turbulence disturbance area; determine whether there is a continuous band of risk voxels in the area where the current original path segment is located in the high-risk paragraph; if so, call the path reconstruction algorithm to plan a new flight path segment with the minimum risk cost; splice the new path segment into the aircraft trajectory control system, send the path points in real time, and synchronously mark the current path segment as a high-risk passage record area.

Images