CN-121982943-A - Low-altitude unmanned aerial vehicle air collision risk alarm method and system

CN121982943ACN 121982943 ACN121982943 ACN 121982943ACN-121982943-A

Abstract

The invention belongs to the technical field of unmanned aerial vehicle safety, and particularly relates to a low-altitude unmanned aerial vehicle air collision risk alarming method and system. The method comprises the steps of predicting future tracks of the unmanned aerial vehicle through an LSTM neural network fused with an Adam optimizer, carrying out collision risk identification by adopting a dynamic Reich protection area model and a Monte Carlo reach domain model respectively aiming at two different scenes of cooperation and non-cooperation, calculating overlapping degree and detecting invasive collision, and outputting a grading alarm signal according to a prediction time window and risk severity. The method effectively solves the problems of high false alarm rate and short early warning time in the prior art, can realize accurate early warning, early warning and differential early warning of the collision risk of the low-altitude unmanned aerial vehicle, and remarkably improves the safety management level of the low-altitude area. The invention is suitable for various unmanned aerial vehicle monitoring platforms and low-altitude traffic management systems.

Inventors

YUAN LIGANG
Chen Kenzhong
LIU JING
CHEN HAIYAN
XIE HUA
LI GUIYI

Assignees

南京航空航天大学

Dates

Publication Date: 20260505
Application Date: 20260121

Claims (10)

1. The low-altitude unmanned aerial vehicle air collision risk alarming method is characterized by comprising the following steps of: The method comprises the steps of constructing a cooperative unmanned aerial vehicle operation scene and a non-cooperative unmanned aerial vehicle operation scene, wherein the cooperative unmanned aerial vehicle operation scene comprises interactive flight of two cooperative unmanned aerial vehicles with clear flight intention on a preset route; The method comprises the steps of predicting tracks and constructing a protection zone, wherein in a cooperative unmanned aerial vehicle running scene, historical track data of the cooperative unmanned aerial vehicle are obtained and input into a prediction model, the predicted tracks are output, and meanwhile, a dynamic Reich model protection zone is constructed; Judging collision according to the predicted track, wherein a spherical crown overlapping degree judgment method is used for judging collision in a cooperative unmanned aerial vehicle operation scene; and sending out an alarm according to the judgment result.
2. A low-altitude unmanned aerial vehicle air collision risk warning method according to claim 1, wherein, The prediction model is an Adam-LSTM model, and the training method comprises the following steps: constructing a data set, and inputting the data set as data of an Adam-LSTM model; Dividing data of a data set into a training set and a testing set, and training and testing an Adam-LSTM model to obtain a trained Adam-LSTM model; The Adam-LSTM model comprises an optimizer Adam and a double-layer LSTM network, wherein the Adam optimizer controls the learning rate, the double-layer LSTM network layer processes and outputs the result, and a coding and decoding structure is adopted to realize single-step time sequence prediction.
3. A low-altitude unmanned aerial vehicle air collision risk warning method according to claim 1, wherein, The method for constructing the dynamic Reich model protection area comprises the following steps: calculating the physical dimension radius of the fuselage : Wherein the method comprises the steps of Represents the diameter of the unmanned aerial vehicle propeller after being unfolded, Representing the diameter of the unmanned plane body; Calculating radius of inner layer protection area : Is a GPS positioning error; The position drift caused by the delay of the control signal under the condition of the maximum flying speed can be obtained by the following formula, V max is the maximum flying speed, T s is the delay time of the control signal, ; Offset under the condition of strong compensation of flight control PID to offset wind power; Calculating the total radius of theoretical dynamic change : Wherein v is the real-time speed of the unmanned aerial vehicle, Is the maximum deceleration of the unmanned aerial vehicle, T is the response time, Indicating the time required for physical braking, and ; ; ; Is the additional compensation required for the actual response time, It is the sensor delay that is a function of the sensor delay, Is the delay of the control signal and, Is the flight control system response delay.
4. A low-altitude unmanned aerial vehicle air collision risk warning method according to claim 1, wherein, The method for constructing the reachable domain protection area for the non-cooperative unmanned aerial vehicle comprises the steps of generating a dynamic reachable domain convex hull for the non-cooperative unmanned aerial vehicle based on an unmanned aerial vehicle dynamic equation and a Monte Carlo random sampling method, and comprises the following steps: Dividing the non-cooperative unmanned aerial vehicle into three motion modes, wherein the first motion mode is a turning mode, the change of the acceleration of the non-cooperative unmanned aerial vehicle in the mode is set to fly along a final random direction, and the track of the unmanned aerial vehicle which flies directly after the turning reaches a preset angle is simulated; Firstly, under the turning condition, a basic kinematic formula and a minimum turning radius are referred to obtain a movement track of the unmanned aerial vehicle under the plane turning condition; Calculating the minimum turning radius: where g=9.8 m/s is the gravitational acceleration, Is the square of the maximum overload factor, Representing the speed of the unmanned aerial vehicle; Secondly, calculating a direct flight process, a directional mode and a braking mode after turning through a basic dynamics formula, wherein the following formula is a calculation formula in the x-axis direction: Wherein the method comprises the steps of Representing a time of flight; And expanding the two-dimensional plane flight track to a three-dimensional state, prescribing the maximum vertical ascending speed and descending speed of the unmanned aerial vehicle, randomly selecting one direction in the vertical direction for direct flight at 180 degrees after the turning link is finished, obtaining a plurality of points of the farthest track, and linking all the points to serve as patches of the reachable domain to form a reachable domain convex hull.
5. A low-altitude unmanned aerial vehicle air collision risk warning method according to claim 1, wherein, The spherical crown overlapping degree judging method comprises the following steps: Assume that the distance between the centers of two spheres is Their radii are respectively And ; When the distance d between the two spherical centers is larger than or equal to r 1 +r 2 , the overlapping degree alpha=0; When d≤r 1 -r 2 |, the overlap α=1, and When |r 1 -r 2 |＜d＜r 1 +r 2 , the overlap α is calculated as follows: Wherein, the The volume of the sphere with radius r 1 , The volume of the sphere with radius r 2 , The volume of the overlapping portion of the two spheres.
6. The method for warning the risk of the air collision of the low-altitude unmanned aerial vehicle according to claim 1, wherein the collision judgment by using the intrusion method is realized by the following steps: Decomposing the convex hull into triangle patch sets face= { , , }, , , Is the vertex coordinates of the triangle; Calculating the outward unit normal vector of each surface piece ; By symbol distance Judging the relative position of the point and the convex hull, wherein the point Is the predicted trace point, if point Satisfy all the following dough pieces <0, Then indicate the point Inside the convex hull if >0, Then point Outside the normal direction of the dough sheet, if Point=0, then Just on the dough sheet; Wherein the method comprises the steps of Representing the vertex of the slave dough sheet Pointing point Is used for the vector of (a), Representing the projected length of the vector in the direction of normal vector n.
7. The method for warning the risk of low-altitude unmanned aerial vehicle air collision according to claim 1, wherein a sliding window mechanism is adopted in the process of predicting the track: Window length 60 seconds, step 1 second; the input being normalized Wherein Is a normalized three-dimensional coordinate of which, Normalized values representing horizontal and vertical velocities, respectively; the output is a future 60 second three-dimensional coordinate sequence.
8. A low-altitude unmanned aerial vehicle air collision risk warning method according to claim 1, wherein, The alarm method according to the judgment result comprises the following steps: In the operation scene of the collaborative unmanned aerial vehicle, the total duration of a prediction window is divided into a plurality of time windows, collision risk levels of different collision situations are defined, alarm types with corresponding severity are generated according to the different predicted collision risk levels of the time windows, and an alarm with the highest severity is selected from a plurality of alarm types; In the operation scene of the non-cooperative unmanned aerial vehicle, if the predicted track point of the cooperative unmanned aerial vehicle invades into the reachable domain protection area of the non-cooperative unmanned aerial vehicle within the preset time, the alarm is immediately given.
9. The low-altitude unmanned aerial vehicle air collision risk warning method of claim 7, further comprising: In a collaborative unmanned aerial vehicle operation scene, performing rolling prediction verification: Re-performing trajectory prediction every 10 seconds; And verifying the risk points outside 30 seconds in the previous prediction, and if the overlapping degree is continuously 15 steps and is lower than 0.15, releasing the alarm.
10. A low-altitude unmanned aerial vehicle air collision risk warning system, comprising: The system comprises a scene construction module, a scene construction module and a control module, wherein the scene construction module is used for constructing a cooperation type unmanned aerial vehicle operation scene and a non-cooperation type unmanned aerial vehicle operation scene, wherein the cooperation type unmanned aerial vehicle operation scene comprises that two cooperation type unmanned aerial vehicles with definite flight intention fly interactively on a preset route; the track prediction module is used for predicting the track of the cooperative unmanned aerial vehicle; The protection zone construction module is used for constructing a dynamic Reich model protection zone in the operation scene of the cooperative unmanned aerial vehicle and constructing a reachable domain protection zone of the non-cooperative unmanned aerial vehicle; The collision judgment module is used for judging collision according to the predicted track, wherein a spherical crown overlapping degree judgment method is used for judging collision in a cooperative unmanned aerial vehicle operation scene; and the alarm issuing module is used for issuing an alarm according to the judgment result.

Description

Low-altitude unmanned aerial vehicle air collision risk alarm method and system Technical Field The invention belongs to the technical field of unmanned aerial vehicle low-altitude safety control, and particularly relates to a low-altitude unmanned aerial vehicle air collision risk alarm method and system. Background Along with the rapid development of low-altitude economy, the unmanned aerial vehicle industry is also increasing at extremely high speed, and low-altitude unmanned aerial vehicles show great potential in the fields of logistics distribution, emergency rescue and the like, but due to the rapid increase of the number of unmanned aerial vehicles and the limited airspace limitation, the risk of air collision becomes a key factor for restricting the development of the industry. The aim of future air traffic is to reduce the collision probability of unmanned aerial vehicle traffic as much as possible, and the occurrence of unmanned aerial vehicle collision accidents is avoided in advance by utilizing observation-prediction. At present, the low-altitude operation route of the unmanned aerial vehicle is usually preset in advance, and although the low-altitude operation route can deviate to a certain extent because of pursuing flexible and dynamic adjustment, the low-altitude unmanned aerial vehicle is generally a pipeline route, and the pipeline route is a three-dimensional channel planned for the unmanned aerial vehicle to fly in a low-altitude or specific area of a city. This concept aims to isolate unmanned aerial vehicle flight activities within a specific airspace, similar to the "overhead road network" established for unmanned aerial vehicles, to ensure that they operate safely, orderly and efficiently. The method has the core aims of realizing physical or virtual isolation and reducing the risk of collision between the unmanned aerial vehicle and the unmanned aerial vehicle, the building, the ground crowd and other unmanned aerial vehicles to the greatest extent. The safety target level (abbreviated as TLS) is a core concept in low-altitude unmanned aerial vehicle collision risk identification and airspace management, provides a quantitative reference for an acceptable risk threshold, and is an ultimate constraint which all risk assessment methods must follow. The necessity of TLS is first manifested in that it provides an explicit acceptance criterion for collision risk models. The operation of the low-altitude unmanned aerial vehicle needs to ensure the safe target level, and the increase of the airspace utilization rate possibly increases the collision risk to make the target safe water product lower than the acceptable range, so the method for finding out the target safe level and improving the airspace utilization rate is a key problem at present. Disclosure of Invention In order to solve the technical problems, the invention provides a low-altitude unmanned aerial vehicle air collision risk alarm method and system for distinguishing cooperation type and cooperation type operation scenes. The first aspect of the invention provides a low-altitude unmanned aerial vehicle air collision risk alarm method, which comprises the following steps: The method comprises the steps of constructing a cooperative unmanned aerial vehicle operation scene and a non-cooperative unmanned aerial vehicle operation scene, wherein the cooperative unmanned aerial vehicle operation scene comprises interactive flight of two cooperative unmanned aerial vehicles with clear flight intention on a preset route; The method comprises the steps of predicting tracks and constructing a protection zone, wherein in a cooperative unmanned aerial vehicle running scene, historical track data of the cooperative unmanned aerial vehicle are obtained and input into a prediction model, the predicted tracks are output, and meanwhile, a dynamic Reich model protection zone is constructed; Judging collision according to the predicted track, wherein a spherical crown overlapping degree judgment method is used for judging collision in a cooperative unmanned aerial vehicle operation scene; and sending out an alarm according to the judgment result. A second aspect of the present invention provides a low-altitude unmanned aerial vehicle air collision risk warning system, comprising: The system comprises a scene construction module, a scene construction module and a control module, wherein the scene construction module is used for constructing a cooperation type unmanned aerial vehicle operation scene and a non-cooperation type unmanned aerial vehicle operation scene, wherein the cooperation type unmanned aerial vehicle operation scene comprises that two cooperation type unmanned aerial vehicles with definite flight intention fly interactively on a preset route; the track prediction module is used for predicting the track of the cooperative unmanned aerial vehicle; The protection zone construction module is used for constructing a dynamic Reich m