CN-122015847-A - Unmanned aerial vehicle cluster conflict decoupling and track reconstruction method and system

CN122015847ACN 122015847 ACN122015847 ACN 122015847ACN-122015847-A

Abstract

The invention provides a method and a system for unmanned aerial vehicle cluster conflict decoupling and flight path reconstruction, wherein the method comprises the steps of constructing an unmanned aerial vehicle-task-airspace heterogeneous relation diagram according to unmanned aerial vehicle nodes, task nodes, airspace nodes, task association sides, airspace conflict sides and energy consumption constraint sides; introducing an attention mechanism when aggregating neighbor information, distributing different importance weights for different neighbors, acquiring attention coefficients, updating node characteristics based on the attention coefficients, carrying out multi-layer stacking and global information fusion to aggregate final characteristics of all nodes, acquiring global characteristic vectors, acquiring control instruction vectors of each unmanned aerial vehicle node and conflict risks between any two unmanned aerial vehicles, judging whether the conflict risks exceed a risk threshold, and if so, triggering a decoupling mechanism. The invention can realize conflict decoupling and flight path collaborative planning for tasks of the unmanned aerial vehicle clusters in a complex low-altitude environment, thereby realizing safe, efficient and collaborative flight of the unmanned aerial vehicle clusters.

Inventors

LUO LUFENG
DENG KAI
CHEN MINGYOU
XIAO HUI

Assignees

佛山大学

Dates

Publication Date: 20260512
Application Date: 20260120

Claims (10)

1. The unmanned aerial vehicle cluster conflict decoupling and track reconstruction method is characterized by comprising the following steps of: Acquiring unmanned plane nodes, task nodes, airspace nodes, task associated edges, airspace conflict edges and energy consumption constraint edges, and constructing an unmanned plane-task-airspace heterogeneous relationship diagram according to the unmanned plane nodes, the task nodes, the airspace nodes, the task associated edges, the airspace conflict edges and the energy consumption constraint edges; based on the graph neural network, neighbor sampling and aggregation are carried out on the nodes, initial updating characteristics are obtained, an attention mechanism is introduced when neighbor information is aggregated, different importance weights are distributed for different neighbors, attention coefficients are obtained, and the node characteristics are updated based on the attention coefficients; performing multi-layer stacking and global information fusion to aggregate the final characteristics of all nodes and obtain a global characteristic vector; and acquiring a control instruction vector of each unmanned aerial vehicle node and conflict risks between any two unmanned aerial vehicles, judging whether the conflict risks exceed a risk threshold, and if so, triggering a decoupling mechanism.
2. A method for cluster conflict decoupling and track reconstruction of unmanned aerial vehicles as claimed in claim 1, the unmanned aerial vehicle cluster conflict decoupling and track reconstruction method is characterized by further comprising the following steps of: acquiring airspace conflict edge weights and distances between the unmanned aerial vehicle and the obstacle, and acquiring safety conflict cost according to the airspace conflict edge weights and the distances between the unmanned aerial vehicle and the obstacle; Acquiring energy consumption constraint side weights and motor power consumption, and acquiring energy consumption constraint costs according to the energy consumption constraint side weights and the motor power consumption; Acquiring task association edge weights and unmanned aerial vehicle states, and acquiring task cooperation costs for ensuring that unmanned aerial vehicles of high-priority tasks reach targets according to the task association edge weights and the unmanned aerial vehicle states; and realizing the track reconstruction optimization strategy according to the safety conflict cost, the energy consumption constraint cost and the task cooperative cost.
3. The unmanned aerial vehicle cluster conflict decoupling and track reconstruction method of claim 2, wherein the specific method for implementing the track reconstruction optimization strategy comprises the following steps: Constructing power constraint, control quantity constraint and electric quantity constraint conditions; Constructing a total cost function according to the safety conflict cost, the energy consumption constraint cost and the task cooperative cost; and based on the constructed power constraint, control quantity constraint and electric quantity constraint conditions, minimizing the total cost function and obtaining an optimal control sequence.
4. A method for cluster conflict decoupling and track reconstruction of an unmanned aerial vehicle as claimed in claim 3, wherein the specific method for implementing the track reconstruction optimization strategy further comprises the steps of: And triggering global track reconstruction when the collision risk is too high, the electric quantity is seriously insufficient or the environment is suddenly changed.
5. The unmanned aerial vehicle cluster conflict decoupling and track reconstruction method of claim 4, wherein the method for implementing the track reconstruction optimization strategy further comprises the step of re-planning only the affected unmanned aerial vehicle subset after triggering the reconstruction, not the entire cluster.
6. A method for cluster conflict decoupling and track reconstruction of unmanned aerial vehicles as claimed in claim 5, the unmanned aerial vehicle cluster conflict decoupling and track reconstruction method is characterized by further comprising the following steps of: judging whether a decoupling mechanism is triggered, if so, taking the unmanned aerial vehicle as a node, constructing a similarity matrix together by using the task association degree and the airspace conflict degree, dividing the cluster into a plurality of subgroups according to the similarity matrix, and executing the next step, otherwise, maintaining the current situation; Based on model predictive control, distributing a corresponding main target for each sub-group, and calculating a cluster center reference track, wherein the main target is used for minimizing the local cost of the sub-group; Introducing an artificial potential field between coarse-granularity track points generated by model predictive control to perform online fine adjustment, and treating burst obstacles; And triggering fusion and recovery of the unmanned aerial vehicle subgroups after conflict risk among the subgroups is relieved or task stages are completed.
7. The unmanned aerial vehicle cluster conflict decoupling and track reconstruction method of claim 6, wherein the specific method for determining whether to trigger the decoupling mechanism comprises the steps of: And detecting whether the energy state of the unmanned aerial vehicle in the cluster is unbalanced, whether the current task modularization degree reaches the standard or not and whether the conflict risk exceeds a risk threshold, if any condition is met, triggering a decoupling mechanism, and otherwise, maintaining the current situation.
8. An unmanned aerial vehicle cluster conflict decoupling and track reconstruction system for implementing the unmanned aerial vehicle cluster conflict decoupling and track reconstruction method according to any one of claims 1 to 7, wherein the unmanned aerial vehicle cluster conflict decoupling and track reconstruction system comprises: The heterogeneous relation diagram construction module is used for acquiring unmanned aerial vehicle nodes, task nodes, airspace nodes, task associated edges, airspace conflict edges and energy consumption constraint edges and constructing an unmanned aerial vehicle-task-airspace heterogeneous relation diagram according to the unmanned aerial vehicle nodes, the task nodes, the airspace nodes, the task associated edges, the airspace conflict edges and the energy consumption constraint edges; The node characteristic updating module is used for carrying out neighbor sampling and aggregation on the nodes based on the graph neural network, acquiring preliminary updating characteristics, introducing an attention mechanism when aggregating neighbor information, distributing different importance weights for different neighbors, acquiring attention coefficients and updating the node characteristics based on the attention coefficients; the feature vector acquisition module is used for carrying out multi-layer stacking and global information fusion so as to aggregate the final features of all the nodes and acquire a global feature vector; The decoupling module is used for acquiring the control instruction vector of each unmanned aerial vehicle node and the conflict risk between any two unmanned aerial vehicles, judging whether the conflict risk exceeds a risk threshold, and if so, triggering a decoupling mechanism.
9. A unmanned aerial vehicle cluster conflict decoupling and flight path reconstruction system as claimed in claim 8, the unmanned aerial vehicle cluster conflict decoupling and track reconstruction system is characterized by further comprising: the safety conflict cost acquisition module is used for acquiring the airspace conflict edge weight and the distance between the unmanned aerial vehicle and the obstacle, and acquiring the safety conflict cost according to the airspace conflict edge weight and the distance between the unmanned aerial vehicle and the obstacle; The energy consumption constraint cost acquisition module is used for acquiring the energy consumption constraint edge weight and the motor power consumption and acquiring the energy consumption constraint cost according to the energy consumption constraint edge weight and the motor power consumption; The task cooperation cost acquisition module is used for acquiring task association side weights and unmanned aerial vehicle states, and acquiring task cooperation costs for ensuring that unmanned aerial vehicles of high-priority tasks reach targets according to the task association side weights and the unmanned aerial vehicle states; And the track reconstruction optimization strategy module is used for realizing a track reconstruction optimization strategy according to the safety conflict cost, the energy consumption constraint cost and the task cooperative cost.
10. The unmanned aerial vehicle cluster conflict decoupling and track reconstruction system of claim 9, wherein the track reconstruction optimization policy module comprises: The constraint condition construction unit is used for constructing power constraint, control quantity constraint and electric quantity constraint conditions; The total cost function construction unit is used for constructing a total cost function according to the safety conflict cost, the energy consumption constraint cost and the task cooperative cost; and the control strategy optimization unit is used for obtaining an optimal control sequence by minimizing the total cost function based on the constructed power constraint, the control quantity constraint and the electric quantity constraint condition.

Description

Unmanned aerial vehicle cluster conflict decoupling and track reconstruction method and system Technical Field The invention relates to the technical field of unmanned aerial vehicle cooperative control and path planning, in particular to a method and a system for unmanned aerial vehicle cluster conflict decoupling and track reconstruction, based on the energy consumption constraint graph neural network, the method integrates airspace structure, task association and energy efficiency to perform multi-objective optimization, and can realize safe, efficient and collaborative flight of the unmanned aerial vehicle cluster. Background Along with the acceleration of urban airspace digitization process and the diversified expansion of unmanned aerial vehicle application scenes, the multi-unmanned aerial vehicle cluster collaborative execution task becomes an important support for urban low-altitude economic development. In typical application scenarios such as logistics distribution, urban inspection, emergency response and the like, the unmanned aerial vehicle cluster can obviously improve task efficiency through collaborative operation. However, in a typical urban environment with a confined, open building, unmanned clusters are severely challenged to operate. The dense building groups form a complex three-dimensional airspace structure, and the channels for flight are limited and the directions are changeable, so that the tracks of multiple unmanned aerial vehicles are extremely easy to cross, wind and even collide. Such a space blocking phenomenon not only causes interruption or delay of task execution, but also causes serious safety hazards. Meanwhile, the problem of energy competition among unmanned aerial vehicles is increasingly remarkable, and the extra maneuvering action required for avoiding conflict can lead to rapid increase of energy consumption, so that the cruising ability and the task completion rate of the whole cluster are affected. Therefore, an intelligent decision method capable of uniformly modeling the air domain structure, the task association and the energy consumption constraint and having online optimization and self-adaption capabilities is urgently needed in the industry. Disclosure of Invention The invention aims to solve the defects in the prior art, and provides an unmanned aerial vehicle cluster conflict decoupling and track reconstruction method and system based on an energy consumption constraint graph neural network, which are mainly used for uniformly coding and learning the topological relation, airspace constraint and energy consumption characteristics of unmanned aerial vehicle clusters through the graph neural network, realizing safe, efficient and energy-saving multi-machine collaborative track planning, and finally improving the overall task completion rate and reliability of a cluster system, and the specific technical scheme is as follows: The unmanned aerial vehicle cluster conflict decoupling and track reconstruction method comprises the following steps: Acquiring unmanned plane nodes, task nodes, airspace nodes, task associated edges, airspace conflict edges and energy consumption constraint edges, and constructing an unmanned plane-task-airspace heterogeneous relationship diagram according to the unmanned plane nodes, the task nodes, the airspace nodes, the task associated edges, the airspace conflict edges and the energy consumption constraint edges; based on the graph neural network, neighbor sampling and aggregation are carried out on the nodes, initial updating characteristics are obtained, an attention mechanism is introduced when neighbor information is aggregated, different importance weights are distributed for different neighbors, attention coefficients are obtained, and the node characteristics are updated based on the attention coefficients; performing multi-layer stacking and global information fusion to aggregate the final characteristics of all nodes and obtain a global characteristic vector; and acquiring a control instruction vector of each unmanned aerial vehicle node and conflict risks between any two unmanned aerial vehicles, judging whether the conflict risks exceed a risk threshold, and if so, triggering a decoupling mechanism. According to the unmanned aerial vehicle cluster conflict decoupling and flight path reconstruction method, the unmanned aerial vehicle-task-airspace heterogeneous relation diagram is constructed, and the distributed decision and optimization are carried out by utilizing the graph neural network fused with energy consumption constraint, so that the conflict decoupling and flight path collaborative planning for multiple unmanned aerial vehicle cluster tasks can be realized in a complex low-altitude environment, and further the safe, efficient and collaborative flight of the unmanned aerial vehicle clusters is realized. Preferably, the unmanned aerial vehicle cluster conflict decoupling and track reconstruction method furt