Self-organizing neural network-based unmanned aerial vehicle mission planning method

Abstract

The invention provides a self-organizing neural network-based unmanned aerial vehicle mission planning method. The method includes a self-organizing neural network initialization step, a winning nodeselection step, a winning node winning neighborhood calculating step, a network parameter updating step and a dynamic response step. According to the method, the structure of a self-organizing neuralnetwork and the weights of nodes are initialized; the weights are normalized; an input vector is received; a node is selected as a winning node, the inner product of the selected node and the input vector being the largest; all nodes in the winning neighborhood of the winning node are calculated; the weights of the nodes are updated differently according to distances from the nodes to the winningnode; and when an emergency situation occurs, the parameters of the network are dynamically adjusted, so that timely response can be made. The weights of the nodes in the winning neighborhood are updated; a multi-constraint condition is introduced during the execution process of the algorithm; a chaotic mechanism is introduced to solve the problem of detouring of unmanned aerial vehicles at a plurality of task points; and a dynamic response mechanism is adopted, so that the adaptability of an unmanned aerial vehicle group in a dynamic environment can be improved, and therefore, more effectiveunmanned aerial vehicle mission planning can be realized.

Description

technical field [0001] The invention belongs to the field of heuristic algorithms, in particular to an unmanned aerial vehicle task planning method based on self-organizing neural network. Mainly aiming at the multi-UAV mission planning problem with constraints in dynamic environment, the SOM algorithm is used to realize the mission planning of UAVs under multi-constraint conditions, which improves the adaptability of UAV fleets in dynamic environments. Background technique [0002] In the context of the modern battlefield, UAVs are developing from a single sortie independent mission to multi-sorties and multi-type fleet operations. UAV mission planning is an important part of UAV fleet cooperative operations. Through mission planning, the pre-war preparation time can be greatly reduced, and the types of missions completed by the multi-UAV system can be more diverse. At the same time, the quality and efficiency of mission completion can also be greatly improved. [0003] At...

AI Technical Summary

Problems solved by technology

[0003] At the same time, multi-UAV swarms are often in an unknown dynamic environment, and the battlefield situation is changing rapidly. During the mission execution process, UAVs will fail or be damaged and withdraw from mission execution. At the same time, mission points may also move dynamically, which requires dynamic The task planning method has strong self-organization and self-adaptive ability

[0004] At present, there are still many challenges in solving such problems, including the inconsistent UAV cooperative mission planning model, the establishment of the model without considering the actual battlefield needs, the time-consuming task planning algorithm, and the poor adaptability of the algorithm in dynamic environments. The real-time performance and accuracy of the algorithm cannot be balanced, etc.

Images