基于经验直觉的无人机威胁规避机动决策方法  

Threat Avoidance Maneuvering Decision‑Making for Unmanned Aerial Vehicles Based on Experiences and Intuitions

作　　者：王盼盼 陈谋[1] 吴庆宪[1] 邵书义[1] WANG Panpan;CHEN Mou;WU Qingxian;SHAO Shuyi(College of Automation Engineering,Nanjing University of Aeronautics&Astronautics,Nanjing 211106,China)

机构地区：[1]南京航空航天大学自动化学院,南京211106

出　　处：《南京航空航天大学学报》2022年第1期77-85,共9页Journal of Nanjing University of Aeronautics & Astronautics

基　　金：国家自然科学基金(U2013201);航空科学基金(201957052001);江苏省博士后基金(2020Z112)。

摘　　要：针对无人机飞行过程中遇到威胁需要及时安全规避的问题,本文提出一种基于经验直觉的无人机机动规避决策算法。首先将无人机威胁安全规避机动过程中所获取的态势信息,按照规避机动区域进行结构化处理并标注相应的机动决策标签,建立基于数据的无人机情景决策知识库。其次,利用无人机情景决策知识库中的数据训练长短时记忆网络以实现情景与机动决策之间的正确映射。然后,根据人类基于直觉进行行动决策的原理,设计无人机基于经验直觉的机动规避算法。最后,通过仿真分析表明所提出无人机威胁规避算法的有效性。In this paper,an unmanned aerial vehicle(UAV)maneuver avoidance decision algorithm based on experiences and intuitions is proposed for threats UAVs encounter during flight processes.Firstly,the situation information obtained during the safety threat avoidance maneuver of the UAV is structured according to the maneuver avoidance area,and the corresponding maneuver decision tag set is recorded.A UAV scenario decision-making knowledge is also established based on the data.Secondly,the long and short-term memory network is used to map the relations between the data scenario and the maneuvering decision in the UAV scenario decision-making knowledge database.Thirdly,according to the principle that human make action decisions based on their intuitions,a UAV maneuver avoidance algorithm is designed based on intuitions.Finally,the simulation analysis is conducted to illustrate the effectiveness of the proposed real-time threat avoidance algorithm for UAVs.

关 键 词：无人机 直觉决策 情景决策知识库 长短时记忆网络 威胁规避机动 

分 类 号：V279[航空宇航科学与技术—飞行器设计]