基于Multivariate BiLSTM-FCNs的机动实时识别方法  

A Real time Identification Method for Maneuvers Based on Multivariate BiLSTM FCNs

作　　者：赵智伟[1] 袁伟伟[1] 关东海[1] ZHAO Zhi-wei;YUAN Wei-wei;GUAN Dong-hai(Nanjing University of Aeronautics and Astronautics,Nanjing 211000,China)

机构地区：[1]南京航空航天大学,江苏南京211000

出　　处：《航空计算技术》2023年第2期65-69,共5页Aeronautical Computing Technique

基　　金：航空科学基金项目资助(ASFC-20200055052005)。

摘　　要：机动的实时、准确识别对飞行员意图识别、空战态势感知具有重要意义。提出了将完整机动划分为若干机动单元,采用基于Multivariate BiLSTM-FCNs的方法自动提取并分析机动单元数据内部的时序特征和依赖关系,实现对机动单元的精确识别。然后通过机动单元窗口过滤噪声机动单元,实时监测机动是否执行或发生变化,实现机动单元到完整机动的实时识别。通过仿真实验识别了筋斗、盘旋、俯冲、爬升和破S机动。实验结果表明,在平均识别延迟率仅为26.19%的情况下,机动识别准确率高达96.67%。Real time and accurate identification of maneuvers is of great significance for pilot intention recognition and air combat situational awareness.It is proposed to divide the complete maneuver into several maneuver units.A method based on Multivariate BiLSTM FCNs is used to automatically extract and analyze the temporal features and dependencies within the maneuver unit data,so as to achieve accurate identification of maneuver units.Then,noisy maneuver units are filtered through the maneuver unit window to monitor whether the maneuver is executed or changed in real time.Finally,the real time identification from maneuver units to complete maneuver is realized.The loop,circling,dive,climb and split S maneuvers are identified through simulation experiments.The experimental results show that the maneuver identification accuracy is as high as 96.67%with an average identification delay rate of 26.19%.

关 键 词：机动识别 空战机动 时间序列分类 双向长短期记忆网络 全卷积网络 

分 类 号：V323[航空宇航科学与技术—人机与环境工程] TP391[自动化与计算机技术—计算机应用技术]