摘要
对飞行员行为进行识别和解码有助于提高航空安全。通过分析飞行员在执行任务时的生理信号变化,可以精确识别其行为。使用DA42飞行模拟器对25名飞行学员进行不同转弯任务的实验,采用近红外脑功能成像(fNIRS)技术收集数据,提取每个通道中氧合血红蛋白(OxyHb)饱和度的变化,并进行统计分析。重点讨论了利用fNIRS采集的△OxyHb对飞行员行为识别的可行性,并提出了基于深度学习的飞行员转弯行为识别模型。结果显示,在执行不同转弯任务时,飞行员大脑的△OxyHb与前额叶皮层、运动皮层和枕叶皮层高度相关。提出的CNN-LSTM-Attention飞行员转弯行为识别模型相较于CNN-LSTM模型具有更好的泛化能力,识别准确率可达96%,明显高于其他单一模型。研究成果为飞行员训练提供了重要参考依据。
The identification and decoding of pilot behaviour can help to improve aviation safety.Pilot behaviour can be accurately identified by analysing changes in their physiological signals while performing tasks.Experimentation of 25 flight trainees on different turning tasks using the DA42 flight simulator.fNIRS(Functional Near Infrared Brain Imaging)is used to collect data,extract the changes in the saturation of Oxyhaemoglobin(OxyHb)in each channel,and statistically analyse the data.The feasibility of usingΔOxyHb collected by fNIRS for pilot behaviour recognition is highlighted,and a deep learning-based pilot turn behaviour recognition model is proposed.The results show that theΔOxyHb of the pilot's brain is highly correlated with prefrontal cortex,motor cortex and occipital cortex when performing different turning tasks.The proposed CNN-LSTM-Attention pilot turning behaviour recognition model has better generalization ability compared to the CNN-LSTM model,and the recognition accuracy can reach 96%,which is significantly higher than other single models.The research results provide an important reference for pilot training.
作者
张庆峰
柯熙俊
张晨阳
陈爽
袁家俊
ZHANG Qingfeng;KE Xijun;ZHANG Chenyang;CHEN Shuang;YUAN Jiajun(Flight Technology Academy,Civil Aviation Flight University of China,Guanghan 618307;School of Transportation and Logistics,Southwest Jiaotong University,Chengdu 611756;Sichuan General Aviation Aircraft Maintenance Engineering Technology Research Centre,Civil Aviation Flight University of China,Guanghan 618307)
出处
《舰船电子工程》
2025年第6期95-100,共6页
Ship Electronic Engineering
基金
2024年研究生科研创新基金项目“基于fNIRS的飞行员不同转弯行为识别研究”(编号:24CAFUC10170)
中国民用航空飞行学院科研处自主课题“DA42NG飞机复合材料损伤检测技术研究”(编号:GAMRC2023YB02)资助。
关键词
飞行安全
行为识别
深度学习
近红外脑功能成像
飞行模拟器
aviation safety
behavioural recognition
deep learning
near infrared functional brain imaging
flight simulators
