Zhou Wen-Gang, Yu Pan-Pan, Wu Liang-Hai, Cao Yu-Fei, Zhou Yue, Yuan Jia-Jun
Flight Technology College, Civil Aviation Flight University of China, Guanghan, China.
Front Neurosci. 2024 Sep 23;18:1450416. doi: 10.3389/fnins.2024.1450416. eCollection 2024.
To identify the cognitive load of different turning tasks in simulated flight, a flight experiment was designed based on real "preliminary screening" training modules for pilots.
Heart Rate Variability (HRV) and flight data were collected during the experiments using a flight simulator and a heart rate sensor bracelet. The turning behaviors in flight were classified into climbing turns, descending turns, and level flight turns. A recognition model for the cognitive load associated with these turning behaviors was developed using machine learning and deep learning algorithms.
pnni_20, range_nni, rmssd, sdsd, nni_20, sd1, triangular_index indicators are negatively correlated with different turning load. The LSTM-Attention model excelled in recognizing turning tasks with varying cognitive load, achieving an F1 score of 0.9491.
Specific HRV characteristics can be used to analyze cognitive load in different turn-ing tasks, and the LSTM-Attention model can provide references for future studies on the selection characteristics of pilot cognitive load, and offer guidance for pilot training, thus having significant implications for pilot training and flight safety.
为识别模拟飞行中不同转向任务的认知负荷，基于真实的飞行员“初步筛选”训练模块设计了一项飞行实验。
在实验过程中，使用飞行模拟器和心率传感器手环收集心率变异性（HRV）和飞行数据。飞行中的转向行为分为爬升转弯、下降转弯和水平飞行转弯。使用机器学习和深度学习算法开发了与这些转向行为相关的认知负荷识别模型。
pnni_20、range_nni、rmssd、sdsd、nni_20、sd1、三角指数指标与不同转向负荷呈负相关。LSTM-Attention模型在识别具有不同认知负荷的转向任务方面表现出色，F1分数达到0.9491。
特定的HRV特征可用于分析不同转向任务中的认知负荷，LSTM-Attention模型可为未来飞行员认知负荷选择特征的研究提供参考，并为飞行员训练提供指导，对飞行员训练和飞行安全具有重要意义。
