Detection and analysis of fatigue flight features using the fusion of pilot motion behavior and EEG information

Pilots are susceptible to fatigue during flight operations, posing significant risks to flight safety. However, single-feature-based detection methods often lack accuracy and robustness.

This study proposes a fatigue classification approach that integrates EEG features and motion behavior features to enhance fatigue recognition and improve aviation safety. The method extracts energy ratios of EEG frequency bands (α, β, θ, δ), incorporates forearm sample entropy and Euler angle standard deviation, and applies Pearson correlation analysis to select key features. Finally, a Support Vector Machine (SVM) classifier is employed to achieve precise fatigue classification.

Experimental findings indicate that the proposed method achieves a test accuracy of 93.67 %, outperforming existing fatigue detection techniques while operating with a reduced computational cost.

This study addresses a gap in current research by integrating physiological and behavioral data for fatigue classification, demonstrating that the fusion of multi-source information significantly enhances detection accuracy and stability compared to single-feature methods. The findings contribute to improved pilot performance and enhanced flight safety by increasing the reliability of fatigue monitoring systems.