Real-time Monitoring of Pilot's Stress Response Based on Multi-modal Emotion Computing

Abstract

Stress responses during flight operations are a critical factor contributing to pilot errors and aviation accidents. Real-time monitoring of pilots' stress responses is essential for maintaining aviation safety. This study proposes an innovative triadic coupling emotional computing theory model integrating physiological, behavioral, and task contexts to dynamically monitor acute stress responses in pilots. The model aims to explore and quantify dynamic characteristics and patterns of stress responses across multiple dimensions and modalities, addressing limitations in existing emotional computing approaches for stress response analysis while enabling precise monitoring under various operational scenarios. By incorporating physiological signals (EEG θ/α waves, ECG LF/HF ratio, EDA skin conductance slope), behavioral features (sweep path entropy, grip strength gradient), and task context parameters (overload bias) with dynamic parameters, the model constructs a dynamic weighting function and triadic coupling output equation. Validation demonstrates that this model achieves adaptive physiological-behavioral weighting adjustments based on flight phases, In the low utetheisa kong penetration mission, the model achieved an accuracy rate of 92.4%, with the stress index providing a 1.8-second early warning for operational errors.