Analysis of Pilot Workload during Power-off Stall Based on HRV Features

Power-off stall training is designed to simulate the emergency situation of engine failure or low-speed flight, which leads to a sudden increase in pilot workload and is a key trigger for operational errors. Based on the monitoring of physiological signals of heart rate variability (HRV), four selected features were screened by correlation test and importance ranking, and it was found that mean_hr, max_hr and min_hr reached the highest level during the recovery from stall, while median_nni decreased to the lowest level. In addition, four machine learning models, namely SVM (linear kernel), SVM (RBF kernel), KNN and RF, were selected to classify and predict workloads, and it was found that the RF model performed optimally in the classification task.