The helicopter pilots' mental workload is a contributing factor that detrimentally affects their performance, which can lead to flight accidents. This study proposes a method to assess and identify helicopter pilots' mental workload in real flight scenarios. Three helicopter pilots were involved in the experiment, and photoplethysmography (PPG) signals were collected during three flight phases (takeoff climb, cruise, and approach landing) in cross-country flight. NASA-TLX scales were used to measure the pilots' mental workload level, and heart rate variability (HRV) features were extracted to further identify mental workload. Statistical analysis results showed that most of the features exhibited significant differences across the three mental workload levels. Three machine learning algorithms, including support vector machines (SVM), k-nearest neighbors (KNN), and random forests (RF), were used for a three-class mental workload classification. The results demonstrated that the HRV feature set achieved the highest classification accuracy of 85.9% using KNN algorithm, and the highest area under the curve (AUC) of 0.934 was obtained using RF algorithm. This research collected the first HRV dataset of real civil helicopter cross-country flight in China. Based on the real flight dataset, the findings of this study could provide a valuable reference for the classification of mental workload among helicopter pilots during real flight.
Assessment of Pilot Mental Workload Based on Physiological Signals: A Real Helicopter Cross-country Flight Study
2023-10-11
2693442 byte
Conference paper
Electronic Resource
English