Pilot workload is a critical factor affecting flight safety, particularly in emergency situations such as engine failure during forced landings. This study aims to assess pilot workload during engine failure-induced forced landings compared to normal landings using EEG signals. EEG data were recorded from 21 pilots using a flight simulator and EEG acquisition equipment during both forced and normal landings. Subjective workload scores were obtained using a subjective workload scale. EEG features were extracted, and nonparametric tests were applied to assess the significance of these features at different workload levels. Subsequently, machine learning algorithms (SVC, KNN, RF and LightGBM) were employed to develop models for workload evaluation. The LightGBM model achieved a peak accuracy of 99.5% using the top 80% of all features (time-domain, frequency-domain and nonlinear features) as inputs. This study provides an effective quantitative approach for assessing pilot workload during emergency situations and offers valuable insights for improving flight safety and optimising pilot training programs.
YUAN et al. (Tue,) studied this question.