Research on a BiLSTM-MHA-based prediction method for pilot's heart rate variability
LIAO Wenyu
LI Limin
PAN Youbin
LIU Hongxiang
YUAN Jiajun
Abstract：To meet the requirements of civil aviation regulations regarding crew fatigue management,a prediction model based on the bidirectional long short-term memory with multi-head attention(BiLSTM-MHA)was designed using heart rate variability(HRV)as the fatigue assessment metric.Overlapping window sampling was used to extract HRV parameter features from pilots'physiological data,which were then utilized as model inputs.Model performance was evaluated using the coefficient of determination(R²),mean absolute error(SAME),and root mean square error(SRMSE).The predictive accuracy of the BiLSTM-MHA model was systematically compared against three other LSTM-based models.Test results showed that the BiLSTM-MHA model consistently outperformed the comparative models in predicting HRV parameters,providing a critical foundation for early intervention in flight fatigue and offering significant potential to enhance aviation safety management.
Keywords：flight safetyfatigue predictionheart rate variabilityBiLSTM-MHA
Publication Date：2026-02-25
Online Publishing Date：2026-03-26(First online date of this platform, not the publication date of the document)
Pages：6( 39-44 )