Research Article

Anomaly Detection in QAR Data Using VAE-LSTM with Multihead Self-Attention Mechanism

Table 1

Comparison of anomaly detection performance based on precision, recall, and F1-score.

MethodsClimbCruiseDescent
RF1RF1RF1

IF0.58330.32930.42110.63330.70390.66670.51080.70550.5926
LSTMS0.69440.98010.81950.44250.94400.61360.82400.63520.7769
LSTM-AE0.88850.94260.91470.87680.94170.91340.72840.85340.7860
LSTM-VAE0.77220.94430.84960.81190.97160.89610.89021.00.9419
VAE-based MHSA-LSTM0.91450.98330.95030.88401.00.93840.94531.00.9718