A Flight Parameter-Based Flight Load Prediction Method for Aircraft Fatigue Life Monitoring via Maneuver Recognition and Deep Learning

The measured load spectrum is essential in evaluating the fatigue life and reliability of an aircraft. To boost the accuracy of flight load estimation, a novel method based on maneuver recognition and deep learning is proposed in this paper. In the aspect of maneuver recognition, a fast and accurate dynamic time warping algorithm is harnessed to effectively identify various maneuvers such as loop and half roll. Furthermore, for a certain determined flight maneuver, a transformer network with self-attention is adopted to establish the prediction model between flight parameters and flight loads. Finally, experimental results manifest that the proposed method can accurately estimate the operational flight loads based on flight parameters. In addition, a comparison with many other neural networks is conducted to demonstrate the advantages of the proposed method. This study provides significant supports in improving the fatigue life monitoring of an aircraft.