Hierarchical encoder-decoder architecture for carrier airwake prediction using attention in frequency domain

The carrier airwake is the result of ocean air flowing across the stern of a moving aircraft carrier. This airwake has a huge impact on how accurately a carrier-based aircraft can land. If it is possible to make accurate predictions for the future carrier airwake, the predicted results will be used to aid the landing process and ensure a more precise landing. In this study, the properties of the carrier airwake are first analyzed to provide a basis for the subsequent design of the neural network. Then we propose a hierarchical encoder-decoder neural network with an attention mechanism in the frequency domain to simultaneously predict the carrier airwake components in all directions. Finally, the full comprehensive experiments show that our model for predicting the carrier airwake is accurate.