Fig. 1: Overview of the proposed WTFTP+ framework.
From: Multi-horizon flight trajectory prediction enabled by time-frequency wavelet transform
The WTFTP+ framework is implemented with an encoder-decoder architecture, in which the encoder is employed to extract the abstract features from the observed trajectory sequence, while the decoder is applied to generate the optimal wavelet coefficients for the predicted horizons. In the decoder, to avoid breaking orthogonality of multi-scale wavelet components, the orthogonal wavelet components are first bridged into the time domain for promotive interaction, and subsequently bridged back into the wavelet domain to obtain enhanced features of inherent dependencies within multi-scale flight patterns.
