A Transformer-Based Model for Stratospheric Wind Field Prediction

Abstract

The stratospheric wind field environment is a key influence on the flight control of stratospheric aerostat, which is a popular field of research in recent years. In this paper, a Transformer-based stratospheric wind field prediction model is proposed for the construction of stratospheric wind field model and prediction, and the feasibility of the model is verified on a 5-year historical data set of the wind field in a place, then the parameters of the model are tuned by ablation studies, and finally, the Transformer prediction model is compared with GRU, LSTM and BP prediction models. The experimental results have shown that the proposed Transformer stratospheric wind prediction model can fully extract global features and its prediction accuracy is significantly higher than those of other prediction models.