CF-LSTM-based post-maneuver position prediction for GEO spacecraft

Abstract

Prediction of the position of Geosynchronous (GEO) spacecraft after a maneuver is crucial for space domain awareness (SDA), as it can enhance the flexibility of the space surveillance network (SSN). The longitude is a unique parameter that can be freely assigned to GEO spacecraft. In this paper, we propose a prediction model for the longitude of a GEO spacecraft based on Causal Feature Long Short-Term Memory networks (CF-LSTM). Initially, we analyze the historical data of GEO spacecraft in a time-series geographic position and identify the causal parameters of longitude using the Gaussian perturbation equation. Subsequently, we present the CF-LSTM to predict the longitude. Our experimental analysis shows that the proposed method reduces the mean absolute error (MAE) and mean square error (MSE) by 60.92% and 18.75%, respectively, compared to conventional LSTM.