Research on Satellite Orbit Prediction Based on Deep Learning

With the rapid development of aerospace technology, accurate prediction of satellite orbits has become the key to ensuring the smooth implementation of space missions. Given the complex characteristics of nonlinear, dynamic, and multi-scale feature coupling in satellite orbit data, this paper proposes a satellite orbit prediction method based on the fusion of long short-term memory network (LSTM) and attention mechanism. Through multiple simulation tests, the LSTM that combined with Attention Model (LAM) proposed in this paper provides accurate orbit position and velocity prediction, and maintains good stability and robustness in complex spatial environments. This study provides a new technological path for real-time and accurate prediction of satellite orbits, which is of great significance for improving satellite operation management and space safety monitoring.