Satellite orbit forecasting method based on artificial neural network algorithm

Abstract

A satellite orbit forecasting method based on an artificial neural network algorithm comprises the steps: reading TLE two-line orbit data containing six elements, and generating a simulation curve; processing the discontinuous curve by adopting a splicing method to form a continuous curve; on the basis of the six elements, adding a TLE data acquisition timestamp T to construct seven elements of TLE prediction suitable for TLE data prediction; training the orbit inclination angle i, the orbit eccentricity e and the average motion n in the six elements of the orbit by adopting a prediction algorithm based on LSTM; performing training by adopting an element prediction algorithm based on linear regression to obtain an optimal function; and repeating the training and correcting to obtain an executable prediction model. When the method provided by the invention is used for forecasting a satellite orbit, the overall position error precision of the model can be controlled within 5km in a short period and a middle period, and the overall position error precision of the model can be controlled within 23km in a long-term forecasting process.

Description

technical field [0001] The invention relates to the field of space technology, in particular to a satellite orbit prediction method based on an artificial neural network algorithm. Background technique [0002] At present, the satellite orbit is mainly predicted by the classical mechanical model in the world. Satellite orbits, especially low-orbit satellites, are greatly affected by perturbation forces, so traditional methods have defined each perturbation force model, including atmospheric drag models, Earth's non-spherical gravity, and solar light pressure radiation. Perturbed power. Due to the low accuracy of the perturbation force model and the errors caused by the parameters introduced into the model, the prediction accuracy of its low-order analytical solution is relatively low, and the process of high-order analytical solution is very cumbersome, resulting in low computational efficiency or even no solution. [0003] North American Aerospace Defense (NORAD) provides...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to provide a satellite orbit prediction method based on artificial neural network algorithm, which solves the problems of cumbersome analysis process, low calculation efficiency and even no solution in the solution process of traditional mechanical methods

Images