This study explores methods to improve the accuracy of satellite maneuver detection by comparing traditional statistical methods with machine learning and neural network-based methods. The two alternative methods explored are a Random Forest algorithm and a Deep Neural Network. Simulation data is generated and appropriate features are selected and computed. Data is split into training and validation. Next, alternative methods are trained and their performance is validated. Optimization techniques are then applied to achieve better results. Finally, the three methods are compared to each other. Conclusions and future work are described.
Satellite Maneuver Detection Using Machine Learning and Neural Network MethodsBehaviors
05.03.2022
5222020 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Geosynchronous Satellite Maneuver Classification via Supervised Machine Learning
British Library Conference Proceedings | 2021
|Satellite Maneuver Detection Using Two-line Elements Data
British Library Conference Proceedings | 2007
|Satellite Maneuver Detection with Optical Survey Observations
TIBKAT | 2021
|Satellite Maneuver Detection with Optical Survey Observations
British Library Conference Proceedings | 2020
|Satellite Maneuver Evaluation Tool
NTIS | 1990
|