Development of a Control and Prediction Method for the Attitude of a 3.5 U CubeSat Optimized with Artificial Neural Networks for Sustainable Use in Low-Earth Orbit Missions
- Paper ID
99429
- DOI
- author
- company
Universidad Nacional de Ingenieria, Peru; Universidad Ricardo Palma; Universidad Nacional Mayor de San Marcos; University of Luxembourg; Universidad Nacional Pedro Ruíz Gallo; Universidad Nacional de Ingeniería (Lima, Perù)
- country
Peru
- year
2025
- abstract
Attitude determination and control in a nanosatellite in Low-Earth Orbit (LEO) is essential as it influences key aspects such as maneuvers and telemetry. When a CubeSat reaches the end of its operational life or suffers failures in its electronic system, its ability to estimate the orientation is compromised. In these cases, it is essential to carry out a deorbiting maneuver for its disintegration in the atmosphere. During this process, the correct prediction of the orientation without the assistance of sensors becomes a crucial factor to be addressed by artificial intelligence and a predictive method, since it allows us to evaluate the state of the resulting fragments, predict the possible impact zone and minimize the risks associated with falling debris in advance. This research analyzes and proposes a method for precision attitude of a 3.5U CubeSat through an optimized control and prediction strategy with artificial neural networks, in the deorbiting stage of low orbit, to avoid Kessler Syndrome. The work process of this article had 3 phases: (1) The analysis of orbital mechanics and modeling of dynamic equations regarding the prediction of the LEO CubeSat attitude, which calculated and described the location and study trajectory for a single use of the actuator. (2) The design of an integration algorithm with extraction, which determined and controlled the orientation, through the fusion of advanced filters and the compensation of the neural network, the error of the torque or attitude was minimized, taking as actuators the application of the magnetorquers, which executed the required turns maximizing the atmospheric drag, which provided the orbital deceleration by varying the ballistic coefficient and producing the estimated descent. (3) The development of rotational time intervals, which made possible the decrease of the orbital speed and thus went from a circular orbit to a trajectory with a destination at Point Nemo, with which the reduction of the risk of impact on inhabited areas of the Earth's surface was tested, since this system is scalable to larger nanosatellites. Using simulation techniques with robust software such as Ansys STK, MATLAB and Python, they showed results that surpass conventional methods, offering a more precise and reliable estimate even under discontinuous measurement conditions. In this way, we have an efficient prediction system with high reliability of controlled maneuverability, which directly impacts the sustainability of new space technologies, to dispense with additional mechanisms.