Digital Twinning in Non-Terrestrial Networks: Increasing the Operational Life of Satellite Batteries
Translated title
Digital Twinning in Non-Terrestrial Networks
Author
Salas Prendes, Ainoa
Term
4. semester
Education
Publication year
2025
Submitted on
2025-06-03
Pages
60
Abstract
Dette projekt udvikler en simulator af en geostationær satellitkonstellation (satellitter, der holder sig over samme punkt på Jorden) og en tilhørende digital tvilling og service med det mål at forlænge satellitbatteriers levetid. En digital tvilling er en virtuel model, der afspejler et fysisk system, så driftsbeslutninger og strategier kan afprøves, før de tages i brug. I den digitale tvillings batterimodel introduceres og sammenlignes to metoder: diskret-tids Markov-kæder (sandsynlighedsmodeller, hvor den næste tilstand afhænger af den nuværende) og autoregressive modeller (statistiske modeller, der bruger tidligere målinger). Markov-tilgangen var mere præcis. Til serviceimplementeringen anvendes forstærkningslæring til at fordele satellitternes ressourcer på en måde, der holder batterierne stabile, hvilket er tæt knyttet til batteriernes holdbarhed. Resultaterne viser, at den digitale tvilling kan bruges til at opretholde batteristabilitet og dermed understøtte det oprindelige mål om at forlænge den forventede batterilevetid.
This project develops a simulator of a geostationary satellite constellation (satellites that remain over the same point on Earth) and a corresponding Digital Twin model and service, with the goal of extending satellite battery life. A Digital Twin is a virtual representation that mirrors a physical system, allowing operating decisions and strategies to be tested before they are applied. For the battery model in the Digital Twin, we introduce and compare two approaches: discrete-time Markov chains (probabilistic models where the next state depends on the current state) and autoregressive models (statistical models that use past measurements). The Markov approach was more accurate. For the service implementation, we use reinforcement learning to allocate satellite resources in a way that keeps battery levels stable, which is closely related to long-term durability. The results show that the Digital Twin can be used to maintain battery stability, supporting the original aim of increasing expected battery lifetime.
[This summary has been rewritten with the help of AI based on the project's original abstract]
Keywords
Documents