Assessing Sensor Technologies and Modelling Approaches for a Small-Scale Life Support System

Abstract

The accuracy and reliability of Life Support System (LSS) simulations, such as V-HAB (Virtual Habitat), are crucial for the design and operation of future space habitats and long-duration missions. To bridge the gap between simulated and real space environment, virtual sensor models were incorporated into V-HAB. These models enable the simulation of realistic sensor characteristics and anomalies, accounting for noise, drift and other imperfections inherent in physical sensors. By integrating virtual sensor, the simulation output will more closely resemble actual system behavior, paving the way for digital twin applications, including the development of a small-scale LSS at the University of Stuttgart. Additional research was conducted to investigate the feasibility of Machine Learning (ML) based virtual sensors to replace their physical counterpart, potentially reducing hardware cost and increasing system flexibility. The ML sensors were trained on real data to predict the measurement of their physical counterparts by leveraging information from other available sensors. This design increases redundancy or allows substitution of a physical sensor when needed. By integrating virtual sensor models into V-HAB and studying the potential of ML-based sensors, this study aims to enhance realism and predictive capabilities of LSS simulations such as V-HAB, contributing to the development of more efficient, resilient and autonomous LSSs for future space exploration missions.