The last few years have shown that artificial intelligence will become an integral part of space in the future to realize the autonomy of the system without ground control and crew interactions. This will give people access to missions that were previously unthinkable and fulfill the dream of mankind to explore deep space. In future space missions AI algorithms have to be applied, which can learn not only offline from single data sets but also using data streams in real time and make corresponding predictions. Future missions foresee the utilization of data processing components with higher performance. These will provide the platform to implement adequate techniques for processing large amounts of data. The ISS Columbus module offers the perfect basis for proof of concept and technology maturing, using available hardware resources, to show how such a system can look like and what would be the benefit for future space missions. Therefore, here we propose the development of a system for the analysis of data streams using AI techniques to give an early indication of the health of a selected COLUMBUS subsystem. Based on this prediction and taking into account further parameters, the system can decide autonomously, which steps have to be done to avoid further damage. In addition to that, the system shall control the housekeeping TM rate and therefore reduce it under nominal conditions. Once the FDIR starts raising early indications of anomalies, the housekeeping TM rate will increase to provide early warnings. The system will learn which additional related data is relevant for Ground (not part of the usual housekeeping TM already sent) and will send it in anticipation to Ground TC commanding if needed. The autonomous FDIR of a subsystem and TM optimization will ensure safety and success of long-term missions in unknown environments through adaptability and reducing human interaction.
