Space explained
Artificial intelligence reduces space risks 12-hour solar warnings
Artificial intelligence works alongside fault management systems incorporated into spacecraft design and small satellite telescopes, known as SmallSats…
In brief
Combining COFFIES’ 12-hour solar forecasts, MBSE-integrated fault management, and repeated SmallSat observations creates a synergy that boosts spacecraft resilience, enables automated protections, and improves exoplanet data quality, supporting safer and more autonomous space missions.
Key points
- Use AI-based 12-hour solar predictions to trigger automated spacecraft protections
- Integrate fault management into MBSE from project inception to improve reliability
- Leverage SmallSats like Pandora for repeated, multi-wavelength observations to separate stellar and planetary signals
- Feed operational and environmental data from constellations into AI models for validation and improvement
- Design workflows that translate predictions directly into fault-management actions to minimize exposure and downtime
Space presents simultaneous invisible risks and technical opportunities: radiation and solar storms can threaten crews and satellites, while star characteristics complicate the reading of planetary spectra. In this context, artificial intelligence works alongside fault management systems incorporated into spacecraft design and small satellite telescopes, known as SmallSats, to reduce these risks. Projects like COFFIES, the Artemis missions, the International Space Station, and the Pandora mission are at the heart of this effort. Artificial intelligence reduces space risks.
When Artificial Intelligence Sees the Problem Before It Appears
The COFFIES team, an acronym for Consequence Of Fields and Flows in the Interior and Exterior of the Sun, has trained an artificial intelligence model capable of predicting the emergence of active regions on the Sun up to 12 hours before they appear on the surface. This prediction provides approximately half a day’s notice. The model observes slight reductions in the Sun’s acoustic power and small changes in magnetic fields that anticipate the rise of sunspots. These spots are manifestations of active regions that generate space weather events, such as solar flares and coronal mass ejections.
The COFFIES team’s technique does not view the entire solar surface at once. It uses a sliding-window transformer architecture to analyze long time sequences of data. In this way, the model focuses on recent changes and remembers past patterns. Data comes from NASA’s Solar Dynamics Observatory space observatory and supercomputers at NASA Ames Research Center. The COFFIES model’s predictions, although not yet ready for real-time operational use, complement current systems that only monitor already visible regions. The idea is to provide useful time to protect sensitive instruments or activate automatic procedures aboard spacecraft. 12-hour warnings allow for active protections.
COFFIES predicts active regions up to 12 hours in advance.
Designing Self-Repairing Vehicles with Artificial Intelligence
Fault management (FM) consists of software and procedures that identify and mitigate problems without human intervention. This capability is a fundamental requirement for autonomous missions. To make it effective, the team at Qualtech Systems Inc., a company specializing in diagnostic and prognostic systems, has linked fault management to the model-based systems engineering (MBSE) design process. This approach translates design models into failure mode and effects analyses, known as FMEA, and fault trees, which are diagrams showing how individual errors can propagate to compromise a system.
Traditionally, fault management was integrated only after the nominal system design. It often acted as a “patch” rather than prevention. Qualtech Systems Inc.’s approach, instead, integrates fault management directly into the MBSE process from the project’s inception. A demonstration on a real project used the model of the HelioSwarm mission. This mission involves a hub and eight small co-orbiting satellites to measure turbulence in the solar wind. The study showed how to translate a SysML v2 model into fault models. FMECAs and FTAs were generated. The company’s commercial TEAMS tool provides practical recommendations, for example, where to place sensors. These recommendations improve diagnosis and reduce risk and costs during the design phase. Integrating fault management from the start improves reliability.
Small Telescopes to Observe Many Worlds and Their Stars
The Pandora mission is a small satellite (SmallSat) designed to observe at least 20 exoplanets and their host stars. Its objective is to simultaneously measure visible and near-infrared light. Pandora, the first mission launched through NASA’s Astrophysics Pioneers program, uses an 18-inch (approximately 45-centimeter diameter) all-aluminum telescope. It also features an infrared sensor, originally developed as a spare for the James Webb Space Telescope.
Separating a planet’s signal from its star’s is crucial. The stellar surface is not uniform. It contains hotter regions, called faculae, and colder regions, similar to sunspots. These characteristics distort the chemical signatures scientists look for in planetary spectra. Pandora is designed to provide long-term, multi-wavelength data. Each target will be observed at least ten times. Each session will have a 24-hour observation window to capture planetary transits and stellar variations within the same interval. This helps to “clean” the signal that larger observatories cannot repeat as often due to high demand for observation time. NASA Ames Research Center also contributes to this mission, managing Pandora’s scientific data processing. Repeated observations improve the quality of planetary data.
Practical Convergences: Anticipate, Resist, and Measure Better
Bringing together 12-hour advance solar predictions, fault management systems designed from the outset, and observing SmallSats generates a chain of operational resilience. A warning from the COFFIES team can be translated into automatic action on a spacecraft executing fault management rules. The software can suspend sensitive instruments, perform redundancies, and reposition components to reduce exposure to solar storms. Concurrently, continuous data from constellations like HelioSwarm provide real-world examples of environmental disturbances and hardware failures. This data is useful for further training and validating artificial intelligence models.
This synergy is vital for the safety of astronauts and equipment in space exploration missions. The Artemis program missions aim to return humans to the Moon and, in the future, to Mars. The International Space Station remains an international orbiting laboratory. Many large-scale verifications remain. The COFFIES algorithm must be validated against many other real solar events. FM and MBSE integration practices need to be adopted in more projects. Pandora must complete its observation program to refine signal separation techniques. If these elements solidify, the probability of a human or robotic mission surviving a solar storm will measurably increase. In that case, missions will be able to operate more autonomously and produce cleaner planetary measurements, pushing the boundaries of exploration further.
The integration of predictions, fault management, and SmallSats builds operational resilience.
Protecting instruments and crews requires automation, data, and integrated design.
Sources: