Description:

One of the significant challenges in Earth observing and space weather satellites is maintaining their health and keeping them operational on orbit.  In many cases, these satellites experience on-orbit anomalies, such as in power, cooling, navigation, or sensors, that require real time detection, diagnosis, and resolution.  These are manpower-intensive, costly, and often addressed on an ad-hoc basis subject to the level of expertise and available resources at the time of the anomaly.  This session solicits presentations and papers that addresses this concern by exploring the use of artificial intelligence and machine learning, AI/ML technologies to handle satellite anomalies through three distinct phases, detection, diagnosis and resolution.  AI/ML technologies have the potential to both improve responsiveness and accuracy of anomaly handling while simultaneously reducing manpower requirements. 

The application of AI/ML to satellite anomaly handling is a nascent effort and while there have been signs that this field is fast maturing, there are still many challenges to overcome before this potential can be fully realized.

We hope in this session to learn about the state of the art in AI/ML technologies for satellite anomaly handling across a broad scientific spectrum including government, industry, and academia.   Specifically, we invite presentations and papers on three distinct phases of on-orbit anomaly handling.  These are:

Anomaly Detection: In this phase, satellite telemetry is continuously monitored to identify anomalies.  Real-time assessment of telemetry probes into related measurands trending in a way would reveal an underlying problem.  Detecting anomalous telemetry requires a deep understanding of both nominal telemetry values as well as how different measurands are related.  

Anomaly Diagnosis – In this phase, the real-time telemetry is analyzed along reference measures that would identify the likely state of the satellite systems that are creating the anomalous telemetry.  These may include rising temperature, sensor malfunction, or battery overcharge.  Reaching the correct diagnosis requires understanding how the telemetry reflects the nominal satellite state.

Anomaly Resolution – In this phase, satellite commands or other actions are taken to resolve the anomaly, e.g., sending a command to stop charging the overheating battery.

By increasing awareness of promising developmental efforts, proofs of concepts, and success stories on all or part of these three phases, we hope to leverage reduced launch costs to transition to proliferated satellite constellations, rapid-response satellite launches, and other new concepts of operation that will dramatically improve our nation’s weather sensing capabilities.