VISIT OUR NETWORK:

tech track papers

Categories: 2019, Space Situational Awareness

Developing Machine Learning Solutions for Space Missions

A space mission with both space and ground assets can be regarded as a dynamic system, which is time dependent and nondeterministic. Situational awareness of a dynamic system is defined as the ability to perceive, analyze, and predict its own behaviors. The architectural model for a machine learning(ML) solution involves four key data processing blocks: data training to create system situational awareness, dynamic monitoring for anomaly detection and filtering to determine true system states, post-training analysis with clustering approach to create the actionable information in assessing the data quality and system operational status, and a decision-making process to generate appropriate actions. The application of the ML framework in an operational environment brings both challenges in implementation and promising benefits for mission operations. ML leads to automated operations and fundamental advances in how satellite and payload health and safety are maintained—spacecraft datasets are monitored within the bounds of learned and predictive dynamic functions rather than within traditional static limits. The challenges for deploying a ML system in operational environments include the large numbers of datasets, very large data volumes, diverse data types, and defective datasets. A ML system for space missions must be scalable and extensible, which requires an enterprise approach that separate the common functions, infrastructure, and services from the mission specific algorithm components. A ML system in a space mission should interface with the ground enterprise, such as GMSEC and EGS, to publish event messages on the operation status from its data monitoring and to provide the machine learning service to other components in a ground enterprise. Our experience in the initial deployment of ML solutions in a ground system shows that the ML approach leads to more rapid detection of anomalies, which leads to more improved mission resiliency.

Downloads

Author: Zhenping Li
Topic: Space Situational Awareness

  • Paper Li, Zhenping - space-symposium-2019-V2.pdf

    Download File
  • Presentation Li, Zhenping - Space_symposium_2019_ZhenpingLi.pdf

    Download File