Bayesian Failure Prognostics Model (BFPM) for Space Networks

ABSTRACT: The Joint Space Operations Center (JSpOC) under the United States Strategic Command employs a network of 29 sensors, known as the Space Surveillance Network (SSN), to track more than 17,000 objects in Earth orbit. Because the number of objects is large compared to the number of sensors, the SSN cannot track every object in the catalog. Decisions must be made to allocate resources to objects. If those resources become unavailable due to equipment failure or hostile activity, high-priority objects will have to be reassigned to other sensors or lost. A tool is needed to process information and predict when a resource is going to become unavailable. The DECISIVE ANALYTICS-Bowman team proposes a dynamic Bayesian network based prognostics approach incorporating anomalies and contextual information. The Bayesian Prognostic Failure Model (BFPM) framework, previously used for failure prognostics on electronics, incorporates a variety of contextual information into a dynamically configurable network to predict the availability of assets to perform tasking. The DECISIVE ANALYTICS-Bowman team will then demonstrate this tool"s ability to predict mission success of a proposed tasking plan using all available information. BENEFIT: The integration and enhancement of DAC"s suite of tools will allow the Air Force operators on the ground to plan missions with more effectiveness. By using available contextual information along with anomaly reports, it can estimate whether an asset can perform a defined mission allowing the best use of resources to monitor space activity. In phase 2, the DECSIVE ANALYTICS-Bowman team will work with Raytheon IDS and ISS to ingrate this technology in to the Joint Space Operations Center Mission System (JMS) and Air Force Space Surveillance System (Space Fence) programs.