Developing a data driven prognostic system with limited system information

The purpose of this research is to determine the feasibility of an automated, self-modifying prognostic system, along with a library of supporting algorithms that could be applied to the Airborne Laser (ABL) (http://www.airbornelaser.com/, 2003). The library will consist of data mining and prognostic algorithms supporting an architecture capable of refining and implementing the algorithms that monitor a critical system without a human-in-the-loop. When the data mined indicates systemic changes, the selfmodifying prognostic system refines the previously developed algorithm. The resulting library architecture will be portable to different platforms and extensible to accommodate advances in data-driven prediction and prognostics techniques. The ABL is part of a highly classified program within the DoD, and as such, detailed system information and data was not available outside the program. This extended abstract describes the approach we are taking to develop a data-driven prognostic system for the ABL, given the limited distribution of ABL information. Our approach can be outlined as follows: 1) Gather available unclassified information about the system; 2) Identify similar, but unclassified, systems; 3) Draw parallels, as appropriate, between the two systems; and 4) Develop the solution architecture using the alternative system as a test case.