Bayesian Prognostic Failure Model for ASoSC2 using a model-of-models approach

Period of Performance: 11/18/2008 - 05/18/2009


Phase 1 SBIR

Recipient Firm

Decisive Analytics Corp.
1400 Crystal Drive Array
Arlington, VA 22202
Principal Investigator


We propose a model-of-models approach for building a Bayesian Prognostic Failure Model that will meet Army IAMD requirements for the ASoSC2. The model-of-models approach closely parallels the system-of-systems approach pursued by the Army for fielding its weapon, sensor, and C2 systems. It is highly modular and will allow warfighters in the field to easily reconfigure the prognostic tool when plug and fight hardware is reconfigured on the battlefield. Our approach models mission critical failures by capturing the (possibly many) way that individual component failures can contribute to a system failure. Our models will exploit component reliability data already available and provides an organized and mathematically principled approach to combining that data. Our team consists of staff who have already contributed to the development and testing of candidate components for the ASoSC2 system as well as mathematicians and computer scientist with extensive experience in Bayesian modeling and reasoning. In addition, Decisive Analytics has a long history of transitioning SBIR technologies to end-users and will work with the prime contractor to integrate the Bayesian Prognostic Failure Model into the deployed ASoSC2.