Data Driven Prognostics

In this SBIR Project we will develop and demonstrate novel general algorithms for anomaly detection in complex systems. Component wear in electrical and mechanical system components causes degradation that occurs on a slow time scale with respect to theobserved system behavior and is evident in very small magnitude changes to system behavior long before a component's eventual failure. An objective of the proposed research is to detect the precursors to slowly approaching failures in order that theremaining life of critical components can be accurately predicted at an early stage. We propose a novel approach for anomaly detection in complex systems using the tools of computational dynamics and pattern discovery. The predictive algorithms will betested on a subsystem of the Airborne Laser. New advances in sensor technology, failure analysis techniques, system predictive modeling, data fusion and automated reasoning algorithms are beginning to make it possible for these predictive technologies tobe developed into a complete Prognostic Health Monitoring system. These same core technologies could be harnessed for almost any type of equipment. As a result of the development of predictive maintenance technologies, we could witness an incrediblerevolution in the way large multi-unit electromechanical systems, such as ships; aircraft or even power plants are maintained.