The integration of anomaly, prognostics, and diagnostics reasoners to optimize overall vehicle health management goals

The challenge is to economically field complex systems in safety critical applications over a service life of at least twenty years. Traditional solutions have included a variety of approaches to implementing diagnostic systems. Just as the health of human beings cannot be maintained entirely through diagnostics, complex safety critical systems cannot either. Diagnoses provide feedback on failure events only after they have occurred. The next major challenge to designing automatic health maintenance systems is to prognose failure events before they occur. The prognostic and diagnostic systems must be integrated into an overall system whose primary customers are the maintainer and the vehicle operator. In this paper we categorize a failure event into three classes: (1) an anomaly, (2) an impending failure, or (3) an active failure. Using the definitions of these three failure classes we develop a reasoner architecture whose integrated solution set provides a framework for next generation vehicle-health-maintenance systems. Moreover, we show that it is theoretically possible to use this reasoner architecture to reduce the physical redundancy in the vehicle design while maintaining safety; thus reducing cost, weight, and power.