Model Updating in Online Aircraft Prognosis Systems

The objective of this research is to develop algorithms for online health monitoring and prognostics (prediction of the remaining life of a component or system) in aerospace applications. The specific areas of need addressed by this project relate to fusion of sensor-based diagnostics with degradation models, management and propagation of uncertainty, autonomous model updating, and practical considerations such as reducing data volume and storage requirements. The algorithms developed in this project represent the generalizable aspects of predictive prognosis; the only application-specific portions are the fault model and the diagnostic signal processing. The Phase I work successfully demonstrated the basic features of the prognosis algorithms using data for several bearing fault examples. In Phase II, Sentient will develop these algorithms into a complete, full-featured prognosis architecture. Application-specific fault models and diagnostics for a NASA relevant application will also be developed in Phase II, and these will be used to demonstrate the complete Phase II prognostic system. Sentient will leverage extensive test data available from other closely related projects to thoroughly evaluate the new prognostic algorithms. This data includes studies of bearing cage instability phenomena conducted in a unique space environment test rig. Combined with the new prognosis algorithms, this will be directly applicable to help understand and predict recent bearing anomalies observed in the CMGs of the International Space Station.