Summary

Description

Impact Technologies, LLC in collaboration with Wright State University and Pratt & Whitney, propose to develop innovative methods to differentiate sensor failure from actual system or component failure for advanced propulsion systems. In sharp contrast to many conventional methods which deal with either sensor failure or component failure but not both, our method considers sensor failure and component failure under one systematic and unified framework. The proposed solution consists of two main components: a bank of real-time nonlinear adaptive fault diagnostic estimators for residual generation and a Transferable Belief Model (TBM) based component for residual evaluation. By employing a nonlinear adaptive learning architecture, the presented approach is capable of directly dealing with nonlinear engine models and nonlinear faults without the need of linearization. Fault sensitivity and robustness to modeling uncertainty is enhanced by several important techniques including adaptive reference nonlinear engine model, adaptive diagnostic thresholds, and TBM based residual evaluation method. Software modules will be developed and integrated into the NASA C-MAPSS engine model for performance evaluation. A subset of core algorithms will be implemented and used in a hardware-in-the-loop demonstration under dSPACE environment to justify a Technology Readiness Level of 4-5 at the conclusion of Phase I.