Projects
Air Force Phase II SBIR on Engine Health Management
MitekAn is working with GE Aviation Systems on Air Force Phase II SBIR project in development of Mixed Data Fusion (MDF) algorithms and software tools for health management of jet engines. The team developes software function for accurate detection isolation of faults in engine LRUs and gas path by combining the gas path sensor data and discrete BIT (built-in-test) codes. The approach relies on the OEM's detailed engine model.
Jet engine health management is the most rigorous of any type of aerospace (or other) hardware. Yet, fielded engines have up to 50-70% Cannot Duplicate (CNDs) and assumed wrongful LRU removals. Mitek Analytics have developed a PHM system to provide a significant accuracy improvement for engine fault and maintenance codes. Called Mixed Data Fusion (MDF), the developed technology combines existing gas path sensor data and existing discrete fault codes to produce improved advisory for engine performance and maintenance. Working in close collaboration with GE Aviation, Mitek Analytics developed MDF ground software function for a legacy military engine using flight data currently recorded by engine avionics. Initial validation with actual engine mission data and GE high-fidelity production simulator for the engine shows that MDF significantly reduces both false alarms and ambiguities of the faulty LRU advisory. The prototype MDF function has been implemented as stand-alone computer software. Fundamentally, the technology is ready for transition into field usage. The SBIR R&D is now focused on additional verification and validation work and on development of scalable software. The server-based software being developed can be accessed as web-based application from any networked computer, and will have more advanced data interface, user interface, and enterprise IT integration capability.
Mixed Data Fusion can substantially improve diagnostics of root cause faults. For the legacy engine deployment of MDF, at least 50% reduction in CNDs (Can Not Duplicate) and wrong LRU removal rates is expected. The deployment system is non-disruptive: MDF inputs currently recorded engine mission data; the outputs are modifications of currently used fault codes. The MDF fault codes can be used within existing maintenance processes. Initial estimates of the savings and deployment costs for the legacy engine show payback in less than a year. The same approach can be implemented in on-ground engine maintenance support systems for other military or civilian aircraft engines. MDF algorithms have provable design traceable to specifications; once validated through the ground deployment, they could be certified for on-board use in engine avioniccs.