A Testbed for Data Fusion for Gas Turbine Engine Diagnostics and Predictive Diagnostics

The key to producing more confident, real-time, on-wing diagnoses resides in the processing of multisensor data. Adding expensive new sensors is not the solution, but data fusion is. Fusion saves cost and weight; no new sensors are required. Fusionreduces false alarm rates; faults to be seen across multiple sensors. Diagnostic performance is improved by allowing detection of unique fault patterns seen on sets of sensors instead of a single sensor. Fusion enables predictive diagnostics; low-levelinformation is integrated across sensors so potential faults can be detected earlier. In Phase IIAC developed a PC-based gas turbine engine health monitoring system that uses a combination of signal and information processing algorithms to perform data fusion for engine fault diagnostics and predictive diagnostics. The system is built around anIntelligent Mechanical Diagnostics System (i-mds) Matlab toolbox that aids in systems development of future and in the integration of new sensors and new information. In Phase II that fusion technology and i-mds will be further developed and included intoan integrated system to support individual aircraft field maintenance, fleet maintenance, as well as research engineers in performing diagnostics and predictive diagnostics on real data. That development includes significant amounts of real data.