Early Detection of System Failure in Complex Chemical or Non-Electrical Based Systems Using a Nerual Network

Health monitoring systems have evolved into complex diagnostic systems. Researchers are attempting to include prediction, or prognostics, into such systems and are resorting to hybrid systems fusing statistics, data mining, expert systems, neural networks, and more into system that perform not only health monitoring and diagnostics but prognostics as well. However, no work has been reported on systems based on chemical or other non-electrical processes where the interactions of the various operating parameters are subtle, complex, and correlated in unknown or difficult to elicit ways. This paper describes the use of neural networks to provide early detection of the onset of operational failure in such devices and suggests ways to use it as part of a prognostic system.