Lab testing of neural networks for improved aircraft onboard-diagnostics on flight-ready hardware

Author

Anderson, Raymond J. ; Aylward, Stephen R.

Author_Institution

McDonnell Douglas Corp., St. Louis, MO, USA

fYear

1993

fDate

26-28 Jan 1993

Firstpage

404

Lastpage

410

Abstract

The use of neural networks (NNs) to enhance onboard diagnostics and provide real-time damage detection for aircraft reconfiguration has been investigated. This research focus was a result of investigating new technologies to improve mission success and reduce life cycle/support cost resulting from a high percentage of `cannot duplicate´ and `retest O.K.´ maintenance actions occurring on some aircraft systems. Laboratory testing results have shown that future onboard diagnostics systems can use NNs to detect intermittent failure and false failure indications. The test instance featured an Ada-based NN running in an advanced vehicle management system computer (VMSC)

Keywords

aerospace testing; aircraft instrumentation; failure analysis; neural nets; real-time systems; technological forecasting; aircraft reconfiguration; false failure indications; flight-ready hardware; future; intermittent failure; laboratory testing; life cycle/support cost; neural networks; onboard diagnostics; real-time damage detection; vehicle management system computer; Aerospace control; Costs; Diagnostic expert systems; Fault detection; Laboratories; Maintenance; Military aircraft; Neural network hardware; Neural networks; System testing;

fLanguage

English

Publisher

ieee

Conference_Titel

Reliability and Maintainability Symposium, 1993. Proceedings., Annual

Conference_Location

Atlanta, GA

Print_ISBN

0-7803-0943-X

Type

conf

DOI

10.1109/RAMS.1993.296823

Filename

296823