Performance Enhanced Navigation Using Neural Network Technology

Barry Griffiths, Anne Orlando, Sandra Berning

Abstract:	PENANT (Performance Enhanced Navigation using Neural Network Technology) represents a new approach to fault detection, identification, and reconfiguration (FDIR) for integrated multi-sensor aircraft navigation systems. This paper reports the major results of the PENANT feasibility study, a Small Business Innovation Research (SBIR) contract sponsored by Air Force Wright Laboratory. In the past, FDIR algorithms for this class of problem have had limited success. There are two principal reason for this: l The random, time-varying, unstable error characteristics of normally-operating aircraft navigation systems tend to obscure failure signatures of the failure modes l Feedback methods for multi-sensor integration, such as Kalman filters, tend to further obscure failure signatures by recursively incorporating the contaminated data into the navigation solution. PENANT is designed to overcome these difficulties by combining advanced statistical methods with a neural network. The statistical methods include the Net Information Approach @IA), a variation on Kalman filtering that partitions information from the navigation sellsots into several local solutions. The neural network then processes the NIA outputs to diagnose the failure state of the system. When a sensor failure is diagnosed, the PENM algorithm reconfigures the integration filter to lock out the failed sensor. The objective of this effort was to determine whether PENANT can provide improved diagnostic capability and navigation accuracy, compared to standard innovations- monitoring techniques. The PENANT concept was tested in a reduced-scale simulation that incorporates many of the critical features of an integrated INS/GPS aircraft navigation system. The results indicate that this approach may be useful for real-time, onboard FDIR.
Published in:	Proceedings of the 1993 National Technical Meeting of The Institute of Navigation January 20 - 22, 1993 Parc 55 Hotel San Francisco, CA
Pages:	469 - 478
Cite this article:	Griffiths, Barry, Orlando, Anne, Berning, Sandra, "Performance Enhanced Navigation Using Neural Network Technology," Proceedings of the 1993 National Technical Meeting of The Institute of Navigation, San Francisco, CA, January 1993, pp. 469-478.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In