West Virginia University
A neural network-based sensor validation scheme within aircraft control laws.
Abstract
dc:description.abstractIn 1992 NASA launched its "faster, better, cheaper" initiative to increase the number of missions while decreasing mission cost and time. Accordingly, NASA began supporting research efforts to replace physically redundant systems on board spacecraft with analytically redundant systems. The sensor failure, detection, identification and accommodation (SFDIA) system eliminates physical redundancy by supplying neural network estimations for the pitch, roll, and yaw rate gyros. A Boolean logic scheme utilizes the estimations to detect and identify gyro sensor failures. Additionally, the SFDIA replaces a faulty sensor with its estimate to ensure nominal flight conditions after a sensor failure. The SFDIA was written, modeled and tested in the Simulink programming environment for a closed loop non-linear model of a De Havilland 2 “Beaver” aircraft. Six types of sensor failures were artificially injected into each parameter. In each case, the SFDIA accurately and timely detected, identified, and accommodated the faulty sensor.
Degree
thesis:*- Name thesis:degree_name
- MS
- Level thesis:degree_level
- Thesis
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Stolarik, Brian Michael
- Contributors dc:contributor
-
- David G. Frazer
Identifiers
dc:identifier.*- Identifier
- https://researchrepository.wvu.edu/etd/10616