Date of Award

12-1996

Document Type

Thesis

Degree Name

Master of Science

Department

Department of Electrical and Computer Engineering

First Advisor

Peter S. Maybeck, PhD

Abstract

Performance characteristics of a Multiple Model Adaptive Estimation and Control Redistribution (MMAE/CR) algorithm are evaluated against single and double actuator and sensor failures. MMAE alone can compensate for sensor failures, whereas Control Redistribution compensates for actuator failures by redistributing commands, initially intended for failed actuators, to the unfailed actuators in such a way that the desired system response is achieved. Both failure detection and compensation capabilities are developed and analyzed through an extensive amount of simulation data, particularly addressing multiple failures. Simulations are performed utilizing the high fidelity, non-linear six degree of freedom Simulation Rapid Prototyping Facility for the VISTA F-16, for both benign and maneuvering scenarios. Methods utilized to incorporate the MMAE/CR techniques are examined and modifications required to enhance performance are also presented. Results are presented which indicate the techniques incorporated provide an excellent means of both failure detection and compensation for the failures of both actuators and sensors. Approximately 98 percent of all secondary failures were successfully detected, and the majority of these detections are shown to occur in less than .5 seconds. The techniques of Multiple Model Adaptive Estimation and Control Redistribution are shown to complement each other well by providing improved failure detection in the face of actuator failures through the redistribution of the dither signal (used to enhance identifiability when there are no maneuvering commands to excite the system), and improved control authority through enhanced state variable estimation in the face of sensor failures.

AFIT Designator

AFIT-GE-ENG-96D-29

DTIC Accession Number

ADA324050

Download

Share

COinS