Date of Award

12-1995

Document Type

Thesis

Degree Name

Master of Science in Electrical Engineering

Department

Department of Electrical and Computer Engineering

First Advisor

Peter Maybeck, PhD

Abstract

Multiple model adaptive control (MMAC) is investigated using the high-fidelity, nonlinear, six-degree-of-freedom Simulation Rapid-Prototyping Facility VISTA F-16. Detection of single actuator and sensor failures is considered, with an MMAC algorithm initially pursued which allows a controller specifically designed for each particular failure condition to replace the standard F-16 Block 40 flight control system (FCS) once the failure is detected. The synthesis of certain discrete-time LQG/PI controllers (those using control variables linearly dependent on state derivatives) is shown to be unattainable due to numerical difficulties. A novel control technique, termed control redistribution, is introduced which redistributes control commands (that would normally be sent to failed actuators) to the non-failed actuators, accomplishing the same control action on the aircraft. Multiple model adaptive estimation-based control redistribution is demonstrated to detect single failures in less than one second and to provide a response nearly identical to that anticipated from a fully functional aircraft in the same environment. Moreover, this method directly employs the proven Block 40 FCS, and no other, thereby guaranteeing desirable closed loop performance. A description of modifications necessary for in-flight testing is also provided. This research represents the most realistic simulation of multiple model adaptive control for flight control to date.

AFIT Designator

AFIT-GE-ENG-95D-26

DTIC Accession Number

ADA309938

Download

Share

COinS