DECENTRALIZED IDENTIFICATION.
Item
-
Title
-
DECENTRALIZED IDENTIFICATION.
-
Identifier
-
AAI8611368
-
identifier
-
8611368
-
Creator
-
MIRECKI, ANTHONY VITO.
-
Contributor
-
Frederick Thau
-
Date
-
1986
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Engineering, Electronics and Electrical
-
Abstract
-
This thesis definitively solves the problem of system identification for linear time-invariant discrete systems when available measurements are "decentralized" as in the theory of decentralized control. In decentralized control a global, usually large scale, system has a total number of outputs. Instead of measuring and processing all outputs together in a "centralized" manner, nonoverlapping subsets of the outputs are measured and are available for feedback by subsystems. The decentralized pole-placement problem was formulated and solved in a 1973 paper by Wang and Davison. Since then other papers have appeared on this problem. In every paper it is given that the global state description is known to every subsystem. This thesis investigates the problem when subsystems do not know the global state-space description and use their decentralized measurements to try and identify it.;First, canonical forms for multi-input, multi-output (MIMO) systems are reviewed with greater emphasis on Luenberger canonical form. This canonical form is especially important because results have been published on how to identify systems in this form from the outputs.;Next, from the decentralized set of outputs of a MIMO system a measurement equation, called the Decentralized Measurement Equation (DME), is derived for a system in any arbitrary state description with measurement noise and state noise multiplying unknown coefficients. The variation of Luenberger canonical form used for identification requires that the system be observable from a given set of outputs. This thesis solves the problem of system identification when the entire system is unobservable either from all outputs taken together or, more relevantly to the thesis title, when any decentralized subset of outputs is measurable.;Three second order systems are formulated in Chapter 4 and identified in accordance with the previously derived decentralized identification results. Chapter 5 again uses these results but now the system, a large flexible beam space structure, is very high order. The motion of the beam has been simulated under a NASA grant and is based on previous research on the Space Shuttle. Finally, Chapter 6 relates identification results to decentralized control results and also discusses further research.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
-
Program
-
Engineering
