 |
 |
Controls and Systems
Associated Laboratories, Groups and Centers:
Center for Rapid Transit Systems
Controls and Systems Group
Digital Signal Processing Research Laboratory
Intelligent Control Engineering Group
Motion Control Systems Research Group
Controls & Systems Faculty
William T. Baumann
A.A. (Louis) Beex
Douglas K. Lindner
Pushkin Kachroo
Richard Moose
Krishnan Ramu
Hugh VanLandingham
Center for Rapid Transit Systems
www.crts.vt.edu
Krishnan Ramu, Director
 Virginia
Tech's PERTS friction-free linear propulsion prototype.
Virginia Tech created the Center for Rapid Transit Systems
(CRTS) to plan, research, design, develop and demonstrate innovative,
energy-efficient, environmentally-friendly technology applications
based on Tech's PERTS linear propulsion technology with electromagnetic
controls. The center will build a national consortium of industries,
federal and state agencies, and other stakeholders to collectively
address technology R&D challenges. Using an innovative business
model, the CRTS will foster public-private partnerships to advance
the development and commercialization of a family of electromagnetically
controlled linear propulsion system applications.
Initial CRTS technical activities will focus on further developing
and demonstrating a low cost, highly energy-efficient personal
electric rapid transit system (PERTS). The PERTS technology represents
a unique friction-free electric propulsion technology used to
move people and their vehicles in highly congested urban areas.
This system will reduce traffic congestion, save energy, improve
air quality, reduce accidents, and improve productivity.
The CRTS will subsequently focus on other potential electromagnetically-controlled
linear propulsion technology applications including: high rise
building cableless elevators, conveyors, airport people movers,
heavy lift cargo and logistics management needs, applications
in meeting space and aerospace industry needs, and applications
responding to a full spectrum of military needs, including electromagnetic
systems for aircraft launch and recovery.
Controls and Systems Group
A new project in cooperation with the computer area involves
feedback control scheduling for distributed processors. Hybrid
feedback control techniques are used to solve the resulting feedback
problems and projection dynamics are used to deal with the convexity
of the state space. A computer middleware installed on a network
has been built for testing and research.
Work continues in applying control theory to problems in intelligent
vehicle highway systems. A current project concerns designing
feedback controllers for ramp metering, developing online travel
time functions, and designing user travel choice models using
learning stochastic automata.
Another project aims to create mathematical models for social
emotional learning in humans dealing with the emotional quotient
(EQ). Future work will deal with developing robotic and web-based
systems for interactive EQ learning.
DC power distribution systems on aircraft are the focus of work
related to the power electronics area. These systems are highly
coupled, nonlinear systems that interact strongly with the generator
and loads. Work is focused on the development of nonlinear analysis
techniques that will allow the stability regions of the power
system to be assessed. Optimization techniques are also being
developed to minimize the weight and size of the components of
the power system while maximizing the performance of the power
system in the presence of two-way power flow.
An interdisciplinary project with the Mechanical Engineering
Department to model and control thermo-acoustic instabilities
in gas turbine and ramjet engines is continuing. The modeling
work involves constructing a nonlinear heat release model from
experimental flame measurements. Coupled with our acoustic models,
this will allow the prediction of limit cycle amplitudes in test
combustors. Additional work in modeling has involved the acoustic
analysis of a new industrial gas turbine combustor being built
by Honeywell Corporation. The control work has centered on applying
and analyzing the performance of simple adaptive algorithms.
Least mean squares (LMS) and time averaged gradient (TAG) algorithms
have been successfully implemented using acoustic actuators and
the work is now focusing on algorithms that use pulse-type inputs
characteristic of fuel injectors. In particular, algorithms using
subharmonically-pulsed injection are being studied.
Digital Signal Processing Research Laboratory
www.ece.vt.edu/~dsprl
A.A. (Louis) Beex, Director
A new research effort, jointly with the Time Domain Reflectometry
Lab, is aimed at identifying the structure and parameters of
a telephone loop from central office measurements. For known
structure and parameters, we can then predict the performance
of such a loop as a digital communication channel. A recently
started effort concerns the DSP implementation of an acoustic
sensor array for adaptive beamforming. The challenge here is
that the array configuration is not known a priori, as the sensors
are only loosely connected, and the array must therefore be self-calibrating.
In addition it would be desirable to track one or more moving
sources. Our interest in explaining the convergence properties
of the class of Affine Projection Algorithms has been extended
from correlated data within a datavector to also include correlation
between the subsequent datavectors being used for adaptation.
The identification of speakers from short utterances, based on
our adaptive forced response inverse filtering approach - for
finding glottal waveform information -combined with linear prediction
parameters, was found to be successful for most speakers exhibiting
long pitch periods. However, this algorithmic approach needs
further development for speakers with short pitch periods. We
have complemented our algorithm for the direct adaptation of
line spectral frequencies in a cascade of second order sections
with the capability of directly using those parameters in resynthesizing
speech. The direct adaptation and synthesis - using line spectral
frequencies in second order sections -is being evaluated as a
speech coder. Hardware implementation of the latter -as part
of a joint research project with the Institute of Technology
in Bandung, Indonesia, is currently being considered. Another
major aspect of this international project is to develop an efficient
image coder based on singularity preservation. The combination
of efficient speech and image coders is hoped to lead to a viable
MPEG-4 coder.
Intelligent Control Engineering Group
Associated ECE Faculty
Hugh VanLandingham
Pushkin Kachroo
A major focus involves biologically inspired learning algorithms,
such as would be found in an artificial immune system, with application
to multiple cooperative robots. The roles of the robots would
mimic those of certain immune system cells with the ultimate
goal of finding adaptive algorithms that model the known processes
in mammals. One facet of this work is the development of a Bayesian
network that exhibits learning (adaptation) ability. Another
facet involves building dynamic-programming-based game theoretic
controllers for pursuit evasion in a distributed robotic application.
In addition, there is a continuing multidisciplinary effort in
the use of fuzzy logic to control ship roll motion and ship-mounted
cranes.
Sources of acoustic noise are under investigation to develop
quiet electric machines and drives with particular reference
to switched reluctance machine drive systems. Machine design
and control-based noise mitigation algorithms have been developed.
Low-cost converter topologies for motor drive systems have been
proposed and experimentally verified together with their controllers.
New control algorithms for ripple-free performance of switched
reluctance and permanent magnet machines have been developed
and implemented. These contributions in motion control are expected
to significantly affect the product development for the emerging
high-volume, but low-cost, variable speed drives market. Due
to the technical merits of the proposed topologies and controllers
and their low cost, they may open some new applications previously
overlooked because of cost considerations.
Representative Projects In
Controls & Systems
Computer Networks Loop Reconstruction, Phase III
Electronics for Recurve Actuator
Regenerative Electronics for Launch Vehicles
US-Japan Seminar on Intelligent Distributed Autonomous Power
Systems
Behavior Self-Organization in Multi-Agent Learning Systems
Enhanced UML Documentation of an Internet Based Energy Information
System
Implications of Automotive/Trucking On-Board Information Systems
on General Aviation Cockpit Weather Systems
System Design Methods for Simultaneous Optimal Control of Combustion
Instabilities and Efficiency
Development of Modular, Reduced-Order Models for Prediction of
Combustion Instabilities
|
