Spring 2004
March
19 | April
2 | April
16 | April 30
| May
11
| June
15
All seminars are held in The
Boeing Classroom, Room 100, Goddard Hall from 3:30 to 4:30
p.m.
|
| |
|
|
March
19
Swarm Intelligence and Ant Colony Optimization |
| |
Mother
Nature has inspired a number of popular computation paradigms:
neural networks, genetic algorithms and simulated annealing.
More recently, the metaphor of emulating ant behavior as
a computational search mechanism has come into mainstream
awareness. The Ant Colony algorithm, embodied by the concept
of swarm intelligence, is interesting because like real
ants, there is no central control or command. However, difficult
optimization and search problems can be readily addressed
via weighted stochastic processes.
This talk will provide an overview of Swarm Intelligence,
describe the mechanisms for constructing an Ant Colony algorithm
and provide practical applications and resources for further
research.
|
| |
Dr.
Philipp A. Djang is a Senior Operations Research
Analyst with the Army Research Labs and is stationed at the
Survivability Lethality Analysis Directorate, White Sands
Missile Range in New Mexico. He earned his Ph.D. in Industrial
Engineering from NMSU in 1998.
|
| |
|
Power
Point Presentation:
|
| |
|
Related
Documents:
- Dorigo,
Marco, et al. "The
Ant System: Optimization by a colony of cooperating agents."
IEEE Transactions on Systems, Man and Cybernetics-Part
B, Vol. 26, No. 1, 1996, pp. 1-13.
-
Parpinelli, Rafael S., et al."An
Ant Colony Algorithm for Classification Rule Discovery."
Idea Group Publishing 2002: 190-208.
- Caro,
Gianni Di, and Marco Dorigo. "An
adaptive multi-agent routing algorithm inspired by ants
behavior."
|
| |
|
|
April
2
Observation-Based Learning and Adaptation |
| |
Understanding
the dynamics of unknown and uncertain systems requires an
observation-based approach in which learning and adaptation
are the key elements. Post 9/11 events have hastened the need
for advanced systems-level approaches to combat terrorism
and at a more general level has created the need for better
information process in methods that can aid in tracking and
control of uncertain dynamical systems.
This presentation will focus on recent advances in learning
and adaptation. Soft computing methodologies, namely fuzzy
logic, neural networks and genetic algorithms comprise the
framework for advanced decision support systems. |
| |
Dr. Nadipuram R. Prasad is an Associate Professor
in the Klipsch School of Electrical and Computer Engineering
Department at New Mexico State University. He is also the
Director of the Rio Grande Institute for Soft Computing at
NMSU. He has authored and co-authored over 100 publications
in journals and conference proceedings, and is the co-author
of two books on fuzzy and neural control. |
|
| |
|
Power
Point Presentation:
|
|
| |
|
|
|
April
16
Reliability Analysis of Bulk Power Systems Using Swarm Intelligence |
| |
This
presentation documents research into the use of an adaptive
cultural model and collective intelligence to model a virtual
team of design engineers in their quest to characterize the
reliability of a complex system. In particular the research
involves the use of swarm intelligence as a means of characterizing
the reliability of bulk power networks. Historically, utilities
support the reliable design and operation of bulk power networks
through first-order contingency analysis. In contingency analyses
the list of candidate elements for disruption are identified
by engineers a priori based on the rate at which the elements
failure through the course of normal grid operation. Using
these naturally occurring failure rates, elements are randomly
chosen from the list and removed from service. The reaction
of the grid to the disruption is analyzed using a computer
model of the power redistribution that results. Reliability
indices are collected and the simulation of contingencies
continue until convergence is reached. Two problems arise
with this with this approach. First, there are over 11,000
generation facilities, in excess of 200,000 miles of very
high voltage transmission line and thousands of substations
that constitute one of the largest machines ever constructed
in the United States: the national power grid. Clearly, the
number of contingencies can explode rapidly. Second, traditional
contingency analyses are based on the natural occurring failure
rates of bulk power system hardware. However, between 1980
and 1989 there were 5,000 worldwide attacks against power
transmission lines and towers; 386 of these were documented
attacks against U.S. energy assets. Through the identification
of critical elements and the quantification of the impact
of their failure on reliability, a more accurate characterization
of the system reliability can be developed. To better identify
critical contingencies, a number of traditional reliability
methods and system theoretic methods were investigated. Of
all the methods, article swarm optimization was found to be
the most promising. Particle swarm optimization (PSO) is regarded
as being of the family of evolutionary strategies for problem
solving. Other members of this family include, for example,
genetic algorithms and evolutionary programming. |
| |
Dr. David Robinson is with the Risk and Reliability
Analysis Department at Sandia National Laboratories, Albuquerque,
New Mexico. He has a BS in Mechanical Engineering from Colorado
State University, a MS in Systems Engineering from the Air
Force Institute of Technology and a PhD in systems theory
and probabilistic methods from University of Arizona. Dr.
Robinson is currently the focal point for research into new
methods for integration of information (real and simulated)
to anticipate the performance/failure of nuclear weapon. |
| |
|
|
| |
|
|
April
30
Brownian Motion in Finance and Biology and Engineering |
| |
Dr.
Smits will discuss properties of Brownian motion and differential
equations driven by a Brownian motion. Models for stock market
dynamics, insurance reserves and bacterial motion will be
introduced and asymptotic properties will be discussed when
possible. Time permitting I will discuss current research
for short term interest rate models, especially time asymptotic
behavior. |
| |
Dr.
Robert Smits, professor of mathematics at New Mexico
State University, earned his Ph.D. from Purdue University
(1996) His research includes stochastic differerential equations,
and mathematical finance. |
| |
|
|
May
11
Stochastic Modeling of Deteriorating Systems |
| |
All
systems for sophisticated electronics to human beings are
subject to deterioration with age and usage. Hence it is of
paramount importance to study optimal repair, maintenance
and replacement of deteriorating systems. In this lecture,
we propose to present a survey of some of the work done in
stochastic modeling of such systems with specific reference
to
(i) General Repair Models
(ii) Shock Models and
(iii) Monotone Process Models |
| |
Dr.
Rangan is a Professor of Mathematics at the Indian
Institute of Technology in Chennai, INDIA. He is a visiting
researcher in the Center for Stochastic Modeling for Summer
2004. |
| |
|
|
| June
15
Stochastic Modeling of Radiation Damage, Repair and Survival
of Cells
|
| |
Radiation
is one of the most important agents amongst all environmental
factors that can damage DNA in a cell, both the formation
of radiation damages and the subsequent repair process are
probabilistic in nature. Understanding the underlying mechanism
behind this important biological problem has not yet been
fully solved experimentally. This lecture presents a comprehensive
stochastic analysis of the damaging effects of radiation on
cells and their possible repair. The optimal scheduling of
cancer drugs for maximum efficacy is covered. |
| |
Dr.
Rangan is a Professor of Mathematics at the Indian
Institute of Technology in Chennai, INDIA. He is a visiting
researcher in the Center for Stochastic Modeling for Summer
2004. |
| |
|
Power
Point Presentation:
|
|
