Technical Sessions – Monday July 11
Technical Sessions – Monday July 11
Technical Sessions – Monday July 11
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
TA-15 IFORS 20<strong>11</strong> - Melbourne<br />
4 - The Back Propagation Algorithm Using the Bi-<br />
Hyperbolic Activation Function<br />
Geraldo Miguez, COPPE / PESC, Universidade Federal do Rio<br />
de Janeiro, Brazil, R Mariz e Barros, 652/502, 20270-002, Rio de<br />
Janeiro, RJ, Brazil, geraldomiguez@yahoo.com, Nelson<br />
Maculan Filho, Adilson Elias Xavier<br />
Back propagation algorithm is one of the most used tools for training artificial<br />
neural networks. However, in some practical applications it may be very<br />
slow. To allow a broader use, many techniques were discussed to speed up<br />
its performance. This paper presents a new strategy based in the use of the<br />
Bi-hyperbolic function that offers more flexibility and a faster evaluation time.<br />
The efficiency and the discrimination capacity of the proposed methodology are<br />
shown through a set of computational experiments with traditional problems of<br />
the literature.<br />
� TA-15<br />
Tuesday, 9:00-10:30<br />
Meeting Room 208<br />
Weapon Systems Analysis<br />
Stream: Military, Defense and Security Applications<br />
Invited session<br />
Chair: Won Joon Jang, Defense Industry Team, Korea Institute for<br />
Industrial Economics and Trade, Hoegi-ro 66, Dongdaemun-gu,<br />
133-771, Seoul, Korea, Republic Of, wjjang47@snu.ac.kr<br />
1 - The RAM Goal Setting Model with the use of OMS/MP<br />
Analysis for the Weapon System Development<br />
Won Joon Jang, Defense Industry Team, Korea Institute for<br />
Industrial Economics and Trade, Hoegi-ro 66, Dongdaemun-gu,<br />
133-771, Seoul, Korea, Republic Of, wjjang47@snu.ac.kr,<br />
Kyung Yong Kim<br />
The paper presents the RAM Goal Setting model with the basis of<br />
wartime/peacetime OMS/MP results and its Total Down Time factors for the<br />
development of the weapon system. Based on both previous studies, the peer<br />
review results and various proven techniques, it presents the RAM goal setting<br />
model with its real implementation case study result. It verifies with ALPHA<br />
simulation tools, too. It could provide the basis of its development of weapon<br />
system and it could contribute both to enhance its operational availability and<br />
to reduce the Total Ownership Cost during its whole service life time.<br />
2 - Integrated Survivability for ADF Land Platforms<br />
Patrick Taliana, Defence, DSTO, PO Box 1500, 5<strong>11</strong>1,<br />
Edinburgh, South Australia, Australia,<br />
patrick.taliana@dsto.defence.gov.au<br />
Land platforms may be engaged by a wide variety of threats which can be difficult<br />
to predict and are constantly evolving. There are numerous technologies<br />
that have the potential to improve the probability of survival of land platforms.<br />
Deciding what mix of technologies maximises probability of survival is a non<br />
trivial task. Many researchers have adopted the Classical Survivability Onion<br />
model to represent the integrated survivability problem space. This report will<br />
outline the limitations of the classical Onion model and preset an alternative<br />
Integrated Survivability strategy called DESIST.<br />
3 - Network Optimization Models for Resource Allocation<br />
in Developing Military Countermeasures<br />
Boaz Golany, Industrial Engineering & Management, Technion -<br />
Israel Institute of Technology, Technion City, 32000, Haifa,<br />
Israel, golany@ie.technion.ac.il, Moshe Kress, Michal Penn,<br />
Uriel G. Rothblum<br />
The paper considers an arms race where an attacker develops new weapons and<br />
a defender develops countermeasures that mitigate the effects of the weapons.<br />
We address the defender’s decision problem: given limited resources, which<br />
countermeasures to develop and how much to invest in their development so as<br />
to minimize the damage caused by the attacker’s weapons over a certain horizon.<br />
The problem is formulated as constrained shortest path model and variants<br />
thereof. The potential applicability and robustness of this approach with respect<br />
to various scenarios is demonstrated.<br />
36<br />
� TA-16<br />
Tuesday, 9:00-10:30<br />
Meeting Room 209<br />
Metaheuristics for Scheduling in<br />
Manufacturing<br />
Stream: Scheduling<br />
Invited session<br />
Chair: Frédéric Dugardin, LOSI, University of Technology of Troyes,<br />
12, rue Marie Curie, 10010, Troyes, France, frederic.dugardin@utt.fr<br />
Chair: Farouk Yalaoui, Institut Charles Delaunay, ICD LOSI,<br />
University of Technology of Troyes, 12, Rue Marie Curie BP 2060,<br />
10000, Troyes, France, farouk.yalaoui@utt.fr<br />
Chair: Lionel Amodeo, Charles Delaunay Institute, University of<br />
Technology of Troyes, 12 Rue Marie Curie BP2060, 10000, Troyes,<br />
France, lionel.amodeo@utt.fr<br />
1 - Fuzzy-Lorenz Algorithm to Solve Multi-objective Reentrant<br />
Scheduling Problem<br />
Frédéric Dugardin, LOSI, University of Technology of Troyes,<br />
12, rue Marie Curie, 10010, Troyes, France,<br />
frederic.dugardin@utt.fr, Farouk Yalaoui, Lionel Amodeo<br />
This paper deals with the multi-objective scheduling of a reentrant hybrid flowshop.<br />
In this study the two different objectives are the makespan and the sum<br />
of the total tardiness minimization. The system is composed of several stages<br />
which involves several parallel identical machines. Moreover each task must<br />
be processed more than once at each stage. This problem is solved using the<br />
Lorenz dominance which involves more parameters than the Pareto one. In this<br />
study we use Fuzzy logic Controller to adapt the value of these parameters to<br />
improve the results of the previous algorithm. This algorithm is tested on several<br />
instances from the literature and compared with some of the best known<br />
algorithms.<br />
2 - New Heuristic for Solving the Minimization of Tool<br />
Switches Problem<br />
Horacio Yanasse, LAC, INPE, Av. dos Astronautas 1758, CP 515<br />
- INPE/CTE, 12227-010, São José dos Campos, SP, Brazil,<br />
horacio@lac.inpe.br, Edson Senne, Rita de Cássia Meneses<br />
Rodrigues<br />
In the minimization of tool switches problem we seek a sequence to process a<br />
set of jobs so that the number of tool switches required is minimized. In this<br />
work we present a new heuristic for solving this problem based on generating<br />
a surrogate potentially smaller sized instance of MTSP, whose solution can be<br />
used to build a solution to the original instance. To obtain this solution to the<br />
MTSP we propose a heuristic based on partial ordered job sequences. Computational<br />
test results are presented showing that the proposed heuristic has an<br />
improved performance compared with previous proposed schemes.<br />
3 - A Fuzzy Logic Controller to Solve a Scheduling Problem<br />
Naim Yalaoui, Institut Charles Delauney, Université de<br />
Technologie de Troyes, 12, Rue Marie Curie, 10000, Troyes,<br />
France, naim.yalaoui@utt.fr, Lionel Amodeo, Farouk Yalaoui,<br />
Halim Mahdi<br />
In this paper, we deal with a specific scheduling problem. This one is an hybrid<br />
flow shop problem. The jobs are processed on parallel unknown machines in<br />
each stage. Those are pre assigned to the machines. In some stages, the jobs are<br />
processed on a fictive machine. The objective function is to minimize the total<br />
tardiness. We propose an exact method based on a complete enumeration and<br />
different metaheuristics such as a genetic algorithm, genetic algorithm under<br />
fuzzy logic-control, a particle swarm algorithm and particle swarm algorithm<br />
under fuzzy logic-control. The tests examples were generated using a specific<br />
protocol. The obtained results are very interesting.