Newsgroups: comp.ai.genetic,sci.op-research,sci.engr.manufacturing
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!news.sprintlink.net!EU.net!sun4nl!ruuinf!csoares
From: csoares@cs.ruu.nl (Carlos Soares)
Subject: Evolutionary computation for job-shop scheduling: report available
Message-ID: <D0FsDM.An3@cs.ruu.nl>
Keywords: job-shop scheduling, genetic algorithms, evolutionary computation
Sender: usenet@cs.ruu.nl (Six O'Clock News)
Organization: Utrecht University, Dept. of Computer Science
Date: Wed, 7 Dec 1994 10:41:45 GMT
Lines: 36
Xref: glinda.oz.cs.cmu.edu comp.ai.genetic:4499 sci.op-research:2343 sci.engr.manufacturing:3158

Report available:

``Evolutionary Computation for the Job-shop Scheduling Problem''

This report describes an Evolutionary Computation algorithm for the
Job-shop Scheduling Problem (JSP). The type of problem approached is
a JSP which includes some of the planning:
- a job can have more than one alternative plans (sets of operations), and
- an operation can be executed in one or more alternative machines.
Some operators are tested.
The system is not only tested with a problem which includes the referred
features but also with two common simple JSP benchmark problems
(Muth and Thompson 6x6 and 10x10).
Results in the 10% margin are achieved with a
population of 30 for 100 generations on the first problem and 200 on
the second (there are no reference results on the third problem).
The system is robust:
- small variances were obtained in the best solutions;
- small sensitivity to parameters variation.

A graphical version of the system is also available (IRIX 5.2 for SGI Indigo).

ftp instructions:

1. ftp to "archive.cs.ruu.nl"
2. login as "anonymous"
3. give your complete email as the password
4. change dir to "pub/RUU/CS/NEURO/Planning"
5. the file "README.JSP" contains information on the files

Comments are welcome:

Carlos Soares
Computer Science Department
Utrecht University
email: csoares@cs.ruu.nl
