文档介绍：UNIVERSITY OF
NEWCASTLE
MULTIPLE CRITERIA IC ALGORITHMS
IN ENGINEERING DESIGN AND OPERATION
A Thesis Submitted for the Degree of
Doctor of Philosophy
October 1997
David Todd
Engineering Design Centre
Supervisor : Prof. Pratyush Sen
Department of Marine Technology
Abstract
This thesis investigates the application of ic Algorithms (GAs) to multiple
criteria problems in engineering design and operation. The GA is an evolutionary
computing technique which applies Darwinian principles such as survival of the fittest,
mating and mutation to a population of individuals to evolve good solutions to a broad
range of problems. GAs are normally used as single criterion optimisers. However, a
Multiple Criteria ic Algorithm (MCGA) has been developed in this thesis which
allows simultaneous maximisation and minimisation across several criteria.
The MCGA is explained, enhanced and modified throughout the thesis as new
requirements are introduced. At each stage the enhancements are tested on basic test
functions to assess their performance. New concepts such as a Pareto Population,
Adaptive Niche Sizing and work Preferencing are introduced.
A broad range of applications have been tackled using the MCGA, all of which are
combinatorial in nature. Combinatorial problems are primarily concerned with
ordering or positioning of elements, the order determining the properties and attributes
of the solution. As such, standard GA operators cannot be used and novel methods
have to be employed.
The applications are drawn from various engineering fields. An ponent
placement problem is tackled initially with the goal of decreasing processing time as a
single binatorial example. A general method of rearranging design
activities so as to maximise parallelisation and reduce lead times is then performed
using the MCGA. plex scheduling model is devised and the MCGA is used to
optimise various operating scenarios. Finally, the arrangement of containers in a
co