碩士 / 國立成功大學 / 交通管理(科學)學系 / 85 / ABSTRACT The Multi-objective
project selection problems have been continuallyexplored for
many years There are lots of possible approaches solving
thiskind of problem. Special contributions and meanings are
reported. Thisstudy tries to rethink another direction for
solution, i.e., the elasticnet (EN) framework of the artificial
neural network family. Evaluating and selecting projects is
to properly cluster the merits orfaults of the projects, in
order to screens better ones. The effect ofunsupervised learning
of EN is employed Initial elastic nodes are randomlydistributed
over the multi-dimensional space. As the project data
repeatedlyflow, knotting of elastic nodes with similar
attributes occurs. Then, elasticnodes gradually match specific
projects. This is the attraction effect thatEN produced. Outputs
can be presented on network structure, be judged by theattribute
of each coordinate. Sequencing of projects searches the best
project arrangement in accordingto prespecified goals. The
projects are then implemented in such order thatleads to the
best effects. The question is similar to the ordinary
travelsalesman problem. However, many benefit and cost items
have to be taken intoconsideration explicitly in project
programming Therefore this study is notlimited to traditional
two-dimension sequencing problems. Basic thinking ismatching a
project to a city. A tour completed repres one feasiblemulti-
objective project sequencing which reveals a goal weighted by
relativevalues of each dimension. The first-choice project is
determined via theabove evaluating process. Then the sequence of
subsequent projects isdetermined by the relative distance
between the representing elastic nodes.Experimental examples
show fairly good results of this method. The project-
sequencing problems are actually the combinatorial
optimizationproblems. It is an attentive problem of many
scientists and engineers for manyyears. It is also an important
issue for management society. Adopting EN tosolve complicated
problems is a new and work trial direction.
| Identifer | oai:union.ndltd.org:TW/085NCKU0119002 |
| Date | January 1997 |
| Creators | Wang, Te-Jun, 王德潤 |
| Contributors | Chien-Hung Wei, 魏健宏 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 106 |
Page generated in 0.0078 seconds