Advances have been made in optimizing quantitative variables within a simulation model, and many methodologies now exist for this purpose. However, many of the design decisions which confront a system's users involve policy alternatives. Often, variables used to represent these alternatives are not only discrete but qualitative. This work seeks to develop a simulation-optimization methodology which can operate on qualitative variables. The proposed approach is to link a genetic algorithm with an object-oriented simulation model generator. The system designs recommended by the genetic algorithm are converted to simulation models and executed. The results then guide the genetic algorithm in its selection of future designs. A simulation model generator for a class of manufacturing systems and a genetic algorithm which can interface with the generator have been developed. The methodology has shown positive results.
[1]
D. E. Goldberg,et al.
Genetic Algorithms in Search
,
1989
.
[2]
David E. Goldberg,et al.
Genetic Algorithms in Search Optimization and Machine Learning
,
1988
.
[3]
John H. Holland,et al.
Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence
,
1992
.
[4]
Subramanian Prakash,et al.
Development of a goal directed simulation environment for discrete part manufacturing systems
,
1993,
Simul..
[5]
Lawrence. Davis,et al.
Handbook Of Genetic Algorithms
,
1990
.
[6]
R. E. Kalman,et al.
Optimum Seeking Methods.
,
1964
.