OPTIMIZATION STRATEGIES FOR SIMULATION MODELS

In most simulation models an attempt is made, often by a trial-and-error search procedure, to adjust the various policy parameters of the model in order to yield a "good" design or tactic. For many different reasons, the usual optimization algorithms available in the published literature can not be used for this purpose. A special algorithm specifically designed for this purpose has been developed. This algorithm requires that the user specify the following three operations for the cases that are generated by his model: Given any number of cases, specify an "average" case for them, given two cases, specify a third "reflected" case so that the second case becomes the "average" of the first and third case, and given two cases specify that either one case is "better" than the other or that they are both equally "good". This algorithm also allows for various forms of on-line user intervention.

• Supplemental Notes:
  • For further information write in care of the author.
• Corporate Authors:

  Logistics Research Conference

  George Washington University
  Washington, DC  United States  20006
• Authors:
  • Fromovitz, S
• Publication Date: 1974-5-9

Subject/Index Terms

• TRT Terms: Computer models; Information systems
• Subject Areas: Data and Information Technology; Railroads;

Filing Info

• Accession Number: 00054771
• Record Type: Publication
• Source Agency: Logistics Research Conference
• Report/Paper Numbers: Paper
• Files: TRIS
• Created Date: Jun 24 1976 12:00AM