Robust Parameter Design for Agent-Based Simulation Models With Application in a Cultural Geography Model

摘要

Robust parameter design (RPD), which has been extensively used in industrial and system design, is used in this thesis to determine where to set controllable factors to achieve the desired response in a social dynamic-driven agent-based simulation model. Specifically, this thesis focuses on the agent- based cultural geography model (CG) of the Helmand Province, Afghanistan, developed by Training and Doctrine Center Monterey. To the author's knowledge, this thesis is the first application of RPD to an irregular warfare (IW) model that incorporates social behavior and population dynamics. Robust parameter settings for CG response were identified using a four-step experimental design and analysis methodology. Results of the developed and applied analysis and robustness study revealed important insights into the CG model. In addition to the robust parameter settings suggested by the analysis, an outcome of the simulation experiments was to illustrate a path that enables military decision makers a method for investigating the population response for different courses of action within the IW theater.