A comparative study between using A*-search, Latin Hypercube and genetic algorithms in design of experiment for simulation of military operational plans

摘要

In this paper we compare three alternative ways of designing a data farming experiment for a military operational planning problem. In this multi-objective optimization problem the goal is to minimize two different measures of effectiveness. Military plans under consideration are evaluated by an event based simulation system. We compare three different approaches for selecting which plans should be simulated; A*-search, genetic algorithms and Latin Hypercube. The more robust Latin Hypercube approach is found to be the better approach for this application compared to the other two more focused approaches.