The goal of this SBIR project is to develop an automated data optimization tool that would derive, from the same data sources used by human systems analysts, near optimal force-deployment input data for use in campaign-level simulation models. Phase I of this project will begin by determining what is required to produce such a system to support the Government acquisition process. It will then evaluate several competing technologies--Genetic Algorithms, Simulated Annealing, and Tabu Search--to find which would produce the best system at the best price.The subject of "weapon-to-mission" allocation (i.e., aircraft, tank, radar, etc.) is one of the main issues THUNDER was designed to. A frequent objection to this analysis technique is that the chosen allocation represents a deployment of unknown effectiveness. Automating the search for highly effective air alloctions can counter such objections in two ways: primarily by finding allocations that are empirically much more effective than those currently being developed by human analysts. All three techniques--GA, SA, and TS--show promise on problems such as this; problems which have poorly behaved objective functions and large and irregular solultion spaces.
Keywords: THUNDER AUTOMATED EFFICIENCY ALGORITHMS SIMULATION ALLOCATIONS OPTIMIZATION EFFECTIVENESS
