Model Identification and Optimization for Operational Simulation
Abstract
The purpose of this research effort was to develop and test frameworks and algorithms for use in air warfare planning systems. Developing planning systems for this problem domain is particularly challenging due to their great complexity and uncertainty. The effort focused on predictive simulation models for generating potential outcomes of proposed operational plans. The planning process was organized as a hierarchy of decisions, with those at the top being broadest and longest term. The algorithms at the highest level of planning use a hill-climbing approach, wherein proposed Blue plans are evaluated, and the average marginal benefits of alternative force reallocations are computed. Evaluation and measurement of each proposed Blue plan is accomplished via a Stochastic Evaluator that draws multiple samples of potential outcomes and Red force levels for a given Blue force structure and a combined target composition. The evaluation metric is the net discounted value from enemy targets hit. Within the evaluator, linear programming and simulation generate optimized Red responses, assumed outcomes, and relative marginal force values. This project successfully demonstrated automated plan optimization, practical embedding of optimization algorithms into an operational planning cycle operating over a multi-period conflict, and use of hierarchical decisionmaking to decomposed planning and on-line optimization problems into computationally practical tasks.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 2004
- Accession Number
- ADA426842
Entities
People
- Douglas A. Popken
- Louis A. Cox