Composite mission variable formulation for real-time mission planning

Abstract

In this thesis, we create a comprehensive model and efficient solution technique for generating air operations plans. Previous air operations models have fallen short in at least one of the following areas: routing. real-time re-planning of aircraft. problem size capability, plan generation speed. and optimal packaging of aircraft. The purpose of the Composite Mission Variable Decomposition (CMVD) approach is to plan and re-plan air operations for a real conflict as it unfolds. Previous model shortcomings were the result of two main reasons: the models were developed for other purposes (typically weapons studies). or developers could not create techniques that can efficiently generate plans while including the listed areas. The application of conventional optimization modeling to an operations problem that includes aspects such as routing and real-time re-planning forms a model that has millions of constraints and a weak linear programming relaxation. The Composite Mission Variable MNodeils the first step in overcoming the above shortcomings because it greatly decreases the number of constraints in the optimization model. and the linear programming relaxation provides tight bounds. The Composite Mission Variable Model combines multiple air operations planning decisions into a composite mission variable. Many complex constraints that are explicitly included in a conventional model are implicitly enforced in the composite mission variables. We apply price coordinated decomposition to generate the composite mission variables. Price coordination reduces the number of variables in the Composite Mission Variable Model and allows for parallel processing of composite mission variable generation. CMVD creates air operations plans in minutes for scenarios with thousands of targets. while including important capabilities such as routing and re-planning of aircraft in air. CMVD is tested in simulated conflicts and its performance validated by comparisons with a heuristic approach for generating plans.