Date of Completion
optimization, shortest path planning, dynamic programming, approximate dynamic programming, rollout, sailing vessel routing, ship routing, weather, meteorology and oceanography
Krishna R. Pattipati
Peter B. Luh
Field of Study
Doctor of Philosophy
This dissertation contains several foci in which optimization-based approaches are developed to obtain optimal or near-optimal solutions to NP-hard dynamic resource management problems that require the integration of spatio-temporally evolving intelligence and weather (meteorology and oceanography) data. Three primary topics of research are addressed: (i) asset allocation for counter-drug trafficking, (ii) multi-objective (ship) path planning, and (iii) fastest-path sailing boat routing. We approach all three problems in a general way for application to multiple domains, while utilizing domain-specific knowledge available in order to help condense the complex problem and decision spaces. The first two topics involve multiple, often competing, objectives for networks with stochastic non-convex edge costs, while the third topic extends this work to bearing-dependent transit times (i.e., two otherwise equal arcs may have differing associated speeds of traversal due to the pointing of the vessel and behavior relative to the true wind speed and angle). The algorithms presented in this dissertation have been transitioned for use by and operationalized at multiple external organizations including, but not limited to, the Naval Research Laboratory - Marine Meteorology Division (NRL-MRY), Joint Interagency Task Force-South (JIATF-South), Space and Naval Warfare Systems Command (SPAWAR) Systems Center-Pacific, and Fleet Numerical Meteorology and Oceanography Center (FNMOC).
Sidoti, David, "Novel Optimization-based Algorithms for Dynamic Resource Management in Complex Systems" (2018). Doctoral Dissertations. 1767.
Available for download on Saturday, May 04, 2019