Novel, multidisciplinary global optimization under uncertainty this project tackles an aeronautics challenge problem with inherently complex and multidisciplinary obstacles to an optimized solution.…

Novel, multidisciplinary global optimization under uncertainty this project tackles an aeronautics challenge problem with inherently complex and multidisciplinary obstacles to an optimized solution. The problem to be considered is the coordination and optimization of a number of advanced ATM concepts in the national airspace system (NAS). Taken individually, such concepts are valuable, but efforts to coordinate and optimize across several of them have not yielded computationally tractable solutions because of a need to (1) balance competing objectives, (2) deal with a variety of NAS network spatial and temporal effects, and (3) account for NAS uncertainties that may even be difficult to predict probabilistically. The multidisciplinary approach that is proposed is to integrate computational agent-based modeling (ABM) and probabilistic graphical models (PBM) to develop system-optimal coordination strategies for managing the simultaneous operation of multiple ATM concepts. This approach is innovative because IT expands applications from modern developments in complexity science (namely, ABM) and probabilistic modeling (namely, PGM) into the ATM domain, along with modeling & simulation capabilities (trajectory models, ATM concept models, etc.). The work directly supports the learn research objective (C), along with overall learn goals of innovation and relevance to NASA needs. The ABM approach is innovative because IT allows a framework for simultaneously optimizing competing objectives and satisfying conflicting constraints of multiple domain-specific planning tools. The PGM approach is innovative because, unlike traditional approaches, IT provides separation between model and solution-method, and IT enables succinct probabilistic modeling of cause-and-effect relationships in the highly interconnected NAS network and fast methods for probabilistically predicting the NAS future state. The results of the work will enable 4D flight trajectory management under uncertainty and directly support NASA, FAA and jpdo nextgen program research.