An Emergency Resource Allocation Research Based On Robust Optimization

Robust optimization can help to achieve practical results in uncertain problermsnot requiring the probability distribution of the parameters, and its applications inmulti-stage uncertain decision problem become the focus in operation research. In thisthesis, I make use of the concept of Adjustable Robust Counterpart (ARC) and AffinelyAdjustable Robust Counterpart (AARC) which are introduced by Ben-Tal andNemirovski, and have a research on emergency resource allocation problem bychoosing polyhedral uncertainty sets, my own responsable works is as following:Firstly, the general mathematical model of emergency resource allocation problemis proposed. I find that the major researches on this problem are under the assumption ofcertain demand or demand obeying some probability distribution through making asurvey of it. On the contrary, the assumption above is removed in this thesis, robustoptimization methodology is employed to constrsuct models ebedding uncertaindemand of disaster points.Secondly, a survey of RLP theory is given, and an improvement of AARC inmulti-stage uncertain decision, named fix recourse assumption, is poeposed consideringthe tractability of computation, along with its proof of function on tractability.Thirdly, I generate a robust logistics plan which can mitigate demand uncertaintyin supply chains and apply robust optimization methodology for dynamically assigningemergency response and evacuation traffc fow problems which have time dependentdemand uncertainty. This thesis studies a Cell Transmission Model (CTM) based systemoptimum dynamic traffc assignment (SO-DTA) model. I employ a min-max criterionand make an application of an extension of the robust optimization method adjusted todynamic optimization problems, named AARC approach as refer above.Finally, simulation experiments of two types of small and large scale networksshow that the AARC solution provides excellent results comparing to DLP solution andSSP solution.