Development of an efficient methodology for robustness-based assignment problem

Abstract

This thesis proposes formulations and algorithms for assignment problem under natural or physical variability, from the perspective of robustness of the assignment. This thesis formulates the robustness-based assignment problem as a non-linear binary programming problem. An equivalent linear binary programming formulation and a modified Hungarian approach are then proposed to achieve computational efficiency. The weighted sum approach is used for the aggregation of multiple objectives and to examine the trade-offs among multiple objectives. The proposed methods are illustrated for two example problems, where the information on the problem parameters is available as their means and standard deviations. Both the non-linear and linear formulations of the proposed robustness-based assignment problem can produce the exact same results, however, with much decreased computational expenses for the later.