2005 年 6 巻 p. 4237-4252
The optimal shelter locations for the flood evacuation planning are studied in this paper. We assume that the authority can control the traffic in certain part of the network while the evacuees choose which shelter to go and by which route. The shelter location problem is posed as a Stackelberg game, consisting of the leader (authority) determining the shelter locations to minimize the total evacuation time and the follower (evacuees) choosing the destination (shelter) and route to evacuate. The problem is formulated as a bi-level programming. The upper level problem is a location problem that models the authority's decision. A combined distribution and assignment (CDA) model is proposed to model the evacuees' decision as the lower level problem. In this study, the bi-level programming problem is solved using genetic algorithm. Numerical example with a real world network is given to demonstrate the application of the proposed model.
