Designing emergency flood evacuation plans using robust optimization and artificial intelligence

Abstract

This paper offers a new robust mathematical model for designing an efficient flood evacuation plan in disasters. The mathematical model takes a set of potential locations for establishing evacuation shelters with limited capacities. We suggest an innovative model to use helicopters to rescue people from the flood for the first time in this field of knowledge. Besides, the helicopters' capacity and maximum flying time are considered limited to adopt real-world conditions. Our goal is to maximize the number of rescued people and to minimize the total cost. By solving the model, we determine the allocation of people to the shelters, the optimal location of shelters, allocation of the helicopters to the evacuation shelters, and flying path and routes of the helicopters. Since the main parameters of the proposed model are due to uncertainty in real-world situations, we implemented robust optimization to formulate uncertainties. Due to the Np-hardness of the suggested formulation, we offer four algorithms to solve the mathematical model. We enhance the efficiency of the algorithms through a robust design of experiments and assess their performance considering several measures via post hoc analysis. At the end, we implement the robust model on real-world data from 2011 Japan’s destructive tsunami in Ishinomaki city. The results reveal that the model is able to provide promising solutions compared to the classical models and leads to higher rescue rates and lower cost.