Genetic Algorithm with Boosting based on Expected Value for Uncertain Routing

Abstract

Floods are natural disasters that seriously affects life, property and infrastructure. People in affected areas need help. Food, water, medicine, and consumer goods are essential relief supplies for victims’ survival in a flood situation. Therefore, relief supplies must be distributed to the victims quickly. Loss of life may occur if performed delayed or long waiting time for rescue. Road transportation is a process used to distributes relief supplies. The core of transportation is vehicle route planning that can reach the victim as quickly as possible. However, roads are often affected by floods. Road conditions may be damaged or flooded after the vehicle is planned or dispatched. Route planning for typical situations is therefore unable to respond to flood situations and results in delay. Therefore, vehicle route planning should be able to cope with the uncertainty. This paper proposes to apply the expected value technique to optimize genetic algorithm (GA) with the uncertainty of the data. Total waiting time is an evaluation of the effectiveness of this research. The proposed model, called GA-BEV model, consists of two components, which are 1) GA is an optimizer to search for the optimal route set, 2) Boosting is elite populations creator and encoding data. The proposed model can search a route set that appropriately responds to the uncertainty of road conditions. The road at risk of flooding or that may take longer to travel is avoided, reducing the waiting time.