Optimization of fuel distribution network using deep reinforcement learning : case of Naftal

Abstract

The goal of this work focuses on the study of real road transportattion problems by proposing models and solutions related to the improvment of performing the gas station distribution network. Inthis study, the focus is on the optimization of demands control and inventory management of station using loT technologies and forecasting tools, and improve the multi-compartment vehicule routing with time windows (MCVRPTW) problem arising in the petroleum products distibution and container tranfer industry, with three different approach using mathematical model, heuristic and deep reinforcement learning approaches.