(338g) Developing a Machine Learning Based Decision-Making Framework for Selecting the Best Locations of Hydrogen Fueling Stations

Hydrogen has been drawing an attention as an alternative sustainable transport energy source. In order to replace the existing gasoline based transportation system, there are much to be done in terms of establishing hydrogen infrastructure. Since hydrogen has been used in petrochemical plants for quite a while, the focus should be given to accelerating the introduction of hydrogen fueling station (HFS). It should be constructed in many geographically spread locations. It is still in the beginning that countries all over the world are considering the potential of hydrogen based transportation system. Practically speaking, only a handful of HFSs can be constructed in the beginning. The location should be carefully selected to maximize the potential of HFS. This paper proposes a new decision-making framework in computing the best location of HFS. The framework is based on the assumption that HFS follows the similar pattern of the fossil fuel stations such as gasoline or LPG stations. At the initial step in the framework, hydrogen demands are evaluated by considering the data including the locations of fossil fuel stations, populations, the number of registered vehicles at each location. The framework thus employs the machine learning techniques to compute the location of HFSs to meet the hydrogen demand. Particularly k-medoids clustering, one of the well-known machine learning techniques is used in this paper. The applicability of the proposed method was illustrated in a numerical case of Seoul. There are much to be done in establishing the hydrogen infrastructure. More works are expected to be followed in the data-based rigorous methodology to accelerating efficient hydrogen infrastructure for the mitigating the impact of climate change.