Space-Time Analysis of Refueling Patterns of Alternative Fuel Vehicles Using GPS Trajectory Data and Machine Learning

The urgency to combat climate change has led countries worldwide to embrace clean energy solutions across various sectors, including transportation, to reduce greenhouse gas emissions. This shift is evident in the growing popularity and adoption of alternative fuel vehicles (AFVs), such as natural gas vehicles and electric vehicles. AFVs have significantly lower carbon footprints compared to conventional petrol-powered vehicles with internal combustion engines. Consequently, there arises a need to gain a deeper understanding of AFV refueling demand to optimize the distribution of refueling stations. To address this, our research proposes an innovative space-time method that integrates GPS trajectory data with the support vector machine technique to accurately identify and analyze patterns in AFV refueling behavior. The results highlight distinct space-time patterns, notably the clustering of refueling activities in areas like Shuiguohu, Shouyilu, Houhu, Hanshuiqiao, Zhoutou, and Yongfeng around noon, influenced by taxi drivers' breaks. This underscores the importance of increasing staff levels at refueling stations in these areas during peak refueling periods, forming alliances with local eateries, and coordinating taxi shift hours to evenly distribute refueling demand throughout the day, ultimately reducing congestion during peak refueling periods in these areas. The proposed method by this research is applicable to urban contexts worldwide and equips policymakers and planners with a powerful tool for effective planning of future AFV refueling stations.