A rule-enhanced clustering approach to planning virtual stations for dockless shared micro-mobility systems

Abstract

Shared micro-mobility (SMM), as a typical service of green mobility, has been widespread in recent years, especially the dockless SMM system. However, the booming development of dockless SMM faces challenges in operations management hindered by disorderly parking. The common management practices for addressing such disadvantages start with partitioning spatial objects, one of which is strategically planning virtual stations. Based on extensive journey data and geospatial information, multi-source data are first integrated and utilized as the data basis. A rule-enhanced clustering approach is proposed for the large-scale virtual station planning of dockless SMM at the city level. The geographical distribution of virtual stations is refined by integrating clustering algorithms and geospatial rules which encompass geospatial object conflicts, enclosed land-use conflicts, traffic conflicts, and adjacent demands fusion. Then, the dockless SMM system in Shenzhen, China is taken as the case study. The results show that (1) the dockless SMM system in Shenzhen is imbalanced both spatially and temporally, requiring further refinement; (2) the optimal number of clusters (namely virtual stations) for K-means, DBSCAN, and OPTICS clustering is 8500, 9194, and 8257 respectively, among which K-means exhibits the best performance over metrics; (3) 5825 virtual stations are eventually located in Shenzhen by rules adoption, of which the alignment between virtual stations and geospatial circumstances is illustrated as well. The findings of this study indicate that the proposed approach can enhance the practicality of virtual station detection results when applied in real-world scenarios.