Geo-fenced parking station planning and user incentive scheme for bike sharing

• Integrates geo-fenced station planning with incentive-based inventory management. • Data-driven clustering optimizes station locations and capacities using historical riding data. • Designs optimal incentives minimizing regulator and firm costs to solve parking issues. • Promotes low-emission, sustainable urban transit via efficient bike-sharing systems. • Proposes a transferable framework readily applicable to diverse urban contexts. Management of dockless bike-sharing systems encounters numerous challenges, including disorderly parking and frequent mismatches between supply and demand in bike inventory, which result in traffic congestion and compromise the efficiency of the entire system. Geo-fenced station restrains the behavior of disorderly parking while eliminating the costs associated with the construction of physical stations. This paper addresses these challenges by investigating two key issues: the location and capacity design of geo-fenced parking stations, and the user incentive scheme for managing the bike-sharing system. The design of station locations and capacities fundamentally serves as a spatial constraint for inventory systems, while inventory levels directly impact station utilization efficiency. This synergistic study improves system’s capability to resolve demand-supply imbalances. Initially, we employ the DBSCAN algorithm to cluster real-world user riding data, which is then utilized as the basis for determining station locations. Furthermore, we develop a consumer choice model aimed at minimizing social costs for the government regulators and operational costs for bike-sharing firms, to derive user incentive strategies based on the dynamic status of the system. Utilizing riding data from a mid-sized Chinese city, extensive experimental results validate the effectiveness in reducing disorderly parking and mismatch of supply and demand applying our proposed methodology.