User-Based Relocation for e-Bike Sharing Systems Through Power-Level Recommendation

Recent years have witnessed the growing popularity of e-bike sharing systems (EBSS) due to their enhanced riding comfort and suitability for longer rides, compared to conventional bike sharing systems (BSS). However, similar to BSS, EBSS also suffer from the spatiotemporal supply-demand imbalance. Additionally, the demand loss in EBSS could be caused by not only the shortage of e-bikes but also the insufficiency of battery power. Currently, operators address this imbalance by employing staff-driven trucks to relocate e-bikes and replace batteries, which is costly and environmentally unfriendly. This paper proposes a user-based relocation policy as a complement to the staff-based relocation, aiming to effectively mitigate current challenges. This policy incentivizes users to select e-bikes with more systematically favored battery power levels, taking both the future system states and user preferences into account. This approach aims to achieve better e-bike distribution and utilization across different power levels within the system. Specifically, a multi-battery-state Markov model is developed to estimate the utilities of incentivized relocations, utilizing an approximate state reduction method to simplify high-dimensional calculations. Additionally, an optimization model considering user choice patterns is formulated to determine incentives and e-bike recommendations for concurrent orders, solved by a variable neighborhood search algorithm. Numerical experiments conducted on a real-world EBSS demonstrate that the proposed policy can increase profits by around 10% when implemented independently or integrated with existing staff operations, with only 1% of the revenue used as incentive costs.