Scheduling and Optimization of Dockless Bike-Sharing Considering Tide Phenomenon in Morning Peak Period

To address the challenges of bike-sharing accumulation and parking difficulties caused by tidal effects, this paper proposes a tidal point recognition algorithm and a user-oriented parking algorithm for tidal areas based on bike-sharing order data in Xiamen, China. Firstly, spatial analysis technology is used to extract multi-dimensional features from the electronic fence of bike-sharing, and clustering is employed to identify tidal point locations. Subsequently, a scheduling algorithm is introduced to optimize the return process of bike-sharing in tidal regions. This algorithm integrates macro-regional incremental equilibrium and micro-dynamic optimization assignment. At the macro level, the equilibrium state of supply and demand for electronic fence groups is analyzed using historical demand data. At the micro level, the weighted supply-demand complementarity and dispatch distance between tide and non-tide fences are calculated. Real-time features are then combined to optimize the scheduling of return orders, achieving a dynamic balance between micro optimization and macro equilibrium. Finally, the proposed algorithm is tested, demonstrating that it can remove 34.6% of congested bike-sharing within an average scheduling distance of 50 m.