共享单车搬迁策略研究

为满足单车用户的用车需求，对共享单车搬迁策略进行研究，提出两阶段共享单车搬迁策略。离线阶段构建单车停放区域提取模型，采用改进的DBSCAN聚类方法（District Pick up Technology， DPT）获取各时段热门用车区域、用车频率与行程结束后的放车区域、放车频率；在线阶段提出搬迁优化模型（Relocation Optimization Plan，OMP），根据当前时段的用车区域，搜索距离其最近的前一时段K近邻放车区域，并结合实时路况为其推荐前K条路况良好的单车搬迁路径。相较于传统的共享单车搬迁策略，在拥堵比例为48%时，OBJ值至少缩小52.4%，AT值至少缩小52.6%。

Moving Strategy of Bike-sharing System

In order to meet the needs of bicycle users， the shared bicycle moving strategy was studied. A two-stage shared bicycle relocation strategy was proposed. The extraction model of bicycle parking area was constructed in the offline phase， and the improved DBSCAN clustering method was adopted to get the popular pick-up area， the frequency of the vehicle and the drop-off area after the end of the trip， and the frequency of the vehicle； in the online stage， the relocation optimization plan is proposed， according to the current time period. In the pick-up area， the nearest K-near drop-off area is searched in the previous period， and the real-time traffic conditions is combined to recommend the top K moving route with good road conditions. Compared with the traditional shared bicycle relocation strategy， the OBJ value is reduced by at least 52.4% and the AT value is reduced by at least 52.6% when the congestion ratio is 48%.