基于网格化的出租车空载寻客路径推荐

GRID-BASED ROUTE RECOMMENDATION OF NO-LOAD TAXI

出租车在城市交通中扮演着十分重要的角色。通过研究出租车空载寻客路径推荐来提高出租车载客效率,具有较大的现实意义。许多城市以"环"进行区域划分,使出租车订单OD(Origin-Destination)数据呈现出环内相似、环间不同的分布特点。基于此,对订单数据进行环形切分,结合区域面积和订单数量建模,计算出租车载客核心点。提出网格化的出租车空载寻客曼哈顿路径算法,将出租车与载客核心点之间的区域进行网格化处理,找出载客概率最大的一条曼哈顿路径推荐给空载出租车司机。实验表明,较直接聚类方法,先进行环形数据切分计算出的载客核心点分布更加均匀、合理。基于网格化方法推荐的最优曼哈顿路径载客概率不低于经典的基于最短距离的路径规划算法获得的路径。

Taxis play a very important role in urban transportation.It is of great practical significance to improve the efficiency of taxi pick up passengers by studying the route recommendation of no-load taxis.Many cities use the ring to divide the area,so that the origin-destination(OD)data of the taxi order have similar features inside rings and different features between rings.Based on this,we did ring segmentation of the order data,and combined the zone area and order quantity modeling to calculate the passenger core point.We proposed a grid-based Manhattan route algorithm.The area between taxi and passenger core was gridded to find a Manhattan path with the greatest passenger carrying probability and recommend it to no-load taxi drivers.Experiments show that compared with the direct clustering method,the distribution of core points calculated by the ring data segmentation method is more uniform and reasonable.The passenger pick up probability of the optimal Manhattan path recommended by the grid-based method is not lower than probability of the path obtained by the classical path planning algorithm based on the shortest distance.