摘要： 为空闲出租车司机推荐有效的闲逛路线在提高出租车司机工作效率、减少乘客等待时间以及缓解交通压力方面具有重要作用。现有的研究工作主要集中于为空闲司机推荐完整的驾驶路线，没有考虑到真实路网环境下某些路段的可等待因素，使得推荐的路线因载客概率较低、行驶距离较长而花费成本较高。提出一种基于候客点规划的路线推荐算法，对出租车轨迹数据进行处理，并设计路径匹配算法将每个轨迹点与真实路段一一匹配。通过统计每个路段历史接载信息，并利用一种改进的多层感知机建立可预测时序接载概率的模型，结合路段的可等待因素设计一种最小花费成本的路线推荐算法。在真实数据集上的实验结果表明，与MNP、InExperence、Random算法相比，所提算法花费成本、巡航时间以及巡航路程均明显减少。

关键词: 空闲出租车, 接载概率, 最小成本, 路线推荐, 多层感知机

Abstract: Effective cruising route recommendation for idle taxi drivers can help to improve drivers' efficiency, reduce passengers' waiting time, and alleviate traffic jam.Most existing studies focus on recommending complete driving routes for idle drivers, and ignore the waiting areas of road segments in real road networks, which causes multiple problems to recommended routes, such as reduced pick-up probability, long driving distance, and increased cost.To address this problem, we propose a route recommendation algorithm based on waiting area planning.The algorithm processes trajectory data of taxis, and employs an algorithm to map each trajectory point to a real road segment.By collecting the historical pick-up probability of each road segment, an improved multi-layer perceptron is used to establish a model for predicting the temporal sequence of pick-up probability.On this basis, we propose a cost-effective route recommendation algorithm that considers the waiting areas of road segments.The experimental results on real-world datasets show that compared with MNP, InExperence, Randombaseline and other algorithms, the proposed algorithm effectively reduces the cruising cost, cruising time, and cruising distance.

Key words: idle taxi, pick-up probability, minimal cost, route recommendation, Multi-Layer Perceptron(MLP)

中图分类号: 

• TP393