基于案例的城市道路行程时间预测

摘要

Most existing travel time prediction methods focus on road segment forecasting for highways, urban expressways and trunk roads, while researches on path travel time prediction for route of urban common roads are relatively few. This paper proposes a case-based travel time prediction method. Based on trajectory data modeling, rich history travel paths with more attributes form into dynamic sample cases database, and a matching case are searched from the case database according to the travel time and environmental information. Finally, certain amendments are made to predict travel time. Experimental results show that the algorithm is weakly influenced by the impact of space network and data difference and can have a strong robustness and portability; in addition, the algorithm shows suitability to low trajectory data coverage, few trajectory data sample cases, as well as time accuracy and reliability, compared with the referenced existing road segment travel time prediction method.%已有旅行时间预测方法多是针对高速公路、城市快速路和主干道路的路段，而针对城市一般道路，及路径旅行时间预测的研究则相对不足。本文提出一种基于案例的旅行时间预测算法。算法的基本原理为，在轨迹数据建模的基础上，建立动态更新的、包含多个属性的历史旅行样本数据库，然后根据出行时间及环境信息从案例库查找匹配案例，其结果经过一定修正用于预测路径旅行时间。与基于路段的旅行时间预测方法相比，该算法具有较强的鲁棒性和可移植性，受空间路网和数据样本量的影响较弱；并且在相同数据样本量和路网空间覆盖率的情况下，该算法预测精度高于对比算法。