中文摘要：

为实现基于公交IC卡数据的大规模OD（origination-destination）矩阵推导,提出了一种不关联公交调度信息和GPS数据的OD矩阵推导算法。提出了站点序号标注算法,通过时间聚类思想设计了自适应调整的时间间隔阈值以判断公交车的行驶状态,将公交站点序列与刷卡记录进行匹配;在此基础上,提出了单个公交车的行驶方向标注算法,通过从已知行驶方向的公交车推导未知方向公交车的方向标注算法。为了最大化解决公交数据的上车站点信息补全问题,将全局公交车行驶方向标注问题映射为图论中的节点遍历问题,利用贪心生长算法和广度优先策略实现了局部最优。最后该算法处理某市的公交IC卡数据,得到了公交出行链假设下的城市居民大规模OD矩阵。结果显示算法可有效推导大规模OD矩阵。

英文摘要：

In order to realize the large-scale OD matrix estimation based on bus IC card data, this paper presented a new method without transit scheduling data and GPS data. First, it proposed a station labeling algorithm to match each record with a sequence number of a bus route. It designed an adaptive adjustment time-interval threshold to determine the driving state of each bus by clustering the time interval of adjacent records. Then, it proposed a single-line direction labeling algorithm to estimate the driving direction of a bus according to a known one. This paper mapped the global driving direction estimating problem to the graph traverse problem and realized the local optimum using the greedy growth algorithm with breadth-first traversal strategy. At last,it estimated the large-scale OD matrix by using the algorithms and real IC card data under the assump- tion of passengers＇ bus-based trip chains. The result shows that the algorithms are effective in estimating large-scale OD matrix.