中文摘要：

旅行时间作为交通系统运行的关键参数，可以为交通诱导系统和出行者路径选择提供决策建议。利用多源数据进行旅行时间的估计是智能交通系统运行的重要支撑。利用基于同一路段的3种检测数据，提出相应的权重分配模型和神经网络模型来进行多源检测数据的融合以获得融合后的旅行时间。对比研究了基于多断面检测器的旅行时间的2种推算方法：速度累进和速度平均。利用北京市典型道路数据对这2种融合技术的融合效果进行了对比分析，结果显示，多源数据融合可以提高旅行时间估计的准确性。

英文摘要：

Travel time, as a main operation parameter of transport system, can provide decision-making suggestions for traffic guidance system and traveler route choice. It is important to estimate travel time based on multi-source detected data in Intelligent Transportation System （ITS）. We proposed the weight distribution model and the neural network model to estimate travel time by fusing multl-source detected data based on 3 types of detected data on the same link, and compared 2 distinct travel time estimation methods based on multiple cross-sectional detectors： the progressive speed model and the mean speed model. The typical data from the roads in Beijing are used for comparing the results of the 2 proposed models. The results show that multi-source data fusion model can improve the accuracy of travel time estimation.