基于非参数回归的路段速度估算方法*

城市道路交通运行状况的及时获取对交通管理和出行有着重要的作用，而准确判断交通状况需要利用准确的交通流速度。针对城市道路配备有一定数量的固定检测器，但是浮动车覆盖率不足的情况，提出了基于非参数回归的路段速度估算方法。该方法有效结合固定检测器数据和浮动车数据，对不同时段、路段、天气对行驶速度的影响分别进行分析,建立状态模式库，利用模式匹配估算出道路交通流速度。通过算例分析比较，可知该算法具有较高的准确度。

Average speed evaluate based on non-parametric regression

Timely access to information about the situation of urban traffic operation has an important impact on traffic mana-gement and daily travel. Meanwhile, it is necessary to use accurate traffic flow speeds to determine traffic status in a high accuracy. As for the problem that the roads are allocated a number of fixed detectors, but the number of floating cars is insufficient or the coverage is not high, this paper proposed an evaluation method of the average speed based on non-parametric regression, in which data from fixed detectors and floating cars were effectively combined. Besides, analyzed the velocity of traffic flow under various time periods, road sections and weather conditions respectively, so could construct a status pattern database based on above analysis, then would compute the velocity of traffic flow through pattern matching. Finally, the results by comparing the estimated speed and actual speed demonstrate the effectiveness of the algorithm.