基于数据融合的城市道路行程时间预测模型研究

发布时间：2018-06-12 06:06

本文选题：道路行程时间 + 固定检测器　；参考：《大连海事大学》2014年硕士论文

【摘要】：道路行程时间是反映道路交通状况的重要指标。一方面,在现实情境中,由于交通需求在一天当中变化很大,使得道路网络交通流的时空分布规律具有时变特性,从而导致道路路段行驶时间很大程度上依赖于交通负荷的变化。因此,准确的路段行程时间动态预测,是交通诱导系统、交通信息服务系统以及交通协调控制系统的重要基础；另一方面,由于道路交通流运行的高度复杂性、随机性和不确定性,传统的基于检测线圈的路段行程时间预测方法、基于GPS浮动车的路段行程时间预测方法等单一方法一直未能取得令人满意的预测效果,这在一定程度上影响了道路交通控制以及道路交通诱导的效果。鉴于以上预测方法的优缺点及其互补性,本文在多源数据融合方法基础上,采用改进的BP神经网络围绕基于数据融合的城市道路行程时间预测方法开展了相关探索性研究工作,以期进一步提高道路行程时间预测的准确性。首先,针对固定检测器和GPS浮动车获取的道路交通参数中存在数据异常和丢失等问题,本文提出了基于相邻时段数据平均值法的故障数据修复改进方法。改进后的方法能够较好地修复故障数据,从而提高数据的质量。其次,针对多辆GPS浮动车数据进行道路行程时间预测过程中存在浮动车在道路上驶过的距离长度不等问题,本文提出了自适应权重系数线性加权融合方法,用于多辆GPS浮动车数据的道路行程时间预测。最后,针对单类型检测器数据存在道路行程时间预测不准确问题,本文提出采用改进的BP神经网路对两种检测器获取的道路行程时间数据进行融合,从而建立基于数据融合的道路行程时间预测模型。通过对上述研究成果进行经过仿真实验,初步达到了预期的研究目标。城市道路行程时间预测问题相对复杂,鉴于作者研究能力有限,论文研究还有许多需要进一步完善的地方,将在后续工作学习中不断改进。
[Abstract]:Road travel time is an important indicator to reflect road traffic situation. On the one hand, in the real situation, because of the great change of traffic demand in one day, the space-time distribution of traffic flow in road network is time-varying. As a result, the driving time of road section depends on the change of traffic load to a great extent. Therefore, accurate road travel time dynamic prediction is the important foundation of traffic guidance system, traffic information service system and traffic coordination control system. On the other hand, because of the high complexity of road traffic flow, Randomness and uncertainty, the traditional method based on detection coil to predict the travel time of road section, and the method based on GPS floating vehicle to predict the travel time of road section have not been able to achieve satisfactory results. To a certain extent, this affects the effect of road traffic control and road traffic guidance. In view of the advantages and disadvantages of the above prediction methods and their complementarities, based on the multi-source data fusion method, the improved BP neural network is used to carry out the related exploratory research work around the urban road travel time prediction method based on the data fusion. In order to further improve the accuracy of road travel time prediction. First of all, there are some problems such as data anomaly and loss in the road traffic parameters obtained by fixed detector and GPS floating vehicle. In this paper, an improved method for repairing fault data based on the mean value method of adjacent interval data is proposed. The improved method can repair the fault data and improve the quality of the data. Secondly, in the course of road travel time prediction based on the data of multiple GPS floating vehicles, the distance length of the floating vehicle passing on the road is not equal. In this paper, an adaptive weighting coefficient linear weighted fusion method is proposed to predict the road travel time of multiple GPS floating vehicle data. In this paper, an improved BP neural network is proposed to fuse the road travel time data obtained by two detectors, and a road travel time prediction model based on data fusion is established. The prediction problem of urban road travel time is relatively complex. In view of the author's limited research ability, there are still many areas that need to be further improved, which will be continuously improved in the follow-up study.
【学位授予单位】：大连海事大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U491;U495;TP202

【参考文献】