基于谱分析的路段行程时间多步预测方法

发布时间：2018-04-03 11:17

本文选题：城市交通　切入点：行程时间　出处：《交通运输系统工程与信息》2015年03期

【摘要】：路段多步行程时间预测数据是动态交通诱导系统的重要参数,但已有研究成果,大多集中于一步预测,且存在适应性不强、计算量大、基础数据需求多等不足.应用谱分析及Karhunen-Loeve(K-L)变换对历史及当前检测行程时间序列进行分解与重构,重构时以历史序列与当前检测序列的欧式距离作为相似性度量指标,优化重构时的特征向量系数,使与当前检测序列相似度高的历史序列信息在重构中占据主要地位,通过重构,实现对后续若干时段的行程时间的预测,实测数据检验显示该方法可实现多步预测,预测精度良好,较以往方法有所提高,且历史数据需求量小,计算量小.
[Abstract]:Multi-step travel time prediction data is an important parameter of dynamic traffic guidance system, but the research results have been mostly focused on one-step prediction, and there are some shortcomings, such as weak adaptability, large amount of calculation and many basic data requirements.The history and current detection travel time series are decomposed and reconstructed by spectral analysis and Karhunen-Loeveen K-Ltransform. The Euclidean distance between the historical sequence and the current detection sequence is taken as the similarity measure to optimize the eigenvector coefficients in the reconstruction.The history sequence information with high similarity to the current detection sequence occupies the main position in the reconstruction. Through the reconstruction, the travel time of several subsequent periods can be predicted. The test of the measured data shows that the method can realize the multi-step prediction.The accuracy of prediction is good, which is higher than that of the previous method, and the demand of historical data is small and the amount of calculation is small.
【作者单位】：西北工业大学自动化学院;华东交通大学土建学院;
【基金】：江西省自然科学基金(20142BAB201015) 江西省科技厅科技计划项目(20123BBE50094)
【分类号】：U491

【参考文献】