季节性ARiMA模型在稀疏交通流下的预测方法

发布时间：2018-06-17 17:22

本文选题：交通工程 + 实时交通状况预测　；参考：《公路交通科技》2014年04期

【摘要】：交通流监测存在普遍的稀疏性,理想的交通流预测模型应该能够充分利用交通流数据的特征,克服稀疏性问题。通过大量的数据分析,城市道路交通流被证实存在时序上的周期性特征。同时,数据分析结果也表明了交通观测数据稀疏性的普遍存在,而且稀疏的分布不均匀,有些极端稀疏道路甚至出现数天的观测缺失。因此,交通流预测模型应该有对稀疏的适应性,而季节性ARiMA交通流预测模型的引入能够很好地利用时序周期特征计算交通观测值的缺失。这种模型的优势在于融合了邻近的交通流观察值和交通流数据的周期性,消除了道路稀疏性导致观测值缺失带来的预测障碍。对比试验的展示表明了这种模型对交通流数据周期性特征的利用和对稀疏性的适应。
[Abstract]:There is a general sparsity in traffic flow monitoring and the ideal traffic flow prediction model should be able to make full use of the characteristics of traffic flow data to overcome the problem of sparsity. Through a large number of data analysis, urban road traffic flow has been proved to have temporal periodicity. At the same time, the results of data analysis also show that the sparsity of traffic observation data is widespread, and the sparse distribution is uneven, and some extremely sparse roads even appear several days of observation missing. Therefore, the traffic flow prediction model should have sparse adaptability, and the introduction of seasonal ARiMA traffic flow forecasting model can make good use of the time series characteristics to calculate the missing of traffic observation value. The advantage of this model is that it combines the observed values of adjacent traffic flows with the periodicity of traffic flow data and eliminates the prediction obstacles caused by the lack of observation values caused by road sparsity. The comparison experiment shows that the model can make use of the periodic characteristics of traffic flow data and adapt to the sparsity of traffic flow data.
【作者单位】：深圳市交通运输委员会;中国科学院深圳先进技术研究院;
【分类号】：U491.14

【参考文献】