Supervised Weighting-Online Learning Algorithm for Short-Term Traffic Flow Prediction
Supervised Weighting-Online Learning Algorithm for Short-Term Traffic Flow Prediction
复制标题
DOI：
10.1109/tits.2013.2267735
复制
发表时间：
2013-12-01
影响因子：
8.5
通讯作者：
Easa, Said M.
中科院分区：
文献类型：
--
作者：
Jeong, Young-Seon;Byon, Young-Ji;Easa, Said M.
短期交通流预测已成为智能交通系统的主要研究领域之一。在动态交通分配的背景下，准确估计交通流量预测对于有效和主动的交通管理系统的运行是非常重要的。对于预测短期交通流量，最近的交通信息显然是近期交通流量的一个更重要的指标。换句话说，应该考虑交通流数据之间的时间差的相对重要性。虽然已经有一些短期交通流量预测的研究工作，但它们都是离线方法。本文提出了一种新的短期交通流预测模型——在线学习加权支持向量回归（OLWSVR）。将OLWSVR模型与人工神经网络模型、局部加权回归模型、传统支持向量回归模型和在线学习支持向量回归模型进行了比较。结果表明，该模型的性能优于现有模型。
Prediction of short-term traffic flow has become one of the major research fields in intelligent transportation systems. Accurately estimated traffic flow forecasts are important for operating effective and proactive traffic management systems in the context of dynamic traffic assignment. For predicting short-term traffic flows, recent traffic information is clearly a more significant indicator of the near-future traffic flow. In other words, the relative significance depending on the time difference between traffic flow data should be considered. Although there have been several research works for short-term traffic flow predictions, they are offline methods. This paper presents a novel prediction model, called online learning weighted support-vector regression (OLWSVR), for short-term traffic flow predictions. The OLWSVR model is compared with several well-known prediction models, including artificial neural network models, locally weighted regression, conventional support-vector regression, and online learning support-vector regression. The results show that the performance of the proposed model is superior to that of existing models.