采用变结构动态贝叶斯网络的交通流量预测

发布时间：2018-01-13 16:22

本文关键词：采用变结构动态贝叶斯网络的交通流量预测　出处：《计算机科学与探索》2017年04期 　论文类型：期刊论文

【摘要】：物联网和大数据流式计算的快速发展为智能交通系统的研究带来新的机遇。交通流量预测一直是智能交通系统的关键问题。针对交通流量预测中一个固定模型无法适应多种环境的问题,以及面向数据流的模型更新问题,提出了一种基于变结构动态贝叶斯网络的交通流量预测方法。该方法以复杂事件处理和事件上下文为基础,通过上下文聚类进行历史数据的划分,并通过事件流在线聚类支持聚簇的更新。面向不同聚簇的数据,采取搜索-打分的方法学习对应的贝叶斯网络结构,基于高斯混合模型实现贝叶斯网络的近似推断。在线预测时根据当前上下文选择合适的模型或模型组合进行预测。真实和仿真数据上的实验结果表明,该方法能够获得比当前常用方法更好的预测效果。
[Abstract]:The rapid development of Internet of things and large data flow has brought new opportunities for the research of intelligent transportation system. Traffic flow forecasting has always been the key problem of intelligent transportation system. A fixed model in traffic flow forecasting is not available. Law adapts to various environments. A traffic flow prediction method based on variable structure dynamic Bayesian network is proposed, which is based on complex event processing and event context. The historical data is partitioned by context clustering, and the updating of clustering is supported by online clustering of event flow. For different clustering data, the corresponding Bayesian network structure is studied by search-scoring method. The approximate inference of Bayesian network is realized based on Gao Si hybrid model. In on-line prediction, the appropriate model or combination of models is selected according to the current context. The experimental results on real and simulation data show that. This method can achieve better prediction results than the current common methods.
【作者单位】：湖南大学信息科学与工程学院;
【基金】：国家自然科学基金No.61371116~~
【分类号】：U491.14;TP18
【正文快照】： 1引言于支持向量机的方法[2-3]、基于深度学习的方法[4-5]、基近年来,随着物联网(Internet of things,Io T)和移于贝叶斯网络(Bayesian network,BN)的方法[6-8]和基动计算技术的迅速发展,人们加速跨进大数据的时于混合模型的方法[9]等。基于贝叶斯统计技术,贝叶代。很多应用快

【相似文献】