基于LS-SVM的交通流时序数据补齐方法

Method for filling time series data of traffic flow based on LS-SVM

实时、准确的交通流数据是实现智能运输系统(Intelligent Transportation Systems,简称ITS)的关键,对交通流的控制和诱导有直接影响。由于种种原因,通过交通检测器获得的数据往往是不完整的,存在丢失现象,影响了后续模型的实际应用效果。针对这一问题,提出一种基于最小二乘支持向量机(Least Squares Support Vector Machines,简称LS-SVM)的交通流时间序列数据补齐模型,利用交通流历史数据对丢失值进行诊断和修补。利用实例仿真验证表明,LS-SVM具有较好的泛化能力和很强的鲁棒性,采用基于LS-SVM的交通流时间序列模型补齐丢失数据能够取得很好的效果。

Real-time and accurate traffic flow data is essential for Intelligent Transportation Systems research.Traffic control and traffic guidance are directly affected by the quality of input data.The detected data is often incomplete and may cause out of or- der.The model for filling lime series data of traffic flow based on LS-SVM is proposed in this paper, missing data can be filled by using traffic flow historical data.The simulation results show that LS-SVM have better generalization ability and strong robustness.