Short-time Traffic Flow Prediction Using Third-order Volterra Filter with Product-decoupled Structure

摘要：

A prediction model for short-time traffic flowseries is proposed in this paper.At first,estimationof the largest Lyapunov exponent is implemented byapplying small data sets method so as to validatethat chaos exists in traffic flow series.Then,throughproperly choosing the delay time and the embeddingdimension using mutual information and falsenearest neighbor methods,respectively,phase spacereconstruction for traffic flow series is performed.Insuccession,aiming at the problem that number ofcoefficients for Volterra filter exponentiallyincreases with the order of the filter,a third-orderVolterra filter with approximately product-decoupledstructure is put forward to reducing computationalcomplexity.And the coefficients of this filter areadaptively adjusted employing an improvednonlinear normalized least mean square(NNLMS)algorithm.Finally,experimental results show thatthe proposed technique can effectively predict trafficflow series and reduce the model complexity.

作者: Yumei Zhang Shiru Qu Kaige Wen

作者单位: Department of Automatic Control,Northwestern Polytechnical University,Xian 710072,P.R.China

会议类型: 国际会议

会议名称: 2008 3rd International Conference on Intelligent System and Knowledge Engineering(第三届智能系统与知识工程国际会议)(ISKE 2008)

会议地点: 厦门

会议语种:英文

页码: 609-614

在线出版日期: 2008-11-17（万方平台首次上网日期，不代表论文的发表时间）