基于混沌时间序列局域法的短时交通流预测

摘要

为提高城市短时交通流预测精度，将混沌时间序列分析应用于城市短时交通流数据，研究混沌时间序列局域预测法中的加权零阶局域法和加权一阶局域法。针对局域预测法在选取邻近相点时采用的欧氏距离和向量夹角两种方法只能片面反映邻近点的特点的问题，提出一种改进邻近相点选取的方法，综合相点相似程度和相点距离来选取邻近相点。再将原有方法和改进后的方法应用于北京市短时交通流预测中。结果表明，混沌时间序列局域法能适用于短时交通流预测，并且改进后的方法比原有方法具有更高的预测精度。%To improve the accuracy of urban short-term traffic flow forecasting,the chaotic time series analysis is applied to urban short-term traffic flow data,study the two local chaotic time series prediction,including adding-weight zero-rank local-region method and adding-weight one-rank local-region method. Euclidean distance method and vector angle method used in selecting neighbor points in local prediction method are being researched,and these two methods can not reflect the overall characteristics of the neighbor points,in view of this problem,an improved neighboring phase point selection method which integrated relative degree of similarity and distance to select neighbor phase points is presented. Then the old methods and the improved method are used in the Beijing short-term traffic flow prediction. The results show that local prediction method in chaotic time series can be used in short-term traffic flow forecasting,and the improved method has higher accuracy in prediction than the old methods.