上海交通大学学报(自然版) ›› 2015, Vol. 49 ›› Issue (08): 1131-1136.

• 自动化技术、计算机技术 • 上一篇    下一篇

二维局部非负矩阵分解的路网态势算法

许榕,吴聪,蒋士正,陈启美   

  1. (南京大学 电子科学与工程学院, 南京 210046)
  • 收稿日期:2014-09-01 出版日期:2015-08-31 发布日期:2015-08-31
  • 基金资助:

    国家科技重大专项(2011ZX0300500403),国家自然科学基金项目(61105015)资助

Evaluation of Network-Level Traffic State Using 2DLNMF Algorithm

XU Rong,WU Cong,JIANG ShiZheng,CHEN QiMei   

  1. (School of Electronic Science and Engineering, Nanjing University, Nanjing 210046, China)
  • Received:2014-09-01 Online:2015-08-31 Published:2015-08-31

摘要:

摘要: 针对路网态势评测算法存在限于断面、依赖单一指标等的不足,在解析测量指标和测量断面的相关性及局部非负矩阵分解(LNMF)算法的基础上,提出了二维局部非负矩阵分解2DLNMF算法,通过选择合适参数对路网数据进行降维处理,提取路网特征数据,从而实现路网态势评测.仿真结果表明,使用2DLNMF算法路网态势评测结果更加准确,而在线评测准确性达到95.69%.
关键词: 路网态势; 聚类; 二维局部非负矩阵分解; 特征提取
中图分类号: TP 18文献标志码: A

Abstract:

Abstract: Network-level traffic state reflects the macroscopical conditions of the road network. The exesting evaluation algorithms have some shortcomings such as their applicable conditions are limited to section and they just depend on a single index. As a result, based on the analysis of the correlation between the measuring index and sections and the local nonnegative matrix factorization(LNMF) algorithm,  the algorithm of 2DLNMF was proposed and the features of the traffic data were extracted by choosing appropriate parameters to reduce the numbers of dimensions of the road network data. The simulation results indicate that the evaluation of 2DLNMF is more accurate and its online accuracy is up to 95.69%.

Key words: network-level traffic state, cluster, 2D-local non-negative matrix factorization (2D-LNMF), feature extraction