01697694

Component

Traffic bayonet system which includes useful traffic information is one of important parts in the Intelligent Transportation System. However, similar to other traffic detectors, missing data in traffic bayonet system is a significant obstacle for its application. To mitigate the consequences of such missing data, numerous tensor-based methods have been proposed in the previous literature. Nevertheless, most of them assume that it is known where and when missing data occurs. This is unpractical because missing data occurs completely at random. In this paper, we propose a novel tensor-based algorithm which utilizes multi-dimensional inherent correlation of traffic data to detect and correct missing data in the bayonet system, namely Iterative Tensor Decomposition (ITD). The proposed algorithm is evaluated using real-world bayonet data sets. Experimental results show that missing states of the bayonet system can be classified into three cases, i.e., no missing, random elements missing and days missing. And the proposed ITD can accurately detect and correct missing data under different missing cases. Furthermore, ITD is compared with other state-of-the-art-methods and the comparison results show that ITD outperforms the other methods.

This paper was sponsored by TRB committee ABJ70 Standing Committee on Artificial Intelligence and Advanced Computing Applications.

19-05437

English

0000-0002-3201-7339

7p

2019

Transportation Research Board 98th Annual Meeting

Digital/other

Figures; References; Tables

Intelligent transportation systems; Tensor analysis; Traffic data

Data and Information Technology; Highways; Operations and Traffic Management

Transportation Research Board Annual Meeting 2019 Paper #19-05437

TRIS, TRB, ATRI

Dec 7 2018 9:34AM