Traffic state estimation through compressed sensing and Markov random field |
| |
Institution: | 1. Queensland University of Technology (QUT), 2 George St GPO Box 2434 Brisbane Qld 4001 Australia;2. Department of Traffic Engineering, Tongji University 4800 Cao''an Road, Shanghai 201804, P.R. China;1. Electrical Engineering Department, King Abdullah University of Science and Technology, Thuwal, Saudi Arabia;2. Department of Civil, Architectural and Environmental Engineering, University of Texas Austin, USA |
| |
Abstract: | This study focuses on information recovery from noisy traffic data and traffic state estimation. The main contributions of this paper are: i) a novel algorithm based on the compressed sensing theory is developed to recover traffic data with Gaussian measurement noise, partial data missing, and corrupted noise; ii) the accuracy of traffic state estimation (TSE) is improved by using Markov random field and total variation (TV) regularization, with introduction of smoothness prior; and iii) a recent TSE method is extended to handle traffic state variables with high dimension. Numerical experiments and field data are used to test performances of these proposed methods; consistent and satisfactory results are obtained. |
| |
Keywords: | |
本文献已被 ScienceDirect 等数据库收录! |
|