A dynamic OD prediction approach for urban networks based on automatic number plate recognition data

Jing Liu, Fangfang Zheng, Henk J. van Zuylen, Jie Li

Research output: Contribution to journalConference articleScientificpeer-review

12 Citations (Scopus)
89 Downloads (Pure)


OD flows provide important information for traffic management and planning. The prediction of dynamic OD matrices gives the possibility to apply anticipatory traffic management measures. In this paper, we propose an OD prediction approach based on the data obtained by Automated Number Plate Recognition (ANPR) cameras. The principal component analysis (PCA) is applied to reduce the dimension of the original OD matrices and to separate the main structure patterns from the noisier components. A state-space model is established for the main structure patterns and the structure deviations, and is incorporated in the Kalman filter framework to make predictions. We further propose three K-Nearest Neighbour (K-NN) based long-term pattern recognition approaches. The proposed approaches are validated with field ANPR data from Changsha city, P.R. China. The results show that the observed OD flows can be accurately predicted by our proposed approaches. Which prediction method performs best depends on the quality of the available data: for regular, periodic OD matrices the Kalman filter is better, for irregular OD matrices the pattern recognition that looks at different time periods in the historical data, gives better results.
Original languageEnglish
Pages (from-to)601 - 608
Number of pages8
JournalTransportation Research Procedia
Publication statusPublished - 2020
Event22nd EURO Working Group on Transportation Meeting, EWGT 2019 - Barcelona, Spain
Duration: 18 Sept 201920 Sept 2019


  • OD matrix prediction
  • pattern recognition
  • principal component analysis
  • state-space kalman filter model


Dive into the research topics of 'A dynamic OD prediction approach for urban networks based on automatic number plate recognition data'. Together they form a unique fingerprint.

Cite this