Flow rate and time mean speed predictions for the urban freeway network using state space models

Abstract

Short-term predictions of traffic parameters such as flow rate and time mean speed is a crucial element of current ITS structures, yet complicated to formulate mathematically. Classifying states of traffic condition as congestion and non-congestion, the present paper is focused on developing flexible and explicitly multivariate state space models for network flow rate and time mean speed predictions. Based on the spatial–temporal patterns of the congested and non-congested traffic, the NSS model and CSS model are developed by solving the macroscopic traffic flow models, conservation equation and Payne–Whitham model for flow rate and time mean speed prediction, respectively. The feeding data of the proposed models are from historical time series and neighboring detector measurements to improve the prediction accuracy and robustness. Using 2-min measurements from urban freeway network in Beijing, we provide some practical guidance on selecting the most appropriate models for congested and non-congested conditions. The result demonstrates that the proposed models are superior to ARIMA models, which ignores the spatial component of the spatial–temporal patterns. Compared to the ARIMA models, the benefit from spatial contribution is much more evident in the proposed models for all cases, and the accuracy can be improved by 5.62% on average. Apart from accuracy improvement, the proposed models are more robust and the predictions can retain a smoother pattern. Our findings suggest that the NSS model is a better alternative for flow rate prediction under non-congestion conditions, and the CSS model is a better alternative for time mean speed prediction under congestion conditions.