A short-term state-space model for free flow prediction based on spatial-temporal characteristics

作者：Dong Chunjiao^*; Shao Chunfu; Zhuge Chengxiang; Meng Meng

来源：China Civil Engineering Journal, 2013, 46(8): 111-118.

• ARMA model
• Free flow
• Spatial-temporal characteristics
• State-space modeling
• Traffic flow prediction

Based on spatial-temporal characteristics of traffic flow, a free-flow short-term state-space prediction (FSSP) model was developed for the network prediction. Firstly, the autocorrelation function was employed to estimate the stability of flow rate, occupancy, and speed time series by using measurement data. Secondly, the spatial transfer characteristics of traffic flow parameters under free flow were investigated. In addition, a short-term traffic flow prediction model was proposed by solving the conservation equation, for which an upwind numerical discretization method was adopted to approximate the solutions. Furthermore, the model was incorporated into FSSP model by considering the impacts of entry and exit ramp, varying number of lanes, and road grade. The model parameters were estimated by combining the qualitative and quantitative analysis methods. The empirical findings show that the FSSP model can achieve the goal of network prediction and the mean deviation of the prediction is 1.73veh/2min. Under the same conditions, the mean deviations predicted by the classical Auto-Regressive Moving Average (ARMA) model and Elman model are 3.19veh/2min and 2.14veh/2min, respectively.

• 出版日期2013
• 单位The University of Tennessee