Short-term traffic flow forecasting has been the most important application of the intelligent transportation system (ITS). This paper presents a model structure with a coupled hidden Markov model (CHMM) for short-term traffic prediction in the highway system with real-time traffic flows data. Data used in this study was gathered from simulation software. The model defines traffic states in a two-dimensional space with speed and volume observations. The decoding function of CHMM is used in this study to estimate the most likely sequence of traffic states. The forecasting model is accessed by predicting errors. The CHMM is compared to autoregressive moving average (ARIMA), which is one of the most widely used regression techniques. These results present that the CHMM outperforms the regression model. Consequently, the paper concludes that CHMM is more robust and successful in modelling unstable traffic conditions.
展开▼
