Error Correction Model of Travel Time for Floating Car Data Based on Co-Integration Theory

The speed for floating car data cannot reflect the actual conditions of traffic flow due to the errors of map-matching algorithm and lack of data samples, which can be solved by data fusion method. An error correction model of floating car travel time based on co-integration theory is constructed by reference to the license plate recognition data, in which co-integration theory and error correction model from econometrics area are applied to data fusion. A case on the elevated roads in Shanghai is provided to reveal the meaning of the parameters in the proposed model. Long-term equilibrium and short-term imbalance are analyzed to explore their effect on the fluctuation of travel time, and their interrelation between each other in terms of parameters' variation with time.