Traffic State Estimation based on Low Frequency Detection Data at Signalized Intersections

Abstract:

A traffic state identification method for intersections is proposed based on detection data with a low frequency from detection on the mid of the urban roads which is applied for urban interrupted flow in medium and small cities of our country. At first, the relationship between occupation, volume and traffic state is analyzed under different parameters of circumstances based on simulation data and a method of curve fitting is proposed to build the boundaries of different traffic state. Then the functional relationship of coefficients of boundaries functions with environment variables is fit out which is later applied to general ones. The methods above is verified by simulation data with the identification rate of over 80%, and empirical data with the identification rate of over 75% , with severe mistake rates less than 2.1%.