Calibration of the empirical fundamental relationship using very large databases

Abstract

This paper describes a procedure for fitting traffic stream models using very large traffic databases. The proposed approach consists of four steps: (1) an initial treatment to eliminate noisy, inaccurate data and to homogenize the information over the density range; (2) a first fitting of the model, based on the sum of squared orthogonal errors; (3) a second filter, to eliminate outliers that survived the initial data treatment; and (4) a second fitting of the model. The proposed approach was tested by fitting the Van Aerde traffic stream model to 104 thousand observations collected by a permanent traffic monitoring station on a freeway in the metropolitan region of São Paulo, Brazil. The model fitting used a genetic algorithm to search for the best values of the model parameters. The results demonstrate the effectiveness of the proposed approach.

Keywords

• Traffic flow models
• very large databases
• genetic algorithm.
• Model fitting