Driving Type Categorizing and Microscopic Simulation Model Calibration

Microsimulation programs, before being used for traffic analysis in practice, should be calibrated to reproduce the traffic characteristics and to represent the driver population in reality. A driver behavior questionnaire survey conducted in China revealed that significant differences in driving behavior existed in drivers. Calibration should be done by taking these differences in driving style into account. In this study, 30 drivers in Changsha, China, were invited to participate in in-car tests and complete a questionnaire. Factor analysis revealed the interdependencies between observed acceleration or deceleration characteristics and other variables. Four driving types were classified according to the four factor scores. Further analysis of repeated trips validated the consistency of driver membership within a class. A censoring method was applied to calibrate the desired acceleration functions and desired speed distributions in a simulation model for these four driving types. Distinct differences were found in desired speed and maximum desired acceleration or deceleration of the four driving types. It is expected that the categorization of driving types will be useful in model calibration.