Advanced Approach to Calibration of Traffic Microsimulation Models Using Travel Times
Pavol Korcek, Lukas Sekanina and Otto Fucik
An effective calibration method of the cellular automaton based traffic microsimulation model is proposed in this paper. It is shown that by utilizing a genetic algorithm it is possible to calibrate different parameters of the model much better than a traffic expert. Moreover, using this process it is also possible to find several model parameters that are extremely difficult to calibrate as relevant data can not be measured using standard monitoring technologies or complete data sets are often not available. The quality of the new calibrated models is discussed in the task of vehicle travel time estimation. The precision of simulations is increased over three times compared to a manually tuned model. The average error rate is 10.75 % in comparison with several field travel time data.
Keywords: Traffic, simulation, cellular automaton model, genetic algorithm, calibration, travel time.