Street and traffic simulation: traffic flow simulation using CORSIM
Proceedings of the 32nd conference on Winter simulation
Genetic Algorithms in Search, Optimization and Machine Learning
Genetic Algorithms in Search, Optimization and Machine Learning
| Hi-index | 0.00 |
A genetic algorithm was used to search for optimum calibration parameter values for the vehicle performance models used by two well-known microscopic traffic simulation models, CORSIM and Integration. The mean absolute error ratio between simulated and empirical performance curves was used as the objective function. Empirical data was obtained using differential GPS transponders installed on trucks travelling on divided highways in Brazil. Optimal parameter values were found for the "average" truck for each truck class and for each vehicle in the sample. The results clearly show the feasibility of the proposed approach. The simulation models calibrated to represent Brazilian trucks individually provided average errors of 2.2%. Average errors around 5.0% were found when using the average truck class parameters.