Application of genetic algorithm in estimating origin-destination matrices of traffic network

Origin-destination (OD) matrices are the essential data in the traffic planning and management activities, and many estimating models and solving methods were proposed. The Newton´s method is a main approach to solving the OD estimating models, but it needs calculate inverse matrices and its converging process strictly depends on the initial values. To overcome these shortcomings of the Newton´s method, a genetic algorithm (GA) is proposed to solve the OD estimating model based on the maximum entropy principle. Finally, this method is proved to be feasible in theory and in practice with a practical example.