GA Based Neural Network for Short-term Traffic Flow Prediction in Urban Signalized Arterials

Short-term traffic flow prediction is complex but important to handle urban traffic congestion. To acquire accurate traffic flow information beforehand, a prediction model based on the sufficient fusion of GA with NN is developed in this study. In this paper, the detailed formulation of the genetic structure is given and procedures for coding the NN architecture, the number of hidden nodes, activation functions in the hidden and output nodes, and the type of minimization algorithms used by the back-propagation are described. Validation of the performance of this model is carried out. The comparison of the prediction results between the GA based NN model and the trial-and-error NN model shows that the model presented in this paper outperforms the trial-and-error NN model