Freeway Travel Time Prediction by Using the GA-Based Hammerstein Recurrent Neural Network

Freeway travel time prediction has become a focus of research in recent years. However, we must understand that most conventional methods are very instinctive. They rely on the small amount of real-time data from the day of travel to look for historical data with similar characteristics and then use the similar data to make predictions. This approach is only applicable for a single day and cannot be used to predict the travel time on a day in the future (such as looking up the travel time for the coming Sunday on a Monday). This study therefore developed a Hammerstein recurrent neural network based on genetic algorithms that learns the freeway travel time for different dates. The trained model can then be used to predict freeway travel time for a future date. The experiment results demonstrated the validity of the proposed approach.