Development of a Java Applet for Generating Truck Trips from Freight Data

Abstract

As freight transportation becomes a more significant concern, the ability to estimate accurately truck trips generated by freight activity at an intermodal facility is important for transportation engineering and planning. Truck trip generation models that used vessel freight data were developed by the University of Central Florida Transportation Systems Institute and have been statistically proved to determine accurately the number of trucks generated at a seaport. To apply these previously developed models more efficiently, a Java applet was developed to execute a selected artificial neural network (ANN) port model and a trainable ANN port model. This applet provides the user with an easy-to-understand interface for entering data and executing models. The two models developed were the Port Everglades ANN model and a hybrid of the Port Everglades model that allowed the user to retrain the ANN model before execution for desired data. Although the trainable model requires additional retraining with new data, the added complexity comes with the benefit of producing a network with a higher degree of accuracy. This applet equips the user with two models and therefore has expanded the capabilities of the previously limited Port Everglades ANN truck trip generation model. The developed Java applet can be expanded to include more ANN models and thus more flexibility, depending on the type of freight data available. This successful adaptation of the ANN model into a Java applet sets the foundation for further applications.