Neural Network-Based Prediction of Import Container Outflows via Trucks: A System-of-Systems Analysis in Port Terminals

This paper proposes a methodology centered on the application of a feed-forward Artificial Neural Network (ANN), with the Jackknife cross validation to predict the curve of import containers leaving a port terminal via trucks. Viewing the terminal as a system of systems, we analyze the interactions among the quay, yard, and gate crucial for finalizing the import process. The performance of our predictive methodology in forecasting the outflows of Twenty-Foot Equivalent Units (TEUs) was evaluated using correlation coefficient R and Mean Squared Error (MSE), and the results demonstrated its effectiveness. However, the task is challenged by the random arrival pattern of trucks at the terminal for TEUs pickups, highlighting the competitive nature of the prediction process. The outcomes of this work can contribute to a more efficient resource allocation and to a better organization of the yard operations.