Knowledge-Based Short-Term Load-Forecasting for Maritime Container Terminals

Short-term load-forecasting for individual industrial customers has become an important issue, as interest in demand response and demand side management in modern energy systems has increased. Integrating knowledge of planned operations at industrial sites into the following day’s energy-consumption forecasting process provides advantages. In the case of a maritime container terminal, these operation plans are based on the list of ship arrivals and departures. In this paper two different approaches to integrating this knowledge are introduced: (i) case-based reasoning, similar to a lazy-learner that uses available knowledge during the forecasting process, and (ii) an Artificial Neural Network that has to be trained before the actual forecasting process occurs. The outcomes show that integrating more knowledge into the forecasting process enables better results in terms of forecast accuracy