Data-Driven Airport Multi-Step Very Short-Term Load Forecasting

Buildings consume around 40% of energy produced in Europe. Building's electrical demand is affected by several external (weather) and internal (indoor environment, occupants) factors. Airport's buildings, particularly the main terminal, require significant amounts of electricity, among other natural resources. For this purpose, this research offers a comparative analysis of several machine learning prediction algorithms (Multi-layer Perceptron (MLP), Long-Short Term Memory (LSTM) and Bidirectional-LSTM), in order to investigate a very short-term electrical demand forecasting accuracy for three and six timesteps ahead of a Greek regional civilian airport during the tourist season.