Medium and short term prognosis of load demand for the Greek Island of Tilos using artificial neural networks and human thermal comfort-discomfort biometeorological data

The objective of the present work is the medium and short term forecasting of load demand (LD) in Tilos Island, Greece. For this purpose, Artificial Neural Network (ANN) models were developed to predict the LD in Tilos Island 24 hours ahead in hourly intervals
(medium-term prognosis) and 24 hours ahead in 10 minutes intervals (short-term prognosis). For training the ANNs, meteorological data covering the 2015-2017 period, were used. These data have been recorded in 1 minute intervals by a meteorological mast which has been installed in a specific location in Tilos Island. Furthermore, a biometeorological human thermal comfortdiscomfort index was calculated and used also during the training procedure. For the validation of the developed ANN forecasting models well established statistical evaluation indices were applied. Results show that in all cases, for both medium and short-term LD prognoses, the developed ANN forecasting models present a remarkable ability to predict LD with high accuracy. The proposed load demand forecasting models enable the design of an energy demand information tool for end-users and transmission system operators.