Abstract:

The short-term electricity forecast for electricity, which is an indispensable element of today, has been among the topics of high importance in recent years. It is important that the price of electricity demand is correctly predicted in order to balance the demand with electricity production. If it is possible, competitive prices can be offered to consumers. Some techniques are used in literature to make correct predictions on electricity demand. The aim of this study is to predict short-term electricity demand by operating the artificial nervous network (YSA) with the long-term memory (LSTM) architecture from the respective techniques. The results obtained by the application of the YSA methodology have been compared with Time Series Analysis (ARIMA) according to the root average square error values. In this context, it was used to make forecasts of the electricity data of Spain between 2015-2016 years. Electricity production and consumption data are collected from ENTSOE, an open-access portal that contains data from the Communication Service Operator (TSO).

Keywords: