Deep learning models for total load and energy production forecasting in smart grids

The increasing demands in electrical load in combination with the high penetration of renewable energy sources have lead the load forecasting to be a difficult procedure. The conventionally means of load forecasting are not any more capable to predict the future load and new approaches are studied. These approaches include the machine learning and statistical methods through the use of multiple variable measurements. This paper aims to present the value of deep learning to the load forecasting procedure, the international standards applied in the field of smart grids and the design of a modern integrated Cloud – Internet of Things forecasting system for application at the Greek electrical transmission system.