A Hybrid Approach to Load Forecast at a Micro Grid level through Machine Learning algorithms

Electric power systems’ operation has been facing new challenges. Intermittent renewable energy production and the consumption side uncertainty has been increasing, not only due to the integration of renewable sources but also flexible loads such as plug-in electric vehicles charging and storage devices. For these reasons, electricity load forecasting is crucial, in the sense of being able to determine the stability of the generation system and maintenance of scalable loads. This paper addresses the forecasts of electricity demand in a Micro Grid context and presents the novel HALOFMI methodology, which includes a Micro Grid scenario, selection and reduction of features and subsequently feeding these entries to the Artificial Neural Network. Final measures include validating the results attained from the developed 24-hour load forecast model defined throughout the work, based on performance metrics.