Fast Fourier Transforms for Microgrid Climate Computing

Solar panels and wind turbines are key technologies for a sustainable low-carbon energy transition. The large diffusion of weather-dependent renewable energy generators face the challenge to fit the demand of uncertain loads with the most appropriate mix of distributed energy resources. The availability of accurate climate variables projections is essential to identify the best combination among distributed generators, energy storages and power loads in each geographical location. From the European Centre for Medium-Range Weather Forecasts (ECMWF) datasets of several climate variables are available for renewable energy resources yield forecasting. This paper presents different approaches to manipulate ECMWF datasets by combining Fast Fourier Transform with polynomial and forest tree regression models to predict climate variables over a one-year period, typical for microgrid simulations. An example of climate datasets forecasts related to the city of Bremen is presented with the evaluation of performances during test and training scenarios phases.