Energy consumption forecasting for emerging technologies

Accurate energy consumption forecasting is critical for Distribution System Operators (DSOs) to optimize grid planning and tariff calculations. Traditional machine learning models, which rely on historical time series data, struggle to account for disruptive changes in energy consumption patterns. With the current climate goals in Switzerland, a significant non-linear growth in electrical energy consumption is expected, mainly due to the increasing electrification of space heating and transportation, i.e. heat pumps and e-mobility. This shift presents challenges that current models based solely on historical data cannot address. In this paper, we propose our multilevel approach to energy consumption forecasting for the next five years that better accounts for these emerging technologies. Our method follows a two-step process in which we predict the emerging technologies separately to the base load and combine them for the final result.