Probabilistic Data-Based Models for a Reliable Energy Management

Reliable forecasts for the energy production of generators in an energy management system become more and more essential. However, forecast models cannot create forecasts with a one hundred percent accuracy. Therefore, additional information is added to the deterministic forecast value. Probabilistic forecasts create some information about the probability distribution of the output value. In this paper data-based models are used to generate a deterministic and probabilistic forecast. A simulation based on real data including a photovoltaic plant and a battery storage is done to show the uncertainties of the individual photovoltaic and battery model as well as the combined uncertainties. The results show that the combined uncertainties for both devices allow a wide range of possible states of charge forecasts for the storage. The results help to answer the question where balancing energy has to be kept available and the analysis shows which amount each device contributes to the uncertainty.