AI-based Load Forecasts - Research and Application

Abstract:
The climate targets require a transformation of the energy supply, which brings new challenges. The volatile generation of renewable energies and dynamic consumption due to the expansion of electricity-consuming systems such as e-mobility and heat pumps make a digitized and intelligent distribution network indispensable. To plan and optimize load distribution in a grid with volatile flexibilities and consumers, accurate load forecasts are needed. According to the current state of the art, these are carried out using Artificial Intelligence (AI) algorithms. An expert group appointed by the European Commission developed guidelines for trustworthy AI in 2019 and augmented them with a list for assessing AI in 2020. The requirements for trustworthy AI, identified in these guidelines, are crucial for the safe application of AI in practice, but they also bring challenges, as new AI methods are often very complex. The challenges, open research questions, and potential solutions are described and discussed in this article.