Electricity price forecasting: from probabilistic to deep learning approaches

Abstract

In recent years, with the increasing penetration of renewable energy sources (RES), the generation of electricity has become more uncertain. As electricity consumption has to equal electricity generation at all times, increasing uncertainty in electricity generation leads to electricity markets that display highly volatile prices with sudden and unexpected price peaks. In this context, accurate price forecasting is paramount to ensure further integration of RES into the electricity grid by guarantying their profitability and reducing the associated market risks.

In this talk, we cover the main approaches and techniques when it comes to forecasting prices. We start the talk with a brief introduction to the field of price forecasting. Then, we talk about the three main types of forecasting methods: point forecasting, probability forecasting, and scenario forecasting. In particular, we discuss the merits and disadvantages of these three classes, and we provide some particular examples of methods belonging to these families. During the talk, we also discuss the importance of deep learning techniques in the context of forecasting.

 

Resources

The talk was part of the PowerWeb lectures at the Technical University of Delft. The slides from the talk are available online.