Exploring Price Classification in Day Ahead Electricity Markets through Graph Theory and PNN Modelling

Abstract

This paper delves into the intersection of graph theory and electricity market data analysis. The introductory section lays the foundation by elucidating the theoretical underpinnings of graph theory and its relevance to understanding the complexities of electricity market dynamics. As an overarching theme, we employ the Probabilistic Neural Network (PNN) model to classify electricity prices into three distinct categories: normal, extremely high, and negative. A pivotal aspect of our approach is the derivation of datasets directly from graphical representations. Specifically, we select datasets that exhibit proximity to the Price vertex, thereby encapsulating crucial insights into price behaviours.These datasets are employed as input for our PNN model, enabling the model to leverage the intricate relationships within the graphs. Next, we select the price data as input for our model. So, we draw a comparison between the price data and datasets that demonstrate proximity to the Price vertex. This comparison allows us to highlight the efficiency of certain datasets over others in capturing important insights. Our analysis extends across multiple electricity markets, encompassing Germany, Finland, and Estonia. By applying the price classification framework to these diverse markets, we gain a comprehensive understanding of how market-specific factors influence price dynamics and volatility. This research not only advances the application of graph theory in electricity market analysis but also offers valuable insights into price classification and forecasting.