Computer as thinking agent: Modeling the European power markets using agent-based simulation

The ongoing liberalisation of the power sector adds a new dimension to the main issues in modelling of power systems. The very complex interactions and interdependencies among power market participants are much like those studied in game theory. However, the strategies used by market participants are often too complex to be conveniently modelled by standard game theoretic techniques. In addition, there has been much less research in the field of dynamic strategic behaviour and their impact on the electricity price in European markets. In this paper, we show new, prosperous combination of computational science and new ideas in evolutionary economics and cognitive science offering appealing extensions to traditional game theoretical modelling. We demonstrate the feasibility of implementing our approach in Matlab using learning algorithm and illustrate its advantages in more detailed and realistic representation of the strategic behaviour of biggest power producers in European power market.