engleski

STATISTICAL AND ARTIFICIAL INTELLIGENCE-BASED APPROACHES TO ELECTRICITY PRICE FORECASTING: A REVIEW

Todays’ electricity markets are undergoing changes and facing new challenges. Namely, there is an ongoing decentralization process which has changed the electricity sector on one side. On the other side, there is an increasing pressure for considering environmental factors everywhere, more efficient use of the limited resources in general and investigating renewable energy resources. Also, the electricity demand is dependent on many factors such as season, particular moment in time, weather conditions and other factors not observed in other markets which makes electricity prices a peculiar commodity with special characteristics. All this requires a growing attention in development of different techniques for forecasting electricity prices. There are different approaches in modelling electricity prices found in the literature. The goal of this work is to systematically review two large groups of them: statistical approaches (applications of classical statistical econometric techniques) and artificial intelligence-based approaches (computational intelligence, include machine learning methods such as neural networks and evolutional computation). Regarding the variables used for creating the forecasts, the models used in these approaches use different combinations of previous prices and previous or current values of selected exogenous variables. Regarding the forecasting horizon, the focus of this work is on short-term forecasting This work contributes to the existing literature by trying to include some of the techniques which are not commonly mentioned in similar reviews and not commonly used in the research in this field. Also, it gives ideas of possible methods which have not been even used so far, but have a potential for possibly good performance. This way, usage of such techniques is promoted with a notion of inspiring future research in this area and improvement on electricity forecast accuracy.

electricity prices, forecasting, review

nije evidentirano