Models for Electricity Demand Forecasting, Classification, and Imbalance Reduction in Competitive Markets

Abstract

In liberalized energy markets, hourly forecasts of consumers and producers are crucial for efficiently using energy resources and reducing environmental impacts. In this study, the countries’ consumption in the ENTSO-E common network between 2006 and 2018 was analyzed using the time series method. With the created model, short, medium, and long-term demand forecasts are made using Fourier Series Expansion. In order to improve the error rate of short-term forecasts, a hybrid model was created with alternatively created feedback and autoregressive methods. While annual forecasts are made with an average error rate of 6%, the error rate in daily forecasts is around 4.5%. With the hybrid models created, hourly estimates can be made with approximately 1.5% and 1% error rates. Accurate estimations are of great importance in terms of the efficiency of energy markets, and the emergence of energy storage opportunities with the developing technology increases this importance. For this reason, the amount of imbalance was estimated by using the forecast result of the hybrid model in the Turkish Energy Market, and a strategy was developed to reduce the imbalance cost accordingly. With this strategy, simulations have been made for situations with and without storage, and the results have been shared.