Because electricity distributors can pre-purchase electric energy in advance at the lower prices than at the time of consumption, they want to forecast their customer's demand for the following day in order to pre-purchase electric energy according to that forecast. A problem arises in the case of an inaccurate forecast, as it brings additional expenses either by buying more expensive energy at the time of consumption either by energy regulator's penalties for overloading the electric network. The described situation is the problem of short-term forecasting of electric energy consumption, which was already addressed by many scientists and corporations using various methods.
Master's thesis tackles the mentioned problem using the latest discoveries in the field of fuzzy modeling. In the literature, fuzzy modeling has been already used for forecasting electric energy consumption, but in the predominantly non-flexible configurations (for example manually predefined clusters), where a good knowledge of the given data and a time-consuming tuning of parameters is required.
Therefore, Master's thesis investigates the possibility of using adaptive fuzzy models for the purpose of simplifying modeling stage in electric energy consumption forecasting. During the research, we also developed upgrades for space partitioning, which could be effective also in other fields where fuzzy modeling is used. The results of the developed model proved to be good and promising for further research, especially when compared with existing methods.