Forecast of electricity supply using adaptive neuro-fuzzy inference system

Abstract

In the balance section Electricity supply, the output quantities are electrical energy and heat generated in heat plants and in combined heat and power plants whereas the input quantities are primary and secondary energy carriers. In the recent years, new sources of primary energy, mostly renewable, appeared next to traditional fuels. In the modelling of energy transformation, energy carriers are treated as endogenous variables and it is postulated that electrical energy generation should be treated as an exogenous variable. An exogenous variable can be introduced as a scenario or it can be stated as a result of a chronological series forecast. Modern forecasting methods employ time series with a large number of samples. The paper presents a method for preparing this variable on the basis of a description of load variation in the power system. The method can yield forecasts at monthly intervals, with the forecasting process using adaptive neuro-fuzzy inference system (ANFIS).