Functional Data Approach for Short-Term Electricity Demand Forecasting

In today’s liberalized electricity markets, modeling and forecasting demand data are highly important for the effective management of power system. However, is a challenging task due to specific features it exhibits. These include presence extreme values, spikes or jumps, multiple periodicities, long trend, bank holiday effect. addition, forecasts required complete day as decided before physical delivery. Therefore, this study aimed investigate performance models based on functional analysis, relatively less explored area in energy research. To end, time series first treated values. The filtered then divided into deterministic stochastic components. generalized additive technique used model component, whereas autoregressive (FAR), FAR with exogenous variable (FARX), classical univariate AR forecast component. Data from Nord Pool market used, one-day-ahead out-of-sample obtained whole year evaluated using different accuracy measures. results indicate that approach produces superior results, while FARX outperforms models. More specifically, NP demand, MAPE value 2.74, 6.27 9.73 values models, respectively.